diff options
Diffstat (limited to 'src/3rdparty/angle/src/libGLESv2/renderer')
172 files changed, 29136 insertions, 14641 deletions
diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/BufferImpl.h b/src/3rdparty/angle/src/libGLESv2/renderer/BufferImpl.h new file mode 100644 index 0000000000..f0b5f02227 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/BufferImpl.h @@ -0,0 +1,34 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// BufferImpl.h: Defines the abstract rx::BufferImpl class. + +#ifndef LIBGLESV2_RENDERER_BUFFERIMPL_H_ +#define LIBGLESV2_RENDERER_BUFFERIMPL_H_ + +#include "common/angleutils.h" +#include "libGLESv2/Buffer.h" + +namespace rx +{ + +class BufferImpl +{ + public: + virtual ~BufferImpl() { } + + virtual gl::Error setData(const void* data, size_t size, GLenum usage) = 0; + virtual void *getData() = 0; + virtual gl::Error setSubData(const void* data, size_t size, size_t offset) = 0; + virtual gl::Error copySubData(BufferImpl* source, GLintptr sourceOffset, GLintptr destOffset, GLsizeiptr size) = 0; + virtual gl::Error map(size_t offset, size_t length, GLbitfield access, GLvoid **mapPtr) = 0; + virtual gl::Error unmap() = 0; + virtual void markTransformFeedbackUsage() = 0; +}; + +} + +#endif // LIBGLESV2_RENDERER_BUFFERIMPL_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/BufferStorage.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/BufferStorage.cpp deleted file mode 100644 index a49b7bab84..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/BufferStorage.cpp +++ /dev/null @@ -1,40 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// BufferStorage.cpp Defines the abstract BufferStorage class. - -#include "libGLESv2/renderer/BufferStorage.h" - -namespace rx -{ - -unsigned int BufferStorage::mNextSerial = 1; - -BufferStorage::BufferStorage() -{ - updateSerial(); -} - -BufferStorage::~BufferStorage() -{ -} - -unsigned int BufferStorage::getSerial() const -{ - return mSerial; -} - -void BufferStorage::updateSerial() -{ - mSerial = mNextSerial++; -} - -void BufferStorage::markBufferUsage() -{ -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/BufferStorage.h b/src/3rdparty/angle/src/libGLESv2/renderer/BufferStorage.h deleted file mode 100644 index ace1a11bae..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/BufferStorage.h +++ /dev/null @@ -1,44 +0,0 @@ -// -// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// BufferStorage.h Defines the abstract BufferStorage class. - -#ifndef LIBGLESV2_RENDERER_BUFFERSTORAGE_H_ -#define LIBGLESV2_RENDERER_BUFFERSTORAGE_H_ - -#include "common/angleutils.h" - -namespace rx -{ - -class BufferStorage -{ - public: - BufferStorage(); - virtual ~BufferStorage(); - - // The data returned is only guaranteed valid until next non-const method. - virtual void *getData() = 0; - virtual void setData(const void* data, unsigned int size, unsigned int offset) = 0; - virtual void clear() = 0; - virtual unsigned int getSize() const = 0; - virtual bool supportsDirectBinding() const = 0; - virtual void markBufferUsage(); - unsigned int getSerial() const; - - protected: - void updateSerial(); - - private: - DISALLOW_COPY_AND_ASSIGN(BufferStorage); - - unsigned int mSerial; - static unsigned int mNextSerial; -}; - -} - -#endif // LIBGLESV2_RENDERER_BUFFERSTORAGE_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/FenceImpl.h b/src/3rdparty/angle/src/libGLESv2/renderer/FenceImpl.h index d7f2102a2e..d54e6becd3 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/FenceImpl.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/FenceImpl.h @@ -17,27 +17,16 @@ namespace rx class FenceImpl { public: - FenceImpl() : mStatus(GL_FALSE), mCondition(GL_NONE) { }; + FenceImpl() { }; virtual ~FenceImpl() { }; - virtual GLboolean isFence() = 0; - virtual void setFence(GLenum condition) = 0; - virtual GLboolean testFence() = 0; - virtual void finishFence() = 0; - virtual void getFenceiv(GLenum pname, GLint *params) = 0; - - protected: - void setStatus(GLboolean status) { mStatus = status; } - GLboolean getStatus() const { return mStatus; } - - void setCondition(GLuint condition) { mCondition = condition; } - GLuint getCondition() const { return mCondition; } + virtual bool isSet() const = 0; + virtual void set() = 0; + virtual bool test(bool flushCommandBuffer) = 0; + virtual bool hasError() const = 0; private: DISALLOW_COPY_AND_ASSIGN(FenceImpl); - - GLboolean mStatus; - GLenum mCondition; }; } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/Image.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/Image.cpp index 57239ef74f..370b086233 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/Image.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/Image.cpp @@ -1,4 +1,3 @@ -#include "precompiled.h" // // Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be @@ -18,531 +17,12 @@ Image::Image() { mWidth = 0; mHeight = 0; + mDepth = 0; mInternalFormat = GL_NONE; mActualFormat = GL_NONE; -} - -void Image::loadAlphaDataToBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned char *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = static_cast<const unsigned char*>(input) + y * inputPitch; - dest = static_cast<unsigned char*>(output) + y * outputPitch; - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = 0; - dest[4 * x + 1] = 0; - dest[4 * x + 2] = 0; - dest[4 * x + 3] = source[x]; - } - } -} - -void Image::loadAlphaDataToNative(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned char *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = static_cast<const unsigned char*>(input) + y * inputPitch; - dest = static_cast<unsigned char*>(output) + y * outputPitch; - memcpy(dest, source, width); - } -} - -void Image::loadAlphaFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const float *source = NULL; - float *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const float*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<float*>(static_cast<unsigned char*>(output) + y * outputPitch); - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = 0; - dest[4 * x + 1] = 0; - dest[4 * x + 2] = 0; - dest[4 * x + 3] = source[x]; - } - } -} - -void Image::loadAlphaHalfFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned short *source = NULL; - unsigned short *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned short*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<unsigned short*>(static_cast<unsigned char*>(output) + y * outputPitch); - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = 0; - dest[4 * x + 1] = 0; - dest[4 * x + 2] = 0; - dest[4 * x + 3] = source[x]; - } - } -} - -void Image::loadLuminanceDataToNativeOrBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output, bool native) -{ - const unsigned char *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = static_cast<const unsigned char*>(input) + y * inputPitch; - dest = static_cast<unsigned char*>(output) + y * outputPitch; - - if (!native) // BGRA8 destination format - { - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = source[x]; - dest[4 * x + 1] = source[x]; - dest[4 * x + 2] = source[x]; - dest[4 * x + 3] = 0xFF; - } - } - else // L8 destination format - { - memcpy(dest, source, width); - } - } -} - -void Image::loadLuminanceFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const float *source = NULL; - float *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const float*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<float*>(static_cast<unsigned char*>(output) + y * outputPitch); - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = source[x]; - dest[4 * x + 1] = source[x]; - dest[4 * x + 2] = source[x]; - dest[4 * x + 3] = 1.0f; - } - } -} - -void Image::loadLuminanceFloatDataToRGB(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const float *source = NULL; - float *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const float*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<float*>(static_cast<unsigned char*>(output) + y * outputPitch); - for (int x = 0; x < width; x++) - { - dest[3 * x + 0] = source[x]; - dest[3 * x + 1] = source[x]; - dest[3 * x + 2] = source[x]; - } - } -} - -void Image::loadLuminanceHalfFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned short *source = NULL; - unsigned short *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned short*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<unsigned short*>(static_cast<unsigned char*>(output) + y * outputPitch); - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = source[x]; - dest[4 * x + 1] = source[x]; - dest[4 * x + 2] = source[x]; - dest[4 * x + 3] = 0x3C00; // SEEEEEMMMMMMMMMM, S = 0, E = 15, M = 0: 16bit flpt representation of 1 - } - } -} - -void Image::loadLuminanceAlphaDataToNativeOrBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output, bool native) -{ - const unsigned char *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = static_cast<const unsigned char*>(input) + y * inputPitch; - dest = static_cast<unsigned char*>(output) + y * outputPitch; - - if (!native) // BGRA8 destination format - { - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = source[2*x+0]; - dest[4 * x + 1] = source[2*x+0]; - dest[4 * x + 2] = source[2*x+0]; - dest[4 * x + 3] = source[2*x+1]; - } - } - else - { - memcpy(dest, source, width * 2); - } - } -} - -void Image::loadLuminanceAlphaFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const float *source = NULL; - float *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const float*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<float*>(static_cast<unsigned char*>(output) + y * outputPitch); - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = source[2*x+0]; - dest[4 * x + 1] = source[2*x+0]; - dest[4 * x + 2] = source[2*x+0]; - dest[4 * x + 3] = source[2*x+1]; - } - } -} - -void Image::loadLuminanceAlphaHalfFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned short *source = NULL; - unsigned short *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned short*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<unsigned short*>(static_cast<unsigned char*>(output) + y * outputPitch); - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = source[2*x+0]; - dest[4 * x + 1] = source[2*x+0]; - dest[4 * x + 2] = source[2*x+0]; - dest[4 * x + 3] = source[2*x+1]; - } - } -} - -void Image::loadRGBUByteDataToBGRX(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned char *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = static_cast<const unsigned char*>(input) + y * inputPitch; - dest = static_cast<unsigned char*>(output) + y * outputPitch; - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = source[x * 3 + 2]; - dest[4 * x + 1] = source[x * 3 + 1]; - dest[4 * x + 2] = source[x * 3 + 0]; - dest[4 * x + 3] = 0xFF; - } - } -} - -void Image::loadRGBUByteDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned char *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = static_cast<const unsigned char*>(input) + y * inputPitch; - dest = static_cast<unsigned char*>(output) + y * outputPitch; - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = source[x * 3 + 0]; - dest[4 * x + 1] = source[x * 3 + 1]; - dest[4 * x + 2] = source[x * 3 + 2]; - dest[4 * x + 3] = 0xFF; - } - } -} - -void Image::loadRGB565DataToBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned short *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned short*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = static_cast<unsigned char*>(output) + y * outputPitch; - for (int x = 0; x < width; x++) - { - unsigned short rgba = source[x]; - dest[4 * x + 0] = ((rgba & 0x001F) << 3) | ((rgba & 0x001F) >> 2); - dest[4 * x + 1] = ((rgba & 0x07E0) >> 3) | ((rgba & 0x07E0) >> 9); - dest[4 * x + 2] = ((rgba & 0xF800) >> 8) | ((rgba & 0xF800) >> 13); - dest[4 * x + 3] = 0xFF; - } - } -} - -void Image::loadRGB565DataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned short *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned short*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = static_cast<unsigned char*>(output) + y * outputPitch; - for (int x = 0; x < width; x++) - { - unsigned short rgba = source[x]; - dest[4 * x + 0] = ((rgba & 0xF800) >> 8) | ((rgba & 0xF800) >> 13); - dest[4 * x + 1] = ((rgba & 0x07E0) >> 3) | ((rgba & 0x07E0) >> 9); - dest[4 * x + 2] = ((rgba & 0x001F) << 3) | ((rgba & 0x001F) >> 2); - dest[4 * x + 3] = 0xFF; - } - } -} - -void Image::loadRGBFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const float *source = NULL; - float *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const float*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<float*>(static_cast<unsigned char*>(output) + y * outputPitch); - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = source[x * 3 + 0]; - dest[4 * x + 1] = source[x * 3 + 1]; - dest[4 * x + 2] = source[x * 3 + 2]; - dest[4 * x + 3] = 1.0f; - } - } -} - -void Image::loadRGBFloatDataToNative(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const float *source = NULL; - float *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const float*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<float*>(static_cast<unsigned char*>(output) + y * outputPitch); - memcpy(dest, source, width * 12); - } -} - -void Image::loadRGBHalfFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned short *source = NULL; - unsigned short *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned short*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<unsigned short*>(static_cast<unsigned char*>(output) + y * outputPitch); - for (int x = 0; x < width; x++) - { - dest[4 * x + 0] = source[x * 3 + 0]; - dest[4 * x + 1] = source[x * 3 + 1]; - dest[4 * x + 2] = source[x * 3 + 2]; - dest[4 * x + 3] = 0x3C00; // SEEEEEMMMMMMMMMM, S = 0, E = 15, M = 0: 16bit flpt representation of 1 - } - } -} - -void Image::loadRGBAUByteDataToBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned int *source = NULL; - unsigned int *dest = NULL; - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned int*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<unsigned int*>(static_cast<unsigned char*>(output) + y * outputPitch); - - for (int x = 0; x < width; x++) - { - unsigned int rgba = source[x]; - dest[x] = (_rotl(rgba, 16) & 0x00ff00ff) | (rgba & 0xff00ff00); - } - } -} - -void Image::loadRGBAUByteDataToNative(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned int *source = NULL; - unsigned int *dest = NULL; - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned int*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<unsigned int*>(static_cast<unsigned char*>(output) + y * outputPitch); - - memcpy(dest, source, width * 4); - } -} - -void Image::loadRGBA4444DataToBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned short *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned short*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = static_cast<unsigned char*>(output) + y * outputPitch; - for (int x = 0; x < width; x++) - { - unsigned short rgba = source[x]; - dest[4 * x + 0] = ((rgba & 0x00F0) << 0) | ((rgba & 0x00F0) >> 4); - dest[4 * x + 1] = ((rgba & 0x0F00) >> 4) | ((rgba & 0x0F00) >> 8); - dest[4 * x + 2] = ((rgba & 0xF000) >> 8) | ((rgba & 0xF000) >> 12); - dest[4 * x + 3] = ((rgba & 0x000F) << 4) | ((rgba & 0x000F) >> 0); - } - } -} - -void Image::loadRGBA4444DataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned short *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned short*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = static_cast<unsigned char*>(output) + y * outputPitch; - for (int x = 0; x < width; x++) - { - unsigned short rgba = source[x]; - dest[4 * x + 0] = ((rgba & 0xF000) >> 8) | ((rgba & 0xF000) >> 12); - dest[4 * x + 1] = ((rgba & 0x0F00) >> 4) | ((rgba & 0x0F00) >> 8); - dest[4 * x + 2] = ((rgba & 0x00F0) << 0) | ((rgba & 0x00F0) >> 4); - dest[4 * x + 3] = ((rgba & 0x000F) << 4) | ((rgba & 0x000F) >> 0); - } - } -} - -void Image::loadRGBA5551DataToBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned short *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned short*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = static_cast<unsigned char*>(output) + y * outputPitch; - for (int x = 0; x < width; x++) - { - unsigned short rgba = source[x]; - dest[4 * x + 0] = ((rgba & 0x003E) << 2) | ((rgba & 0x003E) >> 3); - dest[4 * x + 1] = ((rgba & 0x07C0) >> 3) | ((rgba & 0x07C0) >> 8); - dest[4 * x + 2] = ((rgba & 0xF800) >> 8) | ((rgba & 0xF800) >> 13); - dest[4 * x + 3] = (rgba & 0x0001) ? 0xFF : 0; - } - } -} - -void Image::loadRGBA5551DataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned short *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned short*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = static_cast<unsigned char*>(output) + y * outputPitch; - for (int x = 0; x < width; x++) - { - unsigned short rgba = source[x]; - dest[4 * x + 0] = ((rgba & 0xF800) >> 8) | ((rgba & 0xF800) >> 13); - dest[4 * x + 1] = ((rgba & 0x07C0) >> 3) | ((rgba & 0x07C0) >> 8); - dest[4 * x + 2] = ((rgba & 0x003E) << 2) | ((rgba & 0x003E) >> 3); - dest[4 * x + 3] = (rgba & 0x0001) ? 0xFF : 0; - } - } -} - -void Image::loadRGBAFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const float *source = NULL; - float *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const float*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<float*>(static_cast<unsigned char*>(output) + y * outputPitch); - memcpy(dest, source, width * 16); - } -} - -void Image::loadRGBAHalfFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned char *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = static_cast<const unsigned char*>(input) + y * inputPitch; - dest = static_cast<unsigned char*>(output) + y * outputPitch; - memcpy(dest, source, width * 8); - } -} - -void Image::loadBGRADataToBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned char *source = NULL; - unsigned char *dest = NULL; - - for (int y = 0; y < height; y++) - { - source = static_cast<const unsigned char*>(input) + y * inputPitch; - dest = static_cast<unsigned char*>(output) + y * outputPitch; - memcpy(dest, source, width*4); - } + mTarget = GL_NONE; + mRenderable = false; + mDirty = false; } } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/Image.h b/src/3rdparty/angle/src/libGLESv2/renderer/Image.h index 454e83e21e..3bfc663762 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/Image.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/Image.h @@ -13,6 +13,8 @@ #include "common/debug.h" +#include <GLES2/gl2.h> + namespace gl { class Framebuffer; @@ -20,9 +22,8 @@ class Framebuffer; namespace rx { + class Renderer; -class TextureStorageInterface2D; -class TextureStorageInterfaceCube; class Image { @@ -32,93 +33,33 @@ class Image GLsizei getWidth() const { return mWidth; } GLsizei getHeight() const { return mHeight; } + GLsizei getDepth() const { return mDepth; } GLenum getInternalFormat() const { return mInternalFormat; } GLenum getActualFormat() const { return mActualFormat; } + GLenum getTarget() const { return mTarget; } + bool isRenderableFormat() const { return mRenderable; } void markDirty() {mDirty = true;} void markClean() {mDirty = false;} virtual bool isDirty() const = 0; - virtual void setManagedSurface(TextureStorageInterface2D *storage, int level) {}; - virtual void setManagedSurface(TextureStorageInterfaceCube *storage, int face, int level) {}; - virtual bool updateSurface(TextureStorageInterface2D *storage, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) = 0; - virtual bool updateSurface(TextureStorageInterfaceCube *storage, int face, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) = 0; - - virtual bool redefine(Renderer *renderer, GLint internalformat, GLsizei width, GLsizei height, bool forceRelease) = 0; + virtual bool redefine(Renderer *renderer, GLenum target, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, bool forceRelease) = 0; - virtual bool isRenderableFormat() const = 0; - - virtual void loadData(GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, - GLint unpackAlignment, const void *input) = 0; - virtual void loadCompressedData(GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, + virtual void loadData(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, + GLint unpackAlignment, GLenum type, const void *input) = 0; + virtual void loadCompressedData(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, const void *input) = 0; - virtual void copy(GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) = 0; - - static void loadAlphaDataToBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadAlphaDataToNative(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadAlphaDataToBGRASSE2(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadAlphaFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadAlphaHalfFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadLuminanceDataToNativeOrBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output, bool native); - static void loadLuminanceFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadLuminanceFloatDataToRGB(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadLuminanceHalfFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadLuminanceAlphaDataToNativeOrBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output, bool native); - static void loadLuminanceAlphaFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadLuminanceAlphaHalfFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBUByteDataToBGRX(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBUByteDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGB565DataToBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGB565DataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBFloatDataToNative(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBHalfFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBAUByteDataToBGRASSE2(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBAUByteDataToBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBAUByteDataToNative(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBA4444DataToBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBA4444DataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBA5551DataToBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBA5551DataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBAFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadRGBAHalfFloatDataToRGBA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); - static void loadBGRADataToBGRA(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output); + virtual void copy(GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) = 0; protected: GLsizei mWidth; GLsizei mHeight; - GLint mInternalFormat; + GLsizei mDepth; + GLenum mInternalFormat; GLenum mActualFormat; + bool mRenderable; + GLenum mTarget; bool mDirty; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/ImageSSE2.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/ImageSSE2.cpp deleted file mode 100644 index b2a90ca961..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/ImageSSE2.cpp +++ /dev/null @@ -1,100 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// ImageSSE2.cpp: Implements SSE2-based functions of rx::Image class. It's -// in a separated file for GCC, which can enable SSE usage only per-file, -// not for code blocks that use SSE2 explicitly. - -#include "libGLESv2/Texture.h" -#include "libGLESv2/renderer/Image.h" - -namespace rx -{ - -void Image::loadRGBAUByteDataToBGRASSE2(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned int *source = NULL; - unsigned int *dest = NULL; - __m128i brMask = _mm_set1_epi32(0x00ff00ff); - - for (int y = 0; y < height; y++) - { - source = reinterpret_cast<const unsigned int*>(static_cast<const unsigned char*>(input) + y * inputPitch); - dest = reinterpret_cast<unsigned int*>(static_cast<unsigned char*>(output) + y * outputPitch); - int x = 0; - - // Make output writes aligned - for (x = 0; ((reinterpret_cast<intptr_t>(&dest[x]) & 15) != 0) && x < width; x++) - { - unsigned int rgba = source[x]; - dest[x] = (_rotl(rgba, 16) & 0x00ff00ff) | (rgba & 0xff00ff00); - } - - for (; x + 3 < width; x += 4) - { - __m128i sourceData = _mm_loadu_si128(reinterpret_cast<const __m128i*>(&source[x])); - // Mask out g and a, which don't change - __m128i gaComponents = _mm_andnot_si128(brMask, sourceData); - // Mask out b and r - __m128i brComponents = _mm_and_si128(sourceData, brMask); - // Swap b and r - __m128i brSwapped = _mm_shufflehi_epi16(_mm_shufflelo_epi16(brComponents, _MM_SHUFFLE(2, 3, 0, 1)), _MM_SHUFFLE(2, 3, 0, 1)); - __m128i result = _mm_or_si128(gaComponents, brSwapped); - _mm_store_si128(reinterpret_cast<__m128i*>(&dest[x]), result); - } - - // Perform leftover writes - for (; x < width; x++) - { - unsigned int rgba = source[x]; - dest[x] = (_rotl(rgba, 16) & 0x00ff00ff) | (rgba & 0xff00ff00); - } - } -} - -void Image::loadAlphaDataToBGRASSE2(GLsizei width, GLsizei height, - int inputPitch, const void *input, size_t outputPitch, void *output) -{ - const unsigned char *source = NULL; - unsigned int *dest = NULL; - __m128i zeroWide = _mm_setzero_si128(); - - for (int y = 0; y < height; y++) - { - source = static_cast<const unsigned char*>(input) + y * inputPitch; - dest = reinterpret_cast<unsigned int*>(static_cast<unsigned char*>(output) + y * outputPitch); - - int x; - // Make output writes aligned - for (x = 0; ((reinterpret_cast<intptr_t>(&dest[x]) & 0xF) != 0 && x < width); x++) - { - dest[x] = static_cast<unsigned int>(source[x]) << 24; - } - - for (; x + 7 < width; x += 8) - { - __m128i sourceData = _mm_loadl_epi64(reinterpret_cast<const __m128i*>(&source[x])); - // Interleave each byte to 16bit, make the lower byte to zero - sourceData = _mm_unpacklo_epi8(zeroWide, sourceData); - // Interleave each 16bit to 32bit, make the lower 16bit to zero - __m128i lo = _mm_unpacklo_epi16(zeroWide, sourceData); - __m128i hi = _mm_unpackhi_epi16(zeroWide, sourceData); - - _mm_store_si128(reinterpret_cast<__m128i*>(&dest[x]), lo); - _mm_store_si128(reinterpret_cast<__m128i*>(&dest[x + 4]), hi); - } - - // Handle the remainder - for (; x < width; x++) - { - dest[x] = static_cast<unsigned int>(source[x]) << 24; - } - } -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/IndexDataManager.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/IndexDataManager.cpp deleted file mode 100644 index 49bace8193..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/IndexDataManager.cpp +++ /dev/null @@ -1,339 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// IndexDataManager.cpp: Defines the IndexDataManager, a class that -// runs the Buffer translation process for index buffers. - -#include "libGLESv2/renderer/IndexDataManager.h" -#include "libGLESv2/renderer/BufferStorage.h" - -#include "libGLESv2/Buffer.h" -#include "libGLESv2/main.h" -#include "libGLESv2/utilities.h" -#include "libGLESv2/renderer/IndexBuffer.h" - -namespace rx -{ - -IndexDataManager::IndexDataManager(Renderer *renderer) : mRenderer(renderer) -{ - mStreamingBufferShort = new StreamingIndexBufferInterface(mRenderer); - if (!mStreamingBufferShort->reserveBufferSpace(INITIAL_INDEX_BUFFER_SIZE, GL_UNSIGNED_SHORT)) - { - delete mStreamingBufferShort; - mStreamingBufferShort = NULL; - } - - mStreamingBufferInt = new StreamingIndexBufferInterface(mRenderer); - if (!mStreamingBufferInt->reserveBufferSpace(INITIAL_INDEX_BUFFER_SIZE, GL_UNSIGNED_INT)) - { - delete mStreamingBufferInt; - mStreamingBufferInt = NULL; - } - - if (!mStreamingBufferShort) - { - // Make sure both buffers are deleted. - delete mStreamingBufferInt; - mStreamingBufferInt = NULL; - - ERR("Failed to allocate the streaming index buffer(s)."); - } - - mCountingBuffer = NULL; -} - -IndexDataManager::~IndexDataManager() -{ - delete mStreamingBufferShort; - delete mStreamingBufferInt; - delete mCountingBuffer; -} - -static void convertIndices(GLenum type, const void *input, GLsizei count, void *output) -{ - if (type == GL_UNSIGNED_BYTE) - { - const GLubyte *in = static_cast<const GLubyte*>(input); - GLushort *out = static_cast<GLushort*>(output); - - for (GLsizei i = 0; i < count; i++) - { - out[i] = in[i]; - } - } - else if (type == GL_UNSIGNED_INT) - { - memcpy(output, input, count * sizeof(GLuint)); - } - else if (type == GL_UNSIGNED_SHORT) - { - memcpy(output, input, count * sizeof(GLushort)); - } - else UNREACHABLE(); -} - -template <class IndexType> -static void computeRange(const IndexType *indices, GLsizei count, GLuint *minIndex, GLuint *maxIndex) -{ - *minIndex = indices[0]; - *maxIndex = indices[0]; - - for (GLsizei i = 0; i < count; i++) - { - if (*minIndex > indices[i]) *minIndex = indices[i]; - if (*maxIndex < indices[i]) *maxIndex = indices[i]; - } -} - -static void computeRange(GLenum type, const GLvoid *indices, GLsizei count, GLuint *minIndex, GLuint *maxIndex) -{ - if (type == GL_UNSIGNED_BYTE) - { - computeRange(static_cast<const GLubyte*>(indices), count, minIndex, maxIndex); - } - else if (type == GL_UNSIGNED_INT) - { - computeRange(static_cast<const GLuint*>(indices), count, minIndex, maxIndex); - } - else if (type == GL_UNSIGNED_SHORT) - { - computeRange(static_cast<const GLushort*>(indices), count, minIndex, maxIndex); - } - else UNREACHABLE(); -} - -GLenum IndexDataManager::prepareIndexData(GLenum type, GLsizei count, gl::Buffer *buffer, const GLvoid *indices, TranslatedIndexData *translated) -{ - if (!mStreamingBufferShort) - { - return GL_OUT_OF_MEMORY; - } - - GLenum destinationIndexType = (type == GL_UNSIGNED_INT) ? GL_UNSIGNED_INT : GL_UNSIGNED_SHORT; - unsigned int offset = 0; - bool alignedOffset = false; - - BufferStorage *storage = NULL; - - if (buffer != NULL) - { - if (reinterpret_cast<uintptr_t>(indices) > std::numeric_limits<unsigned int>::max()) - { - return GL_OUT_OF_MEMORY; - } - offset = static_cast<unsigned int>(reinterpret_cast<uintptr_t>(indices)); - - storage = buffer->getStorage(); - - switch (type) - { - case GL_UNSIGNED_BYTE: alignedOffset = (offset % sizeof(GLubyte) == 0); break; - case GL_UNSIGNED_SHORT: alignedOffset = (offset % sizeof(GLushort) == 0); break; - case GL_UNSIGNED_INT: alignedOffset = (offset % sizeof(GLuint) == 0); break; - default: UNREACHABLE(); alignedOffset = false; - } - - unsigned int typeSize = gl::ComputeTypeSize(type); - - // check for integer overflows - if (static_cast<unsigned int>(count) > (std::numeric_limits<unsigned int>::max() / typeSize) || - typeSize * static_cast<unsigned int>(count) + offset < offset) - { - return GL_OUT_OF_MEMORY; - } - - if (typeSize * static_cast<unsigned int>(count) + offset > storage->getSize()) - { - return GL_INVALID_OPERATION; - } - - indices = static_cast<const GLubyte*>(storage->getData()) + offset; - } - - StreamingIndexBufferInterface *streamingBuffer = (type == GL_UNSIGNED_INT) ? mStreamingBufferInt : mStreamingBufferShort; - - StaticIndexBufferInterface *staticBuffer = buffer ? buffer->getStaticIndexBuffer() : NULL; - IndexBufferInterface *indexBuffer = streamingBuffer; - bool directStorage = alignedOffset && storage && storage->supportsDirectBinding() && - destinationIndexType == type; - unsigned int streamOffset = 0; - - if (directStorage) - { - indexBuffer = streamingBuffer; - streamOffset = offset; - storage->markBufferUsage(); - - if (!buffer->getIndexRangeCache()->findRange(type, offset, count, &translated->minIndex, - &translated->maxIndex, NULL)) - { - computeRange(type, indices, count, &translated->minIndex, &translated->maxIndex); - buffer->getIndexRangeCache()->addRange(type, offset, count, translated->minIndex, - translated->maxIndex, offset); - } - } - else if (staticBuffer && staticBuffer->getBufferSize() != 0 && staticBuffer->getIndexType() == type && alignedOffset) - { - indexBuffer = staticBuffer; - if (!staticBuffer->getIndexRangeCache()->findRange(type, offset, count, &translated->minIndex, - &translated->maxIndex, &streamOffset)) - { - streamOffset = (offset / gl::ComputeTypeSize(type)) * gl::ComputeTypeSize(destinationIndexType); - computeRange(type, indices, count, &translated->minIndex, &translated->maxIndex); - staticBuffer->getIndexRangeCache()->addRange(type, offset, count, translated->minIndex, - translated->maxIndex, streamOffset); - } - } - else - { - unsigned int convertCount = count; - - if (staticBuffer) - { - if (staticBuffer->getBufferSize() == 0 && alignedOffset) - { - indexBuffer = staticBuffer; - convertCount = storage->getSize() / gl::ComputeTypeSize(type); - } - else - { - buffer->invalidateStaticData(); - staticBuffer = NULL; - } - } - - if (!indexBuffer) - { - ERR("No valid index buffer."); - return GL_INVALID_OPERATION; - } - - unsigned int indexTypeSize = gl::ComputeTypeSize(destinationIndexType); - if (convertCount > std::numeric_limits<unsigned int>::max() / indexTypeSize) - { - ERR("Reserving %u indicies of %u bytes each exceeds the maximum buffer size.", convertCount, indexTypeSize); - return GL_OUT_OF_MEMORY; - } - - unsigned int bufferSizeRequired = convertCount * indexTypeSize; - if (!indexBuffer->reserveBufferSpace(bufferSizeRequired, type)) - { - ERR("Failed to reserve %u bytes in an index buffer.", bufferSizeRequired); - return GL_OUT_OF_MEMORY; - } - - void* output = NULL; - if (!indexBuffer->mapBuffer(bufferSizeRequired, &output, &streamOffset)) - { - ERR("Failed to map index buffer."); - return GL_OUT_OF_MEMORY; - } - - convertIndices(type, staticBuffer ? storage->getData() : indices, convertCount, output); - - if (!indexBuffer->unmapBuffer()) - { - ERR("Failed to unmap index buffer."); - return GL_OUT_OF_MEMORY; - } - - computeRange(type, indices, count, &translated->minIndex, &translated->maxIndex); - - if (staticBuffer) - { - streamOffset = (offset / gl::ComputeTypeSize(type)) * gl::ComputeTypeSize(destinationIndexType); - staticBuffer->getIndexRangeCache()->addRange(type, offset, count, translated->minIndex, - translated->maxIndex, streamOffset); - } - } - - translated->storage = directStorage ? storage : NULL; - translated->indexBuffer = indexBuffer->getIndexBuffer(); - translated->serial = directStorage ? storage->getSerial() : indexBuffer->getSerial(); - translated->startIndex = streamOffset / gl::ComputeTypeSize(destinationIndexType); - translated->startOffset = streamOffset; - - if (buffer) - { - buffer->promoteStaticUsage(count * gl::ComputeTypeSize(type)); - } - - return GL_NO_ERROR; -} - -StaticIndexBufferInterface *IndexDataManager::getCountingIndices(GLsizei count) -{ - if (count <= 65536) // 16-bit indices - { - const unsigned int spaceNeeded = count * sizeof(unsigned short); - - if (!mCountingBuffer || mCountingBuffer->getBufferSize() < spaceNeeded) - { - delete mCountingBuffer; - mCountingBuffer = new StaticIndexBufferInterface(mRenderer); - mCountingBuffer->reserveBufferSpace(spaceNeeded, GL_UNSIGNED_SHORT); - - void* mappedMemory = NULL; - if (!mCountingBuffer->mapBuffer(spaceNeeded, &mappedMemory, NULL)) - { - ERR("Failed to map counting buffer."); - return NULL; - } - - unsigned short *data = reinterpret_cast<unsigned short*>(mappedMemory); - for(int i = 0; i < count; i++) - { - data[i] = i; - } - - if (!mCountingBuffer->unmapBuffer()) - { - ERR("Failed to unmap counting buffer."); - return NULL; - } - } - } - else if (mStreamingBufferInt) // 32-bit indices supported - { - const unsigned int spaceNeeded = count * sizeof(unsigned int); - - if (!mCountingBuffer || mCountingBuffer->getBufferSize() < spaceNeeded) - { - delete mCountingBuffer; - mCountingBuffer = new StaticIndexBufferInterface(mRenderer); - mCountingBuffer->reserveBufferSpace(spaceNeeded, GL_UNSIGNED_INT); - - void* mappedMemory = NULL; - if (!mCountingBuffer->mapBuffer(spaceNeeded, &mappedMemory, NULL)) - { - ERR("Failed to map counting buffer."); - return NULL; - } - - unsigned int *data = reinterpret_cast<unsigned int*>(mappedMemory); - for(int i = 0; i < count; i++) - { - data[i] = i; - } - - if (!mCountingBuffer->unmapBuffer()) - { - ERR("Failed to unmap counting buffer."); - return NULL; - } - } - } - else - { - return NULL; - } - - return mCountingBuffer; -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/IndexRangeCache.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/IndexRangeCache.cpp index 51d7f0b653..f68ac383de 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/IndexRangeCache.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/IndexRangeCache.cpp @@ -1,4 +1,3 @@ -#include "precompiled.h" // // Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be @@ -9,17 +8,50 @@ // ranges of indices. #include "libGLESv2/renderer/IndexRangeCache.h" +#include "libGLESv2/formatutils.h" + #include "common/debug.h" -#include "libGLESv2/utilities.h" + #include <tuple> namespace rx { -void IndexRangeCache::addRange(GLenum type, unsigned int offset, GLsizei count, unsigned int minIdx, unsigned int maxIdx, +template <class IndexType> +static RangeUI ComputeTypedRange(const IndexType *indices, GLsizei count) +{ + unsigned int minIndex = indices[0]; + unsigned int maxIndex = indices[0]; + + for (GLsizei i = 1; i < count; i++) + { + if (minIndex > indices[i]) minIndex = indices[i]; + if (maxIndex < indices[i]) maxIndex = indices[i]; + } + + return RangeUI(minIndex, maxIndex); +} + +RangeUI IndexRangeCache::ComputeRange(GLenum type, const GLvoid *indices, GLsizei count) +{ + switch (type) + { + case GL_UNSIGNED_BYTE: + return ComputeTypedRange(static_cast<const GLubyte*>(indices), count); + case GL_UNSIGNED_INT: + return ComputeTypedRange(static_cast<const GLuint*>(indices), count); + case GL_UNSIGNED_SHORT: + return ComputeTypedRange(static_cast<const GLushort*>(indices), count); + default: + UNREACHABLE(); + return RangeUI(); + } +} + +void IndexRangeCache::addRange(GLenum type, unsigned int offset, GLsizei count, const RangeUI &range, unsigned int streamOffset) { - mIndexRangeCache[IndexRange(type, offset, count)] = IndexBounds(minIdx, maxIdx, streamOffset); + mIndexRangeCache[IndexRange(type, offset, count)] = IndexBounds(range, streamOffset); } void IndexRangeCache::invalidateRange(unsigned int offset, unsigned int size) @@ -31,7 +63,7 @@ void IndexRangeCache::invalidateRange(unsigned int offset, unsigned int size) while (i != mIndexRangeCache.end()) { unsigned int rangeStart = i->second.streamOffset; - unsigned int rangeEnd = i->second.streamOffset + (gl::ComputeTypeSize(i->first.type) * i->first.count); + unsigned int rangeEnd = i->second.streamOffset + (gl::GetTypeInfo(i->first.type).bytes * i->first.count); if (invalidateEnd < rangeStart || invalidateStart > rangeEnd) { @@ -44,21 +76,19 @@ void IndexRangeCache::invalidateRange(unsigned int offset, unsigned int size) } } -bool IndexRangeCache::findRange(GLenum type, unsigned int offset, GLsizei count, unsigned int *outMinIndex, - unsigned int *outMaxIndex, unsigned int *outStreamOffset) const +bool IndexRangeCache::findRange(GLenum type, unsigned int offset, GLsizei count, + RangeUI *outRange, unsigned int *outStreamOffset) const { IndexRangeMap::const_iterator i = mIndexRangeCache.find(IndexRange(type, offset, count)); if (i != mIndexRangeCache.end()) { - if (outMinIndex) *outMinIndex = i->second.minIndex; - if (outMaxIndex) *outMaxIndex = i->second.maxIndex; + if (outRange) *outRange = i->second.range; if (outStreamOffset) *outStreamOffset = i->second.streamOffset; return true; } else { - if (outMinIndex) *outMinIndex = 0; - if (outMaxIndex) *outMaxIndex = 0; + if (outRange) *outRange = RangeUI(0, 0); if (outStreamOffset) *outStreamOffset = 0; return false; } @@ -89,12 +119,13 @@ bool IndexRangeCache::IndexRange::operator<(const IndexRange& rhs) const } IndexRangeCache::IndexBounds::IndexBounds() - : minIndex(0), maxIndex(0), streamOffset(0) + : range(0, 0), + streamOffset(0) { } -IndexRangeCache::IndexBounds::IndexBounds(unsigned int minIdx, unsigned int maxIdx, unsigned int offset) - : minIndex(minIdx), maxIndex(maxIdx), streamOffset(offset) +IndexRangeCache::IndexBounds::IndexBounds(const RangeUI &rangeIn, unsigned int offset) + : range(rangeIn), streamOffset(offset) { } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/IndexRangeCache.h b/src/3rdparty/angle/src/libGLESv2/renderer/IndexRangeCache.h index 56834306f2..a7d91e035b 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/IndexRangeCache.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/IndexRangeCache.h @@ -11,6 +11,11 @@ #define LIBGLESV2_RENDERER_INDEXRANGECACHE_H_ #include "common/angleutils.h" +#include "common/mathutil.h" + +#include "angle_gl.h" + +#include <map> namespace rx { @@ -18,14 +23,16 @@ namespace rx class IndexRangeCache { public: - void addRange(GLenum type, unsigned int offset, GLsizei count, unsigned int minIdx, unsigned int maxIdx, + void addRange(GLenum type, unsigned int offset, GLsizei count, const RangeUI &range, unsigned int streamOffset); - bool findRange(GLenum type, unsigned int offset, GLsizei count, unsigned int *outMinIndex, - unsigned int *outMaxIndex, unsigned int *outStreamOffset) const; + bool findRange(GLenum type, unsigned int offset, GLsizei count, RangeUI *rangeOut, + unsigned int *outStreamOffset) const; void invalidateRange(unsigned int offset, unsigned int size); void clear(); + static RangeUI ComputeRange(GLenum type, const GLvoid *indices, GLsizei count); + private: struct IndexRange { @@ -41,12 +48,11 @@ class IndexRangeCache struct IndexBounds { - unsigned int minIndex; - unsigned int maxIndex; + RangeUI range; unsigned int streamOffset; IndexBounds(); - IndexBounds(unsigned int minIdx, unsigned int maxIdx, unsigned int offset); + IndexBounds(const RangeUI &range, unsigned int offset); }; typedef std::map<IndexRange, IndexBounds> IndexRangeMap; @@ -55,4 +61,4 @@ class IndexRangeCache } -#endif LIBGLESV2_RENDERER_INDEXRANGECACHE_H +#endif // LIBGLESV2_RENDERER_INDEXRANGECACHE_H diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/ProgramImpl.h b/src/3rdparty/angle/src/libGLESv2/renderer/ProgramImpl.h new file mode 100644 index 0000000000..ba0955fdf8 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/ProgramImpl.h @@ -0,0 +1,58 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// ProgramImpl.h: Defines the abstract rx::ProgramImpl class. + +#ifndef LIBGLESV2_RENDERER_PROGRAMIMPL_H_ +#define LIBGLESV2_RENDERER_PROGRAMIMPL_H_ + +#include "common/angleutils.h" +#include "libGLESv2/BinaryStream.h" +#include "libGLESv2/Constants.h" +#include "libGLESv2/ProgramBinary.h" + +namespace rx +{ + +class DynamicHLSL; +class Renderer; + +class ProgramImpl +{ +public: + virtual ~ProgramImpl() { } + + // TODO: Temporary interfaces to ease migration. Remove soon! + virtual Renderer *getRenderer() = 0; + virtual DynamicHLSL *getDynamicHLSL() = 0; + virtual const std::vector<rx::PixelShaderOutputVariable> &getPixelShaderKey() = 0; + + virtual GLenum getBinaryFormat() = 0; + virtual bool load(gl::InfoLog &infoLog, gl::BinaryInputStream *stream) = 0; + virtual bool save(gl::BinaryOutputStream *stream) = 0; + + virtual rx::ShaderExecutable *getPixelExecutableForOutputLayout(gl::InfoLog &infoLog, const std::vector<GLenum> &outputSignature, + const std::vector<gl::LinkedVarying> &transformFeedbackLinkedVaryings, + bool separatedOutputBuffers) = 0; + virtual rx::ShaderExecutable *getVertexExecutableForInputLayout(gl::InfoLog &infoLog, + const gl::VertexFormat inputLayout[gl::MAX_VERTEX_ATTRIBS], + const sh::Attribute shaderAttributes[], + const std::vector<gl::LinkedVarying> &transformFeedbackLinkedVaryings, + bool separatedOutputBuffers) = 0; + + virtual bool link(gl::InfoLog &infoLog, gl::Shader *fragmentShader, gl::Shader *vertexShader, + const std::vector<std::string> &transformFeedbackVaryings, int *registers, + std::vector<gl::LinkedVarying> *linkedVaryings, std::map<int, + gl::VariableLocation> *outputVariables) = 0; + + virtual void initializeUniformStorage(const std::vector<gl::LinkedUniform*> &uniforms) = 0; + + virtual void reset() = 0; +}; + +} + +#endif // LIBGLESV2_RENDERER_PROGRAMIMPL_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/QueryImpl.h b/src/3rdparty/angle/src/libGLESv2/renderer/QueryImpl.h index a874047b0c..6b45810a3b 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/QueryImpl.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/QueryImpl.h @@ -9,27 +9,27 @@ #ifndef LIBGLESV2_RENDERER_QUERYIMPL_H_ #define LIBGLESV2_RENDERER_QUERYIMPL_H_ +#include "libGLESv2/Error.h" + #include "common/angleutils.h" +#include <GLES2/gl2.h> + namespace rx { class QueryImpl { public: - explicit QueryImpl(GLenum type) : mType(type), mStatus(GL_FALSE), mResult(0) { } + explicit QueryImpl(GLenum type) { mType = type; } virtual ~QueryImpl() { } - virtual void begin() = 0; - virtual void end() = 0; - virtual GLuint getResult() = 0; - virtual GLboolean isResultAvailable() = 0; - - GLenum getType() const { return mType; } + virtual gl::Error begin() = 0; + virtual gl::Error end() = 0; + virtual gl::Error getResult(GLuint *params) = 0; + virtual gl::Error isResultAvailable(GLuint *available) = 0; - protected: - GLuint mResult; - GLboolean mStatus; + GLenum getType() const { return mType; } private: DISALLOW_COPY_AND_ASSIGN(QueryImpl); diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/RenderTarget.h b/src/3rdparty/angle/src/libGLESv2/renderer/RenderTarget.h index 80de39f4f7..44637ec7de 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/RenderTarget.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/RenderTarget.h @@ -11,6 +11,7 @@ #define LIBGLESV2_RENDERER_RENDERTARGET_H_ #include "common/angleutils.h" +#include "libGLESv2/angletypes.h" namespace rx { @@ -21,6 +22,7 @@ class RenderTarget { mWidth = 0; mHeight = 0; + mDepth = 0; mInternalFormat = GL_NONE; mActualFormat = GL_NONE; mSamples = 0; @@ -28,21 +30,27 @@ class RenderTarget virtual ~RenderTarget() {}; - GLsizei getWidth() { return mWidth; } - GLsizei getHeight() { return mHeight; } - GLenum getInternalFormat() { return mInternalFormat; } - GLenum getActualFormat() { return mActualFormat; } - GLsizei getSamples() { return mSamples; } - + GLsizei getWidth() const { return mWidth; } + GLsizei getHeight() const { return mHeight; } + GLsizei getDepth() const { return mDepth; } + GLenum getInternalFormat() const { return mInternalFormat; } + GLenum getActualFormat() const { return mActualFormat; } + GLsizei getSamples() const { return mSamples; } + gl::Extents getExtents() const { return gl::Extents(mWidth, mHeight, mDepth); } + + virtual void invalidate(GLint x, GLint y, GLsizei width, GLsizei height) = 0; + struct Desc { GLsizei width; GLsizei height; + GLsizei depth; GLenum format; }; protected: GLsizei mWidth; GLsizei mHeight; + GLsizei mDepth; GLenum mInternalFormat; GLenum mActualFormat; GLsizei mSamples; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/Renderer.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/Renderer.cpp index 5278113811..910d0285f1 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/Renderer.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/Renderer.cpp @@ -1,222 +1,89 @@ -#include "precompiled.h" // -// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // // Renderer.cpp: Implements EGL dependencies for creating and destroying Renderer instances. -#include <EGL/eglext.h> #include "libGLESv2/main.h" #include "libGLESv2/Program.h" #include "libGLESv2/renderer/Renderer.h" -#if defined(ANGLE_ENABLE_D3D9) -# include "libGLESv2/renderer/d3d9/Renderer9.h" -#endif -#if defined(ANGLE_ENABLE_D3D11) -# include "libGLESv2/renderer/d3d11/Renderer11.h" -#endif -#include "libGLESv2/utilities.h" -#include "third_party/trace_event/trace_event.h" - -#ifndef D3DERR_OUTOFVIDEOMEMORY -#define D3DERR_OUTOFVIDEOMEMORY MAKE_HRESULT(1, 0x876, 380) -#endif - -#if defined(__MINGW32__) || defined(ANGLE_OS_WINPHONE) - -#ifndef D3DCOMPILER_DLL - -// Add define + typedefs for older MinGW-w64 headers (pre 5783) - -#define D3DCOMPILER_DLL L"d3dcompiler_43.dll" +#include "common/utilities.h" +#include "libGLESv2/Shader.h" -HRESULT WINAPI D3DCompile(const void *data, SIZE_T data_size, const char *filename, - const D3D_SHADER_MACRO *defines, ID3DInclude *include, const char *entrypoint, - const char *target, UINT sflags, UINT eflags, ID3DBlob **shader, ID3DBlob **error_messages); -typedef HRESULT (WINAPI *pD3DCompile)(const void *data, SIZE_T data_size, const char *filename, - const D3D_SHADER_MACRO *defines, ID3DInclude *include, const char *entrypoint, - const char *target, UINT sflags, UINT eflags, ID3DBlob **shader, ID3DBlob **error_messages); +#if defined (ANGLE_ENABLE_D3D9) +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" +#endif // ANGLE_ENABLE_D3D9 -#endif // D3DCOMPILER_DLL +#if defined (ANGLE_ENABLE_D3D11) +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" +#endif // ANGLE_ENABLE_D3D11 -#endif // __MINGW32__ || ANGLE_OS_WINPHONE +#if defined (ANGLE_TEST_CONFIG) +#define ANGLE_DEFAULT_D3D11 1 +#endif -#ifndef QT_D3DCOMPILER_DLL -#define QT_D3DCOMPILER_DLL D3DCOMPILER_DLL +#if !defined(ANGLE_DEFAULT_D3D11) +// Enables use of the Direct3D 11 API for a default display, when available +#define ANGLE_DEFAULT_D3D11 0 #endif +#include <EGL/eglext.h> + namespace rx { -Renderer::Renderer(egl::Display *display) : mDisplay(display) +Renderer::Renderer(egl::Display *display) + : mDisplay(display), + mCapsInitialized(false), + mCurrentClientVersion(2) { - mD3dCompilerModule = NULL; - mD3DCompileFunc = NULL; } Renderer::~Renderer() { - if (mD3dCompilerModule) - { - FreeLibrary(mD3dCompilerModule); - mD3dCompilerModule = NULL; - } } -bool Renderer::initializeCompiler() +const gl::Caps &Renderer::getRendererCaps() const { - TRACE_EVENT0("gpu", "initializeCompiler"); -#if defined(ANGLE_PRELOADED_D3DCOMPILER_MODULE_NAMES) - // Find a D3DCompiler module that had already been loaded based on a predefined list of versions. - static TCHAR* d3dCompilerNames[] = ANGLE_PRELOADED_D3DCOMPILER_MODULE_NAMES; - - for (size_t i = 0; i < ArraySize(d3dCompilerNames); ++i) - { - if (GetModuleHandleEx(0, d3dCompilerNames[i], &mD3dCompilerModule)) - { - break; - } - } -#endif // ANGLE_PRELOADED_D3DCOMPILER_MODULE_NAMES - - // Load the compiler DLL specified by the environment, or default to QT_D3DCOMPILER_DLL -#if !defined(ANGLE_OS_WINRT) - const wchar_t *defaultCompiler = _wgetenv(L"QT_D3DCOMPILER_DLL"); - if (!defaultCompiler) - defaultCompiler = QT_D3DCOMPILER_DLL; -#else // !ANGLE_OS_WINRT -# ifdef _DEBUG - const wchar_t *defaultCompiler = L"d3dcompiler_qtd.dll"; -# else - const wchar_t *defaultCompiler = L"d3dcompiler_qt.dll"; -# endif -#endif // ANGLE_OS_WINRT - - const wchar_t *compilerDlls[] = { - defaultCompiler, - L"d3dcompiler_47.dll", - L"d3dcompiler_46.dll", - L"d3dcompiler_45.dll", - L"d3dcompiler_44.dll", - L"d3dcompiler_43.dll", - 0 - }; - - // Load the first available known compiler DLL - for (int i = 0; compilerDlls[i]; ++i) - { - // Load the version of the D3DCompiler DLL associated with the Direct3D version ANGLE was built with. - mD3dCompilerModule = LoadLibrary(compilerDlls[i]); - if (mD3dCompilerModule) - break; - } - - if (!mD3dCompilerModule) + if (!mCapsInitialized) { - ERR("No D3D compiler module found - aborting!\n"); - return false; + generateCaps(&mCaps, &mTextureCaps, &mExtensions); + mCapsInitialized = true; } - mD3DCompileFunc = reinterpret_cast<pCompileFunc>(GetProcAddress(mD3dCompilerModule, "D3DCompile")); - ASSERT(mD3DCompileFunc); - - return mD3DCompileFunc != NULL; + return mCaps; } -// Compiles HLSL code into executable binaries -ShaderBlob *Renderer::compileToBinary(gl::InfoLog &infoLog, const char *hlsl, const char *profile, UINT optimizationFlags, bool alternateFlags) +const gl::TextureCapsMap &Renderer::getRendererTextureCaps() const { - if (!hlsl) + if (!mCapsInitialized) { - return NULL; + generateCaps(&mCaps, &mTextureCaps, &mExtensions); + mCapsInitialized = true; } - HRESULT result = S_OK; - UINT flags = 0; - std::string sourceText; - if (gl::perfActive()) - { - flags |= D3DCOMPILE_DEBUG; - -#ifdef NDEBUG - flags |= optimizationFlags; -#else - flags |= D3DCOMPILE_SKIP_OPTIMIZATION; -#endif + return mTextureCaps; +} - std::string sourcePath = getTempPath(); - sourceText = std::string("#line 2 \"") + sourcePath + std::string("\"\n\n") + std::string(hlsl); - writeFile(sourcePath.c_str(), sourceText.c_str(), sourceText.size()); - } - else +const gl::Extensions &Renderer::getRendererExtensions() const +{ + if (!mCapsInitialized) { - flags |= optimizationFlags; - sourceText = hlsl; + generateCaps(&mCaps, &mTextureCaps, &mExtensions); + mCapsInitialized = true; } - // Sometimes D3DCompile will fail with the default compilation flags for complicated shaders when it would otherwise pass with alternative options. - // Try the default flags first and if compilation fails, try some alternatives. - const static UINT extraFlags[] = - { - 0, - D3DCOMPILE_AVOID_FLOW_CONTROL, - D3DCOMPILE_PREFER_FLOW_CONTROL - }; - - const static char * const extraFlagNames[] = - { - "default", - "avoid flow control", - "prefer flow control" - }; - - int attempts = alternateFlags ? ArraySize(extraFlags) : 1; - pD3DCompile compileFunc = reinterpret_cast<pD3DCompile>(mD3DCompileFunc); - for (int i = 0; i < attempts; ++i) - { - ID3DBlob *errorMessage = NULL; - ID3DBlob *binary = NULL; - - result = compileFunc(hlsl, strlen(hlsl), gl::g_fakepath, NULL, NULL, - "main", profile, flags | extraFlags[i], 0, &binary, &errorMessage); - if (errorMessage) - { - const char *message = (const char*)errorMessage->GetBufferPointer(); - - infoLog.appendSanitized(message); - TRACE("\n%s", hlsl); - TRACE("\n%s", message); - - errorMessage->Release(); - errorMessage = NULL; - } - - if (SUCCEEDED(result)) - { - return (ShaderBlob*)binary; - } - else - { - if (result == D3DERR_OUTOFVIDEOMEMORY || result == E_OUTOFMEMORY) - { - return gl::error(GL_OUT_OF_MEMORY, (ShaderBlob*) NULL); - } + return mExtensions; +} - infoLog.append("Warning: D3D shader compilation failed with "); - infoLog.append(extraFlagNames[i]); - infoLog.append(" flags."); - if (i + 1 < attempts) - { - infoLog.append(" Retrying with "); - infoLog.append(extraFlagNames[i + 1]); - infoLog.append(".\n"); - } - } - } +typedef Renderer *(*CreateRendererFunction)(egl::Display*, EGLNativeDisplayType, EGLint); - return NULL; +template <typename RendererType> +Renderer *CreateRenderer(egl::Display *display, EGLNativeDisplayType nativeDisplay, EGLint requestedDisplayType) +{ + return new RendererType(display, nativeDisplay, requestedDisplayType); } } @@ -224,56 +91,64 @@ ShaderBlob *Renderer::compileToBinary(gl::InfoLog &infoLog, const char *hlsl, co extern "C" { -rx::Renderer *glCreateRenderer(egl::Display *display, EGLNativeDisplayType displayId) +rx::Renderer *glCreateRenderer(egl::Display *display, EGLNativeDisplayType nativeDisplay, EGLint requestedDisplayType) { - rx::Renderer *renderer = NULL; - EGLint status = EGL_BAD_ALLOC; + std::vector<rx::CreateRendererFunction> rendererCreationFunctions; -#if defined(ANGLE_OS_WINRT) - if (displayId == EGL_DEFAULT_DISPLAY) - displayId = EGL_D3D11_ONLY_DISPLAY_ANGLE; -#endif - -#if defined(ANGLE_ENABLE_D3D11) - if (displayId == EGL_DEFAULT_DISPLAY || - displayId == EGL_D3D11_ELSE_D3D9_DISPLAY_ANGLE || - displayId == EGL_D3D11_ONLY_DISPLAY_ANGLE) - { - renderer = new rx::Renderer11(display); - - if (renderer) +# if defined(ANGLE_ENABLE_D3D11) + if (nativeDisplay == EGL_D3D11_ELSE_D3D9_DISPLAY_ANGLE || + nativeDisplay == EGL_D3D11_ONLY_DISPLAY_ANGLE || + requestedDisplayType == EGL_PLATFORM_ANGLE_TYPE_D3D11_ANGLE || + requestedDisplayType == EGL_PLATFORM_ANGLE_TYPE_D3D11_WARP_ANGLE) { - status = renderer->initialize(); + rendererCreationFunctions.push_back(rx::CreateRenderer<rx::Renderer11>); } +# endif - if (status == EGL_SUCCESS) +# if defined(ANGLE_ENABLE_D3D9) + if (nativeDisplay == EGL_D3D11_ELSE_D3D9_DISPLAY_ANGLE || + requestedDisplayType == EGL_PLATFORM_ANGLE_TYPE_D3D9_ANGLE) { - return renderer; + rendererCreationFunctions.push_back(rx::CreateRenderer<rx::Renderer9>); } - else if (displayId == EGL_D3D11_ONLY_DISPLAY_ANGLE) - { - return NULL; - } - - // Failed to create a D3D11 renderer, try creating a D3D9 renderer - delete renderer; - } -#endif // ANGLE_ENABLE_D3D11 - -#if defined(ANGLE_ENABLE_D3D9) - bool softwareDevice = (displayId == EGL_SOFTWARE_DISPLAY_ANGLE); - renderer = new rx::Renderer9(display, displayId, softwareDevice); +# endif - if (renderer) + if (nativeDisplay != EGL_D3D11_ELSE_D3D9_DISPLAY_ANGLE && + nativeDisplay != EGL_D3D11_ONLY_DISPLAY_ANGLE && + requestedDisplayType == EGL_PLATFORM_ANGLE_TYPE_DEFAULT_ANGLE) { - status = renderer->initialize(); + // The default display is requested, try the D3D9 and D3D11 renderers, order them using + // the definition of ANGLE_DEFAULT_D3D11 +# if ANGLE_DEFAULT_D3D11 +# if defined(ANGLE_ENABLE_D3D11) + rendererCreationFunctions.push_back(rx::CreateRenderer<rx::Renderer11>); +# endif +# if defined(ANGLE_ENABLE_D3D9) + rendererCreationFunctions.push_back(rx::CreateRenderer<rx::Renderer9>); +# endif +# else +# if defined(ANGLE_ENABLE_D3D9) + rendererCreationFunctions.push_back(rx::CreateRenderer<rx::Renderer9>); +# endif +# if defined(ANGLE_ENABLE_D3D11) + rendererCreationFunctions.push_back(rx::CreateRenderer<rx::Renderer11>); +# endif +# endif } - if (status == EGL_SUCCESS) + for (size_t i = 0; i < rendererCreationFunctions.size(); i++) { - return renderer; + rx::Renderer *renderer = rendererCreationFunctions[i](display, nativeDisplay, requestedDisplayType); + if (renderer->initialize() == EGL_SUCCESS) + { + return renderer; + } + else + { + // Failed to create the renderer, try the next + SafeDelete(renderer); + } } -#endif // ANGLE_ENABLE_D3D9 return NULL; } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/Renderer.h b/src/3rdparty/angle/src/libGLESv2/renderer/Renderer.h index 79578b2458..b2249741ab 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/Renderer.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/Renderer.h @@ -1,6 +1,5 @@ -#include "../precompiled.h" // -// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // @@ -13,23 +12,18 @@ #include "libGLESv2/Uniform.h" #include "libGLESv2/angletypes.h" +#include "libGLESv2/Caps.h" +#include "libGLESv2/Error.h" -#if !defined(ANGLE_COMPILE_OPTIMIZATION_LEVEL) -#define ANGLE_COMPILE_OPTIMIZATION_LEVEL D3DCOMPILE_OPTIMIZATION_LEVEL3 -#endif +#include <cstdint> -const int versionWindowsVista = MAKEWORD(0x00, 0x06); -const int versionWindows7 = MAKEWORD(0x01, 0x06); +#include <EGL/egl.h> -// Return the version of the operating system in a format suitable for ordering -// comparison. -inline int getComparableOSVersion() -{ - DWORD version = GetVersion(); - int majorVersion = LOBYTE(LOWORD(version)); - int minorVersion = HIBYTE(LOWORD(version)); - return MAKEWORD(minorVersion, majorVersion); -} +#if !defined(ANGLE_COMPILE_OPTIMIZATION_LEVEL) +// WARNING: D3DCOMPILE_OPTIMIZATION_LEVEL3 may lead to a DX9 shader compiler hang. +// It should only be used selectively to work around specific bugs. +#define ANGLE_COMPILE_OPTIMIZATION_LEVEL D3DCOMPILE_OPTIMIZATION_LEVEL1 +#endif namespace egl { @@ -40,31 +34,35 @@ namespace gl { class InfoLog; class ProgramBinary; -class VertexAttribute; +struct LinkedVarying; +struct VertexAttribute; class Buffer; class Texture; class Framebuffer; +struct VertexAttribCurrentValueData; } namespace rx { -class TextureStorageInterface2D; -class TextureStorageInterfaceCube; +class TextureStorage; class VertexBuffer; class IndexBuffer; class QueryImpl; class FenceImpl; +class BufferImpl; +class VertexArrayImpl; class BufferStorage; -class Blit; struct TranslatedIndexData; +class ShaderImpl; +class ProgramImpl; class ShaderExecutable; class SwapChain; class RenderTarget; class Image; class TextureStorage; - -typedef void * ShaderBlob; -typedef void (*pCompileFunc)(); +class UniformStorage; +class TextureImpl; +class TransformFeedbackImpl; struct ConfigDesc { @@ -72,6 +70,7 @@ struct ConfigDesc GLenum depthStencilFormat; GLint multiSample; bool fastConfig; + bool es3Capable; }; struct dx_VertexConstants @@ -94,12 +93,6 @@ enum ShaderType SHADER_GEOMETRY }; -enum D3DWorkaroundType -{ - ANGLE_D3D_WORKAROUND_NONE, - ANGLE_D3D_WORKAROUND_SM3_OPTIMIZER -}; - class Renderer { public: @@ -116,30 +109,37 @@ class Renderer virtual SwapChain *createSwapChain(EGLNativeWindowType window, HANDLE shareHandle, GLenum backBufferFormat, GLenum depthBufferFormat) = 0; - virtual void setSamplerState(gl::SamplerType type, int index, const gl::SamplerState &sampler) = 0; - virtual void setTexture(gl::SamplerType type, int index, gl::Texture *texture) = 0; + virtual gl::Error generateSwizzle(gl::Texture *texture) = 0; + virtual gl::Error setSamplerState(gl::SamplerType type, int index, const gl::SamplerState &sampler) = 0; + virtual gl::Error setTexture(gl::SamplerType type, int index, gl::Texture *texture) = 0; + + virtual gl::Error setUniformBuffers(const gl::Buffer *vertexUniformBuffers[], const gl::Buffer *fragmentUniformBuffers[]) = 0; - virtual void setRasterizerState(const gl::RasterizerState &rasterState) = 0; - virtual void setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState &blendState, const gl::Color &blendColor, - unsigned int sampleMask) = 0; - virtual void setDepthStencilState(const gl::DepthStencilState &depthStencilState, int stencilRef, - int stencilBackRef, bool frontFaceCCW) = 0; + virtual gl::Error setRasterizerState(const gl::RasterizerState &rasterState) = 0; + virtual gl::Error setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState &blendState, const gl::ColorF &blendColor, + unsigned int sampleMask) = 0; + virtual gl::Error setDepthStencilState(const gl::DepthStencilState &depthStencilState, int stencilRef, + int stencilBackRef, bool frontFaceCCW) = 0; virtual void setScissorRectangle(const gl::Rectangle &scissor, bool enabled) = 0; - virtual bool setViewport(const gl::Rectangle &viewport, float zNear, float zFar, GLenum drawMode, GLenum frontFace, + virtual void setViewport(const gl::Rectangle &viewport, float zNear, float zFar, GLenum drawMode, GLenum frontFace, bool ignoreViewport) = 0; - virtual bool applyRenderTarget(gl::Framebuffer *frameBuffer) = 0; - virtual void applyShaders(gl::ProgramBinary *programBinary) = 0; - virtual void applyUniforms(gl::ProgramBinary *programBinary, gl::UniformArray *uniformArray) = 0; + virtual gl::Error applyRenderTarget(gl::Framebuffer *frameBuffer) = 0; + virtual gl::Error applyShaders(gl::ProgramBinary *programBinary, const gl::VertexFormat inputLayout[], const gl::Framebuffer *framebuffer, + bool rasterizerDiscard, bool transformFeedbackActive) = 0; + virtual gl::Error applyUniforms(const gl::ProgramBinary &programBinary) = 0; virtual bool applyPrimitiveType(GLenum primitiveType, GLsizei elementCount) = 0; - virtual GLenum applyVertexBuffer(gl::ProgramBinary *programBinary, gl::VertexAttribute vertexAttributes[], GLint first, GLsizei count, GLsizei instances) = 0; - virtual GLenum applyIndexBuffer(const GLvoid *indices, gl::Buffer *elementArrayBuffer, GLsizei count, GLenum mode, GLenum type, TranslatedIndexData *indexInfo) = 0; + virtual gl::Error applyVertexBuffer(gl::ProgramBinary *programBinary, const gl::VertexAttribute vertexAttributes[], const gl::VertexAttribCurrentValueData currentValues[], + GLint first, GLsizei count, GLsizei instances) = 0; + virtual gl::Error applyIndexBuffer(const GLvoid *indices, gl::Buffer *elementArrayBuffer, GLsizei count, GLenum mode, GLenum type, TranslatedIndexData *indexInfo) = 0; + virtual void applyTransformFeedbackBuffers(gl::Buffer *transformFeedbackBuffers[], GLintptr offsets[]) = 0; - virtual void drawArrays(GLenum mode, GLsizei count, GLsizei instances) = 0; - virtual void drawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, gl::Buffer *elementArrayBuffer, const TranslatedIndexData &indexInfo, GLsizei instances) = 0; + virtual gl::Error drawArrays(GLenum mode, GLsizei count, GLsizei instances, bool transformFeedbackActive) = 0; + virtual gl::Error drawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, + gl::Buffer *elementArrayBuffer, const TranslatedIndexData &indexInfo, GLsizei instances) = 0; - virtual void clear(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer) = 0; + virtual gl::Error clear(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer) = 0; virtual void markAllStateDirty() = 0; @@ -149,102 +149,119 @@ class Renderer virtual bool testDeviceLost(bool notify) = 0; virtual bool testDeviceResettable() = 0; - // Renderer capabilities + // Renderer capabilities (virtual because it is used by egl::Display, do not override) + virtual const gl::Caps &getRendererCaps() const; + virtual const gl::TextureCapsMap &getRendererTextureCaps() const; + virtual const gl::Extensions &getRendererExtensions() const; + virtual DWORD getAdapterVendor() const = 0; virtual std::string getRendererDescription() const = 0; virtual GUID getAdapterIdentifier() const = 0; - virtual bool getBGRATextureSupport() const = 0; - virtual bool getDXT1TextureSupport() = 0; - virtual bool getDXT3TextureSupport() = 0; - virtual bool getDXT5TextureSupport() = 0; - virtual bool getEventQuerySupport() = 0; - virtual bool getFloat32TextureSupport(bool *filtering, bool *renderable) = 0; - virtual bool getFloat16TextureSupport(bool *filtering, bool *renderable) = 0; - virtual bool getLuminanceTextureSupport() = 0; - virtual bool getLuminanceAlphaTextureSupport() = 0; - bool getVertexTextureSupport() const { return getMaxVertexTextureImageUnits() > 0; } - virtual unsigned int getMaxVertexTextureImageUnits() const = 0; - virtual unsigned int getMaxCombinedTextureImageUnits() const = 0; virtual unsigned int getReservedVertexUniformVectors() const = 0; virtual unsigned int getReservedFragmentUniformVectors() const = 0; - virtual unsigned int getMaxVertexUniformVectors() const = 0; - virtual unsigned int getMaxFragmentUniformVectors() const = 0; - virtual unsigned int getMaxVaryingVectors() const = 0; - virtual bool getNonPower2TextureSupport() const = 0; - virtual bool getDepthTextureSupport() const = 0; - virtual bool getOcclusionQuerySupport() const = 0; - virtual bool getInstancingSupport() const = 0; - virtual bool getTextureFilterAnisotropySupport() const = 0; - virtual float getTextureMaxAnisotropy() const = 0; + virtual unsigned int getReservedVertexUniformBuffers() const = 0; + virtual unsigned int getReservedFragmentUniformBuffers() const = 0; virtual bool getShareHandleSupport() const = 0; - virtual bool getDerivativeInstructionSupport() const = 0; virtual bool getPostSubBufferSupport() const = 0; virtual int getMajorShaderModel() const = 0; - virtual float getMaxPointSize() const = 0; - virtual int getMaxViewportDimension() const = 0; - virtual int getMaxTextureWidth() const = 0; - virtual int getMaxTextureHeight() const = 0; - virtual bool get32BitIndexSupport() const = 0; virtual int getMinSwapInterval() const = 0; virtual int getMaxSwapInterval() const = 0; - virtual GLsizei getMaxSupportedSamples() const = 0; - - virtual unsigned int getMaxRenderTargets() const = 0; - // Pixel operations - virtual bool copyToRenderTarget(TextureStorageInterface2D *dest, TextureStorageInterface2D *source) = 0; - virtual bool copyToRenderTarget(TextureStorageInterfaceCube *dest, TextureStorageInterfaceCube *source) = 0; - - virtual bool copyImage(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, - GLint xoffset, GLint yoffset, TextureStorageInterface2D *storage, GLint level) = 0; - virtual bool copyImage(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, - GLint xoffset, GLint yoffset, TextureStorageInterfaceCube *storage, GLenum target, GLint level) = 0; + virtual bool copyToRenderTarget2D(TextureStorage *dest, TextureStorage *source) = 0; + virtual bool copyToRenderTargetCube(TextureStorage *dest, TextureStorage *source) = 0; + virtual bool copyToRenderTarget3D(TextureStorage *dest, TextureStorage *source) = 0; + virtual bool copyToRenderTarget2DArray(TextureStorage *dest, TextureStorage *source) = 0; + + virtual bool copyImage2D(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, TextureStorage *storage, GLint level) = 0; + virtual bool copyImageCube(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, TextureStorage *storage, GLenum target, GLint level) = 0; + virtual bool copyImage3D(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, GLint zOffset, TextureStorage *storage, GLint level) = 0; + virtual bool copyImage2DArray(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, GLint zOffset, TextureStorage *storage, GLint level) = 0; virtual bool blitRect(gl::Framebuffer *readTarget, const gl::Rectangle &readRect, gl::Framebuffer *drawTarget, const gl::Rectangle &drawRect, - bool blitRenderTarget, bool blitDepthStencil) = 0; - virtual void readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, - GLsizei outputPitch, bool packReverseRowOrder, GLint packAlignment, void* pixels) = 0; + const gl::Rectangle *scissor, bool blitRenderTarget, bool blitDepth, bool blitStencil, GLenum filter) = 0; + + virtual gl::Error readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, + GLenum type, GLuint outputPitch, const gl::PixelPackState &pack, uint8_t *pixels) = 0; // RenderTarget creation virtual RenderTarget *createRenderTarget(SwapChain *swapChain, bool depth) = 0; - virtual RenderTarget *createRenderTarget(int width, int height, GLenum format, GLsizei samples, bool depth) = 0; + virtual RenderTarget *createRenderTarget(int width, int height, GLenum format, GLsizei samples) = 0; + + // Shader creation + virtual ShaderImpl *createShader(GLenum type) = 0; + virtual ProgramImpl *createProgram() = 0; // Shader operations - virtual ShaderExecutable *loadExecutable(const void *function, size_t length, rx::ShaderType type) = 0; - virtual ShaderExecutable *compileToExecutable(gl::InfoLog &infoLog, const char *shaderHLSL, rx::ShaderType type, D3DWorkaroundType workaround) = 0; + virtual void releaseShaderCompiler() = 0; + virtual ShaderExecutable *loadExecutable(const void *function, size_t length, rx::ShaderType type, + const std::vector<gl::LinkedVarying> &transformFeedbackVaryings, + bool separatedOutputBuffers) = 0; + virtual ShaderExecutable *compileToExecutable(gl::InfoLog &infoLog, const char *shaderHLSL, rx::ShaderType type, + const std::vector<gl::LinkedVarying> &transformFeedbackVaryings, + bool separatedOutputBuffers, D3DWorkaroundType workaround) = 0; + virtual UniformStorage *createUniformStorage(size_t storageSize) = 0; // Image operations virtual Image *createImage() = 0; virtual void generateMipmap(Image *dest, Image *source) = 0; virtual TextureStorage *createTextureStorage2D(SwapChain *swapChain) = 0; - virtual TextureStorage *createTextureStorage2D(int levels, GLenum internalformat, GLenum usage, bool forceRenderable, GLsizei width, GLsizei height) = 0; - virtual TextureStorage *createTextureStorageCube(int levels, GLenum internalformat, GLenum usage, bool forceRenderable, int size) = 0; + virtual TextureStorage *createTextureStorage2D(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, int levels) = 0; + virtual TextureStorage *createTextureStorageCube(GLenum internalformat, bool renderTarget, int size, int levels) = 0; + virtual TextureStorage *createTextureStorage3D(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, GLsizei depth, int levels) = 0; + virtual TextureStorage *createTextureStorage2DArray(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, GLsizei depth, int levels) = 0; + + // Texture creation + virtual TextureImpl *createTexture(GLenum target) = 0; // Buffer creation + virtual BufferImpl *createBuffer() = 0; virtual VertexBuffer *createVertexBuffer() = 0; virtual IndexBuffer *createIndexBuffer() = 0; - virtual BufferStorage *createBufferStorage() = 0; + + // Vertex Array creation + virtual VertexArrayImpl *createVertexArray() = 0; // Query and Fence creation virtual QueryImpl *createQuery(GLenum type) = 0; virtual FenceImpl *createFence() = 0; + // Transform Feedback creation + virtual TransformFeedbackImpl* createTransformFeedback() = 0; + + // Current GLES client version + void setCurrentClientVersion(int clientVersion) { mCurrentClientVersion = clientVersion; } + int getCurrentClientVersion() const { return mCurrentClientVersion; } + + // Buffer-to-texture and Texture-to-buffer copies + virtual bool supportsFastCopyBufferToTexture(GLenum internalFormat) const = 0; + virtual bool fastCopyBufferToTexture(const gl::PixelUnpackState &unpack, unsigned int offset, RenderTarget *destRenderTarget, + GLenum destinationFormat, GLenum sourcePixelsType, const gl::Box &destArea) = 0; + virtual bool getLUID(LUID *adapterLuid) const = 0; + virtual rx::VertexConversionType getVertexConversionType(const gl::VertexFormat &vertexFormat) const = 0; + virtual GLenum getVertexComponentType(const gl::VertexFormat &vertexFormat) const = 0; protected: - bool initializeCompiler(); - ShaderBlob *compileToBinary(gl::InfoLog &infoLog, const char *hlsl, const char *profile, UINT optimizationFlags, bool alternateFlags); - egl::Display *mDisplay; private: DISALLOW_COPY_AND_ASSIGN(Renderer); - HMODULE mD3dCompilerModule; - pCompileFunc mD3DCompileFunc; + virtual void generateCaps(gl::Caps *outCaps, gl::TextureCapsMap* outTextureCaps, gl::Extensions *outExtensions) const = 0; + + mutable bool mCapsInitialized; + mutable gl::Caps mCaps; + mutable gl::TextureCapsMap mTextureCaps; + mutable gl::Extensions mExtensions; + + int mCurrentClientVersion; }; } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/ShaderExecutable.h b/src/3rdparty/angle/src/libGLESv2/renderer/ShaderExecutable.h index 293e340845..f17195673d 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/ShaderExecutable.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/ShaderExecutable.h @@ -11,6 +11,10 @@ #define LIBGLESV2_RENDERER_SHADEREXECUTABLE_H_ #include "common/angleutils.h" +#include "common/debug.h" + +#include <vector> +#include <cstdint> namespace rx { @@ -18,32 +22,44 @@ namespace rx class ShaderExecutable { public: - ShaderExecutable(const void *function, size_t length) : mLength(length) - { - mFunction = new char[length]; - memcpy(mFunction, function, length); - } - - virtual ~ShaderExecutable() + ShaderExecutable(const void *function, size_t length) + : mFunctionBuffer(length) { - delete[] mFunction; + memcpy(mFunctionBuffer.data(), function, length); } - void *getFunction() const + virtual ~ShaderExecutable() {} + + const uint8_t *getFunction() const { - return mFunction; + return mFunctionBuffer.data(); } size_t getLength() const { - return mLength; + return mFunctionBuffer.size(); } private: DISALLOW_COPY_AND_ASSIGN(ShaderExecutable); - void *mFunction; - const size_t mLength; + std::vector<uint8_t> mFunctionBuffer; +}; + +class UniformStorage +{ + public: + UniformStorage(size_t initialSize) + : mSize(initialSize) + { + } + + virtual ~UniformStorage() {} + + size_t size() const { return mSize; } + + private: + size_t mSize; }; } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/ShaderImpl.h b/src/3rdparty/angle/src/libGLESv2/renderer/ShaderImpl.h new file mode 100644 index 0000000000..de5d30e6fe --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/ShaderImpl.h @@ -0,0 +1,54 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// ShaderImpl.h: Defines the abstract rx::ShaderImpl class. + +#ifndef LIBGLESV2_RENDERER_SHADERIMPL_H_ +#define LIBGLESV2_RENDERER_SHADERIMPL_H_ + +#include <vector> + +#include "common/angleutils.h" +#include "libGLESv2/Shader.h" + +namespace rx +{ + +class ShaderImpl +{ + public: + ShaderImpl() { } + virtual ~ShaderImpl() { } + + virtual bool compile(const std::string &source) = 0; + virtual const std::string &getInfoLog() const = 0; + virtual const std::string &getTranslatedSource() const = 0; + + const std::vector<gl::PackedVarying> &getVaryings() const { return mVaryings; } + const std::vector<sh::Uniform> &getUniforms() const { return mUniforms; } + const std::vector<sh::InterfaceBlock> &getInterfaceBlocks() const { return mInterfaceBlocks; } + const std::vector<sh::Attribute> &getActiveAttributes() const { return mActiveAttributes; } + const std::vector<sh::Attribute> &getActiveOutputVariables() const { return mActiveOutputVariables; } + + std::vector<gl::PackedVarying> &getVaryings() { return mVaryings; } + std::vector<sh::Uniform> &getUniforms() { return mUniforms; } + std::vector<sh::InterfaceBlock> &getInterfaceBlocks() { return mInterfaceBlocks; } + std::vector<sh::Attribute> &getActiveAttributes() { return mActiveAttributes; } + std::vector<sh::Attribute> &getActiveOutputVariables() { return mActiveOutputVariables; } + + protected: + DISALLOW_COPY_AND_ASSIGN(ShaderImpl); + + std::vector<gl::PackedVarying> mVaryings; + std::vector<sh::Uniform> mUniforms; + std::vector<sh::InterfaceBlock> mInterfaceBlocks; + std::vector<sh::Attribute> mActiveAttributes; + std::vector<sh::Attribute> mActiveOutputVariables; +}; + +} + +#endif // LIBGLESV2_RENDERER_SHADERIMPL_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/SwapChain.h b/src/3rdparty/angle/src/libGLESv2/renderer/SwapChain.h index 8231fbcb25..1ec702f299 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/SwapChain.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/SwapChain.h @@ -1,4 +1,3 @@ -#include "../precompiled.h" // // Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be @@ -12,14 +11,23 @@ #define LIBGLESV2_RENDERER_SWAPCHAIN_H_ #include "common/angleutils.h" +#include "common/platform.h" -#if !defined(ANGLE_FORCE_VSYNC_OFF) -#define ANGLE_FORCE_VSYNC_OFF 0 -#endif +#include <GLES2/gl2.h> +#include <EGL/egl.h> +#include <EGL/eglplatform.h> namespace rx { +enum SwapFlags +{ + SWAP_NORMAL = 0, + SWAP_ROTATE_90 = 1, + SWAP_ROTATE_270 = 2, + SWAP_ROTATE_180 = SWAP_ROTATE_90|SWAP_ROTATE_270, +}; + class SwapChain { public: @@ -32,13 +40,13 @@ class SwapChain virtual EGLint resize(EGLint backbufferWidth, EGLint backbufferSize) = 0; virtual EGLint reset(EGLint backbufferWidth, EGLint backbufferHeight, EGLint swapInterval) = 0; - virtual EGLint swapRect(EGLint x, EGLint y, EGLint width, EGLint height) = 0; + virtual EGLint swapRect(EGLint x, EGLint y, EGLint width, EGLint height, EGLint flags) = 0; virtual void recreate() = 0; virtual HANDLE getShareHandle() {return mShareHandle;}; protected: - const EGLNativeWindowType mWindow; // Window that the surface is created for. + const EGLNativeWindowType mWindow; // Window that the surface is created for. const GLenum mBackBufferFormat; const GLenum mDepthBufferFormat; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/TextureImpl.h b/src/3rdparty/angle/src/libGLESv2/renderer/TextureImpl.h new file mode 100644 index 0000000000..e3cc50d680 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/TextureImpl.h @@ -0,0 +1,70 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// TextureImpl.h: Defines the abstract rx::TextureImpl classes. + +#ifndef LIBGLESV2_RENDERER_TEXTUREIMPL_H_ +#define LIBGLESV2_RENDERER_TEXTUREIMPL_H_ + +#include "common/angleutils.h" + +#include "angle_gl.h" + +#include "libGLESv2/ImageIndex.h" + +namespace egl +{ +class Surface; +} + +namespace gl +{ +class Framebuffer; +struct PixelUnpackState; +struct SamplerState; +} + +namespace rx +{ + +class Image; +class Renderer; +class TextureStorage; + +class TextureImpl +{ + public: + virtual ~TextureImpl() {}; + + // TODO: If this methods could go away that would be ideal; + // TextureStorage should only be necessary for the D3D backend, and as such + // higher level code should not rely on it. + virtual TextureStorage *getNativeTexture() = 0; + + // Deprecated in favour of the ImageIndex method + virtual Image *getImage(int level, int layer) const = 0; + virtual Image *getImage(const gl::ImageIndex &index) const = 0; + virtual GLsizei getLayerCount(int level) const = 0; + + virtual void setUsage(GLenum usage) = 0; + + virtual void setImage(GLenum target, GLint level, GLsizei width, GLsizei height, GLsizei depth, GLenum internalFormat, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels) = 0; + virtual void setCompressedImage(GLenum target, GLint level, GLenum format, GLsizei width, GLsizei height, GLsizei depth, GLsizei imageSize, const void *pixels) = 0; + virtual void subImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels) = 0; + virtual void subImageCompressed(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void *pixels) = 0; + virtual void copyImage(GLenum target, GLint level, GLenum format, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) = 0; + virtual void copySubImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) = 0; + virtual void storage(GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth) = 0; + + virtual void generateMipmaps() = 0; + + virtual void bindTexImage(egl::Surface *surface) = 0; + virtual void releaseTexImage() = 0; +}; + +} + +#endif // LIBGLESV2_RENDERER_TEXTUREIMPL_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/TextureStorage.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/TextureStorage.cpp deleted file mode 100644 index 00b316f1cc..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/TextureStorage.cpp +++ /dev/null @@ -1,122 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// TextureStorage.cpp: Implements the abstract rx::TextureStorageInterface class and its concrete derived -// classes TextureStorageInterface2D and TextureStorageInterfaceCube, which act as the interface to the -// GPU-side texture. - -#include "libGLESv2/renderer/TextureStorage.h" -#include "libGLESv2/renderer/Renderer.h" -#include "libGLESv2/Renderbuffer.h" -#include "libGLESv2/Texture.h" - -#include "common/debug.h" - -namespace rx -{ -unsigned int TextureStorageInterface::mCurrentTextureSerial = 1; - -TextureStorageInterface::TextureStorageInterface() - : mTextureSerial(issueTextureSerial()), - mInstance(NULL) -{ -} - -TextureStorageInterface::~TextureStorageInterface() -{ - delete mInstance; -} - -bool TextureStorageInterface::isRenderTarget() const -{ - return mInstance->isRenderTarget(); -} - - -bool TextureStorageInterface::isManaged() const -{ - return mInstance->isManaged(); -} - -unsigned int TextureStorageInterface::getTextureSerial() const -{ - return mTextureSerial; -} - -unsigned int TextureStorageInterface::issueTextureSerial() -{ - return mCurrentTextureSerial++; -} - -int TextureStorageInterface::getLodOffset() const -{ - return mInstance->getLodOffset(); -} - - -int TextureStorageInterface::levelCount() -{ - return mInstance->levelCount(); -} - -TextureStorageInterface2D::TextureStorageInterface2D(Renderer *renderer, SwapChain *swapchain) - : mRenderTargetSerial(gl::RenderbufferStorage::issueSerial()) -{ - mInstance = renderer->createTextureStorage2D(swapchain); -} - -TextureStorageInterface2D::TextureStorageInterface2D(Renderer *renderer, int levels, GLenum internalformat, GLenum usage, bool forceRenderable, GLsizei width, GLsizei height) - : mRenderTargetSerial(gl::RenderbufferStorage::issueSerial()) -{ - mInstance = renderer->createTextureStorage2D(levels, internalformat, usage, forceRenderable, width, height); -} - -TextureStorageInterface2D::~TextureStorageInterface2D() -{ -} - -RenderTarget *TextureStorageInterface2D::getRenderTarget() const -{ - return mInstance->getRenderTarget(); -} - -void TextureStorageInterface2D::generateMipmap(int level) -{ - mInstance->generateMipmap(level); -} - -unsigned int TextureStorageInterface2D::getRenderTargetSerial(GLenum target) const -{ - return mRenderTargetSerial; -} - -TextureStorageInterfaceCube::TextureStorageInterfaceCube(Renderer *renderer, int levels, GLenum internalformat, GLenum usage, bool forceRenderable, int size) - : mFirstRenderTargetSerial(gl::RenderbufferStorage::issueCubeSerials()) -{ - mInstance = renderer->createTextureStorageCube(levels, internalformat, usage, forceRenderable, size); -} - -TextureStorageInterfaceCube::~TextureStorageInterfaceCube() -{ -} - -RenderTarget *TextureStorageInterfaceCube::getRenderTarget(GLenum faceTarget) const -{ - return mInstance->getRenderTarget(faceTarget); -} - -void TextureStorageInterfaceCube::generateMipmap(int face, int level) -{ - mInstance->generateMipmap(face, level); -} - -unsigned int TextureStorageInterfaceCube::getRenderTargetSerial(GLenum target) const -{ - return mFirstRenderTargetSerial + gl::TextureCubeMap::faceIndex(target); -} - -}
\ No newline at end of file diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/TextureStorage.h b/src/3rdparty/angle/src/libGLESv2/renderer/TextureStorage.h deleted file mode 100644 index edddb75f3f..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/TextureStorage.h +++ /dev/null @@ -1,110 +0,0 @@ -// -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// TextureStorage.h: Defines the abstract rx::TextureStorageInterface class and its concrete derived -// classes TextureStorageInterface2D and TextureStorageInterfaceCube, which act as the interface to the -// GPU-side texture. - -#ifndef LIBGLESV2_RENDERER_TEXTURESTORAGE_H_ -#define LIBGLESV2_RENDERER_TEXTURESTORAGE_H_ - -#include "common/debug.h" - -namespace rx -{ -class Renderer; -class SwapChain; -class RenderTarget; -class Blit; - -class TextureStorage -{ - public: - TextureStorage() {}; - virtual ~TextureStorage() {}; - - virtual int getLodOffset() const = 0; - virtual bool isRenderTarget() const = 0; - virtual bool isManaged() const = 0; - virtual int levelCount() = 0; - - virtual RenderTarget *getRenderTarget() = 0; - virtual RenderTarget *getRenderTarget(GLenum faceTarget) = 0; - virtual void generateMipmap(int level) = 0; - virtual void generateMipmap(int face, int level) = 0; - - private: - DISALLOW_COPY_AND_ASSIGN(TextureStorage); - -}; - -class TextureStorageInterface -{ - public: - TextureStorageInterface(); - virtual ~TextureStorageInterface(); - - TextureStorage *getStorageInstance() { return mInstance; } - - unsigned int getTextureSerial() const; - virtual unsigned int getRenderTargetSerial(GLenum target) const = 0; - - virtual int getLodOffset() const; - virtual bool isRenderTarget() const; - virtual bool isManaged() const; - virtual int levelCount(); - - protected: - TextureStorage *mInstance; - - private: - DISALLOW_COPY_AND_ASSIGN(TextureStorageInterface); - - const unsigned int mTextureSerial; - static unsigned int issueTextureSerial(); - - static unsigned int mCurrentTextureSerial; -}; - -class TextureStorageInterface2D : public TextureStorageInterface -{ - public: - TextureStorageInterface2D(Renderer *renderer, SwapChain *swapchain); - TextureStorageInterface2D(Renderer *renderer, int levels, GLenum internalformat, GLenum usage, bool forceRenderable, GLsizei width, GLsizei height); - virtual ~TextureStorageInterface2D(); - - void generateMipmap(int level); - RenderTarget *getRenderTarget() const; - - virtual unsigned int getRenderTargetSerial(GLenum target) const; - - private: - DISALLOW_COPY_AND_ASSIGN(TextureStorageInterface2D); - - const unsigned int mRenderTargetSerial; -}; - -class TextureStorageInterfaceCube : public TextureStorageInterface -{ - public: - TextureStorageInterfaceCube(Renderer *renderer, int levels, GLenum internalformat, GLenum usage, bool forceRenderable, int size); - virtual ~TextureStorageInterfaceCube(); - - void generateMipmap(int face, int level); - RenderTarget *getRenderTarget(GLenum faceTarget) const; - - virtual unsigned int getRenderTargetSerial(GLenum target) const; - - private: - DISALLOW_COPY_AND_ASSIGN(TextureStorageInterfaceCube); - - const unsigned int mFirstRenderTargetSerial; -}; - -} - -#endif // LIBGLESV2_RENDERER_TEXTURESTORAGE_H_ - diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/TransformFeedbackImpl.h b/src/3rdparty/angle/src/libGLESv2/renderer/TransformFeedbackImpl.h new file mode 100644 index 0000000000..8425604d87 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/TransformFeedbackImpl.h @@ -0,0 +1,31 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// TransformFeedbackImpl.h: Defines the abstract rx::TransformFeedbackImpl class. + +#ifndef LIBGLESV2_RENDERER_TRANSFORMFEEDBACKIMPL_H_ +#define LIBGLESV2_RENDERER_TRANSFORMFEEDBACKIMPL_H_ + +#include "common/angleutils.h" +#include "libGLESv2/TransformFeedback.h" + +namespace rx +{ + +class TransformFeedbackImpl +{ + public: + virtual ~TransformFeedbackImpl() { } + + virtual void begin(GLenum primitiveMode) = 0; + virtual void end() = 0; + virtual void pause() = 0; + virtual void resume() = 0; +}; + +} + +#endif // LIBGLESV2_RENDERER_TRANSFORMFEEDBACKIMPL_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/VertexArrayImpl.h b/src/3rdparty/angle/src/libGLESv2/renderer/VertexArrayImpl.h new file mode 100644 index 0000000000..b013f9cdf4 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/VertexArrayImpl.h @@ -0,0 +1,32 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// VertexAttribImpl.h: Defines the abstract rx::VertexAttribImpl class. + +#ifndef LIBGLESV2_RENDERER_VERTEXARRAYIMPL_H_ +#define LIBGLESV2_RENDERER_VERTEXARRAYIMPL_H_ + +#include "common/angleutils.h" +#include "libGLESv2/Buffer.h" +#include "libGLESv2/VertexAttribute.h" + +namespace rx +{ + +class VertexArrayImpl +{ + public: + virtual ~VertexArrayImpl() { } + + virtual void setElementArrayBuffer(const gl::Buffer *buffer) = 0; + virtual void setAttribute(size_t idx, const gl::VertexAttribute &attr) = 0; + virtual void setAttributeDivisor(size_t idx, GLuint divisor) = 0; + virtual void enableAttribute(size_t idx, bool enabledState) = 0; +}; + +} + +#endif // LIBGLESV2_RENDERER_VERTEXARRAYIMPL_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/VertexBuffer.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/VertexBuffer.cpp deleted file mode 100644 index a073d95033..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/VertexBuffer.cpp +++ /dev/null @@ -1,292 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// VertexBuffer.cpp: Defines the abstract VertexBuffer class and VertexBufferInterface -// class with derivations, classes that perform graphics API agnostic vertex buffer operations. - -#include "libGLESv2/renderer/VertexBuffer.h" -#include "libGLESv2/renderer/Renderer.h" -#include "libGLESv2/Context.h" - -namespace rx -{ - -unsigned int VertexBuffer::mNextSerial = 1; - -VertexBuffer::VertexBuffer() -{ - updateSerial(); -} - -VertexBuffer::~VertexBuffer() -{ -} - -void VertexBuffer::updateSerial() -{ - mSerial = mNextSerial++; -} - -unsigned int VertexBuffer::getSerial() const -{ - return mSerial; -} - -VertexBufferInterface::VertexBufferInterface(rx::Renderer *renderer, bool dynamic) : mRenderer(renderer) -{ - mDynamic = dynamic; - mWritePosition = 0; - mReservedSpace = 0; - - mVertexBuffer = renderer->createVertexBuffer(); -} - -VertexBufferInterface::~VertexBufferInterface() -{ - delete mVertexBuffer; -} - -unsigned int VertexBufferInterface::getSerial() const -{ - return mVertexBuffer->getSerial(); -} - -unsigned int VertexBufferInterface::getBufferSize() const -{ - return mVertexBuffer->getBufferSize(); -} - -bool VertexBufferInterface::setBufferSize(unsigned int size) -{ - if (mVertexBuffer->getBufferSize() == 0) - { - return mVertexBuffer->initialize(size, mDynamic); - } - else - { - return mVertexBuffer->setBufferSize(size); - } -} - -unsigned int VertexBufferInterface::getWritePosition() const -{ - return mWritePosition; -} - -void VertexBufferInterface::setWritePosition(unsigned int writePosition) -{ - mWritePosition = writePosition; -} - -bool VertexBufferInterface::discard() -{ - return mVertexBuffer->discard(); -} - -bool VertexBufferInterface::storeVertexAttributes(const gl::VertexAttribute &attrib, GLint start, GLsizei count, GLsizei instances, - unsigned int *outStreamOffset) -{ - unsigned int spaceRequired; - if (!mVertexBuffer->getSpaceRequired(attrib, count, instances, &spaceRequired)) - { - return false; - } - - if (mWritePosition + spaceRequired < mWritePosition) - { - return false; - } - - if (!reserveSpace(mReservedSpace)) - { - return false; - } - mReservedSpace = 0; - - if (!mVertexBuffer->storeVertexAttributes(attrib, start, count, instances, mWritePosition)) - { - return false; - } - - if (outStreamOffset) - { - *outStreamOffset = mWritePosition; - } - - mWritePosition += spaceRequired; - - return true; -} - -bool VertexBufferInterface::storeRawData(const void* data, unsigned int size, unsigned int *outStreamOffset) -{ - if (mWritePosition + size < mWritePosition) - { - return false; - } - - if (!reserveSpace(mReservedSpace)) - { - return false; - } - mReservedSpace = 0; - - if (!mVertexBuffer->storeRawData(data, size, mWritePosition)) - { - return false; - } - - if (outStreamOffset) - { - *outStreamOffset = mWritePosition; - } - - mWritePosition += size; - - return true; -} - -bool VertexBufferInterface::reserveVertexSpace(const gl::VertexAttribute &attribute, GLsizei count, GLsizei instances) -{ - unsigned int requiredSpace; - if (!mVertexBuffer->getSpaceRequired(attribute, count, instances, &requiredSpace)) - { - return false; - } - - // Protect against integer overflow - if (mReservedSpace + requiredSpace < mReservedSpace) - { - return false; - } - - mReservedSpace += requiredSpace; - return true; -} - -bool VertexBufferInterface::reserveRawDataSpace(unsigned int size) -{ - // Protect against integer overflow - if (mReservedSpace + size < mReservedSpace) - { - return false; - } - - mReservedSpace += size; - return true; -} - -VertexBuffer* VertexBufferInterface::getVertexBuffer() const -{ - return mVertexBuffer; -} - - -StreamingVertexBufferInterface::StreamingVertexBufferInterface(rx::Renderer *renderer, std::size_t initialSize) : VertexBufferInterface(renderer, true) -{ - setBufferSize(initialSize); -} - -StreamingVertexBufferInterface::~StreamingVertexBufferInterface() -{ -} - -bool StreamingVertexBufferInterface::reserveSpace(unsigned int size) -{ - bool result = true; - unsigned int curBufferSize = getBufferSize(); - if (size > curBufferSize) - { - result = setBufferSize(std::max(size, 3 * curBufferSize / 2)); - setWritePosition(0); - } - else if (getWritePosition() + size > curBufferSize) - { - if (!discard()) - { - return false; - } - setWritePosition(0); - } - - return result; -} - -StaticVertexBufferInterface::StaticVertexBufferInterface(rx::Renderer *renderer) : VertexBufferInterface(renderer, false) -{ -} - -StaticVertexBufferInterface::~StaticVertexBufferInterface() -{ -} - -bool StaticVertexBufferInterface::lookupAttribute(const gl::VertexAttribute &attribute, unsigned int *outStreamOffset) -{ - for (unsigned int element = 0; element < mCache.size(); element++) - { - if (mCache[element].type == attribute.mType && - mCache[element].size == attribute.mSize && - mCache[element].stride == attribute.stride() && - mCache[element].normalized == attribute.mNormalized) - { - if (mCache[element].attributeOffset == attribute.mOffset % attribute.stride()) - { - if (outStreamOffset) - { - *outStreamOffset = mCache[element].streamOffset; - } - return true; - } - } - } - - return false; -} - -bool StaticVertexBufferInterface::reserveSpace(unsigned int size) -{ - unsigned int curSize = getBufferSize(); - if (curSize == 0) - { - setBufferSize(size); - return true; - } - else if (curSize >= size) - { - return true; - } - else - { - UNREACHABLE(); // Static vertex buffers can't be resized - return false; - } -} - -bool StaticVertexBufferInterface::storeVertexAttributes(const gl::VertexAttribute &attrib, GLint start, GLsizei count, GLsizei instances, - unsigned int *outStreamOffset) -{ - unsigned int streamOffset; - if (VertexBufferInterface::storeVertexAttributes(attrib, start, count, instances, &streamOffset)) - { - int attributeOffset = attrib.mOffset % attrib.stride(); - VertexElement element = { attrib.mType, attrib.mSize, attrib.stride(), attrib.mNormalized, attributeOffset, streamOffset }; - mCache.push_back(element); - - if (outStreamOffset) - { - *outStreamOffset = streamOffset; - } - - return true; - } - else - { - return false; - } -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/VertexDataManager.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/VertexDataManager.cpp deleted file mode 100644 index 8034aed8c9..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/VertexDataManager.cpp +++ /dev/null @@ -1,304 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// VertexDataManager.h: Defines the VertexDataManager, a class that -// runs the Buffer translation process. - -#include "libGLESv2/renderer/VertexDataManager.h" -#include "libGLESv2/renderer/BufferStorage.h" - -#include "libGLESv2/Buffer.h" -#include "libGLESv2/ProgramBinary.h" -#include "libGLESv2/Context.h" -#include "libGLESv2/renderer/VertexBuffer.h" - -namespace -{ - enum { INITIAL_STREAM_BUFFER_SIZE = 1024*1024 }; - // This has to be at least 4k or else it fails on ATI cards. - enum { CONSTANT_VERTEX_BUFFER_SIZE = 4096 }; -} - -namespace rx -{ - -static int elementsInBuffer(const gl::VertexAttribute &attribute, unsigned int size) -{ - // Size cannot be larger than a GLsizei - if (size > static_cast<unsigned int>(std::numeric_limits<int>::max())) - { - size = static_cast<unsigned int>(std::numeric_limits<int>::max()); - } - - GLsizei stride = attribute.stride(); - return (size - attribute.mOffset % stride + (stride - attribute.typeSize())) / stride; -} - -static int StreamingBufferElementCount(const gl::VertexAttribute &attribute, int vertexDrawCount, int instanceDrawCount) -{ - // For instanced rendering, we draw "instanceDrawCount" sets of "vertexDrawCount" vertices. - // - // A vertex attribute with a positive divisor loads one instanced vertex for every set of - // non-instanced vertices, and the instanced vertex index advances once every "mDivisor" instances. - if (instanceDrawCount > 0 && attribute.mDivisor > 0) - { - return instanceDrawCount / attribute.mDivisor; - } - - return vertexDrawCount; -} - -VertexDataManager::VertexDataManager(Renderer *renderer) : mRenderer(renderer) -{ - for (int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) - { - mCurrentValue[i][0] = std::numeric_limits<float>::quiet_NaN(); - mCurrentValue[i][1] = std::numeric_limits<float>::quiet_NaN(); - mCurrentValue[i][2] = std::numeric_limits<float>::quiet_NaN(); - mCurrentValue[i][3] = std::numeric_limits<float>::quiet_NaN(); - mCurrentValueBuffer[i] = NULL; - mCurrentValueOffsets[i] = 0; - } - - mStreamingBuffer = new StreamingVertexBufferInterface(renderer, INITIAL_STREAM_BUFFER_SIZE); - - if (!mStreamingBuffer) - { - ERR("Failed to allocate the streaming vertex buffer."); - } -} - -VertexDataManager::~VertexDataManager() -{ - delete mStreamingBuffer; - - for (int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) - { - delete mCurrentValueBuffer[i]; - } -} - -static bool directStoragePossible(VertexBufferInterface* vb, const gl::VertexAttribute& attrib) -{ - gl::Buffer *buffer = attrib.mBoundBuffer.get(); - BufferStorage *storage = buffer ? buffer->getStorage() : NULL; - - const bool isAligned = (attrib.stride() % 4 == 0) && (attrib.mOffset % 4 == 0); - - return storage && storage->supportsDirectBinding() && !vb->getVertexBuffer()->requiresConversion(attrib) && isAligned; -} - -GLenum VertexDataManager::prepareVertexData(const gl::VertexAttribute attribs[], gl::ProgramBinary *programBinary, GLint start, GLsizei count, TranslatedAttribute *translated, GLsizei instances) -{ - if (!mStreamingBuffer) - { - return GL_OUT_OF_MEMORY; - } - - for (int attributeIndex = 0; attributeIndex < gl::MAX_VERTEX_ATTRIBS; attributeIndex++) - { - translated[attributeIndex].active = (programBinary->getSemanticIndex(attributeIndex) != -1); - } - - // Invalidate static buffers that don't contain matching attributes - for (int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) - { - if (translated[i].active && attribs[i].mArrayEnabled) - { - gl::Buffer *buffer = attribs[i].mBoundBuffer.get(); - StaticVertexBufferInterface *staticBuffer = buffer ? buffer->getStaticVertexBuffer() : NULL; - - if (staticBuffer && staticBuffer->getBufferSize() > 0 && !staticBuffer->lookupAttribute(attribs[i], NULL) && - !directStoragePossible(staticBuffer, attribs[i])) - { - buffer->invalidateStaticData(); - } - } - } - - // Reserve the required space in the buffers - for (int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) - { - if (translated[i].active && attribs[i].mArrayEnabled) - { - gl::Buffer *buffer = attribs[i].mBoundBuffer.get(); - StaticVertexBufferInterface *staticBuffer = buffer ? buffer->getStaticVertexBuffer() : NULL; - VertexBufferInterface *vertexBuffer = staticBuffer ? staticBuffer : static_cast<VertexBufferInterface*>(mStreamingBuffer); - - if (!directStoragePossible(vertexBuffer, attribs[i])) - { - if (staticBuffer) - { - if (staticBuffer->getBufferSize() == 0) - { - int totalCount = elementsInBuffer(attribs[i], buffer->size()); - if (!staticBuffer->reserveVertexSpace(attribs[i], totalCount, 0)) - { - return GL_OUT_OF_MEMORY; - } - } - } - else - { - int totalCount = StreamingBufferElementCount(attribs[i], count, instances); - - // Undefined behaviour: - // We can return INVALID_OPERATION if our vertex attribute does not have enough backing data. - if (buffer && elementsInBuffer(attribs[i], buffer->size()) < totalCount) - { - return GL_INVALID_OPERATION; - } - - if (!mStreamingBuffer->reserveVertexSpace(attribs[i], totalCount, instances)) - { - return GL_OUT_OF_MEMORY; - } - } - } - } - } - - // Perform the vertex data translations - for (int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) - { - if (translated[i].active) - { - if (attribs[i].mArrayEnabled) - { - gl::Buffer *buffer = attribs[i].mBoundBuffer.get(); - - if (!buffer && attribs[i].mPointer == NULL) - { - // This is an application error that would normally result in a crash, but we catch it and return an error - ERR("An enabled vertex array has no buffer and no pointer."); - return GL_INVALID_OPERATION; - } - - StaticVertexBufferInterface *staticBuffer = buffer ? buffer->getStaticVertexBuffer() : NULL; - VertexBufferInterface *vertexBuffer = staticBuffer ? staticBuffer : static_cast<VertexBufferInterface*>(mStreamingBuffer); - - BufferStorage *storage = buffer ? buffer->getStorage() : NULL; - bool directStorage = directStoragePossible(vertexBuffer, attribs[i]); - - unsigned int streamOffset = 0; - unsigned int outputElementSize = 0; - - if (directStorage) - { - outputElementSize = attribs[i].stride(); - streamOffset = attribs[i].mOffset + outputElementSize * start; - storage->markBufferUsage(); - } - else if (staticBuffer) - { - if (!staticBuffer->getVertexBuffer()->getSpaceRequired(attribs[i], 1, 0, &outputElementSize)) - { - return GL_OUT_OF_MEMORY; - } - - if (!staticBuffer->lookupAttribute(attribs[i], &streamOffset)) - { - // Convert the entire buffer - int totalCount = elementsInBuffer(attribs[i], storage->getSize()); - int startIndex = attribs[i].mOffset / attribs[i].stride(); - - if (!staticBuffer->storeVertexAttributes(attribs[i], -startIndex, totalCount, 0, &streamOffset)) - { - return GL_OUT_OF_MEMORY; - } - } - - unsigned int firstElementOffset = (attribs[i].mOffset / attribs[i].stride()) * outputElementSize; - unsigned int startOffset = (instances == 0 || attribs[i].mDivisor == 0) ? start * outputElementSize : 0; - if (streamOffset + firstElementOffset + startOffset < streamOffset) - { - return GL_OUT_OF_MEMORY; - } - - streamOffset += firstElementOffset + startOffset; - } - else - { - int totalCount = StreamingBufferElementCount(attribs[i], count, instances); - if (!mStreamingBuffer->getVertexBuffer()->getSpaceRequired(attribs[i], 1, 0, &outputElementSize) || - !mStreamingBuffer->storeVertexAttributes(attribs[i], start, totalCount, instances, &streamOffset)) - { - return GL_OUT_OF_MEMORY; - } - } - - translated[i].storage = directStorage ? storage : NULL; - translated[i].vertexBuffer = vertexBuffer->getVertexBuffer(); - translated[i].serial = directStorage ? storage->getSerial() : vertexBuffer->getSerial(); - translated[i].divisor = attribs[i].mDivisor; - - translated[i].attribute = &attribs[i]; - translated[i].stride = outputElementSize; - translated[i].offset = streamOffset; - } - else - { - if (!mCurrentValueBuffer[i]) - { - mCurrentValueBuffer[i] = new StreamingVertexBufferInterface(mRenderer, CONSTANT_VERTEX_BUFFER_SIZE); - } - - StreamingVertexBufferInterface *buffer = mCurrentValueBuffer[i]; - - if (mCurrentValue[i][0] != attribs[i].mCurrentValue[0] || - mCurrentValue[i][1] != attribs[i].mCurrentValue[1] || - mCurrentValue[i][2] != attribs[i].mCurrentValue[2] || - mCurrentValue[i][3] != attribs[i].mCurrentValue[3]) - { - unsigned int requiredSpace = sizeof(float) * 4; - if (!buffer->reserveRawDataSpace(requiredSpace)) - { - return GL_OUT_OF_MEMORY; - } - - unsigned int streamOffset; - if (!buffer->storeRawData(attribs[i].mCurrentValue, requiredSpace, &streamOffset)) - { - return GL_OUT_OF_MEMORY; - } - - mCurrentValue[i][0] = attribs[i].mCurrentValue[0]; - mCurrentValue[i][1] = attribs[i].mCurrentValue[1]; - mCurrentValue[i][2] = attribs[i].mCurrentValue[2]; - mCurrentValue[i][3] = attribs[i].mCurrentValue[3]; - mCurrentValueOffsets[i] = streamOffset; - } - - translated[i].storage = NULL; - translated[i].vertexBuffer = mCurrentValueBuffer[i]->getVertexBuffer(); - translated[i].serial = mCurrentValueBuffer[i]->getSerial(); - translated[i].divisor = 0; - - translated[i].attribute = &attribs[i]; - translated[i].stride = 0; - translated[i].offset = mCurrentValueOffsets[i]; - } - } - } - - for (int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) - { - if (translated[i].active && attribs[i].mArrayEnabled) - { - gl::Buffer *buffer = attribs[i].mBoundBuffer.get(); - - if (buffer) - { - buffer->promoteStaticUsage(count * attribs[i].typeSize()); - } - } - } - - return GL_NO_ERROR; -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/VertexDataManager.h b/src/3rdparty/angle/src/libGLESv2/renderer/VertexDataManager.h deleted file mode 100644 index 1a8786552a..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/VertexDataManager.h +++ /dev/null @@ -1,65 +0,0 @@ -// -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// VertexDataManager.h: Defines the VertexDataManager, a class that -// runs the Buffer translation process. - -#ifndef LIBGLESV2_RENDERER_VERTEXDATAMANAGER_H_ -#define LIBGLESV2_RENDERER_VERTEXDATAMANAGER_H_ - -#include "libGLESv2/Constants.h" -#include "common/angleutils.h" - -namespace gl -{ -class VertexAttribute; -class ProgramBinary; -} - -namespace rx -{ -class BufferStorage; -class StreamingVertexBufferInterface; -class VertexBuffer; -class Renderer; - -struct TranslatedAttribute -{ - bool active; - - const gl::VertexAttribute *attribute; - unsigned int offset; - unsigned int stride; // 0 means not to advance the read pointer at all - - VertexBuffer *vertexBuffer; - BufferStorage *storage; - unsigned int serial; - unsigned int divisor; -}; - -class VertexDataManager -{ - public: - VertexDataManager(rx::Renderer *renderer); - virtual ~VertexDataManager(); - - GLenum prepareVertexData(const gl::VertexAttribute attribs[], gl::ProgramBinary *programBinary, GLint start, GLsizei count, TranslatedAttribute *outAttribs, GLsizei instances); - - private: - DISALLOW_COPY_AND_ASSIGN(VertexDataManager); - - rx::Renderer *const mRenderer; - - StreamingVertexBufferInterface *mStreamingBuffer; - - float mCurrentValue[gl::MAX_VERTEX_ATTRIBS][4]; - StreamingVertexBufferInterface *mCurrentValueBuffer[gl::MAX_VERTEX_ATTRIBS]; - std::size_t mCurrentValueOffsets[gl::MAX_VERTEX_ATTRIBS]; -}; - -} - -#endif // LIBGLESV2_RENDERER_VERTEXDATAMANAGER_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/copyimage.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/copyimage.cpp new file mode 100644 index 0000000000..004223d70f --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/copyimage.cpp @@ -0,0 +1,22 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// copyimage.cpp: Defines image copying functions + +#include "libGLESv2/renderer/copyImage.h" + +namespace rx +{ + +void CopyBGRA8ToRGBA8(const uint8_t *source, uint8_t *dest) +{ + uint32_t argb = *reinterpret_cast<const uint32_t*>(source); + *reinterpret_cast<uint32_t*>(dest) = (argb & 0xFF00FF00) | // Keep alpha and green + (argb & 0x00FF0000) >> 16 | // Move red to blue + (argb & 0x000000FF) << 16; // Move blue to red +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/copyimage.h b/src/3rdparty/angle/src/libGLESv2/renderer/copyimage.h new file mode 100644 index 0000000000..513eb5cb3d --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/copyimage.h @@ -0,0 +1,35 @@ +// +// Copyright (c) 2013-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// copyimage.h: Defines image copying functions + +#ifndef LIBGLESV2_RENDERER_COPYIMAGE_H_ +#define LIBGLESV2_RENDERER_COPYIMAGE_H_ + +#include "common/mathutil.h" +#include "libGLESv2/angletypes.h" + +#include <cstdint> + +namespace rx +{ + +template <typename sourceType, typename colorDataType> +void ReadColor(const uint8_t *source, uint8_t *dest); + +template <typename destType, typename colorDataType> +void WriteColor(const uint8_t *source, uint8_t *dest); + +template <typename sourceType, typename destType, typename colorDataType> +void CopyPixel(const uint8_t *source, uint8_t *dest); + +void CopyBGRA8ToRGBA8(const uint8_t *source, uint8_t *dest); + +} + +#include "copyimage.inl" + +#endif // LIBGLESV2_RENDERER_COPYIMAGE_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/copyimage.inl b/src/3rdparty/angle/src/libGLESv2/renderer/copyimage.inl new file mode 100644 index 0000000000..0498cf7750 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/copyimage.inl @@ -0,0 +1,32 @@ +// +// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// copyimage.inl: Defines image copying functions + +namespace rx +{ + +template <typename sourceType, typename colorDataType> +inline void ReadColor(const uint8_t *source, uint8_t *dest) +{ + sourceType::readColor(reinterpret_cast<gl::Color<colorDataType>*>(dest), reinterpret_cast<const sourceType*>(source)); +} + +template <typename destType, typename colorDataType> +inline void WriteColor(const uint8_t *source, uint8_t *dest) +{ + destType::writeColor(reinterpret_cast<destType*>(dest), reinterpret_cast<const gl::Color<colorDataType>*>(source)); +} + +template <typename sourceType, typename destType, typename colorDataType> +inline void CopyPixel(const uint8_t *source, uint8_t *dest) +{ + colorDataType temp; + ReadColor<sourceType, colorDataType>(source, &temp); + WriteColor<destType, colorDataType>(&temp, dest); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/copyvertex.h b/src/3rdparty/angle/src/libGLESv2/renderer/copyvertex.h new file mode 100644 index 0000000000..e0e8af166b --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/copyvertex.h @@ -0,0 +1,35 @@ +// +// Copyright (c) 2013-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// copyvertex.h: Defines vertex buffer copying and conversion functions + +#ifndef LIBGLESV2_RENDERER_COPYVERTEX_H_ +#define LIBGLESV2_RENDERER_COPYVERTEX_H_ + +#include "common/mathutil.h" + +namespace rx +{ + +// 'widenDefaultValueBits' gives the default value for the alpha channel (4th component) +// the sentinel value 0 means we do not want to widen the input or add an alpha channel +template <typename T, size_t componentCount, uint32_t widenDefaultValueBits> +inline void CopyNativeVertexData(const uint8_t *input, size_t stride, size_t count, uint8_t *output); + +template <size_t componentCount> +inline void Copy32FixedTo32FVertexData(const uint8_t *input, size_t stride, size_t count, uint8_t *output); + +template <typename T, size_t componentCount, bool normalized> +inline void CopyTo32FVertexData(const uint8_t *input, size_t stride, size_t count, uint8_t *output); + +template <bool isSigned, bool normalized, bool toFloat> +inline void CopyXYZ10W2ToXYZW32FVertexData(const uint8_t *input, size_t stride, size_t count, uint8_t *output); + +} + +#include "copyvertex.inl" + +#endif // LIBGLESV2_RENDERER_COPYVERTEX_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/copyvertex.inl b/src/3rdparty/angle/src/libGLESv2/renderer/copyvertex.inl new file mode 100644 index 0000000000..7eef17b22b --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/copyvertex.inl @@ -0,0 +1,288 @@ +// +// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +namespace rx +{ + +template <typename T, size_t componentCount, uint32_t widenDefaultValueBits> +inline void CopyNativeVertexData(const uint8_t *input, size_t stride, size_t count, uint8_t *output) +{ + const size_t attribSize = sizeof(T)* componentCount; + const T defaultValue = gl::bitCast<T>(widenDefaultValueBits); + const bool widen = (widenDefaultValueBits != 0); + + if (attribSize == stride && !widen) + { + memcpy(output, input, count * attribSize); + } + else + { + size_t outputStride = widen ? 4 : componentCount; + + for (size_t i = 0; i < count; i++) + { + const T *offsetInput = reinterpret_cast<const T*>(input + (i * stride)); + T *offsetOutput = reinterpret_cast<T*>(output) + i * outputStride; + + for (size_t j = 0; j < componentCount; j++) + { + offsetOutput[j] = offsetInput[j]; + } + + if (widen) + { + offsetOutput[3] = defaultValue; + } + } + } +} + +template <size_t componentCount> +inline void Copy32FixedTo32FVertexData(const uint8_t *input, size_t stride, size_t count, uint8_t *output) +{ + static const float divisor = 1.0f / (1 << 16); + + for (size_t i = 0; i < count; i++) + { + const GLfixed* offsetInput = reinterpret_cast<const GLfixed*>(input + (stride * i)); + float* offsetOutput = reinterpret_cast<float*>(output) + i * componentCount; + + for (size_t j = 0; j < componentCount; j++) + { + offsetOutput[j] = static_cast<float>(offsetInput[j]) * divisor; + } + } +} + +template <typename T, size_t componentCount, bool normalized> +inline void CopyTo32FVertexData(const uint8_t *input, size_t stride, size_t count, uint8_t *output) +{ + typedef std::numeric_limits<T> NL; + + for (size_t i = 0; i < count; i++) + { + const T *offsetInput = reinterpret_cast<const T*>(input + (stride * i)); + float *offsetOutput = reinterpret_cast<float*>(output) + i * componentCount; + + for (size_t j = 0; j < componentCount; j++) + { + if (normalized) + { + if (NL::is_signed) + { + const float divisor = 1.0f / (2 * static_cast<float>(NL::max()) + 1); + offsetOutput[j] = (2 * static_cast<float>(offsetInput[j]) + 1) * divisor; + } + else + { + offsetOutput[j] = static_cast<float>(offsetInput[j]) / NL::max(); + } + } + else + { + offsetOutput[j] = static_cast<float>(offsetInput[j]); + } + } + } +} + +namespace priv +{ + +template <bool isSigned, bool normalized, bool toFloat> +static inline void CopyPackedRGB(uint32_t data, uint8_t *output) +{ + const uint32_t rgbSignMask = 0x200; // 1 set at the 9 bit + const uint32_t negativeMask = 0xFFFFFC00; // All bits from 10 to 31 set to 1 + + if (toFloat) + { + GLfloat *floatOutput = reinterpret_cast<GLfloat*>(output); + if (isSigned) + { + GLfloat finalValue = 0; + if (data & rgbSignMask) + { + int negativeNumber = data | negativeMask; + finalValue = static_cast<GLfloat>(negativeNumber); + } + else + { + finalValue = static_cast<GLfloat>(data); + } + + if (normalized) + { + const int32_t maxValue = 0x1FF; // 1 set in bits 0 through 8 + const int32_t minValue = 0xFFFFFE01; // Inverse of maxValue + + // A 10-bit two's complement number has the possibility of being minValue - 1 but + // OpenGL's normalization rules dictate that it should be clamped to minValue in this + // case. + if (finalValue < minValue) + { + finalValue = minValue; + } + + const int32_t halfRange = (maxValue - minValue) >> 1; + *floatOutput = ((finalValue - minValue) / halfRange) - 1.0f; + } + else + { + *floatOutput = finalValue; + } + } + else + { + if (normalized) + { + const uint32_t maxValue = 0x3FF; // 1 set in bits 0 through 9 + *floatOutput = static_cast<GLfloat>(data) / static_cast<GLfloat>(maxValue); + } + else + { + *floatOutput = static_cast<GLfloat>(data); + } + } + } + else + { + if (isSigned) + { + GLshort *intOutput = reinterpret_cast<GLshort*>(output); + + if (data & rgbSignMask) + { + *intOutput = data | negativeMask; + } + else + { + *intOutput = data; + } + } + else + { + GLushort *uintOutput = reinterpret_cast<GLushort*>(output); + *uintOutput = data; + } + } +} + +template <bool isSigned, bool normalized, bool toFloat> +inline void CopyPackedAlpha(uint32_t data, uint8_t *output) +{ + if (toFloat) + { + GLfloat *floatOutput = reinterpret_cast<GLfloat*>(output); + if (isSigned) + { + if (normalized) + { + switch (data) + { + case 0x0: *floatOutput = 0.0f; break; + case 0x1: *floatOutput = 1.0f; break; + case 0x2: *floatOutput = -1.0f; break; + case 0x3: *floatOutput = -1.0f; break; + default: UNREACHABLE(); + } + } + else + { + switch (data) + { + case 0x0: *floatOutput = 0.0f; break; + case 0x1: *floatOutput = 1.0f; break; + case 0x2: *floatOutput = -2.0f; break; + case 0x3: *floatOutput = -1.0f; break; + default: UNREACHABLE(); + } + } + } + else + { + if (normalized) + { + switch (data) + { + case 0x0: *floatOutput = 0.0f / 3.0f; break; + case 0x1: *floatOutput = 1.0f / 3.0f; break; + case 0x2: *floatOutput = 2.0f / 3.0f; break; + case 0x3: *floatOutput = 3.0f / 3.0f; break; + default: UNREACHABLE(); + } + } + else + { + switch (data) + { + case 0x0: *floatOutput = 0.0f; break; + case 0x1: *floatOutput = 1.0f; break; + case 0x2: *floatOutput = 2.0f; break; + case 0x3: *floatOutput = 3.0f; break; + default: UNREACHABLE(); + } + } + } + } + else + { + if (isSigned) + { + GLshort *intOutput = reinterpret_cast<GLshort*>(output); + switch (data) + { + case 0x0: *intOutput = 0; break; + case 0x1: *intOutput = 1; break; + case 0x2: *intOutput = -2; break; + case 0x3: *intOutput = -1; break; + default: UNREACHABLE(); + } + } + else + { + GLushort *uintOutput = reinterpret_cast<GLushort*>(output); + switch (data) + { + case 0x0: *uintOutput = 0; break; + case 0x1: *uintOutput = 1; break; + case 0x2: *uintOutput = 2; break; + case 0x3: *uintOutput = 3; break; + default: UNREACHABLE(); + } + } + } +} + +} + +template <bool isSigned, bool normalized, bool toFloat> +inline void CopyXYZ10W2ToXYZW32FVertexData(const uint8_t *input, size_t stride, size_t count, uint8_t *output) +{ + const size_t outputComponentSize = toFloat ? 4 : 2; + const size_t componentCount = 4; + + const uint32_t rgbMask = 0x3FF; // 1 set in bits 0 through 9 + const size_t redShift = 0; // red is bits 0 through 9 + const size_t greenShift = 10; // green is bits 10 through 19 + const size_t blueShift = 20; // blue is bits 20 through 29 + + const uint32_t alphaMask = 0x3; // 1 set in bits 0 and 1 + const size_t alphaShift = 30; // Alpha is the 30 and 31 bits + + for (size_t i = 0; i < count; i++) + { + GLuint packedValue = *reinterpret_cast<const GLuint*>(input + (i * stride)); + uint8_t *offsetOutput = output + (i * outputComponentSize * componentCount); + + priv::CopyPackedRGB<isSigned, normalized, toFloat>( (packedValue >> redShift) & rgbMask, offsetOutput + (0 * outputComponentSize)); + priv::CopyPackedRGB<isSigned, normalized, toFloat>( (packedValue >> greenShift) & rgbMask, offsetOutput + (1 * outputComponentSize)); + priv::CopyPackedRGB<isSigned, normalized, toFloat>( (packedValue >> blueShift) & rgbMask, offsetOutput + (2 * outputComponentSize)); + priv::CopyPackedAlpha<isSigned, normalized, toFloat>((packedValue >> alphaShift) & alphaMask, offsetOutput + (3 * outputComponentSize)); + } +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/BufferD3D.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/BufferD3D.cpp new file mode 100644 index 0000000000..a34ef03fb8 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/BufferD3D.cpp @@ -0,0 +1,82 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// BufferD3D.cpp Defines common functionality between the Buffer9 and Buffer11 classes. + +#include "libGLESv2/renderer/d3d/BufferD3D.h" +#include "libGLESv2/renderer/d3d/VertexBuffer.h" +#include "libGLESv2/renderer/d3d/IndexBuffer.h" +#include "libGLESv2/renderer/Renderer.h" +#include "libGLESv2/main.h" + +namespace rx +{ + +unsigned int BufferD3D::mNextSerial = 1; + +BufferD3D::BufferD3D() + : BufferImpl(), + mStaticVertexBuffer(NULL), + mStaticIndexBuffer(NULL) +{ + updateSerial(); +} + +BufferD3D::~BufferD3D() +{ + SafeDelete(mStaticVertexBuffer); + SafeDelete(mStaticIndexBuffer); +} + +BufferD3D *BufferD3D::makeBufferD3D(BufferImpl *buffer) +{ + ASSERT(HAS_DYNAMIC_TYPE(BufferD3D*, buffer)); + return static_cast<BufferD3D*>(buffer); +} + +void BufferD3D::updateSerial() +{ + mSerial = mNextSerial++; +} + +void BufferD3D::initializeStaticData() +{ + if (!mStaticVertexBuffer) + { + mStaticVertexBuffer = new rx::StaticVertexBufferInterface(getRenderer()); + } + if (!mStaticIndexBuffer) + { + mStaticIndexBuffer = new rx::StaticIndexBufferInterface(getRenderer()); + } +} + +void BufferD3D::invalidateStaticData() +{ + if ((mStaticVertexBuffer && mStaticVertexBuffer->getBufferSize() != 0) || (mStaticIndexBuffer && mStaticIndexBuffer->getBufferSize() != 0)) + { + SafeDelete(mStaticVertexBuffer); + SafeDelete(mStaticIndexBuffer); + } + + mUnmodifiedDataUse = 0; +} + +// Creates static buffers if sufficient used data has been left unmodified +void BufferD3D::promoteStaticUsage(int dataSize) +{ + if (!mStaticVertexBuffer && !mStaticIndexBuffer) + { + mUnmodifiedDataUse += dataSize; + + if (mUnmodifiedDataUse > 3 * getSize()) + { + initializeStaticData(); + } + } +} + +}
\ No newline at end of file diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/BufferD3D.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/BufferD3D.h new file mode 100644 index 0000000000..44f14cee58 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/BufferD3D.h @@ -0,0 +1,56 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// BufferImpl.h: Defines the abstract rx::BufferImpl class. + +#ifndef LIBGLESV2_RENDERER_BUFFERD3D_H_ +#define LIBGLESV2_RENDERER_BUFFERD3D_H_ + +#include "libGLESv2/renderer/BufferImpl.h" +#include "libGLESv2/angletypes.h" + +namespace rx +{ + +class Renderer; +class StaticIndexBufferInterface; +class StaticVertexBufferInterface; + +class BufferD3D : public BufferImpl +{ + public: + BufferD3D(); + virtual ~BufferD3D(); + + static BufferD3D *makeBufferD3D(BufferImpl *buffer); + + unsigned int getSerial() const { return mSerial; } + + virtual size_t getSize() const = 0; + virtual bool supportsDirectBinding() const = 0; + virtual Renderer* getRenderer() = 0; + + rx::StaticVertexBufferInterface *getStaticVertexBuffer() { return mStaticVertexBuffer; } + rx::StaticIndexBufferInterface *getStaticIndexBuffer() { return mStaticIndexBuffer; } + + void initializeStaticData(); + void invalidateStaticData(); + void promoteStaticUsage(int dataSize); + + protected: + unsigned int mSerial; + static unsigned int mNextSerial; + + void updateSerial(); + + rx::StaticVertexBufferInterface *mStaticVertexBuffer; + rx::StaticIndexBufferInterface *mStaticIndexBuffer; + unsigned int mUnmodifiedDataUse; +}; + +} + +#endif // LIBGLESV2_RENDERER_BUFFERIMPLD3D_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/DynamicHLSL.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/DynamicHLSL.cpp new file mode 100644 index 0000000000..13411ebe64 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/DynamicHLSL.cpp @@ -0,0 +1,1144 @@ +// +// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// +// DynamicHLSL.cpp: Implementation for link and run-time HLSL generation +// + +#include "libGLESv2/renderer/d3d/DynamicHLSL.h" +#include "libGLESv2/renderer/d3d/ShaderD3D.h" +#include "libGLESv2/renderer/Renderer.h" +#include "libGLESv2/Shader.h" +#include "libGLESv2/Program.h" +#include "libGLESv2/ProgramBinary.h" +#include "libGLESv2/formatutils.h" + +#include "common/utilities.h" +#include "common/blocklayout.h" + +// For use with ArrayString, see angleutils.h +META_ASSERT(GL_INVALID_INDEX == UINT_MAX); + +using namespace gl; + +namespace +{ + +std::string HLSLComponentTypeString(GLenum componentType) +{ + switch (componentType) + { + case GL_UNSIGNED_INT: return "uint"; + case GL_INT: return "int"; + case GL_UNSIGNED_NORMALIZED: + case GL_SIGNED_NORMALIZED: + case GL_FLOAT: return "float"; + default: UNREACHABLE(); return "not-component-type"; + } +} + +std::string HLSLComponentTypeString(GLenum componentType, int componentCount) +{ + return HLSLComponentTypeString(componentType) + (componentCount > 1 ? Str(componentCount) : ""); +} + +std::string HLSLMatrixTypeString(GLenum type) +{ + switch (type) + { + case GL_FLOAT_MAT2: return "float2x2"; + case GL_FLOAT_MAT3: return "float3x3"; + case GL_FLOAT_MAT4: return "float4x4"; + case GL_FLOAT_MAT2x3: return "float2x3"; + case GL_FLOAT_MAT3x2: return "float3x2"; + case GL_FLOAT_MAT2x4: return "float2x4"; + case GL_FLOAT_MAT4x2: return "float4x2"; + case GL_FLOAT_MAT3x4: return "float3x4"; + case GL_FLOAT_MAT4x3: return "float4x3"; + default: UNREACHABLE(); return "not-matrix-type"; + } +} + +std::string HLSLTypeString(GLenum type) +{ + if (gl::IsMatrixType(type)) + { + return HLSLMatrixTypeString(type); + } + + return HLSLComponentTypeString(gl::VariableComponentType(type), gl::VariableComponentCount(type)); +} + +const rx::PixelShaderOutputVariable &GetOutputAtLocation(const std::vector<rx::PixelShaderOutputVariable> &outputVariables, + unsigned int location) +{ + for (size_t variableIndex = 0; variableIndex < outputVariables.size(); ++variableIndex) + { + if (outputVariables[variableIndex].outputIndex == location) + { + return outputVariables[variableIndex]; + } + } + + UNREACHABLE(); + return outputVariables[0]; +} + +} + +namespace rx +{ + +const std::string VERTEX_ATTRIBUTE_STUB_STRING = "@@ VERTEX ATTRIBUTES @@"; +const std::string PIXEL_OUTPUT_STUB_STRING = "@@ PIXEL OUTPUT @@"; + +DynamicHLSL::DynamicHLSL(rx::Renderer *const renderer) + : mRenderer(renderer) +{ +} + +static bool packVarying(PackedVarying *varying, const int maxVaryingVectors, VaryingPacking packing) +{ + GLenum transposedType = TransposeMatrixType(varying->type); + + // matrices within varying structs are not transposed + int registers = (varying->isStruct() ? HLSLVariableRegisterCount(*varying) : VariableRowCount(transposedType)) * varying->elementCount(); + int elements = (varying->isStruct() ? 4 : VariableColumnCount(transposedType)); + + if (elements >= 2 && elements <= 4) + { + for (int r = 0; r <= maxVaryingVectors - registers; r++) + { + bool available = true; + + for (int y = 0; y < registers && available; y++) + { + for (int x = 0; x < elements && available; x++) + { + if (packing[r + y][x]) + { + available = false; + } + } + } + + if (available) + { + varying->registerIndex = r; + varying->columnIndex = 0; + + for (int y = 0; y < registers; y++) + { + for (int x = 0; x < elements; x++) + { + packing[r + y][x] = &*varying; + } + } + + return true; + } + } + + if (elements == 2) + { + for (int r = maxVaryingVectors - registers; r >= 0; r--) + { + bool available = true; + + for (int y = 0; y < registers && available; y++) + { + for (int x = 2; x < 4 && available; x++) + { + if (packing[r + y][x]) + { + available = false; + } + } + } + + if (available) + { + varying->registerIndex = r; + varying->columnIndex = 2; + + for (int y = 0; y < registers; y++) + { + for (int x = 2; x < 4; x++) + { + packing[r + y][x] = &*varying; + } + } + + return true; + } + } + } + } + else if (elements == 1) + { + int space[4] = { 0 }; + + for (int y = 0; y < maxVaryingVectors; y++) + { + for (int x = 0; x < 4; x++) + { + space[x] += packing[y][x] ? 0 : 1; + } + } + + int column = 0; + + for (int x = 0; x < 4; x++) + { + if (space[x] >= registers && (space[column] < registers || space[x] < space[column])) + { + column = x; + } + } + + if (space[column] >= registers) + { + for (int r = 0; r < maxVaryingVectors; r++) + { + if (!packing[r][column]) + { + varying->registerIndex = r; + varying->columnIndex = column; + + for (int y = r; y < r + registers; y++) + { + packing[y][column] = &*varying; + } + + break; + } + } + + return true; + } + } + else UNREACHABLE(); + + return false; +} + +// Packs varyings into generic varying registers, using the algorithm from [OpenGL ES Shading Language 1.00 rev. 17] appendix A section 7 page 111 +// Returns the number of used varying registers, or -1 if unsuccesful +int DynamicHLSL::packVaryings(InfoLog &infoLog, VaryingPacking packing, rx::ShaderD3D *fragmentShader, + rx::ShaderD3D *vertexShader, const std::vector<std::string>& transformFeedbackVaryings) +{ + // TODO (geofflang): Use context's caps + const int maxVaryingVectors = mRenderer->getRendererCaps().maxVaryingVectors; + + vertexShader->resetVaryingsRegisterAssignment(); + fragmentShader->resetVaryingsRegisterAssignment(); + + std::set<std::string> packedVaryings; + + std::vector<gl::PackedVarying> &fragmentVaryings = fragmentShader->getVaryings(); + std::vector<gl::PackedVarying> &vertexVaryings = vertexShader->getVaryings(); + for (unsigned int varyingIndex = 0; varyingIndex < fragmentVaryings.size(); varyingIndex++) + { + PackedVarying *varying = &fragmentVaryings[varyingIndex]; + + // Do not assign registers to built-in or unreferenced varyings + if (varying->isBuiltIn() || !varying->staticUse) + { + continue; + } + + if (packVarying(varying, maxVaryingVectors, packing)) + { + packedVaryings.insert(varying->name); + } + else + { + infoLog.append("Could not pack varying %s", varying->name.c_str()); + return -1; + } + } + + for (unsigned int feedbackVaryingIndex = 0; feedbackVaryingIndex < transformFeedbackVaryings.size(); feedbackVaryingIndex++) + { + const std::string &transformFeedbackVarying = transformFeedbackVaryings[feedbackVaryingIndex]; + if (packedVaryings.find(transformFeedbackVarying) == packedVaryings.end()) + { + bool found = false; + for (unsigned int varyingIndex = 0; varyingIndex < vertexVaryings.size(); varyingIndex++) + { + PackedVarying *varying = &vertexVaryings[varyingIndex]; + if (transformFeedbackVarying == varying->name) + { + if (!packVarying(varying, maxVaryingVectors, packing)) + { + infoLog.append("Could not pack varying %s", varying->name.c_str()); + return -1; + } + + found = true; + break; + } + } + + if (!found && transformFeedbackVarying != "gl_Position" && transformFeedbackVarying != "gl_PointSize") + { + infoLog.append("Transform feedback varying %s does not exist in the vertex shader.", transformFeedbackVarying.c_str()); + return -1; + } + } + } + + // Return the number of used registers + int registers = 0; + + for (int r = 0; r < maxVaryingVectors; r++) + { + if (packing[r][0] || packing[r][1] || packing[r][2] || packing[r][3]) + { + registers++; + } + } + + return registers; +} + +std::string DynamicHLSL::generateVaryingHLSL(const ShaderD3D *shader) const +{ + std::string varyingSemantic = getVaryingSemantic(shader->mUsesPointSize); + std::string varyingHLSL; + + const std::vector<gl::PackedVarying> &varyings = shader->getVaryings(); + + for (unsigned int varyingIndex = 0; varyingIndex < varyings.size(); varyingIndex++) + { + const PackedVarying &varying = varyings[varyingIndex]; + if (varying.registerAssigned()) + { + ASSERT(!varying.isBuiltIn()); + GLenum transposedType = TransposeMatrixType(varying.type); + int variableRows = (varying.isStruct() ? 1 : VariableRowCount(transposedType)); + + for (unsigned int elementIndex = 0; elementIndex < varying.elementCount(); elementIndex++) + { + for (int row = 0; row < variableRows; row++) + { + // TODO: Add checks to ensure D3D interpolation modifiers don't result in too many registers being used. + // For example, if there are N registers, and we have N vec3 varyings and 1 float varying, then D3D will pack them into N registers. + // If the float varying has the 'nointerpolation' modifier on it then we would need N + 1 registers, and D3D compilation will fail. + + switch (varying.interpolation) + { + case sh::INTERPOLATION_SMOOTH: varyingHLSL += " "; break; + case sh::INTERPOLATION_FLAT: varyingHLSL += " nointerpolation "; break; + case sh::INTERPOLATION_CENTROID: varyingHLSL += " centroid "; break; + default: UNREACHABLE(); + } + + unsigned int semanticIndex = elementIndex * variableRows + varying.columnIndex * mRenderer->getRendererCaps().maxVaryingVectors + varying.registerIndex + row; + std::string n = Str(semanticIndex); + + std::string typeString; + + if (varying.isStruct()) + { + // matrices within structs are not transposed, so + // do not use the special struct prefix "rm" + typeString = decorateVariable(varying.structName); + } + else + { + GLenum componentType = VariableComponentType(transposedType); + int columnCount = VariableColumnCount(transposedType); + typeString = HLSLComponentTypeString(componentType, columnCount); + } + varyingHLSL += typeString + " v" + n + " : " + varyingSemantic + n + ";\n"; + } + } + } + } + + return varyingHLSL; +} + +std::string DynamicHLSL::generateVertexShaderForInputLayout(const std::string &sourceShader, + const VertexFormat inputLayout[], + const sh::Attribute shaderAttributes[]) const +{ + std::string structHLSL, initHLSL; + + int semanticIndex = 0; + unsigned int inputIndex = 0; + + for (unsigned int attributeIndex = 0; attributeIndex < MAX_VERTEX_ATTRIBS; attributeIndex++) + { + const sh::Attribute &shaderAttribute = shaderAttributes[attributeIndex]; + if (!shaderAttribute.name.empty()) + { + ASSERT(inputIndex < MAX_VERTEX_ATTRIBS); + const VertexFormat &vertexFormat = inputLayout[inputIndex]; + + // HLSL code for input structure + if (IsMatrixType(shaderAttribute.type)) + { + // Matrix types are always transposed + structHLSL += " " + HLSLMatrixTypeString(TransposeMatrixType(shaderAttribute.type)); + } + else + { + GLenum componentType = mRenderer->getVertexComponentType(vertexFormat); + structHLSL += " " + HLSLComponentTypeString(componentType, VariableComponentCount(shaderAttribute.type)); + } + + structHLSL += " " + decorateVariable(shaderAttribute.name) + " : TEXCOORD" + Str(semanticIndex) + ";\n"; + semanticIndex += VariableRegisterCount(shaderAttribute.type); + + // HLSL code for initialization + initHLSL += " " + decorateVariable(shaderAttribute.name) + " = "; + + // Mismatched vertex attribute to vertex input may result in an undefined + // data reinterpretation (eg for pure integer->float, float->pure integer) + // TODO: issue warning with gl debug info extension, when supported + if (IsMatrixType(shaderAttribute.type) || + (mRenderer->getVertexConversionType(vertexFormat) & rx::VERTEX_CONVERT_GPU) != 0) + { + initHLSL += generateAttributeConversionHLSL(vertexFormat, shaderAttribute); + } + else + { + initHLSL += "input." + decorateVariable(shaderAttribute.name); + } + + initHLSL += ";\n"; + + inputIndex += VariableRowCount(TransposeMatrixType(shaderAttribute.type)); + } + } + + std::string replacementHLSL = "struct VS_INPUT\n" + "{\n" + + structHLSL + + "};\n" + "\n" + "void initAttributes(VS_INPUT input)\n" + "{\n" + + initHLSL + + "}\n"; + + std::string vertexHLSL(sourceShader); + + size_t copyInsertionPos = vertexHLSL.find(VERTEX_ATTRIBUTE_STUB_STRING); + vertexHLSL.replace(copyInsertionPos, VERTEX_ATTRIBUTE_STUB_STRING.length(), replacementHLSL); + + return vertexHLSL; +} + +std::string DynamicHLSL::generatePixelShaderForOutputSignature(const std::string &sourceShader, const std::vector<PixelShaderOutputVariable> &outputVariables, + bool usesFragDepth, const std::vector<GLenum> &outputLayout) const +{ + const int shaderModel = mRenderer->getMajorShaderModel(); + std::string targetSemantic = (shaderModel >= 4) ? "SV_TARGET" : "COLOR"; + std::string depthSemantic = (shaderModel >= 4) ? "SV_Depth" : "DEPTH"; + + std::string declarationHLSL; + std::string copyHLSL; + + for (size_t layoutIndex = 0; layoutIndex < outputLayout.size(); ++layoutIndex) + { + GLenum binding = outputLayout[layoutIndex]; + + if (binding != GL_NONE) + { + unsigned int location = (binding - GL_COLOR_ATTACHMENT0); + + const PixelShaderOutputVariable &outputVariable = GetOutputAtLocation(outputVariables, location); + + declarationHLSL += " " + HLSLTypeString(outputVariable.type) + " " + outputVariable.name + + " : " + targetSemantic + Str(layoutIndex) + ";\n"; + + copyHLSL += " output." + outputVariable.name + " = " + outputVariable.source + ";\n"; + } + } + + if (usesFragDepth) + { + declarationHLSL += " float gl_Depth : " + depthSemantic + ";\n"; + copyHLSL += " output.gl_Depth = gl_Depth; \n"; + } + + std::string replacementHLSL = "struct PS_OUTPUT\n" + "{\n" + + declarationHLSL + + "};\n" + "\n" + "PS_OUTPUT generateOutput()\n" + "{\n" + " PS_OUTPUT output;\n" + + copyHLSL + + " return output;\n" + "}\n"; + + std::string pixelHLSL(sourceShader); + + size_t outputInsertionPos = pixelHLSL.find(PIXEL_OUTPUT_STUB_STRING); + pixelHLSL.replace(outputInsertionPos, PIXEL_OUTPUT_STUB_STRING.length(), replacementHLSL); + + return pixelHLSL; +} + +std::string DynamicHLSL::getVaryingSemantic(bool pointSize) const +{ + // SM3 reserves the TEXCOORD semantic for point sprite texcoords (gl_PointCoord) + // In D3D11 we manually compute gl_PointCoord in the GS. + int shaderModel = mRenderer->getMajorShaderModel(); + return ((pointSize && shaderModel < 4) ? "COLOR" : "TEXCOORD"); +} + +struct DynamicHLSL::SemanticInfo +{ + struct BuiltinInfo + { + BuiltinInfo() + : enabled(false), + index(0), + systemValue(false) + {} + + bool enabled; + std::string semantic; + unsigned int index; + bool systemValue; + + std::string str() const + { + return (systemValue ? semantic : (semantic + Str(index))); + } + + void enableSystem(const std::string &systemValueSemantic) + { + enabled = true; + semantic = systemValueSemantic; + systemValue = true; + } + + void enable(const std::string &semanticVal, unsigned int indexVal) + { + enabled = true; + semantic = semanticVal; + index = indexVal; + } + }; + + BuiltinInfo dxPosition; + BuiltinInfo glPosition; + BuiltinInfo glFragCoord; + BuiltinInfo glPointCoord; + BuiltinInfo glPointSize; +}; + +DynamicHLSL::SemanticInfo DynamicHLSL::getSemanticInfo(int startRegisters, bool fragCoord, bool pointCoord, + bool pointSize, bool pixelShader) const +{ + SemanticInfo info; + bool hlsl4 = (mRenderer->getMajorShaderModel() >= 4); + const std::string &varyingSemantic = getVaryingSemantic(pointSize); + + int reservedRegisterIndex = startRegisters; + + if (hlsl4) + { + info.dxPosition.enableSystem("SV_Position"); + } + else if (pixelShader) + { + info.dxPosition.enableSystem("VPOS"); + } + else + { + info.dxPosition.enableSystem("POSITION"); + } + + info.glPosition.enable(varyingSemantic, reservedRegisterIndex++); + + if (fragCoord) + { + info.glFragCoord.enable(varyingSemantic, reservedRegisterIndex++); + } + + if (pointCoord) + { + // SM3 reserves the TEXCOORD semantic for point sprite texcoords (gl_PointCoord) + // In D3D11 we manually compute gl_PointCoord in the GS. + if (hlsl4) + { + info.glPointCoord.enable(varyingSemantic, reservedRegisterIndex++); + } + else + { + info.glPointCoord.enable("TEXCOORD", 0); + } + } + + // Special case: do not include PSIZE semantic in HLSL 3 pixel shaders + if (pointSize && (!pixelShader || hlsl4)) + { + info.glPointSize.enableSystem("PSIZE"); + } + + return info; +} + +std::string DynamicHLSL::generateVaryingLinkHLSL(const SemanticInfo &info, const std::string &varyingHLSL) const +{ + std::string linkHLSL = "{\n"; + + ASSERT(info.dxPosition.enabled && info.glPosition.enabled); + + linkHLSL += " float4 dx_Position : " + info.dxPosition.str() + ";\n"; + linkHLSL += " float4 gl_Position : " + info.glPosition.str() + ";\n"; + + if (info.glFragCoord.enabled) + { + linkHLSL += " float4 gl_FragCoord : " + info.glFragCoord.str() + ";\n"; + } + + if (info.glPointCoord.enabled) + { + linkHLSL += " float2 gl_PointCoord : " + info.glPointCoord.str() + ";\n"; + } + + linkHLSL += varyingHLSL; + + if (info.glPointSize.enabled) + { + linkHLSL += " float gl_PointSize : " + info.glPointSize.str() + ";\n"; + } + + linkHLSL += "};\n"; + + return linkHLSL; +} + +void DynamicHLSL::storeBuiltinLinkedVaryings(const SemanticInfo &info, + std::vector<LinkedVarying> *linkedVaryings) const +{ + ASSERT(info.glPosition.enabled); + + linkedVaryings->push_back(LinkedVarying("gl_Position", GL_FLOAT_VEC4, 1, info.glPosition.semantic, + info.glPosition.index, 1)); + + if (info.glFragCoord.enabled) + { + linkedVaryings->push_back(LinkedVarying("gl_FragCoord", GL_FLOAT_VEC4, 1, info.glFragCoord.semantic, + info.glFragCoord.index, 1)); + } + + if (info.glPointSize.enabled) + { + linkedVaryings->push_back(LinkedVarying("gl_PointSize", GL_FLOAT, 1, "PSIZE", 0, 1)); + } +} + +void DynamicHLSL::storeUserLinkedVaryings(const rx::ShaderD3D *vertexShader, + std::vector<LinkedVarying> *linkedVaryings) const +{ + const std::string &varyingSemantic = getVaryingSemantic(vertexShader->mUsesPointSize); + const std::vector<PackedVarying> &varyings = vertexShader->getVaryings(); + + for (unsigned int varyingIndex = 0; varyingIndex < varyings.size(); varyingIndex++) + { + const PackedVarying &varying = varyings[varyingIndex]; + + if (varying.registerAssigned()) + { + ASSERT(!varying.isBuiltIn()); + GLenum transposedType = TransposeMatrixType(varying.type); + int variableRows = (varying.isStruct() ? 1 : VariableRowCount(transposedType)); + + linkedVaryings->push_back(LinkedVarying(varying.name, varying.type, varying.elementCount(), + varyingSemantic, varying.registerIndex, + variableRows * varying.elementCount())); + } + } +} + +bool DynamicHLSL::generateShaderLinkHLSL(InfoLog &infoLog, int registers, const VaryingPacking packing, + std::string& pixelHLSL, std::string& vertexHLSL, + rx::ShaderD3D *fragmentShader, rx::ShaderD3D *vertexShader, + const std::vector<std::string>& transformFeedbackVaryings, + std::vector<LinkedVarying> *linkedVaryings, + std::map<int, VariableLocation> *programOutputVars, + std::vector<PixelShaderOutputVariable> *outPixelShaderKey, + bool *outUsesFragDepth) const +{ + if (pixelHLSL.empty() || vertexHLSL.empty()) + { + return false; + } + + bool usesMRT = fragmentShader->mUsesMultipleRenderTargets; + bool usesFragColor = fragmentShader->mUsesFragColor; + bool usesFragData = fragmentShader->mUsesFragData; + bool usesFragCoord = fragmentShader->mUsesFragCoord; + bool usesPointCoord = fragmentShader->mUsesPointCoord; + bool usesPointSize = vertexShader->mUsesPointSize; + + if (usesFragColor && usesFragData) + { + infoLog.append("Cannot use both gl_FragColor and gl_FragData in the same fragment shader."); + return false; + } + + // Write the HLSL input/output declarations + const int shaderModel = mRenderer->getMajorShaderModel(); + + // TODO (geofflang): Use context's caps + const int maxVaryingVectors = mRenderer->getRendererCaps().maxVaryingVectors; + + const int registersNeeded = registers + (usesFragCoord ? 1 : 0) + (usesPointCoord ? 1 : 0); + + // Two cases when writing to gl_FragColor and using ESSL 1.0: + // - with a 3.0 context, the output color is copied to channel 0 + // - with a 2.0 context, the output color is broadcast to all channels + const bool broadcast = (fragmentShader->mUsesFragColor && mRenderer->getCurrentClientVersion() < 3); + const unsigned int numRenderTargets = (broadcast || usesMRT ? mRenderer->getRendererCaps().maxDrawBuffers : 1); + + int shaderVersion = vertexShader->getShaderVersion(); + + if (registersNeeded > maxVaryingVectors) + { + infoLog.append("No varying registers left to support gl_FragCoord/gl_PointCoord"); + return false; + } + + const std::string &varyingHLSL = generateVaryingHLSL(vertexShader); + const SemanticInfo &vertexSemantics = getSemanticInfo(registers, usesFragCoord, + false, usesPointSize, false); + + storeUserLinkedVaryings(vertexShader, linkedVaryings); + storeBuiltinLinkedVaryings(vertexSemantics, linkedVaryings); + + // Add stub string to be replaced when shader is dynamically defined by its layout + vertexHLSL += "\n" + VERTEX_ATTRIBUTE_STUB_STRING + "\n" + "struct VS_OUTPUT\n" + generateVaryingLinkHLSL(vertexSemantics, varyingHLSL) + "\n" + "VS_OUTPUT main(VS_INPUT input)\n" + "{\n" + " initAttributes(input);\n"; + + if (shaderModel >= 4) + { + vertexHLSL += "\n" + " gl_main();\n" + "\n" + " VS_OUTPUT output;\n" + " output.gl_Position = gl_Position;\n" + " output.dx_Position.x = gl_Position.x;\n" + " output.dx_Position.y = -gl_Position.y;\n" + " output.dx_Position.z = (gl_Position.z + gl_Position.w) * 0.5;\n" + " output.dx_Position.w = gl_Position.w;\n"; + } + else + { + vertexHLSL += "\n" + " gl_main();\n" + "\n" + " VS_OUTPUT output;\n" + " output.gl_Position = gl_Position;\n" + " output.dx_Position.x = gl_Position.x * dx_ViewAdjust.z + dx_ViewAdjust.x * gl_Position.w;\n" + " output.dx_Position.y = -(gl_Position.y * dx_ViewAdjust.w + dx_ViewAdjust.y * gl_Position.w);\n" + " output.dx_Position.z = (gl_Position.z + gl_Position.w) * 0.5;\n" + " output.dx_Position.w = gl_Position.w;\n"; + } + + if (usesPointSize && shaderModel >= 3) + { + vertexHLSL += " output.gl_PointSize = gl_PointSize;\n"; + } + + if (usesFragCoord) + { + vertexHLSL += " output.gl_FragCoord = gl_Position;\n"; + } + + const std::vector<PackedVarying> &vertexVaryings = vertexShader->getVaryings(); + for (unsigned int vertVaryingIndex = 0; vertVaryingIndex < vertexVaryings.size(); vertVaryingIndex++) + { + const PackedVarying &varying = vertexVaryings[vertVaryingIndex]; + if (varying.registerAssigned()) + { + for (unsigned int elementIndex = 0; elementIndex < varying.elementCount(); elementIndex++) + { + int variableRows = (varying.isStruct() ? 1 : VariableRowCount(TransposeMatrixType(varying.type))); + + for (int row = 0; row < variableRows; row++) + { + int r = varying.registerIndex + varying.columnIndex * mRenderer->getRendererCaps().maxVaryingVectors + elementIndex * variableRows + row; + vertexHLSL += " output.v" + Str(r); + + vertexHLSL += " = _" + varying.name; + + if (varying.isArray()) + { + vertexHLSL += ArrayString(elementIndex); + } + + if (variableRows > 1) + { + vertexHLSL += ArrayString(row); + } + + vertexHLSL += ";\n"; + } + } + } + } + + vertexHLSL += "\n" + " return output;\n" + "}\n"; + + const SemanticInfo &pixelSemantics = getSemanticInfo(registers, usesFragCoord, usesPointCoord, + usesPointSize, true); + + pixelHLSL += "struct PS_INPUT\n" + generateVaryingLinkHLSL(pixelSemantics, varyingHLSL) + "\n"; + + if (shaderVersion < 300) + { + for (unsigned int renderTargetIndex = 0; renderTargetIndex < numRenderTargets; renderTargetIndex++) + { + PixelShaderOutputVariable outputKeyVariable; + outputKeyVariable.type = GL_FLOAT_VEC4; + outputKeyVariable.name = "gl_Color" + Str(renderTargetIndex); + outputKeyVariable.source = broadcast ? "gl_Color[0]" : "gl_Color[" + Str(renderTargetIndex) + "]"; + outputKeyVariable.outputIndex = renderTargetIndex; + + outPixelShaderKey->push_back(outputKeyVariable); + } + + *outUsesFragDepth = fragmentShader->mUsesFragDepth; + } + else + { + defineOutputVariables(fragmentShader, programOutputVars); + + const std::vector<sh::Attribute> &shaderOutputVars = fragmentShader->getActiveOutputVariables(); + for (auto locationIt = programOutputVars->begin(); locationIt != programOutputVars->end(); locationIt++) + { + const VariableLocation &outputLocation = locationIt->second; + const sh::ShaderVariable &outputVariable = shaderOutputVars[outputLocation.index]; + const std::string &variableName = "out_" + outputLocation.name; + const std::string &elementString = (outputLocation.element == GL_INVALID_INDEX ? "" : Str(outputLocation.element)); + + ASSERT(outputVariable.staticUse); + + PixelShaderOutputVariable outputKeyVariable; + outputKeyVariable.type = outputVariable.type; + outputKeyVariable.name = variableName + elementString; + outputKeyVariable.source = variableName + ArrayString(outputLocation.element); + outputKeyVariable.outputIndex = locationIt->first; + + outPixelShaderKey->push_back(outputKeyVariable); + } + + *outUsesFragDepth = false; + } + + pixelHLSL += PIXEL_OUTPUT_STUB_STRING + "\n"; + + if (fragmentShader->mUsesFrontFacing) + { + if (shaderModel >= 4) + { + pixelHLSL += "PS_OUTPUT main(PS_INPUT input, bool isFrontFace : SV_IsFrontFace)\n" + "{\n"; + } + else + { + pixelHLSL += "PS_OUTPUT main(PS_INPUT input, float vFace : VFACE)\n" + "{\n"; + } + } + else + { + pixelHLSL += "PS_OUTPUT main(PS_INPUT input)\n" + "{\n"; + } + + if (usesFragCoord) + { + pixelHLSL += " float rhw = 1.0 / input.gl_FragCoord.w;\n"; + + if (shaderModel >= 4) + { + pixelHLSL += " gl_FragCoord.x = input.dx_Position.x;\n" + " gl_FragCoord.y = input.dx_Position.y;\n"; + } + else if (shaderModel >= 3) + { + pixelHLSL += " gl_FragCoord.x = input.dx_Position.x + 0.5;\n" + " gl_FragCoord.y = input.dx_Position.y + 0.5;\n"; + } + else + { + // dx_ViewCoords contains the viewport width/2, height/2, center.x and center.y. See Renderer::setViewport() + pixelHLSL += " gl_FragCoord.x = (input.gl_FragCoord.x * rhw) * dx_ViewCoords.x + dx_ViewCoords.z;\n" + " gl_FragCoord.y = (input.gl_FragCoord.y * rhw) * dx_ViewCoords.y + dx_ViewCoords.w;\n"; + } + + pixelHLSL += " gl_FragCoord.z = (input.gl_FragCoord.z * rhw) * dx_DepthFront.x + dx_DepthFront.y;\n" + " gl_FragCoord.w = rhw;\n"; + } + + if (usesPointCoord && shaderModel >= 3) + { + pixelHLSL += " gl_PointCoord.x = input.gl_PointCoord.x;\n"; + pixelHLSL += " gl_PointCoord.y = 1.0 - input.gl_PointCoord.y;\n"; + } + + if (fragmentShader->mUsesFrontFacing) + { + if (shaderModel <= 3) + { + pixelHLSL += " gl_FrontFacing = (vFace * dx_DepthFront.z >= 0.0);\n"; + } + else + { + pixelHLSL += " gl_FrontFacing = isFrontFace;\n"; + } + } + + const std::vector<PackedVarying> &fragmentVaryings = fragmentShader->getVaryings(); + for (unsigned int varyingIndex = 0; varyingIndex < fragmentVaryings.size(); varyingIndex++) + { + const PackedVarying &varying = fragmentVaryings[varyingIndex]; + if (varying.registerAssigned()) + { + ASSERT(!varying.isBuiltIn()); + for (unsigned int elementIndex = 0; elementIndex < varying.elementCount(); elementIndex++) + { + GLenum transposedType = TransposeMatrixType(varying.type); + int variableRows = (varying.isStruct() ? 1 : VariableRowCount(transposedType)); + for (int row = 0; row < variableRows; row++) + { + std::string n = Str(varying.registerIndex + varying.columnIndex * mRenderer->getRendererCaps().maxVaryingVectors + elementIndex * variableRows + row); + pixelHLSL += " _" + varying.name; + + if (varying.isArray()) + { + pixelHLSL += ArrayString(elementIndex); + } + + if (variableRows > 1) + { + pixelHLSL += ArrayString(row); + } + + if (varying.isStruct()) + { + pixelHLSL += " = input.v" + n + ";\n"; break; + } + else + { + switch (VariableColumnCount(transposedType)) + { + case 1: pixelHLSL += " = input.v" + n + ".x;\n"; break; + case 2: pixelHLSL += " = input.v" + n + ".xy;\n"; break; + case 3: pixelHLSL += " = input.v" + n + ".xyz;\n"; break; + case 4: pixelHLSL += " = input.v" + n + ";\n"; break; + default: UNREACHABLE(); + } + } + } + } + } + else + { + ASSERT(varying.isBuiltIn() || !varying.staticUse); + } + } + + pixelHLSL += "\n" + " gl_main();\n" + "\n" + " return generateOutput();\n" + "}\n"; + + return true; +} + +void DynamicHLSL::defineOutputVariables(rx::ShaderD3D *fragmentShader, std::map<int, VariableLocation> *programOutputVars) const +{ + const std::vector<sh::Attribute> &shaderOutputVars = fragmentShader->getActiveOutputVariables(); + + for (unsigned int outputVariableIndex = 0; outputVariableIndex < shaderOutputVars.size(); outputVariableIndex++) + { + const sh::Attribute &outputVariable = shaderOutputVars[outputVariableIndex]; + const int baseLocation = outputVariable.location == -1 ? 0 : outputVariable.location; + + ASSERT(outputVariable.staticUse); + + if (outputVariable.arraySize > 0) + { + for (unsigned int elementIndex = 0; elementIndex < outputVariable.arraySize; elementIndex++) + { + const int location = baseLocation + elementIndex; + ASSERT(programOutputVars->count(location) == 0); + (*programOutputVars)[location] = VariableLocation(outputVariable.name, elementIndex, outputVariableIndex); + } + } + else + { + ASSERT(programOutputVars->count(baseLocation) == 0); + (*programOutputVars)[baseLocation] = VariableLocation(outputVariable.name, GL_INVALID_INDEX, outputVariableIndex); + } + } +} + +std::string DynamicHLSL::generateGeometryShaderHLSL(int registers, rx::ShaderD3D *fragmentShader, rx::ShaderD3D *vertexShader) const +{ + // for now we only handle point sprite emulation + ASSERT(vertexShader->mUsesPointSize && mRenderer->getMajorShaderModel() >= 4); + return generatePointSpriteHLSL(registers, fragmentShader, vertexShader); +} + +std::string DynamicHLSL::generatePointSpriteHLSL(int registers, rx::ShaderD3D *fragmentShader, rx::ShaderD3D *vertexShader) const +{ + ASSERT(registers >= 0); + ASSERT(vertexShader->mUsesPointSize); + ASSERT(mRenderer->getMajorShaderModel() >= 4); + + std::string geomHLSL; + + const SemanticInfo &inSemantics = getSemanticInfo(registers, fragmentShader->mUsesFragCoord, + false, true, false); + const SemanticInfo &outSemantics = getSemanticInfo(registers, fragmentShader->mUsesFragCoord, + fragmentShader->mUsesPointCoord, true, false); + + std::string varyingHLSL = generateVaryingHLSL(vertexShader); + std::string inLinkHLSL = generateVaryingLinkHLSL(inSemantics, varyingHLSL); + std::string outLinkHLSL = generateVaryingLinkHLSL(outSemantics, varyingHLSL); + + // TODO(geofflang): use context's caps + geomHLSL += "uniform float4 dx_ViewCoords : register(c1);\n" + "\n" + "struct GS_INPUT\n" + inLinkHLSL + "\n" + + "struct GS_OUTPUT\n" + outLinkHLSL + "\n" + + "\n" + "static float2 pointSpriteCorners[] = \n" + "{\n" + " float2( 0.5f, -0.5f),\n" + " float2( 0.5f, 0.5f),\n" + " float2(-0.5f, -0.5f),\n" + " float2(-0.5f, 0.5f)\n" + "};\n" + "\n" + "static float2 pointSpriteTexcoords[] = \n" + "{\n" + " float2(1.0f, 1.0f),\n" + " float2(1.0f, 0.0f),\n" + " float2(0.0f, 1.0f),\n" + " float2(0.0f, 0.0f)\n" + "};\n" + "\n" + "static float minPointSize = " + Str(mRenderer->getRendererCaps().minAliasedPointSize) + ".0f;\n" + "static float maxPointSize = " + Str(mRenderer->getRendererCaps().maxAliasedPointSize) + ".0f;\n" + "\n" + "[maxvertexcount(4)]\n" + "void main(point GS_INPUT input[1], inout TriangleStream<GS_OUTPUT> outStream)\n" + "{\n" + " GS_OUTPUT output = (GS_OUTPUT)0;\n" + " output.gl_Position = input[0].gl_Position;\n"; + " output.gl_PointSize = input[0].gl_PointSize;\n"; + + for (int r = 0; r < registers; r++) + { + geomHLSL += " output.v" + Str(r) + " = input[0].v" + Str(r) + ";\n"; + } + + if (fragmentShader->mUsesFragCoord) + { + geomHLSL += " output.gl_FragCoord = input[0].gl_FragCoord;\n"; + } + + geomHLSL += " \n" + " float gl_PointSize = clamp(input[0].gl_PointSize, minPointSize, maxPointSize);\n" + " float4 dx_Position = input[0].dx_Position;\n" + " float2 viewportScale = float2(1.0f / dx_ViewCoords.x, 1.0f / dx_ViewCoords.y) * dx_Position.w;\n"; + + for (int corner = 0; corner < 4; corner++) + { + geomHLSL += " \n" + " output.dx_Position = dx_Position + float4(pointSpriteCorners[" + Str(corner) + "] * viewportScale * gl_PointSize, 0.0f, 0.0f);\n"; + + if (fragmentShader->mUsesPointCoord) + { + geomHLSL += " output.gl_PointCoord = pointSpriteTexcoords[" + Str(corner) + "];\n"; + } + + geomHLSL += " outStream.Append(output);\n"; + } + + geomHLSL += " \n" + " outStream.RestartStrip();\n" + "}\n"; + + return geomHLSL; +} + +// This method needs to match OutputHLSL::decorate +std::string DynamicHLSL::decorateVariable(const std::string &name) +{ + if (name.compare(0, 3, "gl_") != 0) + { + return "_" + name; + } + + return name; +} + +std::string DynamicHLSL::generateAttributeConversionHLSL(const VertexFormat &vertexFormat, const sh::ShaderVariable &shaderAttrib) const +{ + std::string attribString = "input." + decorateVariable(shaderAttrib.name); + + // Matrix + if (IsMatrixType(shaderAttrib.type)) + { + return "transpose(" + attribString + ")"; + } + + GLenum shaderComponentType = VariableComponentType(shaderAttrib.type); + int shaderComponentCount = VariableComponentCount(shaderAttrib.type); + + // Perform integer to float conversion (if necessary) + bool requiresTypeConversion = (shaderComponentType == GL_FLOAT && vertexFormat.mType != GL_FLOAT); + + if (requiresTypeConversion) + { + // TODO: normalization for 32-bit integer formats + ASSERT(!vertexFormat.mNormalized && !vertexFormat.mPureInteger); + return "float" + Str(shaderComponentCount) + "(" + attribString + ")"; + } + + // No conversion necessary + return attribString; +} + +void DynamicHLSL::getInputLayoutSignature(const VertexFormat inputLayout[], GLenum signature[]) const +{ + for (size_t inputIndex = 0; inputIndex < MAX_VERTEX_ATTRIBS; inputIndex++) + { + const VertexFormat &vertexFormat = inputLayout[inputIndex]; + + if (vertexFormat.mType == GL_NONE) + { + signature[inputIndex] = GL_NONE; + } + else + { + bool gpuConverted = ((mRenderer->getVertexConversionType(vertexFormat) & rx::VERTEX_CONVERT_GPU) != 0); + signature[inputIndex] = (gpuConverted ? GL_TRUE : GL_FALSE); + } + } +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/DynamicHLSL.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/DynamicHLSL.h new file mode 100644 index 0000000000..f68ed98401 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/DynamicHLSL.h @@ -0,0 +1,104 @@ +// +// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// +// DynamicHLSL.h: Interface for link and run-time HLSL generation +// + +#ifndef LIBGLESV2_RENDERER_DYNAMIC_HLSL_H_ +#define LIBGLESV2_RENDERER_DYNAMIC_HLSL_H_ + +#include "common/angleutils.h" +#include "libGLESv2/constants.h" + +#include "angle_gl.h" + +#include <vector> +#include <map> + +namespace rx +{ +class Renderer; +} + +namespace sh +{ +struct Attribute; +struct ShaderVariable; +} + +namespace gl +{ +class InfoLog; +struct VariableLocation; +struct LinkedVarying; +struct VertexAttribute; +struct VertexFormat; +struct PackedVarying; +} + +namespace rx +{ +class Renderer; +class ShaderD3D; + +typedef const gl::PackedVarying *VaryingPacking[gl::IMPLEMENTATION_MAX_VARYING_VECTORS][4]; + +struct PixelShaderOutputVariable +{ + GLenum type; + std::string name; + std::string source; + size_t outputIndex; +}; + +class DynamicHLSL +{ + public: + explicit DynamicHLSL(rx::Renderer *const renderer); + + int packVaryings(gl::InfoLog &infoLog, VaryingPacking packing, rx::ShaderD3D *fragmentShader, + rx::ShaderD3D *vertexShader, const std::vector<std::string>& transformFeedbackVaryings); + std::string generateVertexShaderForInputLayout(const std::string &sourceShader, const gl::VertexFormat inputLayout[], + const sh::Attribute shaderAttributes[]) const; + std::string generatePixelShaderForOutputSignature(const std::string &sourceShader, const std::vector<PixelShaderOutputVariable> &outputVariables, + bool usesFragDepth, const std::vector<GLenum> &outputLayout) const; + bool generateShaderLinkHLSL(gl::InfoLog &infoLog, int registers, const VaryingPacking packing, + std::string& pixelHLSL, std::string& vertexHLSL, + rx::ShaderD3D *fragmentShader, rx::ShaderD3D *vertexShader, + const std::vector<std::string>& transformFeedbackVaryings, + std::vector<gl::LinkedVarying> *linkedVaryings, + std::map<int, gl::VariableLocation> *programOutputVars, + std::vector<PixelShaderOutputVariable> *outPixelShaderKey, + bool *outUsesFragDepth) const; + + std::string generateGeometryShaderHLSL(int registers, rx::ShaderD3D *fragmentShader, rx::ShaderD3D *vertexShader) const; + void getInputLayoutSignature(const gl::VertexFormat inputLayout[], GLenum signature[]) const; + + private: + DISALLOW_COPY_AND_ASSIGN(DynamicHLSL); + + rx::Renderer *const mRenderer; + + struct SemanticInfo; + + std::string getVaryingSemantic(bool pointSize) const; + SemanticInfo getSemanticInfo(int startRegisters, bool fragCoord, bool pointCoord, bool pointSize, + bool pixelShader) const; + std::string generateVaryingLinkHLSL(const SemanticInfo &info, const std::string &varyingHLSL) const; + std::string generateVaryingHLSL(const ShaderD3D *shader) const; + void storeUserLinkedVaryings(const rx::ShaderD3D *vertexShader, std::vector<gl::LinkedVarying> *linkedVaryings) const; + void storeBuiltinLinkedVaryings(const SemanticInfo &info, std::vector<gl::LinkedVarying> *linkedVaryings) const; + void defineOutputVariables(rx::ShaderD3D *fragmentShader, std::map<int, gl::VariableLocation> *programOutputVars) const; + std::string generatePointSpriteHLSL(int registers, rx::ShaderD3D *fragmentShader, rx::ShaderD3D *vertexShader) const; + + // Prepend an underscore + static std::string decorateVariable(const std::string &name); + + std::string generateAttributeConversionHLSL(const gl::VertexFormat &vertexFormat, const sh::ShaderVariable &shaderAttrib) const; +}; + +} + +#endif // LIBGLESV2_RENDERER_DYNAMIC_HLSL_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/HLSLCompiler.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/HLSLCompiler.cpp new file mode 100644 index 0000000000..d0131974ee --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/HLSLCompiler.cpp @@ -0,0 +1,164 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +#include "libGLESv2/renderer/d3d/HLSLCompiler.h" +#include "libGLESv2/Program.h" +#include "libGLESv2/main.h" + +#include "common/utilities.h" +#include "common/platform.h" + +#if defined(__MINGW32__) && !defined(D3DCOMPILER_DLL) + +// Add define + typedefs for older MinGW-w64 headers (pre 5783) + +#define D3DCOMPILER_DLL L"d3dcompiler_43.dll" + +HRESULT WINAPI D3DCompile(const void *data, SIZE_T data_size, const char *filename, + const D3D_SHADER_MACRO *defines, ID3DInclude *include, const char *entrypoint, + const char *target, UINT sflags, UINT eflags, ID3DBlob **shader, ID3DBlob **error_messages); +typedef HRESULT (WINAPI *pD3DCompile)(const void *data, SIZE_T data_size, const char *filename, + const D3D_SHADER_MACRO *defines, ID3DInclude *include, const char *entrypoint, + const char *target, UINT sflags, UINT eflags, ID3DBlob **shader, ID3DBlob **error_messages); + +#endif // __MINGW32__ && !D3DCOMPILER_DLL + +#ifndef QT_D3DCOMPILER_DLL +#define QT_D3DCOMPILER_DLL D3DCOMPILER_DLL +#endif + +namespace rx +{ + +HLSLCompiler::HLSLCompiler() + : mD3DCompilerModule(NULL), + mD3DCompileFunc(NULL) +{ +} + +HLSLCompiler::~HLSLCompiler() +{ + release(); +} + +bool HLSLCompiler::initialize() +{ +#if !defined(ANGLE_PLATFORM_WINRT) +#if defined(ANGLE_PRELOADED_D3DCOMPILER_MODULE_NAMES) + // Find a D3DCompiler module that had already been loaded based on a predefined list of versions. + static const char *d3dCompilerNames[] = ANGLE_PRELOADED_D3DCOMPILER_MODULE_NAMES; + + for (size_t i = 0; i < ArraySize(d3dCompilerNames); ++i) + { + if (GetModuleHandleExA(0, d3dCompilerNames[i], &mD3DCompilerModule)) + { + break; + } + } +#endif // ANGLE_PRELOADED_D3DCOMPILER_MODULE_NAMES + + // Load the compiler DLL specified by the environment, or default to QT_D3DCOMPILER_DLL + const wchar_t *defaultCompiler = _wgetenv(L"QT_D3DCOMPILER_DLL"); + if (!defaultCompiler) + defaultCompiler = QT_D3DCOMPILER_DLL; + + const wchar_t *compilerDlls[] = { + defaultCompiler, + L"d3dcompiler_47.dll", + L"d3dcompiler_46.dll", + L"d3dcompiler_45.dll", + L"d3dcompiler_44.dll", + L"d3dcompiler_43.dll", + 0 + }; + + // Load the first available known compiler DLL + for (int i = 0; compilerDlls[i]; ++i) + { + mD3DCompilerModule = LoadLibrary(compilerDlls[i]); + if (mD3DCompilerModule) + break; + } + + if (!mD3DCompilerModule) + { + ERR("No D3D compiler module found - aborting!\n"); + return false; + } + + mD3DCompileFunc = reinterpret_cast<CompileFuncPtr>(GetProcAddress(mD3DCompilerModule, "D3DCompile")); + ASSERT(mD3DCompileFunc); +#else + mD3DCompileFunc = reinterpret_cast<CompileFuncPtr>(&D3DCompile); +#endif + return mD3DCompileFunc != NULL; +} + +void HLSLCompiler::release() +{ + if (mD3DCompilerModule) + { + FreeLibrary(mD3DCompilerModule); + mD3DCompilerModule = NULL; + mD3DCompileFunc = NULL; + } +} + +ShaderBlob *HLSLCompiler::compileToBinary(gl::InfoLog &infoLog, const char *hlsl, const char *profile, + const UINT optimizationFlags[], const char *flagNames[], int attempts) const +{ +#if !defined(ANGLE_PLATFORM_WINRT) + ASSERT(mD3DCompilerModule && mD3DCompileFunc); +#endif + + if (!hlsl) + { + return NULL; + } + + pD3DCompile compileFunc = reinterpret_cast<pD3DCompile>(mD3DCompileFunc); + for (int i = 0; i < attempts; ++i) + { + ID3DBlob *errorMessage = NULL; + ID3DBlob *binary = NULL; + + HRESULT result = compileFunc(hlsl, strlen(hlsl), gl::g_fakepath, NULL, NULL, "main", profile, optimizationFlags[i], 0, &binary, &errorMessage); + + if (errorMessage) + { + const char *message = (const char*)errorMessage->GetBufferPointer(); + + infoLog.appendSanitized(message); + TRACE("\n%s", hlsl); + TRACE("\n%s", message); + + SafeRelease(errorMessage); + } + + if (SUCCEEDED(result)) + { + return (ShaderBlob*)binary; + } + else + { + if (result == E_OUTOFMEMORY) + { + return gl::error(GL_OUT_OF_MEMORY, (ShaderBlob*)NULL); + } + + infoLog.append("Warning: D3D shader compilation failed with %s flags.", flagNames[i]); + + if (i + 1 < attempts) + { + infoLog.append(" Retrying with %s.\n", flagNames[i + 1]); + } + } + } + + return NULL; +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/HLSLCompiler.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/HLSLCompiler.h new file mode 100644 index 0000000000..0ce9e44be5 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/HLSLCompiler.h @@ -0,0 +1,38 @@ +#ifndef LIBGLESV2_RENDERER_HLSL_D3DCOMPILER_H_ +#define LIBGLESV2_RENDERER_HLSL_D3DCOMPILER_H_ + +#include "common/angleutils.h" + +namespace gl +{ +class InfoLog; +} + +namespace rx +{ + +typedef void* ShaderBlob; +typedef void(*CompileFuncPtr)(); + +class HLSLCompiler +{ + public: + HLSLCompiler(); + ~HLSLCompiler(); + + bool initialize(); + void release(); + + ShaderBlob *compileToBinary(gl::InfoLog &infoLog, const char *hlsl, const char *profile, + const UINT optimizationFlags[], const char *flagNames[], int attempts) const; + + private: + DISALLOW_COPY_AND_ASSIGN(HLSLCompiler); + + HMODULE mD3DCompilerModule; + CompileFuncPtr mD3DCompileFunc; +}; + +} + +#endif // LIBGLESV2_RENDERER_HLSL_D3DCOMPILER_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ImageD3D.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ImageD3D.cpp new file mode 100644 index 0000000000..0854b968da --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ImageD3D.cpp @@ -0,0 +1,26 @@ +// +// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Image.h: Implements the rx::Image class, an abstract base class for the +// renderer-specific classes which will define the interface to the underlying +// surfaces or resources. + +#include "libGLESv2/renderer/d3d/ImageD3D.h" + +namespace rx +{ + +ImageD3D::ImageD3D() +{ +} + +ImageD3D *ImageD3D::makeImageD3D(Image *img) +{ + ASSERT(HAS_DYNAMIC_TYPE(rx::ImageD3D*, img)); + return static_cast<rx::ImageD3D*>(img); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ImageD3D.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ImageD3D.h new file mode 100644 index 0000000000..60a6ffdf37 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ImageD3D.h @@ -0,0 +1,51 @@ +// +// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Image.h: Defines the rx::Image class, an abstract base class for the +// renderer-specific classes which will define the interface to the underlying +// surfaces or resources. + +#ifndef LIBGLESV2_RENDERER_IMAGED3D_H_ +#define LIBGLESV2_RENDERER_IMAGED3D_H_ + +#include "common/debug.h" +#include "libGLESv2/renderer/Image.h" + +namespace gl +{ +class Framebuffer; +} + +namespace rx +{ +class TextureStorage; + +class ImageD3D : public Image +{ + public: + ImageD3D(); + virtual ~ImageD3D() {}; + + static ImageD3D *makeImageD3D(Image *img); + + virtual bool isDirty() const = 0; + + virtual void setManagedSurface2D(TextureStorage *storage, int level) {}; + virtual void setManagedSurfaceCube(TextureStorage *storage, int face, int level) {}; + virtual void setManagedSurface3D(TextureStorage *storage, int level) {}; + virtual void setManagedSurface2DArray(TextureStorage *storage, int layer, int level) {}; + virtual bool copyToStorage2D(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) = 0; + virtual bool copyToStorageCube(TextureStorage *storage, int face, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) = 0; + virtual bool copyToStorage3D(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth) = 0; + virtual bool copyToStorage2DArray(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height) = 0; + + private: + DISALLOW_COPY_AND_ASSIGN(ImageD3D); +}; + +} + +#endif // LIBGLESV2_RENDERER_IMAGED3D_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/IndexBuffer.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/IndexBuffer.cpp index 37dbd3e195..1dce1270d8 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/IndexBuffer.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/IndexBuffer.cpp @@ -1,4 +1,3 @@ -#include "precompiled.h" // // Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be @@ -8,7 +7,7 @@ // IndexBuffer.cpp: Defines the abstract IndexBuffer class and IndexBufferInterface // class with derivations, classes that perform graphics API agnostic index buffer operations. -#include "libGLESv2/renderer/IndexBuffer.h" +#include "libGLESv2/renderer/d3d/IndexBuffer.h" #include "libGLESv2/renderer/Renderer.h" namespace rx @@ -67,21 +66,22 @@ unsigned int IndexBufferInterface::getSerial() const return mIndexBuffer->getSerial(); } -bool IndexBufferInterface::mapBuffer(unsigned int size, void** outMappedMemory, unsigned int *streamOffset) +gl::Error IndexBufferInterface::mapBuffer(unsigned int size, void** outMappedMemory, unsigned int *streamOffset) { // Protect against integer overflow if (mWritePosition + size < mWritePosition) { - return false; + return gl::Error(GL_OUT_OF_MEMORY, "Mapping of internal index buffer would cause an integer overflow."); } - if (!mIndexBuffer->mapBuffer(mWritePosition, size, outMappedMemory)) + gl::Error error = mIndexBuffer->mapBuffer(mWritePosition, size, outMappedMemory); + if (error.isError()) { if (outMappedMemory) { *outMappedMemory = NULL; } - return false; + return error; } if (streamOffset) @@ -90,10 +90,10 @@ bool IndexBufferInterface::mapBuffer(unsigned int size, void** outMappedMemory, } mWritePosition += size; - return true; + return gl::Error(GL_NO_ERROR); } -bool IndexBufferInterface::unmapBuffer() +gl::Error IndexBufferInterface::unmapBuffer() { return mIndexBuffer->unmapBuffer(); } @@ -113,12 +113,12 @@ void IndexBufferInterface::setWritePosition(unsigned int writePosition) mWritePosition = writePosition; } -bool IndexBufferInterface::discard() +gl::Error IndexBufferInterface::discard() { return mIndexBuffer->discard(); } -bool IndexBufferInterface::setBufferSize(unsigned int bufferSize, GLenum indexType) +gl::Error IndexBufferInterface::setBufferSize(unsigned int bufferSize, GLenum indexType) { if (mIndexBuffer->getBufferSize() == 0) { @@ -138,26 +138,30 @@ StreamingIndexBufferInterface::~StreamingIndexBufferInterface() { } -bool StreamingIndexBufferInterface::reserveBufferSpace(unsigned int size, GLenum indexType) +gl::Error StreamingIndexBufferInterface::reserveBufferSpace(unsigned int size, GLenum indexType) { - bool result = true; unsigned int curBufferSize = getBufferSize(); unsigned int writePos = getWritePosition(); if (size > curBufferSize) { - result = setBufferSize(std::max(size, 2 * curBufferSize), indexType); + gl::Error error = setBufferSize(std::max(size, 2 * curBufferSize), indexType); + if (error.isError()) + { + return error; + } setWritePosition(0); } else if (writePos + size > curBufferSize || writePos + size < writePos) { - if (!discard()) + gl::Error error = discard(); + if (error.isError()) { - return false; + return error; } setWritePosition(0); } - return result; + return gl::Error(GL_NO_ERROR); } @@ -169,7 +173,7 @@ StaticIndexBufferInterface::~StaticIndexBufferInterface() { } -bool StaticIndexBufferInterface::reserveBufferSpace(unsigned int size, GLenum indexType) +gl::Error StaticIndexBufferInterface::reserveBufferSpace(unsigned int size, GLenum indexType) { unsigned int curSize = getBufferSize(); if (curSize == 0) @@ -178,13 +182,12 @@ bool StaticIndexBufferInterface::reserveBufferSpace(unsigned int size, GLenum in } else if (curSize >= size && indexType == getIndexType()) { - return true; + return gl::Error(GL_NO_ERROR); } else { - ERR("Static index buffers can't be resized"); UNREACHABLE(); - return false; + return gl::Error(GL_INVALID_OPERATION, "Internal static index buffers can't be resized"); } } @@ -194,4 +197,3 @@ IndexRangeCache *StaticIndexBufferInterface::getIndexRangeCache() } } - diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/IndexBuffer.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/IndexBuffer.h index 6fb885a1cd..1bb5ae2c4a 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/IndexBuffer.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/IndexBuffer.h @@ -11,6 +11,7 @@ #define LIBGLESV2_RENDERER_INDEXBUFFER_H_ #include "common/angleutils.h" +#include "libGLESv2/Error.h" #include "libGLESv2/renderer/IndexRangeCache.h" namespace rx @@ -23,16 +24,16 @@ class IndexBuffer IndexBuffer(); virtual ~IndexBuffer(); - virtual bool initialize(unsigned int bufferSize, GLenum indexType, bool dynamic) = 0; + virtual gl::Error initialize(unsigned int bufferSize, GLenum indexType, bool dynamic) = 0; - virtual bool mapBuffer(unsigned int offset, unsigned int size, void** outMappedMemory) = 0; - virtual bool unmapBuffer() = 0; + virtual gl::Error mapBuffer(unsigned int offset, unsigned int size, void** outMappedMemory) = 0; + virtual gl::Error unmapBuffer() = 0; - virtual bool discard() = 0; + virtual gl::Error discard() = 0; virtual GLenum getIndexType() const = 0; virtual unsigned int getBufferSize() const = 0; - virtual bool setSize(unsigned int bufferSize, GLenum indexType) = 0; + virtual gl::Error setSize(unsigned int bufferSize, GLenum indexType) = 0; unsigned int getSerial() const; @@ -52,15 +53,15 @@ class IndexBufferInterface IndexBufferInterface(Renderer *renderer, bool dynamic); virtual ~IndexBufferInterface(); - virtual bool reserveBufferSpace(unsigned int size, GLenum indexType) = 0; + virtual gl::Error reserveBufferSpace(unsigned int size, GLenum indexType) = 0; GLenum getIndexType() const; unsigned int getBufferSize() const; unsigned int getSerial() const; - bool mapBuffer(unsigned int size, void** outMappedMemory, unsigned int *streamOffset); - bool unmapBuffer(); + gl::Error mapBuffer(unsigned int size, void** outMappedMemory, unsigned int *streamOffset); + gl::Error unmapBuffer(); IndexBuffer *getIndexBuffer() const; @@ -68,9 +69,9 @@ class IndexBufferInterface unsigned int getWritePosition() const; void setWritePosition(unsigned int writePosition); - bool discard(); + gl::Error discard(); - bool setBufferSize(unsigned int bufferSize, GLenum indexType); + gl::Error setBufferSize(unsigned int bufferSize, GLenum indexType); private: DISALLOW_COPY_AND_ASSIGN(IndexBufferInterface); @@ -89,7 +90,7 @@ class StreamingIndexBufferInterface : public IndexBufferInterface StreamingIndexBufferInterface(Renderer *renderer); ~StreamingIndexBufferInterface(); - virtual bool reserveBufferSpace(unsigned int size, GLenum indexType); + virtual gl::Error reserveBufferSpace(unsigned int size, GLenum indexType); }; class StaticIndexBufferInterface : public IndexBufferInterface @@ -98,7 +99,7 @@ class StaticIndexBufferInterface : public IndexBufferInterface explicit StaticIndexBufferInterface(Renderer *renderer); ~StaticIndexBufferInterface(); - virtual bool reserveBufferSpace(unsigned int size, GLenum indexType); + virtual gl::Error reserveBufferSpace(unsigned int size, GLenum indexType); IndexRangeCache *getIndexRangeCache(); @@ -108,4 +109,4 @@ class StaticIndexBufferInterface : public IndexBufferInterface } -#endif // LIBGLESV2_RENDERER_INDEXBUFFER_H_
\ No newline at end of file +#endif // LIBGLESV2_RENDERER_INDEXBUFFER_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/IndexDataManager.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/IndexDataManager.cpp new file mode 100644 index 0000000000..8d455b4bf3 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/IndexDataManager.cpp @@ -0,0 +1,255 @@ +// +// Copyright (c) 2002-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// IndexDataManager.cpp: Defines the IndexDataManager, a class that +// runs the Buffer translation process for index buffers. + +#include "libGLESv2/renderer/d3d/IndexDataManager.h" +#include "libGLESv2/renderer/d3d/BufferD3D.h" +#include "libGLESv2/renderer/d3d/IndexBuffer.h" +#include "libGLESv2/renderer/Renderer.h" +#include "libGLESv2/Buffer.h" +#include "libGLESv2/main.h" +#include "libGLESv2/formatutils.h" + +namespace rx +{ + +static void ConvertIndices(GLenum sourceType, GLenum destinationType, const void *input, GLsizei count, void *output) +{ + if (sourceType == GL_UNSIGNED_BYTE) + { + ASSERT(destinationType == GL_UNSIGNED_SHORT); + const GLubyte *in = static_cast<const GLubyte*>(input); + GLushort *out = static_cast<GLushort*>(output); + + for (GLsizei i = 0; i < count; i++) + { + out[i] = in[i]; + } + } + else if (sourceType == GL_UNSIGNED_INT) + { + ASSERT(destinationType == GL_UNSIGNED_INT); + memcpy(output, input, count * sizeof(GLuint)); + } + else if (sourceType == GL_UNSIGNED_SHORT) + { + if (destinationType == GL_UNSIGNED_SHORT) + { + memcpy(output, input, count * sizeof(GLushort)); + } + else if (destinationType == GL_UNSIGNED_INT) + { + const GLushort *in = static_cast<const GLushort*>(input); + GLuint *out = static_cast<GLuint*>(output); + + for (GLsizei i = 0; i < count; i++) + { + out[i] = in[i]; + } + } + else UNREACHABLE(); + } + else UNREACHABLE(); +} + +IndexDataManager::IndexDataManager(Renderer *renderer) + : mRenderer(renderer), + mStreamingBufferShort(NULL), + mStreamingBufferInt(NULL) +{ +} + +IndexDataManager::~IndexDataManager() +{ + SafeDelete(mStreamingBufferShort); + SafeDelete(mStreamingBufferInt); +} + +gl::Error IndexDataManager::prepareIndexData(GLenum type, GLsizei count, gl::Buffer *buffer, const GLvoid *indices, TranslatedIndexData *translated) +{ + const gl::Type &typeInfo = gl::GetTypeInfo(type); + + GLenum destinationIndexType = (type == GL_UNSIGNED_INT) ? GL_UNSIGNED_INT : GL_UNSIGNED_SHORT; + + unsigned int offset = 0; + bool alignedOffset = false; + + BufferD3D *storage = NULL; + + if (buffer != NULL) + { + offset = static_cast<unsigned int>(reinterpret_cast<uintptr_t>(indices)); + + storage = BufferD3D::makeBufferD3D(buffer->getImplementation()); + + switch (type) + { + case GL_UNSIGNED_BYTE: alignedOffset = (offset % sizeof(GLubyte) == 0); break; + case GL_UNSIGNED_SHORT: alignedOffset = (offset % sizeof(GLushort) == 0); break; + case GL_UNSIGNED_INT: alignedOffset = (offset % sizeof(GLuint) == 0); break; + default: UNREACHABLE(); alignedOffset = false; + } + + ASSERT(typeInfo.bytes * static_cast<unsigned int>(count) + offset <= storage->getSize()); + + indices = static_cast<const GLubyte*>(storage->getData()) + offset; + } + + StaticIndexBufferInterface *staticBuffer = storage ? storage->getStaticIndexBuffer() : NULL; + IndexBufferInterface *indexBuffer = NULL; + bool directStorage = alignedOffset && storage && storage->supportsDirectBinding() && + destinationIndexType == type; + unsigned int streamOffset = 0; + + if (directStorage) + { + streamOffset = offset; + + if (!buffer->getIndexRangeCache()->findRange(type, offset, count, NULL, NULL)) + { + buffer->getIndexRangeCache()->addRange(type, offset, count, translated->indexRange, offset); + } + } + else if (staticBuffer && staticBuffer->getBufferSize() != 0 && staticBuffer->getIndexType() == type && alignedOffset) + { + indexBuffer = staticBuffer; + + if (!staticBuffer->getIndexRangeCache()->findRange(type, offset, count, NULL, &streamOffset)) + { + streamOffset = (offset / typeInfo.bytes) * gl::GetTypeInfo(destinationIndexType).bytes; + staticBuffer->getIndexRangeCache()->addRange(type, offset, count, translated->indexRange, streamOffset); + } + } + + // Avoid D3D11's primitive restart index value + // see http://msdn.microsoft.com/en-us/library/windows/desktop/bb205124(v=vs.85).aspx + if (translated->indexRange.end == 0xFFFF && type == GL_UNSIGNED_SHORT && mRenderer->getMajorShaderModel() > 3) + { + destinationIndexType = GL_UNSIGNED_INT; + directStorage = false; + indexBuffer = NULL; + } + + const gl::Type &destTypeInfo = gl::GetTypeInfo(destinationIndexType); + + if (!directStorage && !indexBuffer) + { + gl::Error error = getStreamingIndexBuffer(destinationIndexType, &indexBuffer); + if (error.isError()) + { + return error; + } + + unsigned int convertCount = count; + + if (staticBuffer) + { + if (staticBuffer->getBufferSize() == 0 && alignedOffset) + { + indexBuffer = staticBuffer; + convertCount = storage->getSize() / typeInfo.bytes; + } + else + { + storage->invalidateStaticData(); + staticBuffer = NULL; + } + } + + ASSERT(indexBuffer); + + if (convertCount > std::numeric_limits<unsigned int>::max() / destTypeInfo.bytes) + { + return gl::Error(GL_OUT_OF_MEMORY, "Reserving %u indices of %u bytes each exceeds the maximum buffer size.", + convertCount, destTypeInfo.bytes); + } + + unsigned int bufferSizeRequired = convertCount * destTypeInfo.bytes; + error = indexBuffer->reserveBufferSpace(bufferSizeRequired, type); + if (error.isError()) + { + return error; + } + + void* output = NULL; + error = indexBuffer->mapBuffer(bufferSizeRequired, &output, &streamOffset); + if (error.isError()) + { + return error; + } + + ConvertIndices(type, destinationIndexType, staticBuffer ? storage->getData() : indices, convertCount, output); + + error = indexBuffer->unmapBuffer(); + if (error.isError()) + { + return error; + } + + if (staticBuffer) + { + streamOffset = (offset / typeInfo.bytes) * destTypeInfo.bytes; + staticBuffer->getIndexRangeCache()->addRange(type, offset, count, translated->indexRange, streamOffset); + } + } + + translated->storage = directStorage ? storage : NULL; + translated->indexBuffer = indexBuffer ? indexBuffer->getIndexBuffer() : NULL; + translated->serial = directStorage ? storage->getSerial() : indexBuffer->getSerial(); + translated->startIndex = streamOffset / destTypeInfo.bytes; + translated->startOffset = streamOffset; + translated->indexType = destinationIndexType; + + if (storage) + { + storage->promoteStaticUsage(count * typeInfo.bytes); + } + + return gl::Error(GL_NO_ERROR); +} + +gl::Error IndexDataManager::getStreamingIndexBuffer(GLenum destinationIndexType, IndexBufferInterface **outBuffer) +{ + ASSERT(outBuffer); + if (destinationIndexType == GL_UNSIGNED_INT) + { + if (!mStreamingBufferInt) + { + mStreamingBufferInt = new StreamingIndexBufferInterface(mRenderer); + gl::Error error = mStreamingBufferInt->reserveBufferSpace(INITIAL_INDEX_BUFFER_SIZE, GL_UNSIGNED_INT); + if (error.isError()) + { + SafeDelete(mStreamingBufferInt); + return error; + } + } + + *outBuffer = mStreamingBufferInt; + return gl::Error(GL_NO_ERROR); + } + else + { + ASSERT(destinationIndexType == GL_UNSIGNED_SHORT); + + if (!mStreamingBufferShort) + { + mStreamingBufferShort = new StreamingIndexBufferInterface(mRenderer); + gl::Error error = mStreamingBufferShort->reserveBufferSpace(INITIAL_INDEX_BUFFER_SIZE, GL_UNSIGNED_SHORT); + if (error.isError()) + { + SafeDelete(mStreamingBufferShort); + return error; + } + } + + *outBuffer = mStreamingBufferShort; + return gl::Error(GL_NO_ERROR); + } +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/IndexDataManager.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/IndexDataManager.h index 0e77c81d1b..6d0b89e6d4 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/IndexDataManager.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/IndexDataManager.h @@ -1,5 +1,5 @@ // -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2002-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // @@ -11,6 +11,10 @@ #define LIBGLESV2_INDEXDATAMANAGER_H_ #include "common/angleutils.h" +#include "common/mathutil.h" +#include "libGLESv2/Error.h" + +#include <GLES2/gl2.h> namespace { @@ -24,21 +28,22 @@ class Buffer; namespace rx { +class IndexBufferInterface; class StaticIndexBufferInterface; class StreamingIndexBufferInterface; class IndexBuffer; -class BufferStorage; +class BufferD3D; class Renderer; struct TranslatedIndexData { - unsigned int minIndex; - unsigned int maxIndex; + RangeUI indexRange; unsigned int startIndex; unsigned int startOffset; // In bytes IndexBuffer *indexBuffer; - BufferStorage *storage; + BufferD3D *storage; + GLenum indexType; unsigned int serial; }; @@ -48,17 +53,17 @@ class IndexDataManager explicit IndexDataManager(Renderer *renderer); virtual ~IndexDataManager(); - GLenum prepareIndexData(GLenum type, GLsizei count, gl::Buffer *arrayElementBuffer, const GLvoid *indices, TranslatedIndexData *translated); - StaticIndexBufferInterface *getCountingIndices(GLsizei count); + gl::Error prepareIndexData(GLenum type, GLsizei count, gl::Buffer *arrayElementBuffer, const GLvoid *indices, TranslatedIndexData *translated); private: + gl::Error getStreamingIndexBuffer(GLenum destinationIndexType, IndexBufferInterface **outBuffer); + DISALLOW_COPY_AND_ASSIGN(IndexDataManager); Renderer *const mRenderer; StreamingIndexBufferInterface *mStreamingBufferShort; StreamingIndexBufferInterface *mStreamingBufferInt; - StaticIndexBufferInterface *mCountingBuffer; }; } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/MemoryBuffer.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/MemoryBuffer.cpp new file mode 100644 index 0000000000..2b5b09a324 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/MemoryBuffer.cpp @@ -0,0 +1,74 @@ +// +// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +#include "libGLESv2/renderer/d3d/MemoryBuffer.h" +#include "common/debug.h" + +#include <algorithm> +#include <cstdlib> + +namespace rx +{ + +MemoryBuffer::MemoryBuffer() + : mSize(0), + mData(NULL) +{ +} + +MemoryBuffer::~MemoryBuffer() +{ + free(mData); + mData = NULL; +} + +bool MemoryBuffer::resize(size_t size) +{ + if (size == 0) + { + free(mData); + mData = NULL; + mSize = 0; + } + else + { + uint8_t *newMemory = reinterpret_cast<uint8_t*>(malloc(sizeof(uint8_t) * size)); + if (newMemory == NULL) + { + return false; + } + + if (mData) + { + // Copy the intersection of the old data and the new data + std::copy(mData, mData + std::min(mSize, size), newMemory); + free(mData); + } + + mData = newMemory; + mSize = size; + } + + return true; +} + +size_t MemoryBuffer::size() const +{ + return mSize; +} + +const uint8_t *MemoryBuffer::data() const +{ + return mData; +} + +uint8_t *MemoryBuffer::data() +{ + ASSERT(mData); + return mData; +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/MemoryBuffer.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/MemoryBuffer.h new file mode 100644 index 0000000000..c65f79fe10 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/MemoryBuffer.h @@ -0,0 +1,36 @@ +// +// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +#ifndef LIBGLESV2_RENDERER_D3D_MEMORYBUFFER_H_ +#define LIBGLESV2_RENDERER_D3D_MEMORYBUFFER_H_ + +#include <cstddef> +#include <cstdint> + +namespace rx +{ + +class MemoryBuffer +{ + public: + MemoryBuffer(); + ~MemoryBuffer(); + + bool resize(size_t size); + size_t size() const; + bool empty() const { return mSize == 0; } + + const uint8_t *data() const; + uint8_t *data(); + + private: + size_t mSize; + uint8_t *mData; +}; + +} + +#endif // LIBGLESV2_RENDERER_D3D_MEMORYBUFFER_H diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ProgramD3D.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ProgramD3D.cpp new file mode 100644 index 0000000000..d7d97cc2bd --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ProgramD3D.cpp @@ -0,0 +1,205 @@ +// +// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// ProgramD3D.cpp: Defines the rx::ProgramD3D class which implements rx::ProgramImpl. + +#include "libGLESv2/renderer/d3d/ProgramD3D.h" + +#include "common/utilities.h" +#include "libGLESv2/ProgramBinary.h" +#include "libGLESv2/renderer/Renderer.h" +#include "libGLESv2/renderer/ShaderExecutable.h" +#include "libGLESv2/renderer/d3d/DynamicHLSL.h" +#include "libGLESv2/renderer/d3d/ShaderD3D.h" +#include "libGLESv2/main.h" + +namespace rx +{ + +ProgramD3D::ProgramD3D(rx::Renderer *renderer) + : ProgramImpl(), + mRenderer(renderer), + mDynamicHLSL(NULL), + mVertexWorkarounds(rx::ANGLE_D3D_WORKAROUND_NONE), + mPixelWorkarounds(rx::ANGLE_D3D_WORKAROUND_NONE), + mVertexUniformStorage(NULL), + mFragmentUniformStorage(NULL) +{ + mDynamicHLSL = new rx::DynamicHLSL(renderer); +} + +ProgramD3D::~ProgramD3D() +{ + reset(); + SafeDelete(mDynamicHLSL); +} + +ProgramD3D *ProgramD3D::makeProgramD3D(ProgramImpl *impl) +{ + ASSERT(HAS_DYNAMIC_TYPE(ProgramD3D*, impl)); + return static_cast<ProgramD3D*>(impl); +} + +const ProgramD3D *ProgramD3D::makeProgramD3D(const ProgramImpl *impl) +{ + ASSERT(HAS_DYNAMIC_TYPE(const ProgramD3D*, impl)); + return static_cast<const ProgramD3D*>(impl); +} + +bool ProgramD3D::load(gl::InfoLog &infoLog, gl::BinaryInputStream *stream) +{ + stream->readString(&mVertexHLSL); + stream->readInt(&mVertexWorkarounds); + stream->readString(&mPixelHLSL); + stream->readInt(&mPixelWorkarounds); + stream->readBool(&mUsesFragDepth); + + const size_t pixelShaderKeySize = stream->readInt<unsigned int>(); + mPixelShaderKey.resize(pixelShaderKeySize); + for (size_t pixelShaderKeyIndex = 0; pixelShaderKeyIndex < pixelShaderKeySize; pixelShaderKeyIndex++) + { + stream->readInt(&mPixelShaderKey[pixelShaderKeyIndex].type); + stream->readString(&mPixelShaderKey[pixelShaderKeyIndex].name); + stream->readString(&mPixelShaderKey[pixelShaderKeyIndex].source); + stream->readInt(&mPixelShaderKey[pixelShaderKeyIndex].outputIndex); + } + + return true; +} + +bool ProgramD3D::save(gl::BinaryOutputStream *stream) +{ + stream->writeString(mVertexHLSL); + stream->writeInt(mVertexWorkarounds); + stream->writeString(mPixelHLSL); + stream->writeInt(mPixelWorkarounds); + stream->writeInt(mUsesFragDepth); + + const std::vector<rx::PixelShaderOutputVariable> &pixelShaderKey = mPixelShaderKey; + stream->writeInt(pixelShaderKey.size()); + for (size_t pixelShaderKeyIndex = 0; pixelShaderKeyIndex < pixelShaderKey.size(); pixelShaderKeyIndex++) + { + const rx::PixelShaderOutputVariable &variable = pixelShaderKey[pixelShaderKeyIndex]; + stream->writeInt(variable.type); + stream->writeString(variable.name); + stream->writeString(variable.source); + stream->writeInt(variable.outputIndex); + } + + return true; +} + +rx::ShaderExecutable *ProgramD3D::getPixelExecutableForOutputLayout(gl::InfoLog &infoLog, const std::vector<GLenum> &outputSignature, + const std::vector<gl::LinkedVarying> &transformFeedbackLinkedVaryings, + bool separatedOutputBuffers) +{ + std::string finalPixelHLSL = mDynamicHLSL->generatePixelShaderForOutputSignature(mPixelHLSL, mPixelShaderKey, mUsesFragDepth, + outputSignature); + + // Generate new pixel executable + rx::ShaderExecutable *pixelExecutable = mRenderer->compileToExecutable(infoLog, finalPixelHLSL.c_str(), rx::SHADER_PIXEL, + transformFeedbackLinkedVaryings, separatedOutputBuffers, + mPixelWorkarounds); + + return pixelExecutable; +} + +rx::ShaderExecutable *ProgramD3D::getVertexExecutableForInputLayout(gl::InfoLog &infoLog, + const gl::VertexFormat inputLayout[gl::MAX_VERTEX_ATTRIBS], + const sh::Attribute shaderAttributes[], + const std::vector<gl::LinkedVarying> &transformFeedbackLinkedVaryings, + bool separatedOutputBuffers) +{ + // Generate new dynamic layout with attribute conversions + std::string finalVertexHLSL = mDynamicHLSL->generateVertexShaderForInputLayout(mVertexHLSL, inputLayout, shaderAttributes); + + // Generate new vertex executable + rx::ShaderExecutable *vertexExecutable = mRenderer->compileToExecutable(infoLog, finalVertexHLSL.c_str(), + rx::SHADER_VERTEX, + transformFeedbackLinkedVaryings, separatedOutputBuffers, + mVertexWorkarounds); + + return vertexExecutable; +} + +bool ProgramD3D::link(gl::InfoLog &infoLog, gl::Shader *fragmentShader, gl::Shader *vertexShader, + const std::vector<std::string> &transformFeedbackVaryings, int *registers, + std::vector<gl::LinkedVarying> *linkedVaryings, std::map<int, gl::VariableLocation> *outputVariables) +{ + rx::ShaderD3D *vertexShaderD3D = rx::ShaderD3D::makeShaderD3D(vertexShader->getImplementation()); + rx::ShaderD3D *fragmentShaderD3D = rx::ShaderD3D::makeShaderD3D(fragmentShader->getImplementation()); + + mPixelHLSL = fragmentShaderD3D->getTranslatedSource(); + mPixelWorkarounds = fragmentShaderD3D->getD3DWorkarounds(); + + mVertexHLSL = vertexShaderD3D->getTranslatedSource(); + mVertexWorkarounds = vertexShaderD3D->getD3DWorkarounds(); + + // Map the varyings to the register file + rx::VaryingPacking packing = { NULL }; + *registers = mDynamicHLSL->packVaryings(infoLog, packing, fragmentShaderD3D, vertexShaderD3D, transformFeedbackVaryings); + + if (*registers < 0) + { + return false; + } + + if (!gl::ProgramBinary::linkVaryings(infoLog, fragmentShader, vertexShader)) + { + return false; + } + + if (!mDynamicHLSL->generateShaderLinkHLSL(infoLog, *registers, packing, mPixelHLSL, mVertexHLSL, + fragmentShaderD3D, vertexShaderD3D, transformFeedbackVaryings, + linkedVaryings, outputVariables, &mPixelShaderKey, &mUsesFragDepth)) + { + return false; + } + + return true; +} + +void ProgramD3D::initializeUniformStorage(const std::vector<gl::LinkedUniform*> &uniforms) +{ + // Compute total default block size + unsigned int vertexRegisters = 0; + unsigned int fragmentRegisters = 0; + for (size_t uniformIndex = 0; uniformIndex < uniforms.size(); uniformIndex++) + { + const gl::LinkedUniform &uniform = *uniforms[uniformIndex]; + + if (!gl::IsSampler(uniform.type)) + { + if (uniform.isReferencedByVertexShader()) + { + vertexRegisters = std::max(vertexRegisters, uniform.vsRegisterIndex + uniform.registerCount); + } + if (uniform.isReferencedByFragmentShader()) + { + fragmentRegisters = std::max(fragmentRegisters, uniform.psRegisterIndex + uniform.registerCount); + } + } + } + + mVertexUniformStorage = mRenderer->createUniformStorage(vertexRegisters * 16u); + mFragmentUniformStorage = mRenderer->createUniformStorage(fragmentRegisters * 16u); +} + +void ProgramD3D::reset() +{ + mVertexHLSL.clear(); + mVertexWorkarounds = rx::ANGLE_D3D_WORKAROUND_NONE; + + mPixelHLSL.clear(); + mPixelWorkarounds = rx::ANGLE_D3D_WORKAROUND_NONE; + mUsesFragDepth = false; + mPixelShaderKey.clear(); + + SafeDelete(mVertexUniformStorage); + SafeDelete(mFragmentUniformStorage); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ProgramD3D.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ProgramD3D.h new file mode 100644 index 0000000000..d645c57daa --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ProgramD3D.h @@ -0,0 +1,87 @@ +// +// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// ProgramD3D.h: Defines the rx::ProgramD3D class which implements rx::ProgramImpl. + +#ifndef LIBGLESV2_RENDERER_PROGRAMD3D_H_ +#define LIBGLESV2_RENDERER_PROGRAMD3D_H_ + +#include "libGLESv2/renderer/ProgramImpl.h" + +#include <string> +#include <vector> + +namespace gl +{ +struct LinkedUniform; +struct VariableLocation; +struct VertexFormat; +} + +namespace rx +{ + +class UniformStorage; + +class ProgramD3D : public ProgramImpl +{ + public: + ProgramD3D(rx::Renderer *renderer); + virtual ~ProgramD3D(); + + static ProgramD3D *makeProgramD3D(ProgramImpl *impl); + static const ProgramD3D *makeProgramD3D(const ProgramImpl *impl); + + Renderer *getRenderer() { return mRenderer; } + DynamicHLSL *getDynamicHLSL() { return mDynamicHLSL; } + const std::vector<rx::PixelShaderOutputVariable> &getPixelShaderKey() { return mPixelShaderKey; } + + GLenum getBinaryFormat() { return GL_PROGRAM_BINARY_ANGLE; } + bool load(gl::InfoLog &infoLog, gl::BinaryInputStream *stream); + bool save(gl::BinaryOutputStream *stream); + + ShaderExecutable *getPixelExecutableForOutputLayout(gl::InfoLog &infoLog, const std::vector<GLenum> &outputSignature, + const std::vector<gl::LinkedVarying> &transformFeedbackLinkedVaryings, + bool separatedOutputBuffers); + ShaderExecutable *getVertexExecutableForInputLayout(gl::InfoLog &infoLog, + const gl::VertexFormat inputLayout[gl::MAX_VERTEX_ATTRIBS], + const sh::Attribute shaderAttributes[], + const std::vector<gl::LinkedVarying> &transformFeedbackLinkedVaryings, + bool separatedOutputBuffers); + + bool link(gl::InfoLog &infoLog, gl::Shader *fragmentShader, gl::Shader *vertexShader, + const std::vector<std::string> &transformFeedbackVaryings, int *registers, + std::vector<gl::LinkedVarying> *linkedVaryings, std::map<int, gl::VariableLocation> *outputVariables); + + // D3D only + void initializeUniformStorage(const std::vector<gl::LinkedUniform*> &uniforms); + + const UniformStorage &getVertexUniformStorage() const { return *mVertexUniformStorage; } + const UniformStorage &getFragmentUniformStorage() const { return *mFragmentUniformStorage; } + + void reset(); + + private: + DISALLOW_COPY_AND_ASSIGN(ProgramD3D); + + Renderer *mRenderer; + DynamicHLSL *mDynamicHLSL; + + std::string mVertexHLSL; + rx::D3DWorkaroundType mVertexWorkarounds; + + std::string mPixelHLSL; + rx::D3DWorkaroundType mPixelWorkarounds; + bool mUsesFragDepth; + std::vector<rx::PixelShaderOutputVariable> mPixelShaderKey; + + UniformStorage *mVertexUniformStorage; + UniformStorage *mFragmentUniformStorage; +}; + +} + +#endif // LIBGLESV2_RENDERER_PROGRAMD3D_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ShaderD3D.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ShaderD3D.cpp new file mode 100644 index 0000000000..c472113eba --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ShaderD3D.cpp @@ -0,0 +1,457 @@ +// +// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// ShaderD3D.cpp: Defines the rx::ShaderD3D class which implements rx::ShaderImpl. + +#include "libGLESv2/renderer/d3d/ShaderD3D.h" +#include "libGLESv2/renderer/Renderer.h" +#include "libGLESv2/Shader.h" +#include "libGLESv2/main.h" + +#include "common/utilities.h" + +namespace rx +{ + +template <typename VarT> +void FilterInactiveVariables(std::vector<VarT> *variableList) +{ + ASSERT(variableList); + + for (size_t varIndex = 0; varIndex < variableList->size();) + { + if (!(*variableList)[varIndex].staticUse) + { + variableList->erase(variableList->begin() + varIndex); + } + else + { + varIndex++; + } + } +} + +void *ShaderD3D::mFragmentCompiler = NULL; +void *ShaderD3D::mVertexCompiler = NULL; + +template <typename VarT> +const std::vector<VarT> *GetShaderVariables(const std::vector<VarT> *variableList) +{ + ASSERT(variableList); + return variableList; +} + +ShaderD3D::ShaderD3D(GLenum type, rx::Renderer *renderer) + : mType(type), + mRenderer(renderer), + mShaderVersion(100) +{ + uncompile(); + initializeCompiler(); +} + +ShaderD3D::~ShaderD3D() +{ +} + +ShaderD3D *ShaderD3D::makeShaderD3D(ShaderImpl *impl) +{ + ASSERT(HAS_DYNAMIC_TYPE(ShaderD3D*, impl)); + return static_cast<ShaderD3D*>(impl); +} + +const ShaderD3D *ShaderD3D::makeShaderD3D(const ShaderImpl *impl) +{ + ASSERT(HAS_DYNAMIC_TYPE(const ShaderD3D*, impl)); + return static_cast<const ShaderD3D*>(impl); +} + +// Perform a one-time initialization of the shader compiler (or after being destructed by releaseCompiler) +void ShaderD3D::initializeCompiler() +{ + if (!mFragmentCompiler) + { + int result = ShInitialize(); + + if (result) + { + ShShaderOutput hlslVersion = (mRenderer->getMajorShaderModel() >= 4) ? SH_HLSL11_OUTPUT : SH_HLSL9_OUTPUT; + + ShBuiltInResources resources; + ShInitBuiltInResources(&resources); + + // TODO(geofflang): use context's caps + const gl::Caps &caps = mRenderer->getRendererCaps(); + const gl::Extensions &extensions = mRenderer->getRendererExtensions(); + + resources.MaxVertexAttribs = caps.maxVertexAttributes; + resources.MaxVertexUniformVectors = caps.maxVertexUniformVectors; + resources.MaxVaryingVectors = caps.maxVaryingVectors; + resources.MaxVertexTextureImageUnits = caps.maxVertexTextureImageUnits; + resources.MaxCombinedTextureImageUnits = caps.maxCombinedTextureImageUnits; + resources.MaxTextureImageUnits = caps.maxTextureImageUnits; + resources.MaxFragmentUniformVectors = caps.maxFragmentUniformVectors; + resources.MaxDrawBuffers = caps.maxDrawBuffers; + resources.OES_standard_derivatives = extensions.standardDerivatives; + resources.EXT_draw_buffers = extensions.drawBuffers; + resources.EXT_shader_texture_lod = 1; + // resources.OES_EGL_image_external = mRenderer->getShareHandleSupport() ? 1 : 0; // TODO: commented out until the extension is actually supported. + resources.FragmentPrecisionHigh = 1; // Shader Model 2+ always supports FP24 (s16e7) which corresponds to highp + resources.EXT_frag_depth = 1; // Shader Model 2+ always supports explicit depth output + // GLSL ES 3.0 constants + resources.MaxVertexOutputVectors = caps.maxVertexOutputComponents / 4; + resources.MaxFragmentInputVectors = caps.maxFragmentInputComponents / 4; + resources.MinProgramTexelOffset = caps.minProgramTexelOffset; + resources.MaxProgramTexelOffset = caps.maxProgramTexelOffset; + + mFragmentCompiler = ShConstructCompiler(GL_FRAGMENT_SHADER, SH_GLES2_SPEC, hlslVersion, &resources); + mVertexCompiler = ShConstructCompiler(GL_VERTEX_SHADER, SH_GLES2_SPEC, hlslVersion, &resources); + } + } +} + +void ShaderD3D::releaseCompiler() +{ + ShDestruct(mFragmentCompiler); + ShDestruct(mVertexCompiler); + + mFragmentCompiler = NULL; + mVertexCompiler = NULL; + + ShFinalize(); +} + +void ShaderD3D::parseVaryings(void *compiler) +{ + if (!mHlsl.empty()) + { + const std::vector<sh::Varying> *varyings = ShGetVaryings(compiler); + ASSERT(varyings); + + for (size_t varyingIndex = 0; varyingIndex < varyings->size(); varyingIndex++) + { + mVaryings.push_back(gl::PackedVarying((*varyings)[varyingIndex])); + } + + mUsesMultipleRenderTargets = mHlsl.find("GL_USES_MRT") != std::string::npos; + mUsesFragColor = mHlsl.find("GL_USES_FRAG_COLOR") != std::string::npos; + mUsesFragData = mHlsl.find("GL_USES_FRAG_DATA") != std::string::npos; + mUsesFragCoord = mHlsl.find("GL_USES_FRAG_COORD") != std::string::npos; + mUsesFrontFacing = mHlsl.find("GL_USES_FRONT_FACING") != std::string::npos; + mUsesPointSize = mHlsl.find("GL_USES_POINT_SIZE") != std::string::npos; + mUsesPointCoord = mHlsl.find("GL_USES_POINT_COORD") != std::string::npos; + mUsesDepthRange = mHlsl.find("GL_USES_DEPTH_RANGE") != std::string::npos; + mUsesFragDepth = mHlsl.find("GL_USES_FRAG_DEPTH") != std::string::npos; + mUsesDiscardRewriting = mHlsl.find("ANGLE_USES_DISCARD_REWRITING") != std::string::npos; + mUsesNestedBreak = mHlsl.find("ANGLE_USES_NESTED_BREAK") != std::string::npos; + } +} + +void ShaderD3D::resetVaryingsRegisterAssignment() +{ + for (size_t varyingIndex = 0; varyingIndex < mVaryings.size(); varyingIndex++) + { + mVaryings[varyingIndex].resetRegisterAssignment(); + } +} + +// initialize/clean up previous state +void ShaderD3D::uncompile() +{ + // set by compileToHLSL + mHlsl.clear(); + mInfoLog.clear(); + + mUsesMultipleRenderTargets = false; + mUsesFragColor = false; + mUsesFragData = false; + mUsesFragCoord = false; + mUsesFrontFacing = false; + mUsesPointSize = false; + mUsesPointCoord = false; + mUsesDepthRange = false; + mUsesFragDepth = false; + mShaderVersion = 100; + mUsesDiscardRewriting = false; + mUsesNestedBreak = false; + + mVaryings.clear(); + mUniforms.clear(); + mInterfaceBlocks.clear(); + mActiveAttributes.clear(); + mActiveOutputVariables.clear(); +} + +void ShaderD3D::compileToHLSL(void *compiler, const std::string &source) +{ + // ensure the compiler is loaded + initializeCompiler(); + + int compileOptions = (SH_OBJECT_CODE | SH_VARIABLES); + std::string sourcePath; + if (gl::perfActive()) + { + sourcePath = getTempPath(); + writeFile(sourcePath.c_str(), source.c_str(), source.length()); + compileOptions |= SH_LINE_DIRECTIVES; + } + + int result; + if (sourcePath.empty()) + { + const char* sourceStrings[] = + { + source.c_str(), + }; + + result = ShCompile(compiler, sourceStrings, ArraySize(sourceStrings), compileOptions); + } + else + { + const char* sourceStrings[] = + { + sourcePath.c_str(), + source.c_str(), + }; + + result = ShCompile(compiler, sourceStrings, ArraySize(sourceStrings), compileOptions | SH_SOURCE_PATH); + } + + size_t shaderVersion = 100; + ShGetInfo(compiler, SH_SHADER_VERSION, &shaderVersion); + + mShaderVersion = static_cast<int>(shaderVersion); + + if (shaderVersion == 300 && mRenderer->getCurrentClientVersion() < 3) + { + mInfoLog = "GLSL ES 3.00 is not supported by OpenGL ES 2.0 contexts"; + TRACE("\n%s", mInfoLog.c_str()); + } + else if (result) + { + size_t objCodeLen = 0; + ShGetInfo(compiler, SH_OBJECT_CODE_LENGTH, &objCodeLen); + + char* outputHLSL = new char[objCodeLen]; + ShGetObjectCode(compiler, outputHLSL); + +#ifdef _DEBUG + std::ostringstream hlslStream; + hlslStream << "// GLSL\n"; + hlslStream << "//\n"; + + size_t curPos = 0; + while (curPos != std::string::npos) + { + size_t nextLine = source.find("\n", curPos); + size_t len = (nextLine == std::string::npos) ? std::string::npos : (nextLine - curPos + 1); + + hlslStream << "// " << source.substr(curPos, len); + + curPos = (nextLine == std::string::npos) ? std::string::npos : (nextLine + 1); + } + hlslStream << "\n\n"; + hlslStream << outputHLSL; + mHlsl = hlslStream.str(); +#else + mHlsl = outputHLSL; +#endif + + SafeDeleteArray(outputHLSL); + + mUniforms = *GetShaderVariables(ShGetUniforms(compiler)); + + for (size_t uniformIndex = 0; uniformIndex < mUniforms.size(); uniformIndex++) + { + const sh::Uniform &uniform = mUniforms[uniformIndex]; + + if (uniform.staticUse) + { + unsigned int index = -1; + bool result = ShGetUniformRegister(compiler, uniform.name.c_str(), &index); + UNUSED_ASSERTION_VARIABLE(result); + ASSERT(result); + + mUniformRegisterMap[uniform.name] = index; + } + } + + mInterfaceBlocks = *GetShaderVariables(ShGetInterfaceBlocks(compiler)); + + for (size_t blockIndex = 0; blockIndex < mInterfaceBlocks.size(); blockIndex++) + { + const sh::InterfaceBlock &interfaceBlock = mInterfaceBlocks[blockIndex]; + + if (interfaceBlock.staticUse) + { + unsigned int index = -1; + bool result = ShGetInterfaceBlockRegister(compiler, interfaceBlock.name.c_str(), &index); + UNUSED_ASSERTION_VARIABLE(result); + ASSERT(result); + + mInterfaceBlockRegisterMap[interfaceBlock.name] = index; + } + } + } + else + { + size_t infoLogLen = 0; + ShGetInfo(compiler, SH_INFO_LOG_LENGTH, &infoLogLen); + + char* infoLog = new char[infoLogLen]; + ShGetInfoLog(compiler, infoLog); + mInfoLog = infoLog; + + TRACE("\n%s", mInfoLog.c_str()); + } +} + +rx::D3DWorkaroundType ShaderD3D::getD3DWorkarounds() const +{ + if (mUsesDiscardRewriting) + { + // ANGLE issue 486: + // Work-around a D3D9 compiler bug that presents itself when using conditional discard, by disabling optimization + return rx::ANGLE_D3D_WORKAROUND_SKIP_OPTIMIZATION; + } + + if (mUsesNestedBreak) + { + // ANGLE issue 603: + // Work-around a D3D9 compiler bug that presents itself when using break in a nested loop, by maximizing optimization + // We want to keep the use of ANGLE_D3D_WORKAROUND_MAX_OPTIMIZATION minimal to prevent hangs, so usesDiscard takes precedence + return rx::ANGLE_D3D_WORKAROUND_MAX_OPTIMIZATION; + } + + return rx::ANGLE_D3D_WORKAROUND_NONE; +} + +// true if varying x has a higher priority in packing than y +bool ShaderD3D::compareVarying(const gl::PackedVarying &x, const gl::PackedVarying &y) +{ + if (x.type == y.type) + { + return x.arraySize > y.arraySize; + } + + // Special case for handling structs: we sort these to the end of the list + if (x.type == GL_STRUCT_ANGLEX) + { + return false; + } + + if (y.type == GL_STRUCT_ANGLEX) + { + return true; + } + + return gl::VariableSortOrder(x.type) < gl::VariableSortOrder(y.type); +} + +unsigned int ShaderD3D::getUniformRegister(const std::string &uniformName) const +{ + ASSERT(mUniformRegisterMap.count(uniformName) > 0); + return mUniformRegisterMap.find(uniformName)->second; +} + +unsigned int ShaderD3D::getInterfaceBlockRegister(const std::string &blockName) const +{ + ASSERT(mInterfaceBlockRegisterMap.count(blockName) > 0); + return mInterfaceBlockRegisterMap.find(blockName)->second; +} + +void *ShaderD3D::getCompiler() +{ + if (mType == GL_VERTEX_SHADER) + { + return mVertexCompiler; + } + else + { + ASSERT(mType == GL_FRAGMENT_SHADER); + return mFragmentCompiler; + } +} + +ShShaderOutput ShaderD3D::getCompilerOutputType(GLenum shader) +{ + void *compiler = NULL; + + switch (shader) + { + case GL_VERTEX_SHADER: compiler = mVertexCompiler; break; + case GL_FRAGMENT_SHADER: compiler = mFragmentCompiler; break; + default: UNREACHABLE(); return SH_HLSL9_OUTPUT; + } + + size_t outputType = 0; + ShGetInfo(compiler, SH_OUTPUT_TYPE, &outputType); + + return static_cast<ShShaderOutput>(outputType); +} + +bool ShaderD3D::compile(const std::string &source) +{ + uncompile(); + + void *compiler = getCompiler(); + + compileToHLSL(compiler, source); + + if (mType == GL_VERTEX_SHADER) + { + parseAttributes(compiler); + } + + parseVaryings(compiler); + + if (mType == GL_FRAGMENT_SHADER) + { + std::sort(mVaryings.begin(), mVaryings.end(), compareVarying); + + const std::string &hlsl = getTranslatedSource(); + if (!hlsl.empty()) + { + mActiveOutputVariables = *GetShaderVariables(ShGetOutputVariables(compiler)); + FilterInactiveVariables(&mActiveOutputVariables); + } + } + + return !getTranslatedSource().empty(); +} + +void ShaderD3D::parseAttributes(void *compiler) +{ + const std::string &hlsl = getTranslatedSource(); + if (!hlsl.empty()) + { + mActiveAttributes = *GetShaderVariables(ShGetAttributes(compiler)); + FilterInactiveVariables(&mActiveAttributes); + } +} + +int ShaderD3D::getSemanticIndex(const std::string &attributeName) const +{ + if (!attributeName.empty()) + { + int semanticIndex = 0; + for (size_t attributeIndex = 0; attributeIndex < mActiveAttributes.size(); attributeIndex++) + { + const sh::ShaderVariable &attribute = mActiveAttributes[attributeIndex]; + + if (attribute.name == attributeName) + { + return semanticIndex; + } + + semanticIndex += gl::VariableRegisterCount(attribute.type); + } + } + + return -1; +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ShaderD3D.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ShaderD3D.h new file mode 100644 index 0000000000..40e64cf36c --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/ShaderD3D.h @@ -0,0 +1,94 @@ +// +// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// ShaderD3D.h: Defines the rx::ShaderD3D class which implements rx::ShaderImpl. + +#ifndef LIBGLESV2_RENDERER_SHADERD3D_H_ +#define LIBGLESV2_RENDERER_SHADERD3D_H_ + +#include "libGLESv2/renderer/ShaderImpl.h" +#include "libGLESv2/Shader.h" + +#include <map> + +namespace rx +{ +class DynamicHLSL; +class Renderer; + +class ShaderD3D : public ShaderImpl +{ + friend class DynamicHLSL; + + public: + ShaderD3D(GLenum type, rx::Renderer *renderer); + virtual ~ShaderD3D(); + + static ShaderD3D *makeShaderD3D(ShaderImpl *impl); + static const ShaderD3D *makeShaderD3D(const ShaderImpl *impl); + + // ShaderImpl implementation + const std::string &getInfoLog() const { return mInfoLog; } + const std::string &getTranslatedSource() const { return mHlsl; } + + // D3D-specific methods + virtual void uncompile(); + void resetVaryingsRegisterAssignment(); + unsigned int getUniformRegister(const std::string &uniformName) const; + unsigned int getInterfaceBlockRegister(const std::string &blockName) const; + int getSemanticIndex(const std::string &attributeName) const; + + rx::D3DWorkaroundType getD3DWorkarounds() const; + int getShaderVersion() const { return mShaderVersion; } + bool usesDepthRange() const { return mUsesDepthRange; } + bool usesPointSize() const { return mUsesPointSize; } + + static void releaseCompiler(); + static ShShaderOutput getCompilerOutputType(GLenum shader); + + virtual bool compile(const std::string &source); + + private: + DISALLOW_COPY_AND_ASSIGN(ShaderD3D); + + void compileToHLSL(void *compiler, const std::string &source); + void parseVaryings(void *compiler); + + void initializeCompiler(); + void parseAttributes(void *compiler); + void *getCompiler(); + + static bool compareVarying(const gl::PackedVarying &x, const gl::PackedVarying &y); + + static void *mFragmentCompiler; + static void *mVertexCompiler; + + GLenum mType; + rx::Renderer *mRenderer; + + int mShaderVersion; + + bool mUsesMultipleRenderTargets; + bool mUsesFragColor; + bool mUsesFragData; + bool mUsesFragCoord; + bool mUsesFrontFacing; + bool mUsesPointSize; + bool mUsesPointCoord; + bool mUsesDepthRange; + bool mUsesFragDepth; + bool mUsesDiscardRewriting; + bool mUsesNestedBreak; + + std::string mHlsl; + std::string mInfoLog; + std::map<std::string, unsigned int> mUniformRegisterMap; + std::map<std::string, unsigned int> mInterfaceBlockRegisterMap; +}; + +} + +#endif // LIBGLESV2_RENDERER_SHADERD3D_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TextureD3D.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TextureD3D.cpp new file mode 100644 index 0000000000..96c84977cb --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TextureD3D.cpp @@ -0,0 +1,2260 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// TextureD3D.cpp: Implementations of the Texture interfaces shared betweeen the D3D backends. + +#include "libGLESv2/renderer/d3d/TextureD3D.h" +#include "libGLESv2/renderer/d3d/TextureStorage.h" +#include "libGLESv2/renderer/d3d/ImageD3D.h" +#include "libGLESv2/Buffer.h" +#include "libGLESv2/Framebuffer.h" +#include "libGLESv2/Texture.h" +#include "libGLESv2/main.h" +#include "libGLESv2/formatutils.h" +#include "libGLESv2/renderer/BufferImpl.h" +#include "libGLESv2/renderer/RenderTarget.h" +#include "libGLESv2/renderer/Renderer.h" + +#include "libEGL/Surface.h" + +#include "common/mathutil.h" +#include "common/utilities.h" + +namespace rx +{ + +bool IsRenderTargetUsage(GLenum usage) +{ + return (usage == GL_FRAMEBUFFER_ATTACHMENT_ANGLE); +} + +TextureD3D::TextureD3D(Renderer *renderer) + : mRenderer(renderer), + mUsage(GL_NONE), + mDirtyImages(true), + mImmutable(false) +{ +} + +TextureD3D::~TextureD3D() +{ +} + +TextureD3D *TextureD3D::makeTextureD3D(TextureImpl *texture) +{ + ASSERT(HAS_DYNAMIC_TYPE(TextureD3D*, texture)); + return static_cast<TextureD3D*>(texture); +} + +TextureStorage *TextureD3D::getNativeTexture() +{ + // ensure the underlying texture is created + initializeStorage(false); + + TextureStorage *storage = getBaseLevelStorage(); + if (storage) + { + updateStorage(); + } + + return storage; +} + +GLint TextureD3D::getBaseLevelWidth() const +{ + const Image *baseImage = getBaseLevelImage(); + return (baseImage ? baseImage->getWidth() : 0); +} + +GLint TextureD3D::getBaseLevelHeight() const +{ + const Image *baseImage = getBaseLevelImage(); + return (baseImage ? baseImage->getHeight() : 0); +} + +GLint TextureD3D::getBaseLevelDepth() const +{ + const Image *baseImage = getBaseLevelImage(); + return (baseImage ? baseImage->getDepth() : 0); +} + +// Note: "base level image" is loosely defined to be any image from the base level, +// where in the base of 2D array textures and cube maps there are several. Don't use +// the base level image for anything except querying texture format and size. +GLenum TextureD3D::getBaseLevelInternalFormat() const +{ + const Image *baseImage = getBaseLevelImage(); + return (baseImage ? baseImage->getInternalFormat() : GL_NONE); +} + +void TextureD3D::setImage(const gl::PixelUnpackState &unpack, GLenum type, const void *pixels, Image *image) +{ + // No-op + if (image->getWidth() == 0 || image->getHeight() == 0 || image->getDepth() == 0) + { + return; + } + + // We no longer need the "GLenum format" parameter to TexImage to determine what data format "pixels" contains. + // From our image internal format we know how many channels to expect, and "type" gives the format of pixel's components. + const void *pixelData = pixels; + + if (unpack.pixelBuffer.id() != 0) + { + // Do a CPU readback here, if we have an unpack buffer bound and the fast GPU path is not supported + gl::Buffer *pixelBuffer = unpack.pixelBuffer.get(); + ptrdiff_t offset = reinterpret_cast<ptrdiff_t>(pixels); + // TODO: setImage/subImage is the only place outside of renderer that asks for a buffers raw data. + // This functionality should be moved into renderer and the getData method of BufferImpl removed. + const void *bufferData = pixelBuffer->getImplementation()->getData(); + pixelData = static_cast<const unsigned char *>(bufferData) + offset; + } + + if (pixelData != NULL) + { + image->loadData(0, 0, 0, image->getWidth(), image->getHeight(), image->getDepth(), unpack.alignment, type, pixelData); + mDirtyImages = true; + } +} + +bool TextureD3D::subImage(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, + GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels, const gl::ImageIndex &index) +{ + const void *pixelData = pixels; + + // CPU readback & copy where direct GPU copy is not supported + if (unpack.pixelBuffer.id() != 0) + { + gl::Buffer *pixelBuffer = unpack.pixelBuffer.get(); + ptrdiff_t offset = reinterpret_cast<ptrdiff_t>(pixels); + // TODO: setImage/subImage is the only place outside of renderer that asks for a buffers raw data. + // This functionality should be moved into renderer and the getData method of BufferImpl removed. + const void *bufferData = pixelBuffer->getImplementation()->getData(); + pixelData = static_cast<const unsigned char *>(bufferData) + offset; + } + + if (pixelData != NULL) + { + Image *image = getImage(index); + ASSERT(image); + + image->loadData(xoffset, yoffset, zoffset, width, height, depth, unpack.alignment, type, pixelData); + mDirtyImages = true; + } + + return true; +} + +void TextureD3D::setCompressedImage(GLsizei imageSize, const void *pixels, Image *image) +{ + if (pixels != NULL) + { + image->loadCompressedData(0, 0, 0, image->getWidth(), image->getHeight(), image->getDepth(), pixels); + mDirtyImages = true; + } +} + +bool TextureD3D::subImageCompressed(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, + GLenum format, GLsizei imageSize, const void *pixels, Image *image) +{ + if (pixels != NULL) + { + image->loadCompressedData(xoffset, yoffset, zoffset, width, height, depth, pixels); + mDirtyImages = true; + } + + return true; +} + +bool TextureD3D::isFastUnpackable(const gl::PixelUnpackState &unpack, GLenum sizedInternalFormat) +{ + return unpack.pixelBuffer.id() != 0 && mRenderer->supportsFastCopyBufferToTexture(sizedInternalFormat); +} + +bool TextureD3D::fastUnpackPixels(const gl::PixelUnpackState &unpack, const void *pixels, const gl::Box &destArea, + GLenum sizedInternalFormat, GLenum type, RenderTarget *destRenderTarget) +{ + if (destArea.width <= 0 && destArea.height <= 0 && destArea.depth <= 0) + { + return true; + } + + // In order to perform the fast copy through the shader, we must have the right format, and be able + // to create a render target. + ASSERT(mRenderer->supportsFastCopyBufferToTexture(sizedInternalFormat)); + + ptrdiff_t offset = reinterpret_cast<ptrdiff_t>(pixels); + + return mRenderer->fastCopyBufferToTexture(unpack, offset, destRenderTarget, sizedInternalFormat, type, destArea); +} + +GLint TextureD3D::creationLevels(GLsizei width, GLsizei height, GLsizei depth) const +{ + if ((gl::isPow2(width) && gl::isPow2(height) && gl::isPow2(depth)) || mRenderer->getRendererExtensions().textureNPOT) + { + // Maximum number of levels + return gl::log2(std::max(std::max(width, height), depth)) + 1; + } + else + { + // OpenGL ES 2.0 without GL_OES_texture_npot does not permit NPOT mipmaps. + return 1; + } +} + +int TextureD3D::mipLevels() const +{ + return gl::log2(std::max(std::max(getBaseLevelWidth(), getBaseLevelHeight()), getBaseLevelDepth())) + 1; +} + + +TextureD3D_2D::TextureD3D_2D(Renderer *renderer) + : TextureD3D(renderer), + mTexStorage(NULL) +{ + for (int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; ++i) + { + mImageArray[i] = ImageD3D::makeImageD3D(renderer->createImage()); + } +} + +TextureD3D_2D::~TextureD3D_2D() +{ + // Delete the Images before the TextureStorage. + // Images might be relying on the TextureStorage for some of their data. + // If TextureStorage is deleted before the Images, then their data will be wastefully copied back from the GPU before we delete the Images. + for (int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; ++i) + { + delete mImageArray[i]; + } + + SafeDelete(mTexStorage); +} + +Image *TextureD3D_2D::getImage(int level, int layer) const +{ + ASSERT(level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + ASSERT(layer == 0); + return mImageArray[level]; +} + +Image *TextureD3D_2D::getImage(const gl::ImageIndex &index) const +{ + ASSERT(index.mipIndex < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + ASSERT(!index.hasLayer()); + ASSERT(index.type == GL_TEXTURE_2D); + return mImageArray[index.mipIndex]; +} + +GLsizei TextureD3D_2D::getLayerCount(int level) const +{ + ASSERT(level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + return 1; +} + +GLsizei TextureD3D_2D::getWidth(GLint level) const +{ + if (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + return mImageArray[level]->getWidth(); + else + return 0; +} + +GLsizei TextureD3D_2D::getHeight(GLint level) const +{ + if (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + return mImageArray[level]->getHeight(); + else + return 0; +} + +GLenum TextureD3D_2D::getInternalFormat(GLint level) const +{ + if (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + return mImageArray[level]->getInternalFormat(); + else + return GL_NONE; +} + +GLenum TextureD3D_2D::getActualFormat(GLint level) const +{ + if (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + return mImageArray[level]->getActualFormat(); + else + return GL_NONE; +} + +bool TextureD3D_2D::isDepth(GLint level) const +{ + return gl::GetInternalFormatInfo(getInternalFormat(level)).depthBits > 0; +} + +void TextureD3D_2D::setImage(GLenum target, GLint level, GLsizei width, GLsizei height, GLsizei depth, GLenum internalFormat, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels) +{ + ASSERT(target == GL_TEXTURE_2D && depth == 1); + + GLenum sizedInternalFormat = gl::GetSizedInternalFormat(internalFormat, type); + + bool fastUnpacked = false; + + redefineImage(level, sizedInternalFormat, width, height); + + // Attempt a fast gpu copy of the pixel data to the surface + if (isFastUnpackable(unpack, sizedInternalFormat) && isLevelComplete(level)) + { + gl::ImageIndex index = gl::ImageIndex::Make2D(level); + + // Will try to create RT storage if it does not exist + RenderTarget *destRenderTarget = getRenderTarget(index); + gl::Box destArea(0, 0, 0, getWidth(level), getHeight(level), 1); + + if (destRenderTarget && fastUnpackPixels(unpack, pixels, destArea, sizedInternalFormat, type, destRenderTarget)) + { + // Ensure we don't overwrite our newly initialized data + mImageArray[level]->markClean(); + + fastUnpacked = true; + } + } + + if (!fastUnpacked) + { + TextureD3D::setImage(unpack, type, pixels, mImageArray[level]); + } +} + +void TextureD3D_2D::setCompressedImage(GLenum target, GLint level, GLenum format, GLsizei width, GLsizei height, GLsizei depth, GLsizei imageSize, const void *pixels) +{ + ASSERT(target == GL_TEXTURE_2D && depth == 1); + + // compressed formats don't have separate sized internal formats-- we can just use the compressed format directly + redefineImage(level, format, width, height); + + TextureD3D::setCompressedImage(imageSize, pixels, mImageArray[level]); +} + +void TextureD3D_2D::subImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels) +{ + ASSERT(target == GL_TEXTURE_2D && depth == 1 && zoffset == 0); + + bool fastUnpacked = false; + + gl::ImageIndex index = gl::ImageIndex::Make2D(level); + if (isFastUnpackable(unpack, getInternalFormat(level)) && isLevelComplete(level)) + { + RenderTarget *renderTarget = getRenderTarget(index); + gl::Box destArea(xoffset, yoffset, 0, width, height, 1); + + if (renderTarget && fastUnpackPixels(unpack, pixels, destArea, getInternalFormat(level), type, renderTarget)) + { + // Ensure we don't overwrite our newly initialized data + mImageArray[level]->markClean(); + + fastUnpacked = true; + } + } + + if (!fastUnpacked && TextureD3D::subImage(xoffset, yoffset, 0, width, height, 1, format, type, unpack, pixels, index)) + { + commitRect(level, xoffset, yoffset, width, height); + } +} + +void TextureD3D_2D::subImageCompressed(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void *pixels) +{ + ASSERT(target == GL_TEXTURE_2D && depth == 1 && zoffset == 0); + + if (TextureD3D::subImageCompressed(xoffset, yoffset, 0, width, height, 1, format, imageSize, pixels, mImageArray[level])) + { + commitRect(level, xoffset, yoffset, width, height); + } +} + +void TextureD3D_2D::copyImage(GLenum target, GLint level, GLenum format, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) +{ + ASSERT(target == GL_TEXTURE_2D); + + GLenum sizedInternalFormat = gl::GetSizedInternalFormat(format, GL_UNSIGNED_BYTE); + redefineImage(level, sizedInternalFormat, width, height); + + if (!mImageArray[level]->isRenderableFormat()) + { + mImageArray[level]->copy(0, 0, 0, x, y, width, height, source); + mDirtyImages = true; + } + else + { + ensureRenderTarget(); + mImageArray[level]->markClean(); + + if (width != 0 && height != 0 && isValidLevel(level)) + { + gl::Rectangle sourceRect; + sourceRect.x = x; + sourceRect.width = width; + sourceRect.y = y; + sourceRect.height = height; + + mRenderer->copyImage2D(source, sourceRect, format, 0, 0, mTexStorage, level); + } + } +} + +void TextureD3D_2D::copySubImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) +{ + ASSERT(target == GL_TEXTURE_2D && zoffset == 0); + + // can only make our texture storage to a render target if level 0 is defined (with a width & height) and + // the current level we're copying to is defined (with appropriate format, width & height) + bool canCreateRenderTarget = isLevelComplete(level) && isLevelComplete(0); + + if (!mImageArray[level]->isRenderableFormat() || (!mTexStorage && !canCreateRenderTarget)) + { + mImageArray[level]->copy(xoffset, yoffset, 0, x, y, width, height, source); + mDirtyImages = true; + } + else + { + ensureRenderTarget(); + + if (isValidLevel(level)) + { + updateStorageLevel(level); + + gl::Rectangle sourceRect; + sourceRect.x = x; + sourceRect.width = width; + sourceRect.y = y; + sourceRect.height = height; + + mRenderer->copyImage2D(source, sourceRect, + gl::GetInternalFormatInfo(getBaseLevelInternalFormat()).format, + xoffset, yoffset, mTexStorage, level); + } + } +} + +void TextureD3D_2D::storage(GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth) +{ + ASSERT(target == GL_TEXTURE_2D && depth == 1); + + for (int level = 0; level < levels; level++) + { + GLsizei levelWidth = std::max(1, width >> level); + GLsizei levelHeight = std::max(1, height >> level); + mImageArray[level]->redefine(mRenderer, GL_TEXTURE_2D, internalformat, levelWidth, levelHeight, 1, true); + } + + for (int level = levels; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; level++) + { + mImageArray[level]->redefine(mRenderer, GL_TEXTURE_2D, GL_NONE, 0, 0, 0, true); + } + + mImmutable = true; + + bool renderTarget = IsRenderTargetUsage(mUsage); + TextureStorage *storage = mRenderer->createTextureStorage2D(internalformat, renderTarget, width, height, levels); + setCompleteTexStorage(storage); +} + +void TextureD3D_2D::bindTexImage(egl::Surface *surface) +{ + GLenum internalformat = surface->getFormat(); + + mImageArray[0]->redefine(mRenderer, GL_TEXTURE_2D, internalformat, surface->getWidth(), surface->getHeight(), 1, true); + + if (mTexStorage) + { + SafeDelete(mTexStorage); + } + + mTexStorage = mRenderer->createTextureStorage2D(surface->getSwapChain()); + + mDirtyImages = true; +} + +void TextureD3D_2D::releaseTexImage() +{ + if (mTexStorage) + { + SafeDelete(mTexStorage); + } + + for (int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) + { + mImageArray[i]->redefine(mRenderer, GL_TEXTURE_2D, GL_NONE, 0, 0, 0, true); + } +} + +void TextureD3D_2D::generateMipmaps() +{ + // Purge array levels 1 through q and reset them to represent the generated mipmap levels. + int levelCount = mipLevels(); + for (int level = 1; level < levelCount; level++) + { + redefineImage(level, getBaseLevelInternalFormat(), + std::max(getBaseLevelWidth() >> level, 1), + std::max(getBaseLevelHeight() >> level, 1)); + } + + if (mTexStorage && mTexStorage->isRenderTarget()) + { + mTexStorage->generateMipmaps(); + for (int level = 1; level < levelCount; level++) + { + mImageArray[level]->markClean(); + } + } + else + { + for (int level = 1; level < levelCount; level++) + { + mRenderer->generateMipmap(mImageArray[level], mImageArray[level - 1]); + } + } +} + +unsigned int TextureD3D_2D::getRenderTargetSerial(const gl::ImageIndex &index) +{ + ASSERT(!index.hasLayer()); + return (ensureRenderTarget() ? mTexStorage->getRenderTargetSerial(index) : 0); +} + +RenderTarget *TextureD3D_2D::getRenderTarget(const gl::ImageIndex &index) +{ + ASSERT(!index.hasLayer()); + + // ensure the underlying texture is created + if (!ensureRenderTarget()) + { + return NULL; + } + + updateStorageLevel(index.mipIndex); + return mTexStorage->getRenderTarget(index); +} + +bool TextureD3D_2D::isValidLevel(int level) const +{ + return (mTexStorage ? (level >= 0 && level < mTexStorage->getLevelCount()) : false); +} + +bool TextureD3D_2D::isLevelComplete(int level) const +{ + if (isImmutable()) + { + return true; + } + + const Image *baseImage = getBaseLevelImage(); + + GLsizei width = baseImage->getWidth(); + GLsizei height = baseImage->getHeight(); + + if (width <= 0 || height <= 0) + { + return false; + } + + // The base image level is complete if the width and height are positive + if (level == 0) + { + return true; + } + + ASSERT(level >= 1 && level <= (int)ArraySize(mImageArray) && mImageArray[level] != NULL); + ImageD3D *image = mImageArray[level]; + + if (image->getInternalFormat() != baseImage->getInternalFormat()) + { + return false; + } + + if (image->getWidth() != std::max(1, width >> level)) + { + return false; + } + + if (image->getHeight() != std::max(1, height >> level)) + { + return false; + } + + return true; +} + +// Constructs a native texture resource from the texture images +void TextureD3D_2D::initializeStorage(bool renderTarget) +{ + // Only initialize the first time this texture is used as a render target or shader resource + if (mTexStorage) + { + return; + } + + // do not attempt to create storage for nonexistant data + if (!isLevelComplete(0)) + { + return; + } + + bool createRenderTarget = (renderTarget || IsRenderTargetUsage(mUsage)); + + setCompleteTexStorage(createCompleteStorage(createRenderTarget)); + ASSERT(mTexStorage); + + // flush image data to the storage + updateStorage(); +} + +TextureStorage *TextureD3D_2D::createCompleteStorage(bool renderTarget) const +{ + GLsizei width = getBaseLevelWidth(); + GLsizei height = getBaseLevelHeight(); + GLenum internalFormat = getBaseLevelInternalFormat(); + + ASSERT(width > 0 && height > 0); + + // use existing storage level count, when previously specified by TexStorage*D + GLint levels = (mTexStorage ? mTexStorage->getLevelCount() : creationLevels(width, height, 1)); + + return mRenderer->createTextureStorage2D(internalFormat, renderTarget, width, height, levels); +} + +void TextureD3D_2D::setCompleteTexStorage(TextureStorage *newCompleteTexStorage) +{ + SafeDelete(mTexStorage); + mTexStorage = newCompleteTexStorage; + + if (mTexStorage && mTexStorage->isManaged()) + { + for (int level = 0; level < mTexStorage->getLevelCount(); level++) + { + mImageArray[level]->setManagedSurface2D(mTexStorage, level); + } + } + + mDirtyImages = true; +} + +void TextureD3D_2D::updateStorage() +{ + ASSERT(mTexStorage != NULL); + GLint storageLevels = mTexStorage->getLevelCount(); + for (int level = 0; level < storageLevels; level++) + { + if (mImageArray[level]->isDirty() && isLevelComplete(level)) + { + updateStorageLevel(level); + } + } +} + +bool TextureD3D_2D::ensureRenderTarget() +{ + initializeStorage(true); + + if (getBaseLevelWidth() > 0 && getBaseLevelHeight() > 0) + { + ASSERT(mTexStorage); + if (!mTexStorage->isRenderTarget()) + { + TextureStorage *newRenderTargetStorage = createCompleteStorage(true); + + if (!mRenderer->copyToRenderTarget2D(newRenderTargetStorage, mTexStorage)) + { + delete newRenderTargetStorage; + return gl::error(GL_OUT_OF_MEMORY, false); + } + + setCompleteTexStorage(newRenderTargetStorage); + } + } + + return (mTexStorage && mTexStorage->isRenderTarget()); +} + +TextureStorage *TextureD3D_2D::getBaseLevelStorage() +{ + return mTexStorage; +} + +const ImageD3D *TextureD3D_2D::getBaseLevelImage() const +{ + return mImageArray[0]; +} + +void TextureD3D_2D::updateStorageLevel(int level) +{ + ASSERT(level <= (int)ArraySize(mImageArray) && mImageArray[level] != NULL); + ASSERT(isLevelComplete(level)); + + if (mImageArray[level]->isDirty()) + { + commitRect(level, 0, 0, getWidth(level), getHeight(level)); + } +} + +void TextureD3D_2D::redefineImage(GLint level, GLenum internalformat, GLsizei width, GLsizei height) +{ + // If there currently is a corresponding storage texture image, it has these parameters + const int storageWidth = std::max(1, getBaseLevelWidth() >> level); + const int storageHeight = std::max(1, getBaseLevelHeight() >> level); + const GLenum storageFormat = getBaseLevelInternalFormat(); + + mImageArray[level]->redefine(mRenderer, GL_TEXTURE_2D, internalformat, width, height, 1, false); + + if (mTexStorage) + { + const int storageLevels = mTexStorage->getLevelCount(); + + if ((level >= storageLevels && storageLevels != 0) || + width != storageWidth || + height != storageHeight || + internalformat != storageFormat) // Discard mismatched storage + { + for (int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) + { + mImageArray[i]->markDirty(); + } + + SafeDelete(mTexStorage); + mDirtyImages = true; + } + } +} + +void TextureD3D_2D::commitRect(GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) +{ + if (isValidLevel(level)) + { + ImageD3D *image = mImageArray[level]; + if (image->copyToStorage2D(mTexStorage, level, xoffset, yoffset, width, height)) + { + image->markClean(); + } + } +} + + +TextureD3D_Cube::TextureD3D_Cube(Renderer *renderer) + : TextureD3D(renderer), + mTexStorage(NULL) +{ + for (int i = 0; i < 6; i++) + { + for (int j = 0; j < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; ++j) + { + mImageArray[i][j] = ImageD3D::makeImageD3D(renderer->createImage()); + } + } +} + +TextureD3D_Cube::~TextureD3D_Cube() +{ + // Delete the Images before the TextureStorage. + // Images might be relying on the TextureStorage for some of their data. + // If TextureStorage is deleted before the Images, then their data will be wastefully copied back from the GPU before we delete the Images. + for (int i = 0; i < 6; i++) + { + for (int j = 0; j < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; ++j) + { + SafeDelete(mImageArray[i][j]); + } + } + + SafeDelete(mTexStorage); +} + +Image *TextureD3D_Cube::getImage(int level, int layer) const +{ + ASSERT(level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + ASSERT(layer < 6); + return mImageArray[layer][level]; +} + +Image *TextureD3D_Cube::getImage(const gl::ImageIndex &index) const +{ + ASSERT(index.mipIndex < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + ASSERT(index.layerIndex < 6); + return mImageArray[index.layerIndex][index.mipIndex]; +} + +GLsizei TextureD3D_Cube::getLayerCount(int level) const +{ + ASSERT(level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + return 6; +} + +GLenum TextureD3D_Cube::getInternalFormat(GLint level, GLint layer) const +{ + if (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + return mImageArray[layer][level]->getInternalFormat(); + else + return GL_NONE; +} + +bool TextureD3D_Cube::isDepth(GLint level, GLint layer) const +{ + return gl::GetInternalFormatInfo(getInternalFormat(level, layer)).depthBits > 0; +} + +void TextureD3D_Cube::setImage(GLenum target, GLint level, GLsizei width, GLsizei height, GLsizei depth, GLenum internalFormat, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels) +{ + ASSERT(depth == 1); + + int faceIndex = gl::TextureCubeMap::targetToLayerIndex(target); + GLenum sizedInternalFormat = gl::GetSizedInternalFormat(internalFormat, type); + + redefineImage(faceIndex, level, sizedInternalFormat, width, height); + + TextureD3D::setImage(unpack, type, pixels, mImageArray[faceIndex][level]); +} + +void TextureD3D_Cube::setCompressedImage(GLenum target, GLint level, GLenum format, GLsizei width, GLsizei height, GLsizei depth, GLsizei imageSize, const void *pixels) +{ + ASSERT(depth == 1); + + // compressed formats don't have separate sized internal formats-- we can just use the compressed format directly + int faceIndex = gl::TextureCubeMap::targetToLayerIndex(target); + + redefineImage(faceIndex, level, format, width, height); + + TextureD3D::setCompressedImage(imageSize, pixels, mImageArray[faceIndex][level]); +} + +void TextureD3D_Cube::subImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels) +{ + ASSERT(depth == 1 && zoffset == 0); + + int faceIndex = gl::TextureCubeMap::targetToLayerIndex(target); + + gl::ImageIndex index = gl::ImageIndex::MakeCube(target, level); + if (TextureD3D::subImage(xoffset, yoffset, 0, width, height, 1, format, type, unpack, pixels, index)) + { + commitRect(faceIndex, level, xoffset, yoffset, width, height); + } +} + +void TextureD3D_Cube::subImageCompressed(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void *pixels) +{ + ASSERT(depth == 1 && zoffset == 0); + + int faceIndex = gl::TextureCubeMap::targetToLayerIndex(target); + + if (TextureD3D::subImageCompressed(xoffset, yoffset, 0, width, height, 1, format, imageSize, pixels, mImageArray[faceIndex][level])) + { + commitRect(faceIndex, level, xoffset, yoffset, width, height); + } +} + +void TextureD3D_Cube::copyImage(GLenum target, GLint level, GLenum format, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) +{ + int faceIndex = gl::TextureCubeMap::targetToLayerIndex(target); + GLenum sizedInternalFormat = gl::GetSizedInternalFormat(format, GL_UNSIGNED_BYTE); + + redefineImage(faceIndex, level, sizedInternalFormat, width, height); + + if (!mImageArray[faceIndex][level]->isRenderableFormat()) + { + mImageArray[faceIndex][level]->copy(0, 0, 0, x, y, width, height, source); + mDirtyImages = true; + } + else + { + ensureRenderTarget(); + mImageArray[faceIndex][level]->markClean(); + + ASSERT(width == height); + + if (width > 0 && isValidFaceLevel(faceIndex, level)) + { + gl::Rectangle sourceRect; + sourceRect.x = x; + sourceRect.width = width; + sourceRect.y = y; + sourceRect.height = height; + + mRenderer->copyImageCube(source, sourceRect, format, 0, 0, mTexStorage, target, level); + } + } +} + +void TextureD3D_Cube::copySubImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) +{ + int faceIndex = gl::TextureCubeMap::targetToLayerIndex(target); + + // We can only make our texture storage to a render target if the level we're copying *to* is complete + // and the base level is cube-complete. The base level must be cube complete (common case) because we cannot + // rely on the "getBaseLevel*" methods reliably otherwise. + bool canCreateRenderTarget = isFaceLevelComplete(faceIndex, level) && isCubeComplete(); + + if (!mImageArray[faceIndex][level]->isRenderableFormat() || (!mTexStorage && !canCreateRenderTarget)) + { + mImageArray[faceIndex][level]->copy(0, 0, 0, x, y, width, height, source); + mDirtyImages = true; + } + else + { + ensureRenderTarget(); + + if (isValidFaceLevel(faceIndex, level)) + { + updateStorageFaceLevel(faceIndex, level); + + gl::Rectangle sourceRect; + sourceRect.x = x; + sourceRect.width = width; + sourceRect.y = y; + sourceRect.height = height; + + mRenderer->copyImageCube(source, sourceRect, gl::GetInternalFormatInfo(getBaseLevelInternalFormat()).format, + xoffset, yoffset, mTexStorage, target, level); + } + } +} + +void TextureD3D_Cube::storage(GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth) +{ + ASSERT(width == height); + ASSERT(depth == 1); + + for (int level = 0; level < levels; level++) + { + GLsizei mipSize = std::max(1, width >> level); + for (int faceIndex = 0; faceIndex < 6; faceIndex++) + { + mImageArray[faceIndex][level]->redefine(mRenderer, GL_TEXTURE_CUBE_MAP, internalformat, mipSize, mipSize, 1, true); + } + } + + for (int level = levels; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; level++) + { + for (int faceIndex = 0; faceIndex < 6; faceIndex++) + { + mImageArray[faceIndex][level]->redefine(mRenderer, GL_TEXTURE_CUBE_MAP, GL_NONE, 0, 0, 0, true); + } + } + + mImmutable = true; + + bool renderTarget = IsRenderTargetUsage(mUsage); + TextureStorage *storage = mRenderer->createTextureStorageCube(internalformat, renderTarget, width, levels); + setCompleteTexStorage(storage); +} + +// Tests for cube texture completeness. [OpenGL ES 2.0.24] section 3.7.10 page 81. +bool TextureD3D_Cube::isCubeComplete() const +{ + int baseWidth = getBaseLevelWidth(); + int baseHeight = getBaseLevelHeight(); + GLenum baseFormat = getBaseLevelInternalFormat(); + + if (baseWidth <= 0 || baseWidth != baseHeight) + { + return false; + } + + for (int faceIndex = 1; faceIndex < 6; faceIndex++) + { + const ImageD3D &faceBaseImage = *mImageArray[faceIndex][0]; + + if (faceBaseImage.getWidth() != baseWidth || + faceBaseImage.getHeight() != baseHeight || + faceBaseImage.getInternalFormat() != baseFormat ) + { + return false; + } + } + + return true; +} + +void TextureD3D_Cube::bindTexImage(egl::Surface *surface) +{ + UNREACHABLE(); +} + +void TextureD3D_Cube::releaseTexImage() +{ + UNREACHABLE(); +} + + +void TextureD3D_Cube::generateMipmaps() +{ + // Purge array levels 1 through q and reset them to represent the generated mipmap levels. + int levelCount = mipLevels(); + for (int faceIndex = 0; faceIndex < 6; faceIndex++) + { + for (int level = 1; level < levelCount; level++) + { + int faceLevelSize = (std::max(mImageArray[faceIndex][0]->getWidth() >> level, 1)); + redefineImage(faceIndex, level, mImageArray[faceIndex][0]->getInternalFormat(), faceLevelSize, faceLevelSize); + } + } + + if (mTexStorage && mTexStorage->isRenderTarget()) + { + mTexStorage->generateMipmaps(); + + for (int faceIndex = 0; faceIndex < 6; faceIndex++) + { + for (int level = 1; level < levelCount; level++) + { + mImageArray[faceIndex][level]->markClean(); + } + } + } + else + { + for (int faceIndex = 0; faceIndex < 6; faceIndex++) + { + for (int level = 1; level < levelCount; level++) + { + mRenderer->generateMipmap(mImageArray[faceIndex][level], mImageArray[faceIndex][level - 1]); + } + } + } +} + +unsigned int TextureD3D_Cube::getRenderTargetSerial(const gl::ImageIndex &index) +{ + return (ensureRenderTarget() ? mTexStorage->getRenderTargetSerial(index) : 0); +} + +RenderTarget *TextureD3D_Cube::getRenderTarget(const gl::ImageIndex &index) +{ + ASSERT(gl::IsCubemapTextureTarget(index.type)); + + // ensure the underlying texture is created + if (!ensureRenderTarget()) + { + return NULL; + } + + updateStorageFaceLevel(index.layerIndex, index.mipIndex); + return mTexStorage->getRenderTarget(index); +} + +void TextureD3D_Cube::initializeStorage(bool renderTarget) +{ + // Only initialize the first time this texture is used as a render target or shader resource + if (mTexStorage) + { + return; + } + + // do not attempt to create storage for nonexistant data + if (!isFaceLevelComplete(0, 0)) + { + return; + } + + bool createRenderTarget = (renderTarget || IsRenderTargetUsage(mUsage)); + + setCompleteTexStorage(createCompleteStorage(createRenderTarget)); + ASSERT(mTexStorage); + + // flush image data to the storage + updateStorage(); +} + +TextureStorage *TextureD3D_Cube::createCompleteStorage(bool renderTarget) const +{ + GLsizei size = getBaseLevelWidth(); + + ASSERT(size > 0); + + // use existing storage level count, when previously specified by TexStorage*D + GLint levels = (mTexStorage ? mTexStorage->getLevelCount() : creationLevels(size, size, 1)); + + return mRenderer->createTextureStorageCube(getBaseLevelInternalFormat(), renderTarget, size, levels); +} + +void TextureD3D_Cube::setCompleteTexStorage(TextureStorage *newCompleteTexStorage) +{ + SafeDelete(mTexStorage); + mTexStorage = newCompleteTexStorage; + + if (mTexStorage && mTexStorage->isManaged()) + { + for (int faceIndex = 0; faceIndex < 6; faceIndex++) + { + for (int level = 0; level < mTexStorage->getLevelCount(); level++) + { + mImageArray[faceIndex][level]->setManagedSurfaceCube(mTexStorage, faceIndex, level); + } + } + } + + mDirtyImages = true; +} + +void TextureD3D_Cube::updateStorage() +{ + ASSERT(mTexStorage != NULL); + GLint storageLevels = mTexStorage->getLevelCount(); + for (int face = 0; face < 6; face++) + { + for (int level = 0; level < storageLevels; level++) + { + if (mImageArray[face][level]->isDirty() && isFaceLevelComplete(face, level)) + { + updateStorageFaceLevel(face, level); + } + } + } +} + +bool TextureD3D_Cube::ensureRenderTarget() +{ + initializeStorage(true); + + if (getBaseLevelWidth() > 0) + { + ASSERT(mTexStorage); + if (!mTexStorage->isRenderTarget()) + { + TextureStorage *newRenderTargetStorage = createCompleteStorage(true); + + if (!mRenderer->copyToRenderTargetCube(newRenderTargetStorage, mTexStorage)) + { + delete newRenderTargetStorage; + return gl::error(GL_OUT_OF_MEMORY, false); + } + + setCompleteTexStorage(newRenderTargetStorage); + } + } + + return (mTexStorage && mTexStorage->isRenderTarget()); +} + +TextureStorage *TextureD3D_Cube::getBaseLevelStorage() +{ + return mTexStorage; +} + +const ImageD3D *TextureD3D_Cube::getBaseLevelImage() const +{ + // Note: if we are not cube-complete, there is no single base level image that can describe all + // cube faces, so this method is only well-defined for a cube-complete base level. + return mImageArray[0][0]; +} + +bool TextureD3D_Cube::isValidFaceLevel(int faceIndex, int level) const +{ + return (mTexStorage ? (level >= 0 && level < mTexStorage->getLevelCount()) : 0); +} + +bool TextureD3D_Cube::isFaceLevelComplete(int faceIndex, int level) const +{ + ASSERT(level >= 0 && faceIndex < 6 && level < (int)ArraySize(mImageArray[faceIndex]) && mImageArray[faceIndex][level] != NULL); + + if (isImmutable()) + { + return true; + } + + int baseSize = getBaseLevelWidth(); + + if (baseSize <= 0) + { + return false; + } + + // "isCubeComplete" checks for base level completeness and we must call that + // to determine if any face at level 0 is complete. We omit that check here + // to avoid re-checking cube-completeness for every face at level 0. + if (level == 0) + { + return true; + } + + // Check that non-zero levels are consistent with the base level. + const ImageD3D *faceLevelImage = mImageArray[faceIndex][level]; + + if (faceLevelImage->getInternalFormat() != getBaseLevelInternalFormat()) + { + return false; + } + + if (faceLevelImage->getWidth() != std::max(1, baseSize >> level)) + { + return false; + } + + return true; +} + +void TextureD3D_Cube::updateStorageFaceLevel(int faceIndex, int level) +{ + ASSERT(level >= 0 && faceIndex < 6 && level < (int)ArraySize(mImageArray[faceIndex]) && mImageArray[faceIndex][level] != NULL); + ImageD3D *image = mImageArray[faceIndex][level]; + + if (image->isDirty()) + { + commitRect(faceIndex, level, 0, 0, image->getWidth(), image->getHeight()); + } +} + +void TextureD3D_Cube::redefineImage(int faceIndex, GLint level, GLenum internalformat, GLsizei width, GLsizei height) +{ + // If there currently is a corresponding storage texture image, it has these parameters + const int storageWidth = std::max(1, getBaseLevelWidth() >> level); + const int storageHeight = std::max(1, getBaseLevelHeight() >> level); + const GLenum storageFormat = getBaseLevelInternalFormat(); + + mImageArray[faceIndex][level]->redefine(mRenderer, GL_TEXTURE_CUBE_MAP, internalformat, width, height, 1, false); + + if (mTexStorage) + { + const int storageLevels = mTexStorage->getLevelCount(); + + if ((level >= storageLevels && storageLevels != 0) || + width != storageWidth || + height != storageHeight || + internalformat != storageFormat) // Discard mismatched storage + { + for (int level = 0; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; level++) + { + for (int faceIndex = 0; faceIndex < 6; faceIndex++) + { + mImageArray[faceIndex][level]->markDirty(); + } + } + + SafeDelete(mTexStorage); + + mDirtyImages = true; + } + } +} + +void TextureD3D_Cube::commitRect(int faceIndex, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) +{ + if (isValidFaceLevel(faceIndex, level)) + { + ImageD3D *image = mImageArray[faceIndex][level]; + if (image->copyToStorageCube(mTexStorage, faceIndex, level, xoffset, yoffset, width, height)) + image->markClean(); + } +} + + +TextureD3D_3D::TextureD3D_3D(Renderer *renderer) + : TextureD3D(renderer), + mTexStorage(NULL) +{ + for (int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; ++i) + { + mImageArray[i] = ImageD3D::makeImageD3D(renderer->createImage()); + } +} + +TextureD3D_3D::~TextureD3D_3D() +{ + // Delete the Images before the TextureStorage. + // Images might be relying on the TextureStorage for some of their data. + // If TextureStorage is deleted before the Images, then their data will be wastefully copied back from the GPU before we delete the Images. + for (int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; ++i) + { + delete mImageArray[i]; + } + + SafeDelete(mTexStorage); +} + +Image *TextureD3D_3D::getImage(int level, int layer) const +{ + ASSERT(level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + ASSERT(layer == 0); + return mImageArray[level]; +} + +Image *TextureD3D_3D::getImage(const gl::ImageIndex &index) const +{ + ASSERT(index.mipIndex < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + ASSERT(!index.hasLayer()); + ASSERT(index.type == GL_TEXTURE_3D); + return mImageArray[index.mipIndex]; +} + +GLsizei TextureD3D_3D::getLayerCount(int level) const +{ + ASSERT(level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + return 1; +} + +GLsizei TextureD3D_3D::getWidth(GLint level) const +{ + if (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + return mImageArray[level]->getWidth(); + else + return 0; +} + +GLsizei TextureD3D_3D::getHeight(GLint level) const +{ + if (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + return mImageArray[level]->getHeight(); + else + return 0; +} + +GLsizei TextureD3D_3D::getDepth(GLint level) const +{ + if (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + return mImageArray[level]->getDepth(); + else + return 0; +} + +GLenum TextureD3D_3D::getInternalFormat(GLint level) const +{ + if (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + return mImageArray[level]->getInternalFormat(); + else + return GL_NONE; +} + +bool TextureD3D_3D::isDepth(GLint level) const +{ + return gl::GetInternalFormatInfo(getInternalFormat(level)).depthBits > 0; +} + +void TextureD3D_3D::setImage(GLenum target, GLint level, GLsizei width, GLsizei height, GLsizei depth, GLenum internalFormat, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels) +{ + ASSERT(target == GL_TEXTURE_3D); + GLenum sizedInternalFormat = gl::GetSizedInternalFormat(internalFormat, type); + + redefineImage(level, sizedInternalFormat, width, height, depth); + + bool fastUnpacked = false; + + // Attempt a fast gpu copy of the pixel data to the surface if the app bound an unpack buffer + if (isFastUnpackable(unpack, sizedInternalFormat)) + { + // Will try to create RT storage if it does not exist + gl::ImageIndex index = gl::ImageIndex::Make3D(level); + RenderTarget *destRenderTarget = getRenderTarget(index); + gl::Box destArea(0, 0, 0, getWidth(level), getHeight(level), getDepth(level)); + + if (destRenderTarget && fastUnpackPixels(unpack, pixels, destArea, sizedInternalFormat, type, destRenderTarget)) + { + // Ensure we don't overwrite our newly initialized data + mImageArray[level]->markClean(); + + fastUnpacked = true; + } + } + + if (!fastUnpacked) + { + TextureD3D::setImage(unpack, type, pixels, mImageArray[level]); + } +} + +void TextureD3D_3D::setCompressedImage(GLenum target, GLint level, GLenum format, GLsizei width, GLsizei height, GLsizei depth, GLsizei imageSize, const void *pixels) +{ + ASSERT(target == GL_TEXTURE_3D); + + // compressed formats don't have separate sized internal formats-- we can just use the compressed format directly + redefineImage(level, format, width, height, depth); + + TextureD3D::setCompressedImage(imageSize, pixels, mImageArray[level]); +} + +void TextureD3D_3D::subImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels) +{ + ASSERT(target == GL_TEXTURE_3D); + + bool fastUnpacked = false; + + gl::ImageIndex index = gl::ImageIndex::Make3D(level); + + // Attempt a fast gpu copy of the pixel data to the surface if the app bound an unpack buffer + if (isFastUnpackable(unpack, getInternalFormat(level))) + { + RenderTarget *destRenderTarget = getRenderTarget(index); + gl::Box destArea(xoffset, yoffset, zoffset, width, height, depth); + + if (destRenderTarget && fastUnpackPixels(unpack, pixels, destArea, getInternalFormat(level), type, destRenderTarget)) + { + // Ensure we don't overwrite our newly initialized data + mImageArray[level]->markClean(); + + fastUnpacked = true; + } + } + + if (!fastUnpacked && TextureD3D::subImage(xoffset, yoffset, zoffset, width, height, depth, format, type, unpack, pixels, index)) + { + commitRect(level, xoffset, yoffset, zoffset, width, height, depth); + } +} + +void TextureD3D_3D::subImageCompressed(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void *pixels) +{ + ASSERT(target == GL_TEXTURE_3D); + + if (TextureD3D::subImageCompressed(xoffset, yoffset, zoffset, width, height, depth, format, imageSize, pixels, mImageArray[level])) + { + commitRect(level, xoffset, yoffset, zoffset, width, height, depth); + } +} + +void TextureD3D_3D::copyImage(GLenum target, GLint level, GLenum format, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) +{ + UNIMPLEMENTED(); +} + +void TextureD3D_3D::copySubImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) +{ + ASSERT(target == GL_TEXTURE_3D); + + // can only make our texture storage to a render target if level 0 is defined (with a width & height) and + // the current level we're copying to is defined (with appropriate format, width & height) + bool canCreateRenderTarget = isLevelComplete(level) && isLevelComplete(0); + + if (!mImageArray[level]->isRenderableFormat() || (!mTexStorage && !canCreateRenderTarget)) + { + mImageArray[level]->copy(xoffset, yoffset, zoffset, x, y, width, height, source); + mDirtyImages = true; + } + else + { + ensureRenderTarget(); + + if (isValidLevel(level)) + { + updateStorageLevel(level); + + gl::Rectangle sourceRect; + sourceRect.x = x; + sourceRect.width = width; + sourceRect.y = y; + sourceRect.height = height; + + mRenderer->copyImage3D(source, sourceRect, + gl::GetInternalFormatInfo(getBaseLevelInternalFormat()).format, + xoffset, yoffset, zoffset, mTexStorage, level); + } + } +} + +void TextureD3D_3D::storage(GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth) +{ + ASSERT(target == GL_TEXTURE_3D); + + for (int level = 0; level < levels; level++) + { + GLsizei levelWidth = std::max(1, width >> level); + GLsizei levelHeight = std::max(1, height >> level); + GLsizei levelDepth = std::max(1, depth >> level); + mImageArray[level]->redefine(mRenderer, GL_TEXTURE_3D, internalformat, levelWidth, levelHeight, levelDepth, true); + } + + for (int level = levels; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; level++) + { + mImageArray[level]->redefine(mRenderer, GL_TEXTURE_3D, GL_NONE, 0, 0, 0, true); + } + + mImmutable = true; + + bool renderTarget = IsRenderTargetUsage(mUsage); + TextureStorage *storage = mRenderer->createTextureStorage3D(internalformat, renderTarget, width, height, depth, levels); + setCompleteTexStorage(storage); +} + +void TextureD3D_3D::bindTexImage(egl::Surface *surface) +{ + UNREACHABLE(); +} + +void TextureD3D_3D::releaseTexImage() +{ + UNREACHABLE(); +} + + +void TextureD3D_3D::generateMipmaps() +{ + // Purge array levels 1 through q and reset them to represent the generated mipmap levels. + int levelCount = mipLevels(); + for (int level = 1; level < levelCount; level++) + { + redefineImage(level, getBaseLevelInternalFormat(), + std::max(getBaseLevelWidth() >> level, 1), + std::max(getBaseLevelHeight() >> level, 1), + std::max(getBaseLevelDepth() >> level, 1)); + } + + if (mTexStorage && mTexStorage->isRenderTarget()) + { + mTexStorage->generateMipmaps(); + + for (int level = 1; level < levelCount; level++) + { + mImageArray[level]->markClean(); + } + } + else + { + for (int level = 1; level < levelCount; level++) + { + mRenderer->generateMipmap(mImageArray[level], mImageArray[level - 1]); + } + } +} + +unsigned int TextureD3D_3D::getRenderTargetSerial(const gl::ImageIndex &index) +{ + return (ensureRenderTarget() ? mTexStorage->getRenderTargetSerial(index) : 0); +} + +RenderTarget *TextureD3D_3D::getRenderTarget(const gl::ImageIndex &index) +{ + // ensure the underlying texture is created + if (!ensureRenderTarget()) + { + return NULL; + } + + if (index.hasLayer()) + { + updateStorage(); + } + else + { + updateStorageLevel(index.mipIndex); + } + + return mTexStorage->getRenderTarget(index); +} + +void TextureD3D_3D::initializeStorage(bool renderTarget) +{ + // Only initialize the first time this texture is used as a render target or shader resource + if (mTexStorage) + { + return; + } + + // do not attempt to create storage for nonexistant data + if (!isLevelComplete(0)) + { + return; + } + + bool createRenderTarget = (renderTarget || mUsage == GL_FRAMEBUFFER_ATTACHMENT_ANGLE); + + setCompleteTexStorage(createCompleteStorage(createRenderTarget)); + ASSERT(mTexStorage); + + // flush image data to the storage + updateStorage(); +} + +TextureStorage *TextureD3D_3D::createCompleteStorage(bool renderTarget) const +{ + GLsizei width = getBaseLevelWidth(); + GLsizei height = getBaseLevelHeight(); + GLsizei depth = getBaseLevelDepth(); + GLenum internalFormat = getBaseLevelInternalFormat(); + + ASSERT(width > 0 && height > 0 && depth > 0); + + // use existing storage level count, when previously specified by TexStorage*D + GLint levels = (mTexStorage ? mTexStorage->getLevelCount() : creationLevels(width, height, depth)); + + return mRenderer->createTextureStorage3D(internalFormat, renderTarget, width, height, depth, levels); +} + +void TextureD3D_3D::setCompleteTexStorage(TextureStorage *newCompleteTexStorage) +{ + SafeDelete(mTexStorage); + mTexStorage = newCompleteTexStorage; + mDirtyImages = true; + + // We do not support managed 3D storage, as that is D3D9/ES2-only + ASSERT(!mTexStorage->isManaged()); +} + +void TextureD3D_3D::updateStorage() +{ + ASSERT(mTexStorage != NULL); + GLint storageLevels = mTexStorage->getLevelCount(); + for (int level = 0; level < storageLevels; level++) + { + if (mImageArray[level]->isDirty() && isLevelComplete(level)) + { + updateStorageLevel(level); + } + } +} + +bool TextureD3D_3D::ensureRenderTarget() +{ + initializeStorage(true); + + if (getBaseLevelWidth() > 0 && getBaseLevelHeight() > 0 && getBaseLevelDepth() > 0) + { + ASSERT(mTexStorage); + if (!mTexStorage->isRenderTarget()) + { + TextureStorage *newRenderTargetStorage = createCompleteStorage(true); + + if (!mRenderer->copyToRenderTarget3D(newRenderTargetStorage, mTexStorage)) + { + delete newRenderTargetStorage; + return gl::error(GL_OUT_OF_MEMORY, false); + } + + setCompleteTexStorage(newRenderTargetStorage); + } + } + + return (mTexStorage && mTexStorage->isRenderTarget()); +} + +TextureStorage *TextureD3D_3D::getBaseLevelStorage() +{ + return mTexStorage; +} + +const ImageD3D *TextureD3D_3D::getBaseLevelImage() const +{ + return mImageArray[0]; +} + +bool TextureD3D_3D::isValidLevel(int level) const +{ + return (mTexStorage ? (level >= 0 && level < mTexStorage->getLevelCount()) : 0); +} + +bool TextureD3D_3D::isLevelComplete(int level) const +{ + ASSERT(level >= 0 && level < (int)ArraySize(mImageArray) && mImageArray[level] != NULL); + + if (isImmutable()) + { + return true; + } + + GLsizei width = getBaseLevelWidth(); + GLsizei height = getBaseLevelHeight(); + GLsizei depth = getBaseLevelDepth(); + + if (width <= 0 || height <= 0 || depth <= 0) + { + return false; + } + + if (level == 0) + { + return true; + } + + ImageD3D *levelImage = mImageArray[level]; + + if (levelImage->getInternalFormat() != getBaseLevelInternalFormat()) + { + return false; + } + + if (levelImage->getWidth() != std::max(1, width >> level)) + { + return false; + } + + if (levelImage->getHeight() != std::max(1, height >> level)) + { + return false; + } + + if (levelImage->getDepth() != std::max(1, depth >> level)) + { + return false; + } + + return true; +} + +void TextureD3D_3D::updateStorageLevel(int level) +{ + ASSERT(level >= 0 && level < (int)ArraySize(mImageArray) && mImageArray[level] != NULL); + ASSERT(isLevelComplete(level)); + + if (mImageArray[level]->isDirty()) + { + commitRect(level, 0, 0, 0, getWidth(level), getHeight(level), getDepth(level)); + } +} + +void TextureD3D_3D::redefineImage(GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth) +{ + // If there currently is a corresponding storage texture image, it has these parameters + const int storageWidth = std::max(1, getBaseLevelWidth() >> level); + const int storageHeight = std::max(1, getBaseLevelHeight() >> level); + const int storageDepth = std::max(1, getBaseLevelDepth() >> level); + const GLenum storageFormat = getBaseLevelInternalFormat(); + + mImageArray[level]->redefine(mRenderer, GL_TEXTURE_3D, internalformat, width, height, depth, false); + + if (mTexStorage) + { + const int storageLevels = mTexStorage->getLevelCount(); + + if ((level >= storageLevels && storageLevels != 0) || + width != storageWidth || + height != storageHeight || + depth != storageDepth || + internalformat != storageFormat) // Discard mismatched storage + { + for (int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) + { + mImageArray[i]->markDirty(); + } + + SafeDelete(mTexStorage); + mDirtyImages = true; + } + } +} + +void TextureD3D_3D::commitRect(GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth) +{ + if (isValidLevel(level)) + { + ImageD3D *image = mImageArray[level]; + if (image->copyToStorage3D(mTexStorage, level, xoffset, yoffset, zoffset, width, height, depth)) + { + image->markClean(); + } + } +} + + +TextureD3D_2DArray::TextureD3D_2DArray(Renderer *renderer) + : TextureD3D(renderer), + mTexStorage(NULL) +{ + for (int level = 0; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; ++level) + { + mLayerCounts[level] = 0; + mImageArray[level] = NULL; + } +} + +TextureD3D_2DArray::~TextureD3D_2DArray() +{ + // Delete the Images before the TextureStorage. + // Images might be relying on the TextureStorage for some of their data. + // If TextureStorage is deleted before the Images, then their data will be wastefully copied back from the GPU before we delete the Images. + deleteImages(); + SafeDelete(mTexStorage); +} + +Image *TextureD3D_2DArray::getImage(int level, int layer) const +{ + ASSERT(level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + ASSERT(layer < mLayerCounts[level]); + return mImageArray[level][layer]; +} + +Image *TextureD3D_2DArray::getImage(const gl::ImageIndex &index) const +{ + ASSERT(index.mipIndex < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + ASSERT(index.layerIndex < mLayerCounts[index.mipIndex]); + ASSERT(index.type == GL_TEXTURE_2D_ARRAY); + return mImageArray[index.mipIndex][index.layerIndex]; +} + +GLsizei TextureD3D_2DArray::getLayerCount(int level) const +{ + ASSERT(level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + return mLayerCounts[level]; +} + +GLsizei TextureD3D_2DArray::getWidth(GLint level) const +{ + return (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS && mLayerCounts[level] > 0) ? mImageArray[level][0]->getWidth() : 0; +} + +GLsizei TextureD3D_2DArray::getHeight(GLint level) const +{ + return (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS && mLayerCounts[level] > 0) ? mImageArray[level][0]->getHeight() : 0; +} + +GLsizei TextureD3D_2DArray::getLayers(GLint level) const +{ + return (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) ? mLayerCounts[level] : 0; +} + +GLenum TextureD3D_2DArray::getInternalFormat(GLint level) const +{ + return (level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS && mLayerCounts[level] > 0) ? mImageArray[level][0]->getInternalFormat() : GL_NONE; +} + +bool TextureD3D_2DArray::isDepth(GLint level) const +{ + return gl::GetInternalFormatInfo(getInternalFormat(level)).depthBits > 0; +} + +void TextureD3D_2DArray::setImage(GLenum target, GLint level, GLsizei width, GLsizei height, GLsizei depth, GLenum internalFormat, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels) +{ + ASSERT(target == GL_TEXTURE_2D_ARRAY); + + GLenum sizedInternalFormat = gl::GetSizedInternalFormat(internalFormat, type); + + redefineImage(level, sizedInternalFormat, width, height, depth); + + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(sizedInternalFormat); + GLsizei inputDepthPitch = formatInfo.computeDepthPitch(type, width, height, unpack.alignment); + + for (int i = 0; i < depth; i++) + { + const void *layerPixels = pixels ? (reinterpret_cast<const unsigned char*>(pixels) + (inputDepthPitch * i)) : NULL; + TextureD3D::setImage(unpack, type, layerPixels, mImageArray[level][i]); + } +} + +void TextureD3D_2DArray::setCompressedImage(GLenum target, GLint level, GLenum format, GLsizei width, GLsizei height, GLsizei depth, GLsizei imageSize, const void *pixels) +{ + ASSERT(target == GL_TEXTURE_2D_ARRAY); + + // compressed formats don't have separate sized internal formats-- we can just use the compressed format directly + redefineImage(level, format, width, height, depth); + + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(format); + GLsizei inputDepthPitch = formatInfo.computeDepthPitch(GL_UNSIGNED_BYTE, width, height, 1); + + for (int i = 0; i < depth; i++) + { + const void *layerPixels = pixels ? (reinterpret_cast<const unsigned char*>(pixels) + (inputDepthPitch * i)) : NULL; + TextureD3D::setCompressedImage(imageSize, layerPixels, mImageArray[level][i]); + } +} + +void TextureD3D_2DArray::subImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels) +{ + ASSERT(target == GL_TEXTURE_2D_ARRAY); + + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(getInternalFormat(level)); + GLsizei inputDepthPitch = formatInfo.computeDepthPitch(type, width, height, unpack.alignment); + + for (int i = 0; i < depth; i++) + { + int layer = zoffset + i; + const void *layerPixels = pixels ? (reinterpret_cast<const unsigned char*>(pixels) + (inputDepthPitch * i)) : NULL; + + gl::ImageIndex index = gl::ImageIndex::Make2DArray(level, layer); + if (TextureD3D::subImage(xoffset, yoffset, zoffset, width, height, 1, format, type, unpack, layerPixels, index)) + { + commitRect(level, xoffset, yoffset, layer, width, height); + } + } +} + +void TextureD3D_2DArray::subImageCompressed(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void *pixels) +{ + ASSERT(target == GL_TEXTURE_2D_ARRAY); + + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(format); + GLsizei inputDepthPitch = formatInfo.computeDepthPitch(GL_UNSIGNED_BYTE, width, height, 1); + + for (int i = 0; i < depth; i++) + { + int layer = zoffset + i; + const void *layerPixels = pixels ? (reinterpret_cast<const unsigned char*>(pixels) + (inputDepthPitch * i)) : NULL; + + if (TextureD3D::subImageCompressed(xoffset, yoffset, zoffset, width, height, 1, format, imageSize, layerPixels, mImageArray[level][layer])) + { + commitRect(level, xoffset, yoffset, layer, width, height); + } + } +} + +void TextureD3D_2DArray::copyImage(GLenum target, GLint level, GLenum format, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) +{ + UNIMPLEMENTED(); +} + +void TextureD3D_2DArray::copySubImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) +{ + ASSERT(target == GL_TEXTURE_2D_ARRAY); + + // can only make our texture storage to a render target if level 0 is defined (with a width & height) and + // the current level we're copying to is defined (with appropriate format, width & height) + bool canCreateRenderTarget = isLevelComplete(level) && isLevelComplete(0); + + if (!mImageArray[level][0]->isRenderableFormat() || (!mTexStorage && !canCreateRenderTarget)) + { + mImageArray[level][zoffset]->copy(xoffset, yoffset, 0, x, y, width, height, source); + mDirtyImages = true; + } + else + { + ensureRenderTarget(); + + if (isValidLevel(level)) + { + updateStorageLevel(level); + + gl::Rectangle sourceRect; + sourceRect.x = x; + sourceRect.width = width; + sourceRect.y = y; + sourceRect.height = height; + + mRenderer->copyImage2DArray(source, sourceRect, gl::GetInternalFormatInfo(getInternalFormat(0)).format, + xoffset, yoffset, zoffset, mTexStorage, level); + } + } +} + +void TextureD3D_2DArray::storage(GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth) +{ + ASSERT(target == GL_TEXTURE_2D_ARRAY); + + deleteImages(); + + for (int level = 0; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; level++) + { + GLsizei levelWidth = std::max(1, width >> level); + GLsizei levelHeight = std::max(1, height >> level); + + mLayerCounts[level] = (level < levels ? depth : 0); + + if (mLayerCounts[level] > 0) + { + // Create new images for this level + mImageArray[level] = new ImageD3D*[mLayerCounts[level]]; + + for (int layer = 0; layer < mLayerCounts[level]; layer++) + { + mImageArray[level][layer] = ImageD3D::makeImageD3D(mRenderer->createImage()); + mImageArray[level][layer]->redefine(mRenderer, GL_TEXTURE_2D_ARRAY, internalformat, levelWidth, + levelHeight, 1, true); + } + } + } + + mImmutable = true; + + bool renderTarget = IsRenderTargetUsage(mUsage); + TextureStorage *storage = mRenderer->createTextureStorage2DArray(internalformat, renderTarget, width, height, depth, levels); + setCompleteTexStorage(storage); +} + +void TextureD3D_2DArray::bindTexImage(egl::Surface *surface) +{ + UNREACHABLE(); +} + +void TextureD3D_2DArray::releaseTexImage() +{ + UNREACHABLE(); +} + + +void TextureD3D_2DArray::generateMipmaps() +{ + int baseWidth = getBaseLevelWidth(); + int baseHeight = getBaseLevelHeight(); + int baseDepth = getBaseLevelDepth(); + GLenum baseFormat = getBaseLevelInternalFormat(); + + // Purge array levels 1 through q and reset them to represent the generated mipmap levels. + int levelCount = mipLevels(); + for (int level = 1; level < levelCount; level++) + { + redefineImage(level, baseFormat, std::max(baseWidth >> level, 1), std::max(baseHeight >> level, 1), baseDepth); + } + + if (mTexStorage && mTexStorage->isRenderTarget()) + { + mTexStorage->generateMipmaps(); + + for (int level = 1; level < levelCount; level++) + { + for (int layer = 0; layer < mLayerCounts[level]; layer++) + { + mImageArray[level][layer]->markClean(); + } + } + } + else + { + for (int level = 1; level < levelCount; level++) + { + for (int layer = 0; layer < mLayerCounts[level]; layer++) + { + mRenderer->generateMipmap(mImageArray[level][layer], mImageArray[level - 1][layer]); + } + } + } +} + +unsigned int TextureD3D_2DArray::getRenderTargetSerial(const gl::ImageIndex &index) +{ + return (ensureRenderTarget() ? mTexStorage->getRenderTargetSerial(index) : 0); +} + +RenderTarget *TextureD3D_2DArray::getRenderTarget(const gl::ImageIndex &index) +{ + // ensure the underlying texture is created + if (!ensureRenderTarget()) + { + return NULL; + } + + updateStorageLevel(index.mipIndex); + return mTexStorage->getRenderTarget(index); +} + +void TextureD3D_2DArray::initializeStorage(bool renderTarget) +{ + // Only initialize the first time this texture is used as a render target or shader resource + if (mTexStorage) + { + return; + } + + // do not attempt to create storage for nonexistant data + if (!isLevelComplete(0)) + { + return; + } + + bool createRenderTarget = (renderTarget || mUsage == GL_FRAMEBUFFER_ATTACHMENT_ANGLE); + + setCompleteTexStorage(createCompleteStorage(createRenderTarget)); + ASSERT(mTexStorage); + + // flush image data to the storage + updateStorage(); +} + +TextureStorage *TextureD3D_2DArray::createCompleteStorage(bool renderTarget) const +{ + GLsizei width = getBaseLevelWidth(); + GLsizei height = getBaseLevelHeight(); + GLsizei depth = getLayers(0); + GLenum internalFormat = getBaseLevelInternalFormat(); + + ASSERT(width > 0 && height > 0 && depth > 0); + + // use existing storage level count, when previously specified by TexStorage*D + GLint levels = (mTexStorage ? mTexStorage->getLevelCount() : creationLevels(width, height, 1)); + + return mRenderer->createTextureStorage2DArray(internalFormat, renderTarget, width, height, depth, levels); +} + +void TextureD3D_2DArray::setCompleteTexStorage(TextureStorage *newCompleteTexStorage) +{ + SafeDelete(mTexStorage); + mTexStorage = newCompleteTexStorage; + mDirtyImages = true; + + // We do not support managed 2D array storage, as managed storage is ES2/D3D9 only + ASSERT(!mTexStorage->isManaged()); +} + +void TextureD3D_2DArray::updateStorage() +{ + ASSERT(mTexStorage != NULL); + GLint storageLevels = mTexStorage->getLevelCount(); + for (int level = 0; level < storageLevels; level++) + { + if (isLevelComplete(level)) + { + updateStorageLevel(level); + } + } +} + +bool TextureD3D_2DArray::ensureRenderTarget() +{ + initializeStorage(true); + + if (getBaseLevelWidth() > 0 && getBaseLevelHeight() > 0 && getLayers(0) > 0) + { + ASSERT(mTexStorage); + if (!mTexStorage->isRenderTarget()) + { + TextureStorage *newRenderTargetStorage = createCompleteStorage(true); + + if (!mRenderer->copyToRenderTarget2DArray(newRenderTargetStorage, mTexStorage)) + { + delete newRenderTargetStorage; + return gl::error(GL_OUT_OF_MEMORY, false); + } + + setCompleteTexStorage(newRenderTargetStorage); + } + } + + return (mTexStorage && mTexStorage->isRenderTarget()); +} + +const ImageD3D *TextureD3D_2DArray::getBaseLevelImage() const +{ + return (mLayerCounts[0] > 0 ? mImageArray[0][0] : NULL); +} + +TextureStorage *TextureD3D_2DArray::getBaseLevelStorage() +{ + return mTexStorage; +} + +bool TextureD3D_2DArray::isValidLevel(int level) const +{ + return (mTexStorage ? (level >= 0 && level < mTexStorage->getLevelCount()) : 0); +} + +bool TextureD3D_2DArray::isLevelComplete(int level) const +{ + ASSERT(level >= 0 && level < (int)ArraySize(mImageArray)); + + if (isImmutable()) + { + return true; + } + + GLsizei width = getBaseLevelWidth(); + GLsizei height = getBaseLevelHeight(); + GLsizei layers = getLayers(0); + + if (width <= 0 || height <= 0 || layers <= 0) + { + return false; + } + + if (level == 0) + { + return true; + } + + if (getInternalFormat(level) != getInternalFormat(0)) + { + return false; + } + + if (getWidth(level) != std::max(1, width >> level)) + { + return false; + } + + if (getHeight(level) != std::max(1, height >> level)) + { + return false; + } + + if (getLayers(level) != layers) + { + return false; + } + + return true; +} + +void TextureD3D_2DArray::updateStorageLevel(int level) +{ + ASSERT(level >= 0 && level < (int)ArraySize(mLayerCounts)); + ASSERT(isLevelComplete(level)); + + for (int layer = 0; layer < mLayerCounts[level]; layer++) + { + ASSERT(mImageArray[level] != NULL && mImageArray[level][layer] != NULL); + if (mImageArray[level][layer]->isDirty()) + { + commitRect(level, 0, 0, layer, getWidth(level), getHeight(level)); + } + } +} + +void TextureD3D_2DArray::deleteImages() +{ + for (int level = 0; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; ++level) + { + for (int layer = 0; layer < mLayerCounts[level]; ++layer) + { + delete mImageArray[level][layer]; + } + delete[] mImageArray[level]; + mImageArray[level] = NULL; + mLayerCounts[level] = 0; + } +} + +void TextureD3D_2DArray::redefineImage(GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth) +{ + // If there currently is a corresponding storage texture image, it has these parameters + const int storageWidth = std::max(1, getBaseLevelWidth() >> level); + const int storageHeight = std::max(1, getBaseLevelHeight() >> level); + const int storageDepth = getLayers(0); + const GLenum storageFormat = getBaseLevelInternalFormat(); + + for (int layer = 0; layer < mLayerCounts[level]; layer++) + { + delete mImageArray[level][layer]; + } + delete[] mImageArray[level]; + mImageArray[level] = NULL; + mLayerCounts[level] = depth; + + if (depth > 0) + { + mImageArray[level] = new ImageD3D*[depth](); + + for (int layer = 0; layer < mLayerCounts[level]; layer++) + { + mImageArray[level][layer] = ImageD3D::makeImageD3D(mRenderer->createImage()); + mImageArray[level][layer]->redefine(mRenderer, GL_TEXTURE_2D_ARRAY, internalformat, width, height, 1, false); + } + } + + if (mTexStorage) + { + const int storageLevels = mTexStorage->getLevelCount(); + + if ((level >= storageLevels && storageLevels != 0) || + width != storageWidth || + height != storageHeight || + depth != storageDepth || + internalformat != storageFormat) // Discard mismatched storage + { + for (int level = 0; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; level++) + { + for (int layer = 0; layer < mLayerCounts[level]; layer++) + { + mImageArray[level][layer]->markDirty(); + } + } + + delete mTexStorage; + mTexStorage = NULL; + mDirtyImages = true; + } + } +} + +void TextureD3D_2DArray::commitRect(GLint level, GLint xoffset, GLint yoffset, GLint layerTarget, GLsizei width, GLsizei height) +{ + if (isValidLevel(level) && layerTarget < getLayers(level)) + { + ImageD3D *image = mImageArray[level][layerTarget]; + if (image->copyToStorage2DArray(mTexStorage, level, xoffset, yoffset, layerTarget, width, height)) + { + image->markClean(); + } + } +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TextureD3D.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TextureD3D.h new file mode 100644 index 0000000000..41c73180de --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TextureD3D.h @@ -0,0 +1,320 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// TextureD3D.h: Implementations of the Texture interfaces shared betweeen the D3D backends. + +#ifndef LIBGLESV2_RENDERER_TEXTURED3D_H_ +#define LIBGLESV2_RENDERER_TEXTURED3D_H_ + +#include "libGLESv2/renderer/TextureImpl.h" +#include "libGLESv2/angletypes.h" +#include "libGLESv2/constants.h" + +namespace gl +{ +class Framebuffer; +} + +namespace rx +{ + +class Image; +class ImageD3D; +class Renderer; +class RenderTarget; +class TextureStorage; + +class TextureD3D : public TextureImpl +{ + public: + TextureD3D(Renderer *renderer); + virtual ~TextureD3D(); + + static TextureD3D *makeTextureD3D(TextureImpl *texture); + + virtual TextureStorage *getNativeTexture(); + + virtual void setUsage(GLenum usage) { mUsage = usage; } + bool hasDirtyImages() const { return mDirtyImages; } + void resetDirty() { mDirtyImages = false; } + + GLint getBaseLevelWidth() const; + GLint getBaseLevelHeight() const; + GLint getBaseLevelDepth() const; + GLenum getBaseLevelInternalFormat() const; + + bool isImmutable() const { return mImmutable; } + + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index) = 0; + virtual unsigned int getRenderTargetSerial(const gl::ImageIndex &index) = 0; + + protected: + void setImage(const gl::PixelUnpackState &unpack, GLenum type, const void *pixels, Image *image); + bool subImage(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, + GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels, const gl::ImageIndex &index); + void setCompressedImage(GLsizei imageSize, const void *pixels, Image *image); + bool subImageCompressed(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, + GLenum format, GLsizei imageSize, const void *pixels, Image *image); + bool isFastUnpackable(const gl::PixelUnpackState &unpack, GLenum sizedInternalFormat); + bool fastUnpackPixels(const gl::PixelUnpackState &unpack, const void *pixels, const gl::Box &destArea, + GLenum sizedInternalFormat, GLenum type, RenderTarget *destRenderTarget); + + GLint creationLevels(GLsizei width, GLsizei height, GLsizei depth) const; + int mipLevels() const; + + Renderer *mRenderer; + + GLenum mUsage; + + bool mDirtyImages; + + bool mImmutable; + + private: + DISALLOW_COPY_AND_ASSIGN(TextureD3D); + + virtual void initializeStorage(bool renderTarget) = 0; + + virtual void updateStorage() = 0; + virtual TextureStorage *getBaseLevelStorage() = 0; + virtual const ImageD3D *getBaseLevelImage() const = 0; +}; + +class TextureD3D_2D : public TextureD3D +{ + public: + TextureD3D_2D(Renderer *renderer); + virtual ~TextureD3D_2D(); + + virtual Image *getImage(int level, int layer) const; + virtual Image *getImage(const gl::ImageIndex &index) const; + virtual GLsizei getLayerCount(int level) const; + + GLsizei getWidth(GLint level) const; + GLsizei getHeight(GLint level) const; + GLenum getInternalFormat(GLint level) const; + GLenum getActualFormat(GLint level) const; + bool isDepth(GLint level) const; + + virtual void setImage(GLenum target, GLint level, GLsizei width, GLsizei height, GLsizei depth, GLenum internalFormat, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels); + virtual void setCompressedImage(GLenum target, GLint level, GLenum format, GLsizei width, GLsizei height, GLsizei depth, GLsizei imageSize, const void *pixels); + virtual void subImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels); + virtual void subImageCompressed(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void *pixels); + virtual void copyImage(GLenum target, GLint level, GLenum format, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source); + virtual void copySubImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source); + virtual void storage(GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth); + + virtual void bindTexImage(egl::Surface *surface); + virtual void releaseTexImage(); + + virtual void generateMipmaps(); + + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index); + virtual unsigned int getRenderTargetSerial(const gl::ImageIndex &index); + + private: + DISALLOW_COPY_AND_ASSIGN(TextureD3D_2D); + + virtual void initializeStorage(bool renderTarget); + TextureStorage *createCompleteStorage(bool renderTarget) const; + void setCompleteTexStorage(TextureStorage *newCompleteTexStorage); + + virtual void updateStorage(); + bool ensureRenderTarget(); + virtual TextureStorage *getBaseLevelStorage(); + virtual const ImageD3D *getBaseLevelImage() const; + + bool isValidLevel(int level) const; + bool isLevelComplete(int level) const; + + void updateStorageLevel(int level); + + void redefineImage(GLint level, GLenum internalformat, GLsizei width, GLsizei height); + void commitRect(GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); + + TextureStorage *mTexStorage; + ImageD3D *mImageArray[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; +}; + +class TextureD3D_Cube : public TextureD3D +{ + public: + TextureD3D_Cube(Renderer *renderer); + virtual ~TextureD3D_Cube(); + + virtual Image *getImage(int level, int layer) const; + virtual Image *getImage(const gl::ImageIndex &index) const; + virtual GLsizei getLayerCount(int level) const; + + virtual bool hasDirtyImages() const { return mDirtyImages; } + virtual void resetDirty() { mDirtyImages = false; } + virtual void setUsage(GLenum usage) { mUsage = usage; } + + GLenum getInternalFormat(GLint level, GLint layer) const; + bool isDepth(GLint level, GLint layer) const; + + virtual void setImage(GLenum target, GLint level, GLsizei width, GLsizei height, GLsizei depth, GLenum internalFormat, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels); + virtual void setCompressedImage(GLenum target, GLint level, GLenum format, GLsizei width, GLsizei height, GLsizei depth, GLsizei imageSize, const void *pixels); + virtual void subImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels); + virtual void subImageCompressed(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void *pixels); + virtual void copyImage(GLenum target, GLint level, GLenum format, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source); + virtual void copySubImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source); + virtual void storage(GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth); + + virtual void bindTexImage(egl::Surface *surface); + virtual void releaseTexImage(); + + virtual void generateMipmaps(); + + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index); + virtual unsigned int getRenderTargetSerial(const gl::ImageIndex &index); + + private: + DISALLOW_COPY_AND_ASSIGN(TextureD3D_Cube); + + virtual void initializeStorage(bool renderTarget); + TextureStorage *createCompleteStorage(bool renderTarget) const; + void setCompleteTexStorage(TextureStorage *newCompleteTexStorage); + + virtual void updateStorage(); + bool ensureRenderTarget(); + virtual TextureStorage *getBaseLevelStorage(); + virtual const ImageD3D *getBaseLevelImage() const; + + bool isValidFaceLevel(int faceIndex, int level) const; + bool isFaceLevelComplete(int faceIndex, int level) const; + bool isCubeComplete() const; + void updateStorageFaceLevel(int faceIndex, int level); + + void redefineImage(int faceIndex, GLint level, GLenum internalformat, GLsizei width, GLsizei height); + void commitRect(int faceIndex, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); + + ImageD3D *mImageArray[6][gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; + + TextureStorage *mTexStorage; +}; + +class TextureD3D_3D : public TextureD3D +{ + public: + TextureD3D_3D(Renderer *renderer); + virtual ~TextureD3D_3D(); + + virtual Image *getImage(int level, int layer) const; + virtual Image *getImage(const gl::ImageIndex &index) const; + virtual GLsizei getLayerCount(int level) const; + + GLsizei getWidth(GLint level) const; + GLsizei getHeight(GLint level) const; + GLsizei getDepth(GLint level) const; + GLenum getInternalFormat(GLint level) const; + bool isDepth(GLint level) const; + + virtual void setImage(GLenum target, GLint level, GLsizei width, GLsizei height, GLsizei depth, GLenum internalFormat, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels); + virtual void setCompressedImage(GLenum target, GLint level, GLenum format, GLsizei width, GLsizei height, GLsizei depth, GLsizei imageSize, const void *pixels); + virtual void subImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels); + virtual void subImageCompressed(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void *pixels); + virtual void copyImage(GLenum target, GLint level, GLenum format, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source); + virtual void copySubImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source); + virtual void storage(GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth); + + virtual void bindTexImage(egl::Surface *surface); + virtual void releaseTexImage(); + + virtual void generateMipmaps(); + + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index); + virtual unsigned int getRenderTargetSerial(const gl::ImageIndex &index); + + private: + DISALLOW_COPY_AND_ASSIGN(TextureD3D_3D); + + virtual void initializeStorage(bool renderTarget); + TextureStorage *createCompleteStorage(bool renderTarget) const; + void setCompleteTexStorage(TextureStorage *newCompleteTexStorage); + + virtual void updateStorage(); + bool ensureRenderTarget(); + virtual TextureStorage *getBaseLevelStorage(); + virtual const ImageD3D *getBaseLevelImage() const; + + bool isValidLevel(int level) const; + bool isLevelComplete(int level) const; + void updateStorageLevel(int level); + + void redefineImage(GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth); + void commitRect(GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth); + + ImageD3D *mImageArray[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; + + TextureStorage *mTexStorage; +}; + +class TextureD3D_2DArray : public TextureD3D +{ + public: + TextureD3D_2DArray(Renderer *renderer); + virtual ~TextureD3D_2DArray(); + + virtual Image *getImage(int level, int layer) const; + virtual Image *getImage(const gl::ImageIndex &index) const; + virtual GLsizei getLayerCount(int level) const; + + GLsizei getWidth(GLint level) const; + GLsizei getHeight(GLint level) const; + GLsizei getLayers(GLint level) const; + GLenum getInternalFormat(GLint level) const; + bool isDepth(GLint level) const; + + virtual void setImage(GLenum target, GLint level, GLsizei width, GLsizei height, GLsizei depth, GLenum internalFormat, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels); + virtual void setCompressedImage(GLenum target, GLint level, GLenum format, GLsizei width, GLsizei height, GLsizei depth, GLsizei imageSize, const void *pixels); + virtual void subImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const gl::PixelUnpackState &unpack, const void *pixels); + virtual void subImageCompressed(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void *pixels); + virtual void copyImage(GLenum target, GLint level, GLenum format, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source); + virtual void copySubImage(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source); + virtual void storage(GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth); + + virtual void bindTexImage(egl::Surface *surface); + virtual void releaseTexImage(); + + virtual void generateMipmaps(); + + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index); + virtual unsigned int getRenderTargetSerial(const gl::ImageIndex &index); + + private: + DISALLOW_COPY_AND_ASSIGN(TextureD3D_2DArray); + + virtual void initializeStorage(bool renderTarget); + TextureStorage *createCompleteStorage(bool renderTarget) const; + void setCompleteTexStorage(TextureStorage *newCompleteTexStorage); + + virtual void updateStorage(); + bool ensureRenderTarget(); + virtual TextureStorage *getBaseLevelStorage(); + virtual const ImageD3D *getBaseLevelImage() const; + + bool isValidLevel(int level) const; + bool isLevelComplete(int level) const; + void updateStorageLevel(int level); + + void deleteImages(); + void redefineImage(GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth); + void commitRect(GLint level, GLint xoffset, GLint yoffset, GLint layerTarget, GLsizei width, GLsizei height); + + // Storing images as an array of single depth textures since D3D11 treats each array level of a + // Texture2D object as a separate subresource. Each layer would have to be looped over + // to update all the texture layers since they cannot all be updated at once and it makes the most + // sense for the Image class to not have to worry about layer subresource as well as mip subresources. + GLsizei mLayerCounts[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; + ImageD3D **mImageArray[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; + + TextureStorage *mTexStorage; +}; + +} + +#endif // LIBGLESV2_RENDERER_TEXTURED3D_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TextureStorage.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TextureStorage.cpp new file mode 100644 index 0000000000..dedd266c09 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TextureStorage.cpp @@ -0,0 +1,51 @@ +// +// Copyright (c) 2002-2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// TextureStorage.cpp: Shared members of abstract rx::TextureStorage class. + +#include "libGLESv2/renderer/d3d/TextureStorage.h" +#include "libGLESv2/renderer/d3d/TextureD3D.h" +#include "libGLESv2/renderer/Renderer.h" +#include "libGLESv2/Renderbuffer.h" +#include "libGLESv2/Texture.h" + +#include "common/debug.h" +#include "common/mathutil.h" + +namespace rx +{ + +unsigned int TextureStorage::mCurrentTextureSerial = 1; + +TextureStorage::TextureStorage() + : mTextureSerial(issueTextureSerial()), + mFirstRenderTargetSerial(0), + mRenderTargetSerialsLayerStride(0) +{} + +void TextureStorage::initializeSerials(unsigned int rtSerialsToReserve, unsigned int rtSerialsLayerStride) +{ + mFirstRenderTargetSerial = gl::RenderbufferStorage::issueSerials(rtSerialsToReserve); + mRenderTargetSerialsLayerStride = rtSerialsLayerStride; +} + +unsigned int TextureStorage::getRenderTargetSerial(const gl::ImageIndex &index) const +{ + unsigned int layerOffset = (index.hasLayer() ? (static_cast<unsigned int>(index.layerIndex) * mRenderTargetSerialsLayerStride) : 0); + return mFirstRenderTargetSerial + static_cast<unsigned int>(index.mipIndex) + layerOffset; +} + +unsigned int TextureStorage::getTextureSerial() const +{ + return mTextureSerial; +} + +unsigned int TextureStorage::issueTextureSerial() +{ + return mCurrentTextureSerial++; +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TextureStorage.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TextureStorage.h new file mode 100644 index 0000000000..9cc2c2977b --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TextureStorage.h @@ -0,0 +1,61 @@ +// +// Copyright (c) 2002-2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// TextureStorage.h: Defines the abstract rx::TextureStorage class. + +#ifndef LIBGLESV2_RENDERER_TEXTURESTORAGE_H_ +#define LIBGLESV2_RENDERER_TEXTURESTORAGE_H_ + +#include "common/debug.h" + +#include <GLES2/gl2.h> + +namespace gl +{ +struct ImageIndex; +} + +namespace rx +{ +class Renderer; +class SwapChain; +class RenderTarget; + +class TextureStorage +{ + public: + TextureStorage(); + virtual ~TextureStorage() {}; + + virtual int getTopLevel() const = 0; + virtual bool isRenderTarget() const = 0; + virtual bool isManaged() const = 0; + virtual int getLevelCount() const = 0; + + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index) = 0; + virtual void generateMipmaps() = 0; + + unsigned int getRenderTargetSerial(const gl::ImageIndex &index) const; + unsigned int getTextureSerial() const; + + protected: + void initializeSerials(unsigned int rtSerialsToReserve, unsigned int rtSerialsLayerStride); + + private: + DISALLOW_COPY_AND_ASSIGN(TextureStorage); + + const unsigned int mTextureSerial; + static unsigned int issueTextureSerial(); + + static unsigned int mCurrentTextureSerial; + + unsigned int mFirstRenderTargetSerial; + unsigned int mRenderTargetSerialsLayerStride; +}; + +} + +#endif // LIBGLESV2_RENDERER_TEXTURESTORAGE_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TransformFeedbackD3D.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TransformFeedbackD3D.cpp new file mode 100644 index 0000000000..11596006d0 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TransformFeedbackD3D.cpp @@ -0,0 +1,38 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// TransformFeedbackD3D.cpp is a no-op implementation for both the D3D9 and D3D11 renderers. + +#include "libGLESv2/renderer/d3d/TransformFeedbackD3D.h" + +namespace rx +{ + +TransformFeedbackD3D::TransformFeedbackD3D() +{ +} + +TransformFeedbackD3D::~TransformFeedbackD3D() +{ +} + +void TransformFeedbackD3D::begin(GLenum primitiveMode) +{ +} + +void TransformFeedbackD3D::end() +{ +} + +void TransformFeedbackD3D::pause() +{ +} + +void TransformFeedbackD3D::resume() +{ +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TransformFeedbackD3D.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TransformFeedbackD3D.h new file mode 100644 index 0000000000..7c367aba1d --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/TransformFeedbackD3D.h @@ -0,0 +1,32 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// TransformFeedbackD3D.h: Implements the abstract rx::TransformFeedbackImpl class. + +#ifndef LIBGLESV2_RENDERER_D3D_TRANSFORMFEEDBACKD3D_H_ +#define LIBGLESV2_RENDERER_D3D_TRANSFORMFEEDBACKD3D_H_ + +#include "libGLESv2/renderer/TransformFeedbackImpl.h" +#include "libGLESv2/angletypes.h" + +namespace rx +{ + +class TransformFeedbackD3D : public TransformFeedbackImpl +{ + public: + TransformFeedbackD3D(); + virtual ~TransformFeedbackD3D(); + + virtual void begin(GLenum primitiveMode); + virtual void end(); + virtual void pause(); + virtual void resume(); +}; + +} + +#endif // LIBGLESV2_RENDERER_D3D_TRANSFORMFEEDBACKD3D_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/VertexBuffer.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/VertexBuffer.cpp new file mode 100644 index 0000000000..4f85eb94fa --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/VertexBuffer.cpp @@ -0,0 +1,307 @@ +// +// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// VertexBuffer.cpp: Defines the abstract VertexBuffer class and VertexBufferInterface +// class with derivations, classes that perform graphics API agnostic vertex buffer operations. + +#include "libGLESv2/renderer/d3d/VertexBuffer.h" +#include "libGLESv2/renderer/d3d/BufferD3D.h" +#include "libGLESv2/renderer/Renderer.h" +#include "libGLESv2/VertexAttribute.h" + +#include "common/mathutil.h" + +namespace rx +{ + +unsigned int VertexBuffer::mNextSerial = 1; + +VertexBuffer::VertexBuffer() +{ + updateSerial(); +} + +VertexBuffer::~VertexBuffer() +{ +} + +void VertexBuffer::updateSerial() +{ + mSerial = mNextSerial++; +} + +unsigned int VertexBuffer::getSerial() const +{ + return mSerial; +} + +VertexBufferInterface::VertexBufferInterface(rx::Renderer *renderer, bool dynamic) : mRenderer(renderer) +{ + mDynamic = dynamic; + mWritePosition = 0; + mReservedSpace = 0; + + mVertexBuffer = renderer->createVertexBuffer(); +} + +VertexBufferInterface::~VertexBufferInterface() +{ + delete mVertexBuffer; +} + +unsigned int VertexBufferInterface::getSerial() const +{ + return mVertexBuffer->getSerial(); +} + +unsigned int VertexBufferInterface::getBufferSize() const +{ + return mVertexBuffer->getBufferSize(); +} + +gl::Error VertexBufferInterface::setBufferSize(unsigned int size) +{ + if (mVertexBuffer->getBufferSize() == 0) + { + return mVertexBuffer->initialize(size, mDynamic); + } + else + { + return mVertexBuffer->setBufferSize(size); + } +} + +unsigned int VertexBufferInterface::getWritePosition() const +{ + return mWritePosition; +} + +void VertexBufferInterface::setWritePosition(unsigned int writePosition) +{ + mWritePosition = writePosition; +} + +gl::Error VertexBufferInterface::discard() +{ + return mVertexBuffer->discard(); +} + +gl::Error VertexBufferInterface::storeVertexAttributes(const gl::VertexAttribute &attrib, const gl::VertexAttribCurrentValueData ¤tValue, + GLint start, GLsizei count, GLsizei instances, unsigned int *outStreamOffset) +{ + gl::Error error(GL_NO_ERROR); + + unsigned int spaceRequired; + error = mVertexBuffer->getSpaceRequired(attrib, count, instances, &spaceRequired); + if (error.isError()) + { + return error; + } + + if (mWritePosition + spaceRequired < mWritePosition) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal error, new vertex buffer write position would overflow."); + } + + error = reserveSpace(mReservedSpace); + if (error.isError()) + { + return error; + } + mReservedSpace = 0; + + error = mVertexBuffer->storeVertexAttributes(attrib, currentValue, start, count, instances, mWritePosition); + if (error.isError()) + { + return error; + } + + if (outStreamOffset) + { + *outStreamOffset = mWritePosition; + } + + mWritePosition += spaceRequired; + + // Align to 16-byte boundary + mWritePosition = rx::roundUp(mWritePosition, 16u); + + return gl::Error(GL_NO_ERROR); +} + +gl::Error VertexBufferInterface::reserveVertexSpace(const gl::VertexAttribute &attrib, GLsizei count, GLsizei instances) +{ + gl::Error error(GL_NO_ERROR); + + unsigned int requiredSpace; + error = mVertexBuffer->getSpaceRequired(attrib, count, instances, &requiredSpace); + if (error.isError()) + { + return error; + } + + // Protect against integer overflow + if (mReservedSpace + requiredSpace < mReservedSpace) + { + return gl::Error(GL_OUT_OF_MEMORY, "Unable to reserve %u extra bytes in internal vertex buffer, " + "it would result in an overflow.", requiredSpace); + } + + mReservedSpace += requiredSpace; + + // Align to 16-byte boundary + mReservedSpace = rx::roundUp(mReservedSpace, 16u); + + return gl::Error(GL_NO_ERROR); +} + +VertexBuffer* VertexBufferInterface::getVertexBuffer() const +{ + return mVertexBuffer; +} + +bool VertexBufferInterface::directStoragePossible(const gl::VertexAttribute &attrib, + const gl::VertexAttribCurrentValueData ¤tValue) const +{ + gl::Buffer *buffer = attrib.buffer.get(); + BufferD3D *storage = buffer ? BufferD3D::makeBufferD3D(buffer->getImplementation()) : NULL; + + if (!storage || !storage->supportsDirectBinding()) + { + return false; + } + + // Alignment restrictions: In D3D, vertex data must be aligned to + // the format stride, or to a 4-byte boundary, whichever is smaller. + // (Undocumented, and experimentally confirmed) + size_t alignment = 4; + bool requiresConversion = false; + + if (attrib.type != GL_FLOAT) + { + gl::VertexFormat vertexFormat(attrib, currentValue.Type); + + unsigned int outputElementSize; + getVertexBuffer()->getSpaceRequired(attrib, 1, 0, &outputElementSize); + alignment = std::min<size_t>(outputElementSize, 4); + + requiresConversion = (mRenderer->getVertexConversionType(vertexFormat) & VERTEX_CONVERT_CPU) != 0; + } + + bool isAligned = (static_cast<size_t>(ComputeVertexAttributeStride(attrib)) % alignment == 0) && + (static_cast<size_t>(attrib.offset) % alignment == 0); + + return !requiresConversion && isAligned; +} + +StreamingVertexBufferInterface::StreamingVertexBufferInterface(rx::Renderer *renderer, std::size_t initialSize) : VertexBufferInterface(renderer, true) +{ + setBufferSize(initialSize); +} + +StreamingVertexBufferInterface::~StreamingVertexBufferInterface() +{ +} + +gl::Error StreamingVertexBufferInterface::reserveSpace(unsigned int size) +{ + unsigned int curBufferSize = getBufferSize(); + if (size > curBufferSize) + { + gl::Error error = setBufferSize(std::max(size, 3 * curBufferSize / 2)); + if (error.isError()) + { + return error; + } + setWritePosition(0); + } + else if (getWritePosition() + size > curBufferSize) + { + gl::Error error = discard(); + if (error.isError()) + { + return error; + } + setWritePosition(0); + } + + return gl::Error(GL_NO_ERROR); +} + +StaticVertexBufferInterface::StaticVertexBufferInterface(rx::Renderer *renderer) : VertexBufferInterface(renderer, false) +{ +} + +StaticVertexBufferInterface::~StaticVertexBufferInterface() +{ +} + +bool StaticVertexBufferInterface::lookupAttribute(const gl::VertexAttribute &attrib, unsigned int *outStreamOffset) +{ + for (unsigned int element = 0; element < mCache.size(); element++) + { + if (mCache[element].type == attrib.type && + mCache[element].size == attrib.size && + mCache[element].stride == ComputeVertexAttributeStride(attrib) && + mCache[element].normalized == attrib.normalized && + mCache[element].pureInteger == attrib.pureInteger) + { + size_t offset = (static_cast<size_t>(attrib.offset) % ComputeVertexAttributeStride(attrib)); + if (mCache[element].attributeOffset == offset) + { + if (outStreamOffset) + { + *outStreamOffset = mCache[element].streamOffset; + } + return true; + } + } + } + + return false; +} + +gl::Error StaticVertexBufferInterface::reserveSpace(unsigned int size) +{ + unsigned int curSize = getBufferSize(); + if (curSize == 0) + { + return setBufferSize(size); + } + else if (curSize >= size) + { + return gl::Error(GL_NO_ERROR); + } + else + { + UNREACHABLE(); + return gl::Error(GL_INVALID_OPERATION, "Internal error, Static vertex buffers can't be resized."); + } +} + +gl::Error StaticVertexBufferInterface::storeVertexAttributes(const gl::VertexAttribute &attrib, const gl::VertexAttribCurrentValueData ¤tValue, + GLint start, GLsizei count, GLsizei instances, unsigned int *outStreamOffset) +{ + unsigned int streamOffset; + gl::Error error = VertexBufferInterface::storeVertexAttributes(attrib, currentValue, start, count, instances, &streamOffset); + if (error.isError()) + { + return error; + } + + size_t attributeOffset = static_cast<size_t>(attrib.offset) % ComputeVertexAttributeStride(attrib); + VertexElement element = { attrib.type, attrib.size, ComputeVertexAttributeStride(attrib), attrib.normalized, attrib.pureInteger, attributeOffset, streamOffset }; + mCache.push_back(element); + + if (outStreamOffset) + { + *outStreamOffset = streamOffset; + } + + return gl::Error(GL_NO_ERROR); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/VertexBuffer.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/VertexBuffer.h index cbafdd20f6..fa747d9cb4 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/VertexBuffer.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/VertexBuffer.h @@ -11,10 +11,17 @@ #define LIBGLESV2_RENDERER_VERTEXBUFFER_H_ #include "common/angleutils.h" +#include "libGLESv2/Error.h" + +#include <GLES2/gl2.h> + +#include <cstddef> +#include <vector> namespace gl { -class VertexAttribute; +struct VertexAttribute; +struct VertexAttribCurrentValueData; } namespace rx @@ -27,20 +34,16 @@ class VertexBuffer VertexBuffer(); virtual ~VertexBuffer(); - virtual bool initialize(unsigned int size, bool dynamicUsage) = 0; - - virtual bool storeVertexAttributes(const gl::VertexAttribute &attrib, GLint start, GLsizei count, - GLsizei instances, unsigned int offset) = 0; - virtual bool storeRawData(const void* data, unsigned int size, unsigned int offset) = 0; + virtual gl::Error initialize(unsigned int size, bool dynamicUsage) = 0; - virtual bool getSpaceRequired(const gl::VertexAttribute &attrib, GLsizei count, GLsizei instances, - unsigned int *outSpaceRequired) const = 0; - - virtual bool requiresConversion(const gl::VertexAttribute &attrib) const = 0; + virtual gl::Error storeVertexAttributes(const gl::VertexAttribute &attrib, const gl::VertexAttribCurrentValueData ¤tValue, + GLint start, GLsizei count, GLsizei instances, unsigned int offset) = 0; + virtual gl::Error getSpaceRequired(const gl::VertexAttribute &attrib, GLsizei count, GLsizei instances, + unsigned int *outSpaceRequired) const = 0; virtual unsigned int getBufferSize() const = 0; - virtual bool setBufferSize(unsigned int size) = 0; - virtual bool discard() = 0; + virtual gl::Error setBufferSize(unsigned int size) = 0; + virtual gl::Error discard() = 0; unsigned int getSerial() const; @@ -60,28 +63,29 @@ class VertexBufferInterface VertexBufferInterface(rx::Renderer *renderer, bool dynamic); virtual ~VertexBufferInterface(); - bool reserveVertexSpace(const gl::VertexAttribute &attribute, GLsizei count, GLsizei instances); - bool reserveRawDataSpace(unsigned int size); + gl::Error reserveVertexSpace(const gl::VertexAttribute &attribute, GLsizei count, GLsizei instances); unsigned int getBufferSize() const; unsigned int getSerial() const; - virtual bool storeVertexAttributes(const gl::VertexAttribute &attrib, GLint start, GLsizei count, GLsizei instances, - unsigned int *outStreamOffset); - virtual bool storeRawData(const void* data, unsigned int size, unsigned int *outStreamOffset); + virtual gl::Error storeVertexAttributes(const gl::VertexAttribute &attrib, const gl::VertexAttribCurrentValueData ¤tValue, + GLint start, GLsizei count, GLsizei instances, unsigned int *outStreamOffset); + + bool directStoragePossible(const gl::VertexAttribute &attrib, + const gl::VertexAttribCurrentValueData ¤tValue) const; VertexBuffer* getVertexBuffer() const; protected: - virtual bool reserveSpace(unsigned int size) = 0; + virtual gl::Error reserveSpace(unsigned int size) = 0; unsigned int getWritePosition() const; void setWritePosition(unsigned int writePosition); - bool discard(); + gl::Error discard(); - bool setBufferSize(unsigned int size); + gl::Error setBufferSize(unsigned int size); private: DISALLOW_COPY_AND_ASSIGN(VertexBufferInterface); @@ -102,7 +106,7 @@ class StreamingVertexBufferInterface : public VertexBufferInterface ~StreamingVertexBufferInterface(); protected: - bool reserveSpace(unsigned int size); + gl::Error reserveSpace(unsigned int size); }; class StaticVertexBufferInterface : public VertexBufferInterface @@ -111,22 +115,23 @@ class StaticVertexBufferInterface : public VertexBufferInterface explicit StaticVertexBufferInterface(rx::Renderer *renderer); ~StaticVertexBufferInterface(); - bool storeVertexAttributes(const gl::VertexAttribute &attrib, GLint start, GLsizei count, GLsizei instances, - unsigned int *outStreamOffset); + gl::Error storeVertexAttributes(const gl::VertexAttribute &attrib, const gl::VertexAttribCurrentValueData ¤tValue, + GLint start, GLsizei count, GLsizei instances, unsigned int *outStreamOffset); - bool lookupAttribute(const gl::VertexAttribute &attribute, unsigned int* outStreamOffset); + bool lookupAttribute(const gl::VertexAttribute &attribute, unsigned int* outStreamFffset); protected: - bool reserveSpace(unsigned int size); + gl::Error reserveSpace(unsigned int size); private: struct VertexElement { GLenum type; - GLint size; - GLsizei stride; + GLuint size; + GLuint stride; bool normalized; - int attributeOffset; + bool pureInteger; + size_t attributeOffset; unsigned int streamOffset; }; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/VertexDataManager.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/VertexDataManager.cpp new file mode 100644 index 0000000000..7034b78eab --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/VertexDataManager.cpp @@ -0,0 +1,348 @@ +// +// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// VertexDataManager.h: Defines the VertexDataManager, a class that +// runs the Buffer translation process. + +#include "libGLESv2/renderer/d3d/VertexDataManager.h" +#include "libGLESv2/renderer/d3d/BufferD3D.h" +#include "libGLESv2/renderer/d3d/VertexBuffer.h" +#include "libGLESv2/renderer/Renderer.h" +#include "libGLESv2/Buffer.h" +#include "libGLESv2/ProgramBinary.h" +#include "libGLESv2/VertexAttribute.h" + +namespace +{ + enum { INITIAL_STREAM_BUFFER_SIZE = 1024*1024 }; + // This has to be at least 4k or else it fails on ATI cards. + enum { CONSTANT_VERTEX_BUFFER_SIZE = 4096 }; +} + +namespace rx +{ + +static int ElementsInBuffer(const gl::VertexAttribute &attrib, unsigned int size) +{ + // Size cannot be larger than a GLsizei + if (size > static_cast<unsigned int>(std::numeric_limits<int>::max())) + { + size = static_cast<unsigned int>(std::numeric_limits<int>::max()); + } + + GLsizei stride = ComputeVertexAttributeStride(attrib); + return (size - attrib.offset % stride + (stride - ComputeVertexAttributeTypeSize(attrib))) / stride; +} + +static int StreamingBufferElementCount(const gl::VertexAttribute &attrib, int vertexDrawCount, int instanceDrawCount) +{ + // For instanced rendering, we draw "instanceDrawCount" sets of "vertexDrawCount" vertices. + // + // A vertex attribute with a positive divisor loads one instanced vertex for every set of + // non-instanced vertices, and the instanced vertex index advances once every "mDivisor" instances. + if (instanceDrawCount > 0 && attrib.divisor > 0) + { + return instanceDrawCount / attrib.divisor; + } + + return vertexDrawCount; +} + +VertexDataManager::VertexDataManager(Renderer *renderer) : mRenderer(renderer) +{ + for (int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) + { + mCurrentValue[i].FloatValues[0] = std::numeric_limits<float>::quiet_NaN(); + mCurrentValue[i].FloatValues[1] = std::numeric_limits<float>::quiet_NaN(); + mCurrentValue[i].FloatValues[2] = std::numeric_limits<float>::quiet_NaN(); + mCurrentValue[i].FloatValues[3] = std::numeric_limits<float>::quiet_NaN(); + mCurrentValue[i].Type = GL_FLOAT; + mCurrentValueBuffer[i] = NULL; + mCurrentValueOffsets[i] = 0; + } + + mStreamingBuffer = new StreamingVertexBufferInterface(renderer, INITIAL_STREAM_BUFFER_SIZE); + + if (!mStreamingBuffer) + { + ERR("Failed to allocate the streaming vertex buffer."); + } +} + +VertexDataManager::~VertexDataManager() +{ + delete mStreamingBuffer; + + for (int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) + { + delete mCurrentValueBuffer[i]; + } +} + +gl::Error VertexDataManager::prepareVertexData(const gl::VertexAttribute attribs[], const gl::VertexAttribCurrentValueData currentValues[], + gl::ProgramBinary *programBinary, GLint start, GLsizei count, TranslatedAttribute *translated, GLsizei instances) +{ + if (!mStreamingBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal streaming vertex buffer is unexpectedly NULL."); + } + + // Invalidate static buffers that don't contain matching attributes + for (int attributeIndex = 0; attributeIndex < gl::MAX_VERTEX_ATTRIBS; attributeIndex++) + { + translated[attributeIndex].active = (programBinary->getSemanticIndex(attributeIndex) != -1); + + if (translated[attributeIndex].active && attribs[attributeIndex].enabled) + { + invalidateMatchingStaticData(attribs[attributeIndex], currentValues[attributeIndex]); + } + } + + // Reserve the required space in the buffers + for (int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) + { + if (translated[i].active && attribs[i].enabled) + { + gl::Error error = reserveSpaceForAttrib(attribs[i], currentValues[i], count, instances); + if (error.isError()) + { + return error; + } + } + } + + // Perform the vertex data translations + for (int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) + { + if (translated[i].active) + { + if (attribs[i].enabled) + { + gl::Error error = storeAttribute(attribs[i], currentValues[i], &translated[i], + start, count, instances); + if (error.isError()) + { + return error; + } + } + else + { + if (!mCurrentValueBuffer[i]) + { + mCurrentValueBuffer[i] = new StreamingVertexBufferInterface(mRenderer, CONSTANT_VERTEX_BUFFER_SIZE); + } + + gl::Error error = storeCurrentValue(attribs[i], currentValues[i], &translated[i], + &mCurrentValue[i], &mCurrentValueOffsets[i], + mCurrentValueBuffer[i]); + if (error.isError()) + { + return error; + } + } + } + } + + for (int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) + { + if (translated[i].active && attribs[i].enabled) + { + gl::Buffer *buffer = attribs[i].buffer.get(); + + if (buffer) + { + BufferD3D *bufferImpl = BufferD3D::makeBufferD3D(buffer->getImplementation()); + bufferImpl->promoteStaticUsage(count * ComputeVertexAttributeTypeSize(attribs[i])); + } + } + } + + return gl::Error(GL_NO_ERROR); +} + +void VertexDataManager::invalidateMatchingStaticData(const gl::VertexAttribute &attrib, + const gl::VertexAttribCurrentValueData ¤tValue) const +{ + gl::Buffer *buffer = attrib.buffer.get(); + + if (buffer) + { + BufferD3D *bufferImpl = BufferD3D::makeBufferD3D(buffer->getImplementation()); + StaticVertexBufferInterface *staticBuffer = bufferImpl->getStaticVertexBuffer(); + + if (staticBuffer && + staticBuffer->getBufferSize() > 0 && + !staticBuffer->lookupAttribute(attrib, NULL) && + !staticBuffer->directStoragePossible(attrib, currentValue)) + { + bufferImpl->invalidateStaticData(); + } + } +} + +gl::Error VertexDataManager::reserveSpaceForAttrib(const gl::VertexAttribute &attrib, + const gl::VertexAttribCurrentValueData ¤tValue, + GLsizei count, + GLsizei instances) const +{ + gl::Buffer *buffer = attrib.buffer.get(); + BufferD3D *bufferImpl = buffer ? BufferD3D::makeBufferD3D(buffer->getImplementation()) : NULL; + StaticVertexBufferInterface *staticBuffer = bufferImpl ? bufferImpl->getStaticVertexBuffer() : NULL; + VertexBufferInterface *vertexBuffer = staticBuffer ? staticBuffer : static_cast<VertexBufferInterface*>(mStreamingBuffer); + + if (!vertexBuffer->directStoragePossible(attrib, currentValue)) + { + if (staticBuffer) + { + if (staticBuffer->getBufferSize() == 0) + { + int totalCount = ElementsInBuffer(attrib, bufferImpl->getSize()); + gl::Error error = staticBuffer->reserveVertexSpace(attrib, totalCount, 0); + if (error.isError()) + { + return error; + } + } + } + else + { + int totalCount = StreamingBufferElementCount(attrib, count, instances); + ASSERT(!bufferImpl || ElementsInBuffer(attrib, bufferImpl->getSize()) >= totalCount); + + gl::Error error = mStreamingBuffer->reserveVertexSpace(attrib, totalCount, instances); + if (error.isError()) + { + return error; + } + } + } + + return gl::Error(GL_NO_ERROR); +} + +gl::Error VertexDataManager::storeAttribute(const gl::VertexAttribute &attrib, + const gl::VertexAttribCurrentValueData ¤tValue, + TranslatedAttribute *translated, + GLint start, + GLsizei count, + GLsizei instances) +{ + gl::Buffer *buffer = attrib.buffer.get(); + ASSERT(buffer || attrib.pointer); + + BufferD3D *storage = buffer ? BufferD3D::makeBufferD3D(buffer->getImplementation()) : NULL; + StaticVertexBufferInterface *staticBuffer = storage ? storage->getStaticVertexBuffer() : NULL; + VertexBufferInterface *vertexBuffer = staticBuffer ? staticBuffer : static_cast<VertexBufferInterface*>(mStreamingBuffer); + bool directStorage = vertexBuffer->directStoragePossible(attrib, currentValue); + + unsigned int streamOffset = 0; + unsigned int outputElementSize = 0; + + if (directStorage) + { + outputElementSize = ComputeVertexAttributeStride(attrib); + streamOffset = attrib.offset + outputElementSize * start; + } + else if (staticBuffer) + { + gl::Error error = staticBuffer->getVertexBuffer()->getSpaceRequired(attrib, 1, 0, &outputElementSize); + if (error.isError()) + { + return error; + } + + if (!staticBuffer->lookupAttribute(attrib, &streamOffset)) + { + // Convert the entire buffer + int totalCount = ElementsInBuffer(attrib, storage->getSize()); + int startIndex = attrib.offset / ComputeVertexAttributeStride(attrib); + + gl::Error error = staticBuffer->storeVertexAttributes(attrib, currentValue, -startIndex, totalCount, + 0, &streamOffset); + if (error.isError()) + { + return error; + } + } + + unsigned int firstElementOffset = (attrib.offset / ComputeVertexAttributeStride(attrib)) * outputElementSize; + unsigned int startOffset = (instances == 0 || attrib.divisor == 0) ? start * outputElementSize : 0; + if (streamOffset + firstElementOffset + startOffset < streamOffset) + { + return gl::Error(GL_OUT_OF_MEMORY); + } + + streamOffset += firstElementOffset + startOffset; + } + else + { + int totalCount = StreamingBufferElementCount(attrib, count, instances); + gl::Error error = mStreamingBuffer->getVertexBuffer()->getSpaceRequired(attrib, 1, 0, &outputElementSize); + if (error.isError()) + { + return error; + } + + error = mStreamingBuffer->storeVertexAttributes(attrib, currentValue, start, totalCount, instances, &streamOffset); + if (error.isError()) + { + return error; + } + } + + translated->storage = directStorage ? storage : NULL; + translated->vertexBuffer = vertexBuffer->getVertexBuffer(); + translated->serial = directStorage ? storage->getSerial() : vertexBuffer->getSerial(); + translated->divisor = attrib.divisor; + + translated->attribute = &attrib; + translated->currentValueType = currentValue.Type; + translated->stride = outputElementSize; + translated->offset = streamOffset; + + return gl::Error(GL_NO_ERROR); +} + +gl::Error VertexDataManager::storeCurrentValue(const gl::VertexAttribute &attrib, + const gl::VertexAttribCurrentValueData ¤tValue, + TranslatedAttribute *translated, + gl::VertexAttribCurrentValueData *cachedValue, + size_t *cachedOffset, + StreamingVertexBufferInterface *buffer) +{ + if (*cachedValue != currentValue) + { + gl::Error error = buffer->reserveVertexSpace(attrib, 1, 0); + if (error.isError()) + { + return error; + } + + unsigned int streamOffset; + error = buffer->storeVertexAttributes(attrib, currentValue, 0, 1, 0, &streamOffset); + if (error.isError()) + { + return error; + } + + *cachedValue = currentValue; + *cachedOffset = streamOffset; + } + + translated->storage = NULL; + translated->vertexBuffer = buffer->getVertexBuffer(); + translated->serial = buffer->getSerial(); + translated->divisor = 0; + + translated->attribute = &attrib; + translated->currentValueType = currentValue.Type; + translated->stride = 0; + translated->offset = *cachedOffset; + + return gl::Error(GL_NO_ERROR); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/VertexDataManager.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/VertexDataManager.h new file mode 100644 index 0000000000..7728722246 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/VertexDataManager.h @@ -0,0 +1,95 @@ +// +// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// VertexDataManager.h: Defines the VertexDataManager, a class that +// runs the Buffer translation process. + +#ifndef LIBGLESV2_RENDERER_VERTEXDATAMANAGER_H_ +#define LIBGLESV2_RENDERER_VERTEXDATAMANAGER_H_ + +#include "libGLESv2/Constants.h" +#include "libGLESv2/VertexAttribute.h" +#include "common/angleutils.h" + +namespace gl +{ +struct VertexAttribute; +class ProgramBinary; +struct VertexAttribCurrentValueData; +} + +namespace rx +{ +class BufferD3D; +class StreamingVertexBufferInterface; +class VertexBuffer; +class Renderer; + +struct TranslatedAttribute +{ + TranslatedAttribute() : active(false), attribute(NULL), currentValueType(GL_NONE), + offset(0), stride(0), vertexBuffer(NULL), storage(NULL), + serial(0), divisor(0) {}; + bool active; + + const gl::VertexAttribute *attribute; + GLenum currentValueType; + unsigned int offset; + unsigned int stride; // 0 means not to advance the read pointer at all + + VertexBuffer *vertexBuffer; + BufferD3D *storage; + unsigned int serial; + unsigned int divisor; +}; + +class VertexDataManager +{ + public: + VertexDataManager(rx::Renderer *renderer); + virtual ~VertexDataManager(); + + gl::Error prepareVertexData(const gl::VertexAttribute attribs[], const gl::VertexAttribCurrentValueData currentValues[], + gl::ProgramBinary *programBinary, GLint start, GLsizei count, TranslatedAttribute *outAttribs, GLsizei instances); + + private: + DISALLOW_COPY_AND_ASSIGN(VertexDataManager); + + gl::Error reserveSpaceForAttrib(const gl::VertexAttribute &attrib, + const gl::VertexAttribCurrentValueData ¤tValue, + GLsizei count, + GLsizei instances) const; + + void invalidateMatchingStaticData(const gl::VertexAttribute &attrib, + const gl::VertexAttribCurrentValueData ¤tValue) const; + + gl::Error storeAttribute(const gl::VertexAttribute &attrib, + const gl::VertexAttribCurrentValueData ¤tValue, + TranslatedAttribute *translated, + GLint start, + GLsizei count, + GLsizei instances); + + gl::Error storeCurrentValue(const gl::VertexAttribute &attrib, + const gl::VertexAttribCurrentValueData ¤tValue, + TranslatedAttribute *translated, + gl::VertexAttribCurrentValueData *cachedValue, + size_t *cachedOffset, + StreamingVertexBufferInterface *buffer); + + rx::Renderer *const mRenderer; + + StreamingVertexBufferInterface *mStreamingBuffer; + + gl::VertexAttribCurrentValueData mCurrentValue[gl::MAX_VERTEX_ATTRIBS]; + + StreamingVertexBufferInterface *mCurrentValueBuffer[gl::MAX_VERTEX_ATTRIBS]; + std::size_t mCurrentValueOffsets[gl::MAX_VERTEX_ATTRIBS]; +}; + +} + +#endif // LIBGLESV2_RENDERER_VERTEXDATAMANAGER_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Blit11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Blit11.cpp new file mode 100644 index 0000000000..d43e65ea78 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Blit11.cpp @@ -0,0 +1,1049 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Blit11.cpp: Texture copy utility class. + +#include "libGLESv2/renderer/d3d/d3d11/Blit11.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" +#include "libGLESv2/renderer/d3d/d3d11/renderer11_utils.h" +#include "libGLESv2/renderer/d3d/d3d11/formatutils11.h" +#include "libGLESv2/main.h" +#include "libGLESv2/formatutils.h" + +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthrough2dvs.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughdepth2dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgba2dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgba2duips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgba2dips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgb2dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgb2duips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgb2dips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrg2dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrg2duips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrg2dips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughr2dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughr2duips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughr2dips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughlum2dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughlumalpha2dps.h" + +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthrough3dvs.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthrough3dgs.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgba3dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgba3duips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgba3dips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgb3dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgb3duips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgb3dips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrg3dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrg3duips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrg3dips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughr3dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughr3duips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughr3dips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughlum3dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughlumalpha3dps.h" + +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/swizzlef2dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/swizzlei2dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/swizzleui2dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/swizzlef3dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/swizzlei3dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/swizzleui3dps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/swizzlef2darrayps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/swizzlei2darrayps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/swizzleui2darrayps.h" + +namespace rx +{ + +static DXGI_FORMAT GetTextureFormat(ID3D11Resource *resource) +{ + ID3D11Texture2D *texture = d3d11::DynamicCastComObject<ID3D11Texture2D>(resource); + if (!texture) + { + return DXGI_FORMAT_UNKNOWN; + } + + D3D11_TEXTURE2D_DESC desc; + texture->GetDesc(&desc); + + SafeRelease(texture); + + return desc.Format; +} + +static ID3D11Resource *CreateStagingTexture(ID3D11Device *device, ID3D11DeviceContext *context, + ID3D11Resource *source, unsigned int subresource, + const gl::Extents &size, unsigned int cpuAccessFlags) +{ + D3D11_TEXTURE2D_DESC stagingDesc; + stagingDesc.Width = size.width; + stagingDesc.Height = size.height; + stagingDesc.MipLevels = 1; + stagingDesc.ArraySize = 1; + stagingDesc.Format = GetTextureFormat(source); + stagingDesc.SampleDesc.Count = 1; + stagingDesc.SampleDesc.Quality = 0; + stagingDesc.Usage = D3D11_USAGE_STAGING; + stagingDesc.CPUAccessFlags = cpuAccessFlags; + stagingDesc.MiscFlags = 0; + stagingDesc.BindFlags = 0; + + ID3D11Texture2D *stagingTexture = NULL; + HRESULT result = device->CreateTexture2D(&stagingDesc, NULL, &stagingTexture); + if (FAILED(result)) + { + ERR("Failed to create staging texture for depth stencil blit. HRESULT: 0x%X.", result); + return NULL; + } + + context->CopySubresourceRegion(stagingTexture, 0, 0, 0, 0, source, subresource, NULL); + + return stagingTexture; +} + +inline static void GenerateVertexCoords(const gl::Box &sourceArea, const gl::Extents &sourceSize, + const gl::Box &destArea, const gl::Extents &destSize, + float *x1, float *y1, float *x2, float *y2, + float *u1, float *v1, float *u2, float *v2) +{ + *x1 = (destArea.x / float(destSize.width)) * 2.0f - 1.0f; + *y1 = ((destSize.height - destArea.y - destArea.height) / float(destSize.height)) * 2.0f - 1.0f; + *x2 = ((destArea.x + destArea.width) / float(destSize.width)) * 2.0f - 1.0f; + *y2 = ((destSize.height - destArea.y) / float(destSize.height)) * 2.0f - 1.0f; + + *u1 = sourceArea.x / float(sourceSize.width); + *v1 = sourceArea.y / float(sourceSize.height); + *u2 = (sourceArea.x + sourceArea.width) / float(sourceSize.width); + *v2 = (sourceArea.y + sourceArea.height) / float(sourceSize.height); +} + +static void Write2DVertices(const gl::Box &sourceArea, const gl::Extents &sourceSize, + const gl::Box &destArea, const gl::Extents &destSize, + void *outVertices, unsigned int *outStride, unsigned int *outVertexCount, + D3D11_PRIMITIVE_TOPOLOGY *outTopology) +{ + float x1, y1, x2, y2, u1, v1, u2, v2; + GenerateVertexCoords(sourceArea, sourceSize, destArea, destSize, &x1, &y1, &x2, &y2, &u1, &v1, &u2, &v2); + + d3d11::PositionTexCoordVertex *vertices = static_cast<d3d11::PositionTexCoordVertex*>(outVertices); + + d3d11::SetPositionTexCoordVertex(&vertices[0], x1, y1, u1, v2); + d3d11::SetPositionTexCoordVertex(&vertices[1], x1, y2, u1, v1); + d3d11::SetPositionTexCoordVertex(&vertices[2], x2, y1, u2, v2); + d3d11::SetPositionTexCoordVertex(&vertices[3], x2, y2, u2, v1); + + *outStride = sizeof(d3d11::PositionTexCoordVertex); + *outVertexCount = 4; + *outTopology = D3D_PRIMITIVE_TOPOLOGY_TRIANGLESTRIP; +} + +static void Write3DVertices(const gl::Box &sourceArea, const gl::Extents &sourceSize, + const gl::Box &destArea, const gl::Extents &destSize, + void *outVertices, unsigned int *outStride, unsigned int *outVertexCount, + D3D11_PRIMITIVE_TOPOLOGY *outTopology) +{ + ASSERT(sourceSize.depth > 0 && destSize.depth > 0); + + float x1, y1, x2, y2, u1, v1, u2, v2; + GenerateVertexCoords(sourceArea, sourceSize, destArea, destSize, &x1, &y1, &x2, &y2, &u1, &v1, &u2, &v2); + + d3d11::PositionLayerTexCoord3DVertex *vertices = static_cast<d3d11::PositionLayerTexCoord3DVertex*>(outVertices); + + for (int i = 0; i < destSize.depth; i++) + { + float readDepth = (float)i / std::max(destSize.depth - 1, 1); + + d3d11::SetPositionLayerTexCoord3DVertex(&vertices[i * 6 + 0], x1, y1, i, u1, v2, readDepth); + d3d11::SetPositionLayerTexCoord3DVertex(&vertices[i * 6 + 1], x1, y2, i, u1, v1, readDepth); + d3d11::SetPositionLayerTexCoord3DVertex(&vertices[i * 6 + 2], x2, y1, i, u2, v2, readDepth); + + d3d11::SetPositionLayerTexCoord3DVertex(&vertices[i * 6 + 3], x1, y2, i, u1, v1, readDepth); + d3d11::SetPositionLayerTexCoord3DVertex(&vertices[i * 6 + 4], x2, y2, i, u2, v1, readDepth); + d3d11::SetPositionLayerTexCoord3DVertex(&vertices[i * 6 + 5], x2, y1, i, u2, v2, readDepth); + } + + *outStride = sizeof(d3d11::PositionLayerTexCoord3DVertex); + *outVertexCount = destSize.depth * 6; + *outTopology = D3D_PRIMITIVE_TOPOLOGY_TRIANGLELIST; +} + +Blit11::Blit11(rx::Renderer11 *renderer) + : mRenderer(renderer), mBlitShaderMap(compareBlitParameters), mSwizzleShaderMap(compareSwizzleParameters), + mVertexBuffer(NULL), mPointSampler(NULL), mLinearSampler(NULL), mScissorEnabledRasterizerState(NULL), + mScissorDisabledRasterizerState(NULL), mDepthStencilState(NULL), + mQuad2DIL(NULL), mQuad2DVS(NULL), mDepthPS(NULL), + mQuad3DIL(NULL), mQuad3DVS(NULL), mQuad3DGS(NULL), + mSwizzleCB(NULL) +{ + HRESULT result; + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_BUFFER_DESC vbDesc; + vbDesc.ByteWidth = std::max(sizeof(d3d11::PositionLayerTexCoord3DVertex), sizeof(d3d11::PositionTexCoordVertex)) * + 6 * renderer->getRendererCaps().max3DTextureSize; + vbDesc.Usage = D3D11_USAGE_DYNAMIC; + vbDesc.BindFlags = D3D11_BIND_VERTEX_BUFFER; + vbDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; + vbDesc.MiscFlags = 0; + vbDesc.StructureByteStride = 0; + + result = device->CreateBuffer(&vbDesc, NULL, &mVertexBuffer); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mVertexBuffer, "Blit11 vertex buffer"); + + D3D11_SAMPLER_DESC pointSamplerDesc; + pointSamplerDesc.Filter = D3D11_FILTER_MIN_MAG_POINT_MIP_LINEAR; + pointSamplerDesc.AddressU = D3D11_TEXTURE_ADDRESS_CLAMP; + pointSamplerDesc.AddressV = D3D11_TEXTURE_ADDRESS_CLAMP; + pointSamplerDesc.AddressW = D3D11_TEXTURE_ADDRESS_CLAMP; + pointSamplerDesc.MipLODBias = 0.0f; + pointSamplerDesc.MaxAnisotropy = 0; + pointSamplerDesc.ComparisonFunc = D3D11_COMPARISON_NEVER; + pointSamplerDesc.BorderColor[0] = 0.0f; + pointSamplerDesc.BorderColor[1] = 0.0f; + pointSamplerDesc.BorderColor[2] = 0.0f; + pointSamplerDesc.BorderColor[3] = 0.0f; + pointSamplerDesc.MinLOD = 0.0f; + pointSamplerDesc.MaxLOD = mRenderer->isLevel9() ? FLT_MAX : 0.0f; + + result = device->CreateSamplerState(&pointSamplerDesc, &mPointSampler); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mPointSampler, "Blit11 point sampler"); + + D3D11_SAMPLER_DESC linearSamplerDesc; + linearSamplerDesc.Filter = D3D11_FILTER_MIN_MAG_MIP_LINEAR; + linearSamplerDesc.AddressU = D3D11_TEXTURE_ADDRESS_CLAMP; + linearSamplerDesc.AddressV = D3D11_TEXTURE_ADDRESS_CLAMP; + linearSamplerDesc.AddressW = D3D11_TEXTURE_ADDRESS_CLAMP; + linearSamplerDesc.MipLODBias = 0.0f; + linearSamplerDesc.MaxAnisotropy = 0; + linearSamplerDesc.ComparisonFunc = D3D11_COMPARISON_NEVER; + linearSamplerDesc.BorderColor[0] = 0.0f; + linearSamplerDesc.BorderColor[1] = 0.0f; + linearSamplerDesc.BorderColor[2] = 0.0f; + linearSamplerDesc.BorderColor[3] = 0.0f; + linearSamplerDesc.MinLOD = 0.0f; + linearSamplerDesc.MaxLOD = mRenderer->isLevel9() ? FLT_MAX : 0.0f; + + result = device->CreateSamplerState(&linearSamplerDesc, &mLinearSampler); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mLinearSampler, "Blit11 linear sampler"); + + // Use a rasterizer state that will not cull so that inverted quads will not be culled + D3D11_RASTERIZER_DESC rasterDesc; + rasterDesc.FillMode = D3D11_FILL_SOLID; + rasterDesc.CullMode = D3D11_CULL_NONE; + rasterDesc.FrontCounterClockwise = FALSE; + rasterDesc.DepthBias = 0; + rasterDesc.SlopeScaledDepthBias = 0.0f; + rasterDesc.DepthBiasClamp = 0.0f; + rasterDesc.DepthClipEnable = TRUE; + rasterDesc.MultisampleEnable = FALSE; + rasterDesc.AntialiasedLineEnable = FALSE; + + rasterDesc.ScissorEnable = TRUE; + result = device->CreateRasterizerState(&rasterDesc, &mScissorEnabledRasterizerState); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mScissorEnabledRasterizerState, "Blit11 scissoring rasterizer state"); + + rasterDesc.ScissorEnable = FALSE; + result = device->CreateRasterizerState(&rasterDesc, &mScissorDisabledRasterizerState); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mScissorDisabledRasterizerState, "Blit11 no scissoring rasterizer state"); + + D3D11_DEPTH_STENCIL_DESC depthStencilDesc; + depthStencilDesc.DepthEnable = true; + depthStencilDesc.DepthWriteMask = D3D11_DEPTH_WRITE_MASK_ALL; + depthStencilDesc.DepthFunc = D3D11_COMPARISON_ALWAYS; + depthStencilDesc.StencilEnable = FALSE; + depthStencilDesc.StencilReadMask = D3D11_DEFAULT_STENCIL_READ_MASK; + depthStencilDesc.StencilWriteMask = D3D11_DEFAULT_STENCIL_WRITE_MASK; + depthStencilDesc.FrontFace.StencilFailOp = D3D11_STENCIL_OP_KEEP; + depthStencilDesc.FrontFace.StencilDepthFailOp = D3D11_STENCIL_OP_KEEP; + depthStencilDesc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_KEEP; + depthStencilDesc.FrontFace.StencilFunc = D3D11_COMPARISON_ALWAYS; + depthStencilDesc.BackFace.StencilFailOp = D3D11_STENCIL_OP_KEEP; + depthStencilDesc.BackFace.StencilDepthFailOp = D3D11_STENCIL_OP_KEEP; + depthStencilDesc.BackFace.StencilPassOp = D3D11_STENCIL_OP_KEEP; + depthStencilDesc.BackFace.StencilFunc = D3D11_COMPARISON_ALWAYS; + + result = device->CreateDepthStencilState(&depthStencilDesc, &mDepthStencilState); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mDepthStencilState, "Blit11 depth stencil state"); + + D3D11_INPUT_ELEMENT_DESC quad2DLayout[] = + { + { "POSITION", 0, DXGI_FORMAT_R32G32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0 }, + { "TEXCOORD", 0, DXGI_FORMAT_R32G32_FLOAT, 0, 8, D3D11_INPUT_PER_VERTEX_DATA, 0 }, + }; + + result = device->CreateInputLayout(quad2DLayout, ArraySize(quad2DLayout), g_VS_Passthrough2D, ArraySize(g_VS_Passthrough2D), &mQuad2DIL); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mQuad2DIL, "Blit11 2D input layout"); + + result = device->CreateVertexShader(g_VS_Passthrough2D, ArraySize(g_VS_Passthrough2D), NULL, &mQuad2DVS); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mQuad2DVS, "Blit11 2D vertex shader"); + + if (!renderer->isLevel9()) + { + result = device->CreatePixelShader(g_PS_PassthroughDepth2D, ArraySize(g_PS_PassthroughDepth2D), NULL, &mDepthPS); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mDepthPS, "Blit11 2D depth pixel shader"); + + D3D11_INPUT_ELEMENT_DESC quad3DLayout[] = + { + { "POSITION", 0, DXGI_FORMAT_R32G32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0 }, + { "LAYER", 0, DXGI_FORMAT_R32_UINT, 0, 8, D3D11_INPUT_PER_VERTEX_DATA, 0 }, + { "TEXCOORD", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 12, D3D11_INPUT_PER_VERTEX_DATA, 0 }, + }; + + result = device->CreateInputLayout(quad3DLayout, ArraySize(quad3DLayout), g_VS_Passthrough3D, ArraySize(g_VS_Passthrough3D), &mQuad3DIL); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mQuad3DIL, "Blit11 3D input layout"); + + result = device->CreateVertexShader(g_VS_Passthrough3D, ArraySize(g_VS_Passthrough3D), NULL, &mQuad3DVS); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mQuad3DVS, "Blit11 3D vertex shader"); + + result = device->CreateGeometryShader(g_GS_Passthrough3D, ArraySize(g_GS_Passthrough3D), NULL, &mQuad3DGS); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mQuad3DGS, "Renderer11 copy 3D texture geometry shader"); + } + + buildShaderMap(); + + D3D11_BUFFER_DESC swizzleBufferDesc; + swizzleBufferDesc.ByteWidth = sizeof(unsigned int) * 4; + swizzleBufferDesc.Usage = D3D11_USAGE_DYNAMIC; + swizzleBufferDesc.BindFlags = D3D11_BIND_CONSTANT_BUFFER; + swizzleBufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; + swizzleBufferDesc.MiscFlags = 0; + swizzleBufferDesc.StructureByteStride = 0; + + result = device->CreateBuffer(&swizzleBufferDesc, NULL, &mSwizzleCB); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mSwizzleCB, "Blit11 swizzle constant buffer"); +} + +Blit11::~Blit11() +{ + SafeRelease(mVertexBuffer); + SafeRelease(mPointSampler); + SafeRelease(mLinearSampler); + SafeRelease(mScissorEnabledRasterizerState); + SafeRelease(mScissorDisabledRasterizerState); + SafeRelease(mDepthStencilState); + + SafeRelease(mQuad2DIL); + SafeRelease(mQuad2DVS); + SafeRelease(mDepthPS); + + SafeRelease(mQuad3DIL); + SafeRelease(mQuad3DVS); + SafeRelease(mQuad3DGS); + + SafeRelease(mSwizzleCB); + + clearShaderMap(); +} + +static inline unsigned int GetSwizzleIndex(GLenum swizzle) +{ + unsigned int colorIndex = 0; + + switch (swizzle) + { + case GL_RED: colorIndex = 0; break; + case GL_GREEN: colorIndex = 1; break; + case GL_BLUE: colorIndex = 2; break; + case GL_ALPHA: colorIndex = 3; break; + case GL_ZERO: colorIndex = 4; break; + case GL_ONE: colorIndex = 5; break; + default: UNREACHABLE(); break; + } + + return colorIndex; +} + +gl::Error Blit11::swizzleTexture(ID3D11ShaderResourceView *source, ID3D11RenderTargetView *dest, const gl::Extents &size, + GLenum swizzleRed, GLenum swizzleGreen, GLenum swizzleBlue, GLenum swizzleAlpha) +{ + HRESULT result; + ID3D11DeviceContext *deviceContext = mRenderer->getDeviceContext(); + + D3D11_SHADER_RESOURCE_VIEW_DESC sourceSRVDesc; + source->GetDesc(&sourceSRVDesc); + + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(sourceSRVDesc.Format); + const gl::InternalFormat &sourceFormatInfo = gl::GetInternalFormatInfo(dxgiFormatInfo.internalFormat); + + GLenum shaderType = GL_NONE; + switch (sourceFormatInfo.componentType) + { + case GL_UNSIGNED_NORMALIZED: + case GL_SIGNED_NORMALIZED: + case GL_FLOAT: + shaderType = GL_FLOAT; + break; + case GL_INT: + shaderType = GL_INT; + break; + case GL_UNSIGNED_INT: + shaderType = GL_UNSIGNED_INT; + break; + default: + UNREACHABLE(); + break; + } + + SwizzleParameters parameters = { 0 }; + parameters.mDestinationType = shaderType; + parameters.mViewDimension = sourceSRVDesc.ViewDimension; + + SwizzleShaderMap::const_iterator i = mSwizzleShaderMap.find(parameters); + if (i == mSwizzleShaderMap.end()) + { + UNREACHABLE(); + return gl::Error(GL_INVALID_OPERATION, "Internal error, missing swizzle shader."); + } + + const Shader &shader = i->second; + + // Set vertices + D3D11_MAPPED_SUBRESOURCE mappedResource; + result = deviceContext->Map(mVertexBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to map internal vertex buffer for swizzle, HRESULT: 0x%X.", result); + } + + UINT stride = 0; + UINT startIdx = 0; + UINT drawCount = 0; + D3D11_PRIMITIVE_TOPOLOGY topology; + + gl::Box area(0, 0, 0, size.width, size.height, size.depth); + shader.mVertexWriteFunction(area, size, area, size, mappedResource.pData, &stride, &drawCount, &topology); + + deviceContext->Unmap(mVertexBuffer, 0); + + // Set constant buffer + result = deviceContext->Map(mSwizzleCB, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to map internal constant buffer for swizzle, HRESULT: 0x%X.", result); + } + + unsigned int *swizzleIndices = reinterpret_cast<unsigned int*>(mappedResource.pData); + swizzleIndices[0] = GetSwizzleIndex(swizzleRed); + swizzleIndices[1] = GetSwizzleIndex(swizzleGreen); + swizzleIndices[2] = GetSwizzleIndex(swizzleBlue); + swizzleIndices[3] = GetSwizzleIndex(swizzleAlpha); + + deviceContext->Unmap(mSwizzleCB, 0); + + // Apply vertex buffer + deviceContext->IASetVertexBuffers(0, 1, &mVertexBuffer, &stride, &startIdx); + + // Apply constant buffer + deviceContext->PSSetConstantBuffers(0, 1, &mSwizzleCB); + + // Apply state + deviceContext->OMSetBlendState(NULL, NULL, 0xFFFFFFF); + deviceContext->OMSetDepthStencilState(NULL, 0xFFFFFFFF); + deviceContext->RSSetState(mScissorDisabledRasterizerState); + + // Apply shaders + deviceContext->IASetInputLayout(shader.mInputLayout); + deviceContext->IASetPrimitiveTopology(topology); + deviceContext->VSSetShader(shader.mVertexShader, NULL, 0); + + deviceContext->PSSetShader(shader.mPixelShader, NULL, 0); + deviceContext->GSSetShader(shader.mGeometryShader, NULL, 0); + + // Unset the currently bound shader resource to avoid conflicts + ID3D11ShaderResourceView *const nullSRV = NULL; + deviceContext->PSSetShaderResources(0, 1, &nullSRV); + + // Apply render target + mRenderer->setOneTimeRenderTarget(dest); + + // Set the viewport + D3D11_VIEWPORT viewport; + viewport.TopLeftX = 0; + viewport.TopLeftY = 0; + viewport.Width = size.width; + viewport.Height = size.height; + viewport.MinDepth = 0.0f; + viewport.MaxDepth = 1.0f; + deviceContext->RSSetViewports(1, &viewport); + + // Apply textures + deviceContext->PSSetShaderResources(0, 1, &source); + + // Apply samplers + deviceContext->PSSetSamplers(0, 1, &mPointSampler); + + // Draw the quad + deviceContext->Draw(drawCount, 0); + + // Unbind textures and render targets and vertex buffer + deviceContext->PSSetShaderResources(0, 1, &nullSRV); + + mRenderer->unapplyRenderTargets(); + + UINT zero = 0; + ID3D11Buffer *const nullBuffer = NULL; + deviceContext->IASetVertexBuffers(0, 1, &nullBuffer, &zero, &zero); + + mRenderer->markAllStateDirty(); + + return gl::Error(GL_NO_ERROR); +} + +bool Blit11::copyTexture(ID3D11ShaderResourceView *source, const gl::Box &sourceArea, const gl::Extents &sourceSize, + ID3D11RenderTargetView *dest, const gl::Box &destArea, const gl::Extents &destSize, + const gl::Rectangle *scissor, GLenum destFormat, GLenum filter) +{ + HRESULT result; + ID3D11DeviceContext *deviceContext = mRenderer->getDeviceContext(); + + // Determine if the source format is a signed integer format, the destFormat will already + // be GL_XXXX_INTEGER but it does not tell us if it is signed or unsigned. + D3D11_SHADER_RESOURCE_VIEW_DESC sourceSRVDesc; + source->GetDesc(&sourceSRVDesc); + + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(sourceSRVDesc.Format); + const gl::InternalFormat &internalFormatInfo = gl::GetInternalFormatInfo(dxgiFormatInfo.internalFormat); + + BlitParameters parameters = { 0 }; + parameters.mDestinationFormat = destFormat; + parameters.mSignedInteger = (internalFormatInfo.componentType == GL_INT); + parameters.m3DBlit = sourceArea.depth > 1; + + BlitShaderMap::const_iterator i = mBlitShaderMap.find(parameters); + if (i == mBlitShaderMap.end()) + { + UNREACHABLE(); + return false; + } + + const Shader& shader = i->second; + + // Set vertices + D3D11_MAPPED_SUBRESOURCE mappedResource; + result = deviceContext->Map(mVertexBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + ERR("Failed to map vertex buffer for texture copy, HRESULT: 0x%X.", result); + return false; + } + + UINT stride = 0; + UINT startIdx = 0; + UINT drawCount = 0; + D3D11_PRIMITIVE_TOPOLOGY topology; + + shader.mVertexWriteFunction(sourceArea, sourceSize, destArea, destSize, mappedResource.pData, + &stride, &drawCount, &topology); + + deviceContext->Unmap(mVertexBuffer, 0); + + // Apply vertex buffer + deviceContext->IASetVertexBuffers(0, 1, &mVertexBuffer, &stride, &startIdx); + + // Apply state + deviceContext->OMSetBlendState(NULL, NULL, 0xFFFFFFF); + deviceContext->OMSetDepthStencilState(NULL, 0xFFFFFFFF); + + if (scissor) + { + D3D11_RECT scissorRect; + scissorRect.left = scissor->x; + scissorRect.right = scissor->x + scissor->width; + scissorRect.top = scissor->y; + scissorRect.bottom = scissor->y + scissor->height; + + deviceContext->RSSetScissorRects(1, &scissorRect); + deviceContext->RSSetState(mScissorEnabledRasterizerState); + } + else + { + deviceContext->RSSetState(mScissorDisabledRasterizerState); + } + + // Apply shaders + deviceContext->IASetInputLayout(shader.mInputLayout); + deviceContext->IASetPrimitiveTopology(topology); + deviceContext->VSSetShader(shader.mVertexShader, NULL, 0); + + deviceContext->PSSetShader(shader.mPixelShader, NULL, 0); + deviceContext->GSSetShader(shader.mGeometryShader, NULL, 0); + + // Unset the currently bound shader resource to avoid conflicts + ID3D11ShaderResourceView *const nullSRV = NULL; + deviceContext->PSSetShaderResources(0, 1, &nullSRV); + + // Apply render target + mRenderer->setOneTimeRenderTarget(dest); + + // Set the viewport + D3D11_VIEWPORT viewport; + viewport.TopLeftX = 0; + viewport.TopLeftY = 0; + viewport.Width = destSize.width; + viewport.Height = destSize.height; + viewport.MinDepth = 0.0f; + viewport.MaxDepth = 1.0f; + deviceContext->RSSetViewports(1, &viewport); + + // Apply textures + deviceContext->PSSetShaderResources(0, 1, &source); + + // Apply samplers + ID3D11SamplerState *sampler = NULL; + switch (filter) + { + case GL_NEAREST: sampler = mPointSampler; break; + case GL_LINEAR: sampler = mLinearSampler; break; + default: UNREACHABLE(); return false; + } + deviceContext->PSSetSamplers(0, 1, &sampler); + + // Draw the quad + deviceContext->Draw(drawCount, 0); + + // Unbind textures and render targets and vertex buffer + deviceContext->PSSetShaderResources(0, 1, &nullSRV); + + mRenderer->unapplyRenderTargets(); + + UINT zero = 0; + ID3D11Buffer *const nullBuffer = NULL; + deviceContext->IASetVertexBuffers(0, 1, &nullBuffer, &zero, &zero); + + mRenderer->markAllStateDirty(); + + return true; +} + +bool Blit11::copyStencil(ID3D11Resource *source, unsigned int sourceSubresource, const gl::Box &sourceArea, const gl::Extents &sourceSize, + ID3D11Resource *dest, unsigned int destSubresource, const gl::Box &destArea, const gl::Extents &destSize, + const gl::Rectangle *scissor) +{ + return copyDepthStencil(source, sourceSubresource, sourceArea, sourceSize, + dest, destSubresource, destArea, destSize, + scissor, true); +} + +bool Blit11::copyDepth(ID3D11ShaderResourceView *source, const gl::Box &sourceArea, const gl::Extents &sourceSize, + ID3D11DepthStencilView *dest, const gl::Box &destArea, const gl::Extents &destSize, + const gl::Rectangle *scissor) +{ + HRESULT result; + ID3D11DeviceContext *deviceContext = mRenderer->getDeviceContext(); + + // Set vertices + D3D11_MAPPED_SUBRESOURCE mappedResource; + result = deviceContext->Map(mVertexBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + ERR("Failed to map vertex buffer for texture copy, HRESULT: 0x%X.", result); + return false; + } + + UINT stride = 0; + UINT startIdx = 0; + UINT drawCount = 0; + D3D11_PRIMITIVE_TOPOLOGY topology; + + Write2DVertices(sourceArea, sourceSize, destArea, destSize, mappedResource.pData, + &stride, &drawCount, &topology); + + deviceContext->Unmap(mVertexBuffer, 0); + + // Apply vertex buffer + deviceContext->IASetVertexBuffers(0, 1, &mVertexBuffer, &stride, &startIdx); + + // Apply state + deviceContext->OMSetBlendState(NULL, NULL, 0xFFFFFFF); + deviceContext->OMSetDepthStencilState(mDepthStencilState, 0xFFFFFFFF); + + if (scissor) + { + D3D11_RECT scissorRect; + scissorRect.left = scissor->x; + scissorRect.right = scissor->x + scissor->width; + scissorRect.top = scissor->y; + scissorRect.bottom = scissor->y + scissor->height; + + deviceContext->RSSetScissorRects(1, &scissorRect); + deviceContext->RSSetState(mScissorEnabledRasterizerState); + } + else + { + deviceContext->RSSetState(mScissorDisabledRasterizerState); + } + + // Apply shaders + deviceContext->IASetInputLayout(mQuad2DIL); + deviceContext->IASetPrimitiveTopology(topology); + deviceContext->VSSetShader(mQuad2DVS, NULL, 0); + + deviceContext->PSSetShader(mDepthPS, NULL, 0); + deviceContext->GSSetShader(NULL, NULL, 0); + + // Unset the currently bound shader resource to avoid conflicts + ID3D11ShaderResourceView *const nullSRV = NULL; + deviceContext->PSSetShaderResources(0, 1, &nullSRV); + + // Apply render target + deviceContext->OMSetRenderTargets(0, NULL, dest); + + // Set the viewport + D3D11_VIEWPORT viewport; + viewport.TopLeftX = 0; + viewport.TopLeftY = 0; + viewport.Width = destSize.width; + viewport.Height = destSize.height; + viewport.MinDepth = 0.0f; + viewport.MaxDepth = 1.0f; + deviceContext->RSSetViewports(1, &viewport); + + // Apply textures + deviceContext->PSSetShaderResources(0, 1, &source); + + // Apply samplers + deviceContext->PSSetSamplers(0, 1, &mPointSampler); + + // Draw the quad + deviceContext->Draw(drawCount, 0); + + // Unbind textures and render targets and vertex buffer + deviceContext->PSSetShaderResources(0, 1, &nullSRV); + + mRenderer->unapplyRenderTargets(); + + UINT zero = 0; + ID3D11Buffer *const nullBuffer = NULL; + deviceContext->IASetVertexBuffers(0, 1, &nullBuffer, &zero, &zero); + + mRenderer->markAllStateDirty(); + + return true; +} + +bool Blit11::copyDepthStencil(ID3D11Resource *source, unsigned int sourceSubresource, const gl::Box &sourceArea, const gl::Extents &sourceSize, + ID3D11Resource *dest, unsigned int destSubresource, const gl::Box &destArea, const gl::Extents &destSize, + const gl::Rectangle *scissor) +{ + return copyDepthStencil(source, sourceSubresource, sourceArea, sourceSize, + dest, destSubresource, destArea, destSize, + scissor, false); +} + +bool Blit11::copyDepthStencil(ID3D11Resource *source, unsigned int sourceSubresource, const gl::Box &sourceArea, const gl::Extents &sourceSize, + ID3D11Resource *dest, unsigned int destSubresource, const gl::Box &destArea, const gl::Extents &destSize, + const gl::Rectangle *scissor, bool stencilOnly) +{ + ID3D11Device *device = mRenderer->getDevice(); + ID3D11DeviceContext *deviceContext = mRenderer->getDeviceContext(); + + ID3D11Resource *sourceStaging = CreateStagingTexture(device, deviceContext, source, sourceSubresource, sourceSize, D3D11_CPU_ACCESS_READ); + // HACK: Create the destination staging buffer as a read/write texture so ID3D11DevicContext::UpdateSubresource can be called + // using it's mapped data as a source + ID3D11Resource *destStaging = CreateStagingTexture(device, deviceContext, dest, destSubresource, destSize, D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE); + + if (!sourceStaging || !destStaging) + { + SafeRelease(sourceStaging); + SafeRelease(destStaging); + return false; + } + + DXGI_FORMAT format = GetTextureFormat(source); + ASSERT(format == GetTextureFormat(dest)); + + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(format); + unsigned int pixelSize = dxgiFormatInfo.pixelBytes; + unsigned int copyOffset = 0; + unsigned int copySize = pixelSize; + if (stencilOnly) + { + copyOffset = dxgiFormatInfo.depthBits / 8; + copySize = dxgiFormatInfo.stencilBits / 8; + + // It would be expensive to have non-byte sized stencil sizes since it would + // require reading from the destination, currently there aren't any though. + ASSERT(dxgiFormatInfo.stencilBits % 8 == 0 && + dxgiFormatInfo.depthBits % 8 == 0); + } + + D3D11_MAPPED_SUBRESOURCE sourceMapping, destMapping; + deviceContext->Map(sourceStaging, 0, D3D11_MAP_READ, 0, &sourceMapping); + deviceContext->Map(destStaging, 0, D3D11_MAP_WRITE, 0, &destMapping); + + if (!sourceMapping.pData || !destMapping.pData) + { + if (!sourceMapping.pData) + { + deviceContext->Unmap(sourceStaging, 0); + } + if (!destMapping.pData) + { + deviceContext->Unmap(destStaging, 0); + } + SafeRelease(sourceStaging); + SafeRelease(destStaging); + return false; + } + + gl::Rectangle clippedDestArea(destArea.x, destArea.y, destArea.width, destArea.height); + + // Clip dest area to the destination size + gl::ClipRectangle(clippedDestArea, gl::Rectangle(0, 0, destSize.width, destSize.height), &clippedDestArea); + + // Clip dest area to the scissor + if (scissor) + { + gl::ClipRectangle(clippedDestArea, *scissor, &clippedDestArea); + } + + // Determine if entire rows can be copied at once instead of each individual pixel, requires that there is + // no out of bounds lookups required, the entire pixel is copied and no stretching + bool wholeRowCopy = sourceArea.width == clippedDestArea.width && + sourceArea.x >= 0 && sourceArea.x + sourceArea.width <= sourceSize.width && + copySize == pixelSize; + + for (int y = clippedDestArea.y; y < clippedDestArea.y + clippedDestArea.height; y++) + { + float yPerc = static_cast<float>(y - destArea.y) / (destArea.height - 1); + + // Interpolate using the original source rectangle to determine which row to sample from while clamping to the edges + unsigned int readRow = gl::clamp(sourceArea.y + floor(yPerc * (sourceArea.height - 1) + 0.5f), 0, sourceSize.height - 1); + unsigned int writeRow = y; + + if (wholeRowCopy) + { + void *sourceRow = reinterpret_cast<char*>(sourceMapping.pData) + + readRow * sourceMapping.RowPitch + + sourceArea.x * pixelSize; + + void *destRow = reinterpret_cast<char*>(destMapping.pData) + + writeRow * destMapping.RowPitch + + destArea.x * pixelSize; + + memcpy(destRow, sourceRow, pixelSize * destArea.width); + } + else + { + for (int x = clippedDestArea.x; x < clippedDestArea.x + clippedDestArea.width; x++) + { + float xPerc = static_cast<float>(x - destArea.x) / (destArea.width - 1); + + // Interpolate the original source rectangle to determine which column to sample from while clamping to the edges + unsigned int readColumn = gl::clamp(sourceArea.x + floor(xPerc * (sourceArea.width - 1) + 0.5f), 0, sourceSize.width - 1); + unsigned int writeColumn = x; + + void *sourcePixel = reinterpret_cast<char*>(sourceMapping.pData) + + readRow * sourceMapping.RowPitch + + readColumn * pixelSize + + copyOffset; + + void *destPixel = reinterpret_cast<char*>(destMapping.pData) + + writeRow * destMapping.RowPitch + + writeColumn * pixelSize + + copyOffset; + + memcpy(destPixel, sourcePixel, copySize); + } + } + } + + // HACK: Use ID3D11DevicContext::UpdateSubresource which causes an extra copy compared to ID3D11DevicContext::CopySubresourceRegion + // according to MSDN. + deviceContext->UpdateSubresource(dest, destSubresource, NULL, destMapping.pData, destMapping.RowPitch, destMapping.DepthPitch); + + deviceContext->Unmap(sourceStaging, 0); + deviceContext->Unmap(destStaging, 0); + + // TODO: Determine why this call to ID3D11DevicContext::CopySubresourceRegion causes a TDR timeout on some + // systems when called repeatedly. + // deviceContext->CopySubresourceRegion(dest, destSubresource, 0, 0, 0, destStaging, 0, NULL); + + SafeRelease(sourceStaging); + SafeRelease(destStaging); + + return true; +} + +bool Blit11::compareBlitParameters(const Blit11::BlitParameters &a, const Blit11::BlitParameters &b) +{ + return memcmp(&a, &b, sizeof(Blit11::BlitParameters)) < 0; +} + +bool Blit11::compareSwizzleParameters(const SwizzleParameters &a, const SwizzleParameters &b) +{ + return memcmp(&a, &b, sizeof(Blit11::SwizzleParameters)) < 0; +} + +void Blit11::add2DBlitShaderToMap(GLenum destFormat, bool signedInteger, ID3D11PixelShader *ps) +{ + BlitParameters params = { 0 }; + params.mDestinationFormat = destFormat; + params.mSignedInteger = signedInteger; + params.m3DBlit = false; + + ASSERT(mBlitShaderMap.find(params) == mBlitShaderMap.end()); + ASSERT(ps); + + Shader shader; + shader.mVertexWriteFunction = Write2DVertices; + shader.mInputLayout = mQuad2DIL; + shader.mVertexShader = mQuad2DVS; + shader.mGeometryShader = NULL; + shader.mPixelShader = ps; + + mBlitShaderMap[params] = shader; +} + +void Blit11::add3DBlitShaderToMap(GLenum destFormat, bool signedInteger, ID3D11PixelShader *ps) +{ + BlitParameters params = { 0 }; + params.mDestinationFormat = destFormat; + params.mSignedInteger = signedInteger; + params.m3DBlit = true; + + ASSERT(mBlitShaderMap.find(params) == mBlitShaderMap.end()); + ASSERT(ps); + + Shader shader; + shader.mVertexWriteFunction = Write3DVertices; + shader.mInputLayout = mQuad3DIL; + shader.mVertexShader = mQuad3DVS; + shader.mGeometryShader = mQuad3DGS; + shader.mPixelShader = ps; + + mBlitShaderMap[params] = shader; +} + +void Blit11::addSwizzleShaderToMap(GLenum destType, D3D11_SRV_DIMENSION viewDimension, ID3D11PixelShader *ps) +{ + SwizzleParameters params = { 0 }; + params.mDestinationType = destType; + params.mViewDimension = viewDimension; + + ASSERT(mSwizzleShaderMap.find(params) == mSwizzleShaderMap.end()); + ASSERT(ps); + + Shader shader; + switch (viewDimension) + { + case D3D_SRV_DIMENSION_TEXTURE2D: + shader.mVertexWriteFunction = Write2DVertices; + shader.mInputLayout = mQuad2DIL; + shader.mVertexShader = mQuad2DVS; + shader.mGeometryShader = NULL; + break; + + case D3D_SRV_DIMENSION_TEXTURE3D: + case D3D_SRV_DIMENSION_TEXTURE2DARRAY: + case D3D_SRV_DIMENSION_TEXTURECUBE: + shader.mVertexWriteFunction = Write3DVertices; + shader.mInputLayout = mQuad3DIL; + shader.mVertexShader = mQuad3DVS; + shader.mGeometryShader = mQuad3DGS; + break; + + default: + UNREACHABLE(); + break; + } + shader.mPixelShader = ps; + + mSwizzleShaderMap[params] = shader; +} + +void Blit11::buildShaderMap() +{ + ID3D11Device *device = mRenderer->getDevice(); + + add2DBlitShaderToMap(GL_RGBA, false, d3d11::CompilePS(device, g_PS_PassthroughRGBA2D, "Blit11 2D RGBA pixel shader" )); + add2DBlitShaderToMap(GL_BGRA_EXT, false, d3d11::CompilePS(device, g_PS_PassthroughRGBA2D, "Blit11 2D BGRA pixel shader" )); + add2DBlitShaderToMap(GL_RGB, false, d3d11::CompilePS(device, g_PS_PassthroughRGB2D, "Blit11 2D RGB pixel shader" )); + add2DBlitShaderToMap(GL_RG, false, d3d11::CompilePS(device, g_PS_PassthroughRG2D, "Blit11 2D RG pixel shader" )); + add2DBlitShaderToMap(GL_RED, false, d3d11::CompilePS(device, g_PS_PassthroughR2D, "Blit11 2D R pixel shader" )); + add2DBlitShaderToMap(GL_ALPHA, false, d3d11::CompilePS(device, g_PS_PassthroughRGBA2D, "Blit11 2D alpha pixel shader" )); + add2DBlitShaderToMap(GL_LUMINANCE, false, d3d11::CompilePS(device, g_PS_PassthroughLum2D, "Blit11 2D lum pixel shader" )); + add2DBlitShaderToMap(GL_LUMINANCE_ALPHA, false, d3d11::CompilePS(device, g_PS_PassthroughLumAlpha2D, "Blit11 2D luminance alpha pixel shader")); + + addSwizzleShaderToMap(GL_FLOAT, D3D_SRV_DIMENSION_TEXTURE2D, d3d11::CompilePS(device, g_PS_SwizzleF2D, "Blit11 2D F swizzle pixel shader" )); + + if (mRenderer->isLevel9()) + return; + + add2DBlitShaderToMap(GL_RGBA_INTEGER, false, d3d11::CompilePS(device, g_PS_PassthroughRGBA2DUI, "Blit11 2D RGBA UI pixel shader" )); + add2DBlitShaderToMap(GL_RGBA_INTEGER, true, d3d11::CompilePS(device, g_PS_PassthroughRGBA2DI, "Blit11 2D RGBA I pixel shader" )); + add2DBlitShaderToMap(GL_RGB_INTEGER, false, d3d11::CompilePS(device, g_PS_PassthroughRGB2DUI, "Blit11 2D RGB UI pixel shader" )); + add2DBlitShaderToMap(GL_RGB_INTEGER, true, d3d11::CompilePS(device, g_PS_PassthroughRGB2DI, "Blit11 2D RGB I pixel shader" )); + add2DBlitShaderToMap(GL_RG_INTEGER, false, d3d11::CompilePS(device, g_PS_PassthroughRG2DUI, "Blit11 2D RG UI pixel shader" )); + add2DBlitShaderToMap(GL_RG_INTEGER, true, d3d11::CompilePS(device, g_PS_PassthroughRG2DI, "Blit11 2D RG I pixel shader" )); + add2DBlitShaderToMap(GL_RED_INTEGER, false, d3d11::CompilePS(device, g_PS_PassthroughR2DUI, "Blit11 2D R UI pixel shader" )); + add2DBlitShaderToMap(GL_RED_INTEGER, true, d3d11::CompilePS(device, g_PS_PassthroughR2DI, "Blit11 2D R I pixel shader" )); + add3DBlitShaderToMap(GL_RGBA, false, d3d11::CompilePS(device, g_PS_PassthroughRGBA3D, "Blit11 3D RGBA pixel shader" )); + add3DBlitShaderToMap(GL_RGBA_INTEGER, false, d3d11::CompilePS(device, g_PS_PassthroughRGBA3DUI, "Blit11 3D UI RGBA pixel shader" )); + add3DBlitShaderToMap(GL_RGBA_INTEGER, true, d3d11::CompilePS(device, g_PS_PassthroughRGBA3DI, "Blit11 3D I RGBA pixel shader" )); + add3DBlitShaderToMap(GL_BGRA_EXT, false, d3d11::CompilePS(device, g_PS_PassthroughRGBA3D, "Blit11 3D BGRA pixel shader" )); + add3DBlitShaderToMap(GL_RGB, false, d3d11::CompilePS(device, g_PS_PassthroughRGB3D, "Blit11 3D RGB pixel shader" )); + add3DBlitShaderToMap(GL_RG, false, d3d11::CompilePS(device, g_PS_PassthroughRG3D, "Blit11 3D RG pixel shader" )); + add3DBlitShaderToMap(GL_RGB_INTEGER, false, d3d11::CompilePS(device, g_PS_PassthroughRGB3DUI, "Blit11 3D RGB UI pixel shader" )); + add3DBlitShaderToMap(GL_RGB_INTEGER, true, d3d11::CompilePS(device, g_PS_PassthroughRGB3DI, "Blit11 3D RGB I pixel shader" )); + add3DBlitShaderToMap(GL_RED, false, d3d11::CompilePS(device, g_PS_PassthroughR3D, "Blit11 3D R pixel shader" )); + add3DBlitShaderToMap(GL_RG_INTEGER, false, d3d11::CompilePS(device, g_PS_PassthroughRG3DUI, "Blit11 3D RG UI pixel shader" )); + add3DBlitShaderToMap(GL_RG_INTEGER, true, d3d11::CompilePS(device, g_PS_PassthroughRG3DI, "Blit11 3D RG I pixel shader" )); + add3DBlitShaderToMap(GL_RED_INTEGER, false, d3d11::CompilePS(device, g_PS_PassthroughR3DUI, "Blit11 3D R UI pixel shader" )); + add3DBlitShaderToMap(GL_RED_INTEGER, true, d3d11::CompilePS(device, g_PS_PassthroughR3DI, "Blit11 3D R I pixel shader" )); + add3DBlitShaderToMap(GL_ALPHA, false, d3d11::CompilePS(device, g_PS_PassthroughRGBA3D, "Blit11 3D alpha pixel shader" )); + add3DBlitShaderToMap(GL_LUMINANCE, false, d3d11::CompilePS(device, g_PS_PassthroughLum3D, "Blit11 3D luminance pixel shader" )); + add3DBlitShaderToMap(GL_LUMINANCE_ALPHA, false, d3d11::CompilePS(device, g_PS_PassthroughLumAlpha3D, "Blit11 3D luminance alpha pixel shader")); + + addSwizzleShaderToMap(GL_UNSIGNED_INT, D3D_SRV_DIMENSION_TEXTURE2D, d3d11::CompilePS(device, g_PS_SwizzleUI2D, "Blit11 2D UI swizzle pixel shader")); + addSwizzleShaderToMap(GL_INT, D3D_SRV_DIMENSION_TEXTURE2D, d3d11::CompilePS(device, g_PS_SwizzleI2D, "Blit11 2D I swizzle pixel shader" )); + + addSwizzleShaderToMap(GL_FLOAT, D3D_SRV_DIMENSION_TEXTURECUBE, d3d11::CompilePS(device, g_PS_SwizzleF2DArray, "Blit11 2D Cube F swizzle pixel shader" )); + addSwizzleShaderToMap(GL_UNSIGNED_INT, D3D_SRV_DIMENSION_TEXTURECUBE, d3d11::CompilePS(device, g_PS_SwizzleUI2DArray, "Blit11 2D Cube UI swizzle pixel shader")); + addSwizzleShaderToMap(GL_INT, D3D_SRV_DIMENSION_TEXTURECUBE, d3d11::CompilePS(device, g_PS_SwizzleI2DArray, "Blit11 2D Cube I swizzle pixel shader" )); + + addSwizzleShaderToMap(GL_FLOAT, D3D_SRV_DIMENSION_TEXTURE3D, d3d11::CompilePS(device, g_PS_SwizzleF3D, "Blit11 3D F swizzle pixel shader" )); + addSwizzleShaderToMap(GL_UNSIGNED_INT, D3D_SRV_DIMENSION_TEXTURE3D, d3d11::CompilePS(device, g_PS_SwizzleUI3D, "Blit11 3D UI swizzle pixel shader")); + addSwizzleShaderToMap(GL_INT, D3D_SRV_DIMENSION_TEXTURE3D, d3d11::CompilePS(device, g_PS_SwizzleI3D, "Blit11 3D I swizzle pixel shader" )); + + addSwizzleShaderToMap(GL_FLOAT, D3D_SRV_DIMENSION_TEXTURE2DARRAY, d3d11::CompilePS(device, g_PS_SwizzleF2DArray, "Blit11 2D Array F swizzle pixel shader" )); + addSwizzleShaderToMap(GL_UNSIGNED_INT, D3D_SRV_DIMENSION_TEXTURE2DARRAY, d3d11::CompilePS(device, g_PS_SwizzleUI2DArray, "Blit11 2D Array UI swizzle pixel shader")); + addSwizzleShaderToMap(GL_INT, D3D_SRV_DIMENSION_TEXTURE2DARRAY, d3d11::CompilePS(device, g_PS_SwizzleI2DArray, "Blit11 2D Array I swizzle pixel shader" )); +} + +void Blit11::clearShaderMap() +{ + for (BlitShaderMap::iterator i = mBlitShaderMap.begin(); i != mBlitShaderMap.end(); ++i) + { + Shader &shader = i->second; + SafeRelease(shader.mPixelShader); + } + mBlitShaderMap.clear(); + + for (SwizzleShaderMap::iterator i = mSwizzleShaderMap.begin(); i != mSwizzleShaderMap.end(); ++i) + { + Shader &shader = i->second; + SafeRelease(shader.mPixelShader); + } + mSwizzleShaderMap.clear(); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Blit11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Blit11.h new file mode 100644 index 0000000000..d6a0b795f4 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Blit11.h @@ -0,0 +1,129 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Blit11.cpp: Texture copy utility class. + +#ifndef LIBGLESV2_BLIT11_H_ +#define LIBGLESV2_BLIT11_H_ + +#include "common/angleutils.h" +#include "libGLESv2/angletypes.h" +#include "libGLESv2/Error.h" + +#include <map> + +namespace rx +{ +class Renderer11; + +enum Filter +{ + Point, + Linear, +}; + +class Blit11 +{ + public: + explicit Blit11(Renderer11 *renderer); + ~Blit11(); + + gl::Error swizzleTexture(ID3D11ShaderResourceView *source, ID3D11RenderTargetView *dest, const gl::Extents &size, + GLenum swizzleRed, GLenum swizzleGreen, GLenum swizzleBlue, GLenum swizzleAlpha); + + bool copyTexture(ID3D11ShaderResourceView *source, const gl::Box &sourceArea, const gl::Extents &sourceSize, + ID3D11RenderTargetView *dest, const gl::Box &destArea, const gl::Extents &destSize, + const gl::Rectangle *scissor, GLenum destFormat, GLenum filter); + + bool copyStencil(ID3D11Resource *source, unsigned int sourceSubresource, const gl::Box &sourceArea, const gl::Extents &sourceSize, + ID3D11Resource *dest, unsigned int destSubresource, const gl::Box &destArea, const gl::Extents &destSize, + const gl::Rectangle *scissor); + + bool copyDepth(ID3D11ShaderResourceView *source, const gl::Box &sourceArea, const gl::Extents &sourceSize, + ID3D11DepthStencilView *dest, const gl::Box &destArea, const gl::Extents &destSize, + const gl::Rectangle *scissor); + + bool copyDepthStencil(ID3D11Resource *source, unsigned int sourceSubresource, const gl::Box &sourceArea, const gl::Extents &sourceSize, + ID3D11Resource *dest, unsigned int destSubresource, const gl::Box &destArea, const gl::Extents &destSize, + const gl::Rectangle *scissor); + + private: + rx::Renderer11 *mRenderer; + + struct BlitParameters + { + GLenum mDestinationFormat; + bool mSignedInteger; + bool m3DBlit; + }; + + bool copyDepthStencil(ID3D11Resource *source, unsigned int sourceSubresource, const gl::Box &sourceArea, const gl::Extents &sourceSize, + ID3D11Resource *dest, unsigned int destSubresource, const gl::Box &destArea, const gl::Extents &destSize, + const gl::Rectangle *scissor, bool stencilOnly); + + static bool compareBlitParameters(const BlitParameters &a, const BlitParameters &b); + + typedef void (*WriteVertexFunction)(const gl::Box &sourceArea, const gl::Extents &sourceSize, + const gl::Box &destArea, const gl::Extents &destSize, + void *outVertices, unsigned int *outStride, unsigned int *outVertexCount, + D3D11_PRIMITIVE_TOPOLOGY *outTopology); + + struct Shader + { + WriteVertexFunction mVertexWriteFunction; + ID3D11InputLayout *mInputLayout; + ID3D11VertexShader *mVertexShader; + ID3D11GeometryShader *mGeometryShader; + ID3D11PixelShader *mPixelShader; + }; + + typedef bool (*BlitParametersComparisonFunction)(const BlitParameters&, const BlitParameters &); + typedef std::map<BlitParameters, Shader, BlitParametersComparisonFunction> BlitShaderMap; + BlitShaderMap mBlitShaderMap; + + void add2DBlitShaderToMap(GLenum destFormat, bool signedInteger, ID3D11PixelShader *ps); + void add3DBlitShaderToMap(GLenum destFormat, bool signedInteger, ID3D11PixelShader *ps); + + struct SwizzleParameters + { + GLenum mDestinationType; + D3D11_SRV_DIMENSION mViewDimension; + }; + + static bool compareSwizzleParameters(const SwizzleParameters &a, const SwizzleParameters &b); + + typedef bool (*SwizzleParametersComparisonFunction)(const SwizzleParameters&, const SwizzleParameters &); + typedef std::map<SwizzleParameters, Shader, SwizzleParametersComparisonFunction> SwizzleShaderMap; + SwizzleShaderMap mSwizzleShaderMap; + + void addSwizzleShaderToMap(GLenum destType, D3D11_SRV_DIMENSION viewDimension, ID3D11PixelShader *ps); + + void buildShaderMap(); + void clearShaderMap(); + + ID3D11Buffer *mVertexBuffer; + ID3D11SamplerState *mPointSampler; + ID3D11SamplerState *mLinearSampler; + ID3D11RasterizerState *mScissorEnabledRasterizerState; + ID3D11RasterizerState *mScissorDisabledRasterizerState; + ID3D11DepthStencilState *mDepthStencilState; + + ID3D11InputLayout *mQuad2DIL; + ID3D11VertexShader *mQuad2DVS; + ID3D11PixelShader *mDepthPS; + + ID3D11InputLayout *mQuad3DIL; + ID3D11VertexShader *mQuad3DVS; + ID3D11GeometryShader *mQuad3DGS; + + ID3D11Buffer *mSwizzleCB; + + DISALLOW_COPY_AND_ASSIGN(Blit11); +}; + +} + +#endif // LIBGLESV2_BLIT11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Buffer11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Buffer11.cpp new file mode 100644 index 0000000000..ecd4d4672b --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Buffer11.cpp @@ -0,0 +1,961 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Buffer11.cpp Defines the Buffer11 class. + +#include "libGLESv2/renderer/d3d/d3d11/Buffer11.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" +#include "libGLESv2/renderer/d3d/d3d11/formatutils11.h" +#include "libGLESv2/main.h" + +namespace rx +{ + +PackPixelsParams::PackPixelsParams() + : format(GL_NONE), + type(GL_NONE), + outputPitch(0), + packBuffer(NULL), + offset(0) +{} + +PackPixelsParams::PackPixelsParams(const gl::Rectangle &areaIn, GLenum formatIn, GLenum typeIn, GLuint outputPitchIn, + const gl::PixelPackState &packIn, ptrdiff_t offsetIn) + : area(areaIn), + format(formatIn), + type(typeIn), + outputPitch(outputPitchIn), + packBuffer(packIn.pixelBuffer.get()), + pack(packIn.alignment, packIn.reverseRowOrder), + offset(offsetIn) +{} + +namespace gl_d3d11 +{ + +D3D11_MAP GetD3DMapTypeFromBits(GLbitfield access) +{ + bool readBit = ((access & GL_MAP_READ_BIT) != 0); + bool writeBit = ((access & GL_MAP_WRITE_BIT) != 0); + + ASSERT(readBit || writeBit); + + // Note : we ignore the discard bit, because in D3D11, staging buffers + // don't accept the map-discard flag (discard only works for DYNAMIC usage) + + if (readBit && !writeBit) + { + return D3D11_MAP_READ; + } + else if (writeBit && !readBit) + { + return D3D11_MAP_WRITE; + } + else if (writeBit && readBit) + { + return D3D11_MAP_READ_WRITE; + } + else + { + UNREACHABLE(); + return D3D11_MAP_READ; + } +} + +} + +// Each instance of Buffer11::BufferStorage11 is specialized for a class of D3D binding points +// - vertex/transform feedback buffers +// - index buffers +// - pixel unpack buffers +// - uniform buffers +class Buffer11::BufferStorage11 +{ + public: + virtual ~BufferStorage11() {} + + DataRevision getDataRevision() const { return mRevision; } + BufferUsage getUsage() const { return mUsage; } + size_t getSize() const { return mBufferSize; } + bool isMappable() const { return (mUsage == BUFFER_USAGE_STAGING || mUsage == BUFFER_USAGE_PIXEL_PACK); } + + void setDataRevision(DataRevision rev) { mRevision = rev; } + + virtual bool copyFromStorage(BufferStorage11 *source, size_t sourceOffset, + size_t size, size_t destOffset) = 0; + virtual bool resize(size_t size, bool preserveData) = 0; + + virtual void *map(size_t offset, size_t length, GLbitfield access) = 0; + virtual void unmap() = 0; + + protected: + BufferStorage11(Renderer11 *renderer, BufferUsage usage); + + Renderer11 *mRenderer; + DataRevision mRevision; + const BufferUsage mUsage; + size_t mBufferSize; +}; + +// A native buffer storage represents an underlying D3D11 buffer for a particular +// type of storage. +class Buffer11::NativeBuffer11 : public Buffer11::BufferStorage11 +{ + public: + NativeBuffer11(Renderer11 *renderer, BufferUsage usage); + ~NativeBuffer11(); + + ID3D11Buffer *getNativeBuffer() const { return mNativeBuffer; } + + virtual bool copyFromStorage(BufferStorage11 *source, size_t sourceOffset, + size_t size, size_t destOffset); + virtual bool resize(size_t size, bool preserveData); + + virtual void *map(size_t offset, size_t length, GLbitfield access); + virtual void unmap(); + + bool setData(D3D11_MAP mapMode, const uint8_t *data, size_t size, size_t offset); + + private: + ID3D11Buffer *mNativeBuffer; + + static void fillBufferDesc(D3D11_BUFFER_DESC* bufferDesc, Renderer *renderer, BufferUsage usage, unsigned int bufferSize); +}; + +// Pack storage represents internal storage for pack buffers. We implement pack buffers +// as CPU memory, tied to a staging texture, for asynchronous texture readback. +class Buffer11::PackStorage11 : public Buffer11::BufferStorage11 +{ + public: + PackStorage11(Renderer11 *renderer); + ~PackStorage11(); + + virtual bool copyFromStorage(BufferStorage11 *source, size_t sourceOffset, + size_t size, size_t destOffset); + virtual bool resize(size_t size, bool preserveData); + + virtual void *map(size_t offset, size_t length, GLbitfield access); + virtual void unmap(); + + gl::Error packPixels(ID3D11Texture2D *srcTexure, UINT srcSubresource, const PackPixelsParams ¶ms); + + private: + + void flushQueuedPackCommand(); + + ID3D11Texture2D *mStagingTexture; + DXGI_FORMAT mTextureFormat; + gl::Extents mTextureSize; + MemoryBuffer mMemoryBuffer; + PackPixelsParams *mQueuedPackCommand; + PackPixelsParams mPackParams; + bool mDataModified; +}; + + +Buffer11::Buffer11(Renderer11 *renderer) + : BufferD3D(), + mRenderer(renderer), + mSize(0), + mMappedStorage(NULL), + mResolvedDataRevision(0), + mReadUsageCount(0) +{} + +Buffer11::~Buffer11() +{ + for (auto it = mBufferStorages.begin(); it != mBufferStorages.end(); it++) + { + SafeDelete(it->second); + } +} + +Buffer11 *Buffer11::makeBuffer11(BufferImpl *buffer) +{ + ASSERT(HAS_DYNAMIC_TYPE(Buffer11*, buffer)); + return static_cast<Buffer11*>(buffer); +} + +gl::Error Buffer11::setData(const void *data, size_t size, GLenum usage) +{ + gl::Error error = setSubData(data, size, 0); + if (error.isError()) + { + return error; + } + + if (usage == GL_STATIC_DRAW) + { + initializeStaticData(); + } + + return error; +} + +void *Buffer11::getData() +{ + NativeBuffer11 *stagingBuffer = getStagingBuffer(); + + if (!stagingBuffer) + { + // Out-of-memory + return NULL; + } + + if (stagingBuffer->getDataRevision() > mResolvedDataRevision) + { + if (stagingBuffer->getSize() > mResolvedData.size()) + { + if (!mResolvedData.resize(stagingBuffer->getSize())) + { + return gl::error(GL_OUT_OF_MEMORY, (void*)NULL); + } + } + + ID3D11DeviceContext *context = mRenderer->getDeviceContext(); + + D3D11_MAPPED_SUBRESOURCE mappedResource; + HRESULT result = context->Map(stagingBuffer->getNativeBuffer(), 0, D3D11_MAP_READ, 0, &mappedResource); + if (FAILED(result)) + { + return gl::error(GL_OUT_OF_MEMORY, (void*)NULL); + } + + memcpy(mResolvedData.data(), mappedResource.pData, stagingBuffer->getSize()); + + context->Unmap(stagingBuffer->getNativeBuffer(), 0); + + mResolvedDataRevision = stagingBuffer->getDataRevision(); + } + + mReadUsageCount = 0; + + // Only happens if we initialized the buffer with no data (NULL) + if (mResolvedData.empty()) + { + if (!mResolvedData.resize(mSize)) + { + return gl::error(GL_OUT_OF_MEMORY, (void*)NULL); + } + } + + ASSERT(mResolvedData.size() >= mSize); + + return mResolvedData.data(); +} + +gl::Error Buffer11::setSubData(const void *data, size_t size, size_t offset) +{ + size_t requiredSize = size + offset; + + if (data && size > 0) + { + NativeBuffer11 *stagingBuffer = getStagingBuffer(); + + if (!stagingBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to allocate internal staging buffer."); + } + + // Explicitly resize the staging buffer, preserving data if the new data will not + // completely fill the buffer + if (stagingBuffer->getSize() < requiredSize) + { + bool preserveData = (offset > 0); + if (!stagingBuffer->resize(requiredSize, preserveData)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to resize internal staging buffer."); + } + } + + if (!stagingBuffer->setData(D3D11_MAP_WRITE, reinterpret_cast<const uint8_t *>(data), size, offset)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to set data on internal staging buffer."); + } + + stagingBuffer->setDataRevision(stagingBuffer->getDataRevision() + 1); + } + + mSize = std::max(mSize, requiredSize); + invalidateStaticData(); + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Buffer11::copySubData(BufferImpl* source, GLintptr sourceOffset, GLintptr destOffset, GLsizeiptr size) +{ + Buffer11 *sourceBuffer = makeBuffer11(source); + ASSERT(sourceBuffer != NULL); + + BufferStorage11 *copyDest = getLatestBufferStorage(); + if (!copyDest) + { + copyDest = getStagingBuffer(); + } + + BufferStorage11 *copySource = sourceBuffer->getLatestBufferStorage(); + + if (!copySource || !copyDest) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to allocate internal staging buffer."); + } + + // If copying to/from a pixel pack buffer, we must have a staging or + // pack buffer partner, because other native buffers can't be mapped + if (copyDest->getUsage() == BUFFER_USAGE_PIXEL_PACK && !copySource->isMappable()) + { + copySource = sourceBuffer->getStagingBuffer(); + } + else if (copySource->getUsage() == BUFFER_USAGE_PIXEL_PACK && !copyDest->isMappable()) + { + copyDest = getStagingBuffer(); + } + + // D3D11 does not allow overlapped copies until 11.1, and only if the + // device supports D3D11_FEATURE_DATA_D3D11_OPTIONS::CopyWithOverlap + // Get around this via a different source buffer + if (copySource == copyDest) + { + if (copySource->getUsage() == BUFFER_USAGE_STAGING) + { + copySource = getBufferStorage(BUFFER_USAGE_VERTEX_OR_TRANSFORM_FEEDBACK); + } + else + { + copySource = getStagingBuffer(); + } + } + + copyDest->copyFromStorage(copySource, sourceOffset, size, destOffset); + copyDest->setDataRevision(copyDest->getDataRevision() + 1); + + mSize = std::max<size_t>(mSize, destOffset + size); + invalidateStaticData(); + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Buffer11::map(size_t offset, size_t length, GLbitfield access, GLvoid **mapPtr) +{ + ASSERT(!mMappedStorage); + + BufferStorage11 *latestStorage = getLatestBufferStorage(); + if (latestStorage && + (latestStorage->getUsage() == BUFFER_USAGE_PIXEL_PACK || + latestStorage->getUsage() == BUFFER_USAGE_STAGING)) + { + // Latest storage is mappable. + mMappedStorage = latestStorage; + } + else + { + // Fall back to using the staging buffer if the latest storage does + // not exist or is not CPU-accessible. + mMappedStorage = getStagingBuffer(); + } + + if (!mMappedStorage) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to allocate mappable internal buffer."); + } + + if ((access & GL_MAP_WRITE_BIT) > 0) + { + // Update the data revision immediately, since the data might be changed at any time + mMappedStorage->setDataRevision(mMappedStorage->getDataRevision() + 1); + } + + void *mappedBuffer = mMappedStorage->map(offset, length, access); + if (!mappedBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to map internal buffer."); + } + + *mapPtr = mappedBuffer; + return gl::Error(GL_NO_ERROR); +} + +gl::Error Buffer11::unmap() +{ + ASSERT(mMappedStorage); + mMappedStorage->unmap(); + mMappedStorage = NULL; + return gl::Error(GL_NO_ERROR); +} + +void Buffer11::markTransformFeedbackUsage() +{ + BufferStorage11 *transformFeedbackStorage = getBufferStorage(BUFFER_USAGE_VERTEX_OR_TRANSFORM_FEEDBACK); + + if (transformFeedbackStorage) + { + transformFeedbackStorage->setDataRevision(transformFeedbackStorage->getDataRevision() + 1); + } + + invalidateStaticData(); +} + +void Buffer11::markBufferUsage() +{ + mReadUsageCount++; + + const unsigned int usageLimit = 5; + + if (mReadUsageCount > usageLimit && mResolvedData.size() > 0) + { + mResolvedData.resize(0); + mResolvedDataRevision = 0; + } +} + +Renderer* Buffer11::getRenderer() +{ + return mRenderer; +} + +ID3D11Buffer *Buffer11::getBuffer(BufferUsage usage) +{ + markBufferUsage(); + + BufferStorage11 *bufferStorage = getBufferStorage(usage); + + if (!bufferStorage) + { + // Storage out-of-memory + return NULL; + } + + ASSERT(HAS_DYNAMIC_TYPE(NativeBuffer11*, bufferStorage)); + + return static_cast<NativeBuffer11*>(bufferStorage)->getNativeBuffer(); +} + +ID3D11ShaderResourceView *Buffer11::getSRV(DXGI_FORMAT srvFormat) +{ + BufferStorage11 *storage = getBufferStorage(BUFFER_USAGE_PIXEL_UNPACK); + + if (!storage) + { + // Storage out-of-memory + return NULL; + } + + ASSERT(HAS_DYNAMIC_TYPE(NativeBuffer11*, storage)); + ID3D11Buffer *buffer = static_cast<NativeBuffer11*>(storage)->getNativeBuffer(); + + auto bufferSRVIt = mBufferResourceViews.find(srvFormat); + + if (bufferSRVIt != mBufferResourceViews.end()) + { + if (bufferSRVIt->second.first == buffer) + { + return bufferSRVIt->second.second; + } + else + { + // The underlying buffer has changed since the SRV was created: recreate the SRV. + SafeRelease(bufferSRVIt->second.second); + } + } + + ID3D11Device *device = mRenderer->getDevice(); + ID3D11ShaderResourceView *bufferSRV = NULL; + + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(srvFormat); + + D3D11_SHADER_RESOURCE_VIEW_DESC bufferSRVDesc; + bufferSRVDesc.Buffer.ElementOffset = 0; + bufferSRVDesc.Buffer.ElementWidth = mSize / dxgiFormatInfo.pixelBytes; + bufferSRVDesc.ViewDimension = D3D11_SRV_DIMENSION_BUFFER; + bufferSRVDesc.Format = srvFormat; + + HRESULT result = device->CreateShaderResourceView(buffer, &bufferSRVDesc, &bufferSRV); + UNUSED_ASSERTION_VARIABLE(result); + ASSERT(SUCCEEDED(result)); + + mBufferResourceViews[srvFormat] = BufferSRVPair(buffer, bufferSRV); + + return bufferSRV; +} + +gl::Error Buffer11::packPixels(ID3D11Texture2D *srcTexture, UINT srcSubresource, const PackPixelsParams ¶ms) +{ + PackStorage11 *packStorage = getPackStorage(); + + BufferStorage11 *latestStorage = getLatestBufferStorage(); + + if (packStorage) + { + gl::Error error = packStorage->packPixels(srcTexture, srcSubresource, params); + if (error.isError()) + { + return error; + } + packStorage->setDataRevision(latestStorage ? latestStorage->getDataRevision() + 1 : 1); + } + + return gl::Error(GL_NO_ERROR); +} + +Buffer11::BufferStorage11 *Buffer11::getBufferStorage(BufferUsage usage) +{ + BufferStorage11 *directBuffer = NULL; + auto directBufferIt = mBufferStorages.find(usage); + if (directBufferIt != mBufferStorages.end()) + { + directBuffer = directBufferIt->second; + } + + if (!directBuffer) + { + if (usage == BUFFER_USAGE_PIXEL_PACK) + { + directBuffer = new PackStorage11(mRenderer); + } + else + { + // buffer is not allocated, create it + directBuffer = new NativeBuffer11(mRenderer, usage); + } + + mBufferStorages.insert(std::make_pair(usage, directBuffer)); + } + + // resize buffer + if (directBuffer->getSize() < mSize) + { + if (!directBuffer->resize(mSize, true)) + { + // Out of memory error + return NULL; + } + } + + BufferStorage11 *latestBuffer = getLatestBufferStorage(); + if (latestBuffer && latestBuffer->getDataRevision() > directBuffer->getDataRevision()) + { + // if copying from a pack buffer to a non-staging native buffer, we must first + // copy through the staging buffer, because other native buffers can't be mapped + if (latestBuffer->getUsage() == BUFFER_USAGE_PIXEL_PACK && !directBuffer->isMappable()) + { + NativeBuffer11 *stagingBuffer = getStagingBuffer(); + + stagingBuffer->copyFromStorage(latestBuffer, 0, latestBuffer->getSize(), 0); + directBuffer->setDataRevision(latestBuffer->getDataRevision()); + + latestBuffer = stagingBuffer; + } + + // if copyFromStorage returns true, the D3D buffer has been recreated + // and we should update our serial + if (directBuffer->copyFromStorage(latestBuffer, 0, latestBuffer->getSize(), 0)) + { + updateSerial(); + } + directBuffer->setDataRevision(latestBuffer->getDataRevision()); + } + + return directBuffer; +} + +Buffer11::BufferStorage11 *Buffer11::getLatestBufferStorage() const +{ + // Even though we iterate over all the direct buffers, it is expected that only + // 1 or 2 will be present. + BufferStorage11 *latestStorage = NULL; + DataRevision latestRevision = 0; + for (auto it = mBufferStorages.begin(); it != mBufferStorages.end(); it++) + { + BufferStorage11 *storage = it->second; + if (!latestStorage || storage->getDataRevision() > latestRevision) + { + latestStorage = storage; + latestRevision = storage->getDataRevision(); + } + } + + return latestStorage; +} + +Buffer11::NativeBuffer11 *Buffer11::getStagingBuffer() +{ + BufferStorage11 *stagingStorage = getBufferStorage(BUFFER_USAGE_STAGING); + + if (!stagingStorage) + { + // Out-of-memory + return NULL; + } + + ASSERT(HAS_DYNAMIC_TYPE(NativeBuffer11*, stagingStorage)); + return static_cast<NativeBuffer11*>(stagingStorage); +} + +Buffer11::PackStorage11 *Buffer11::getPackStorage() +{ + BufferStorage11 *packStorage = getBufferStorage(BUFFER_USAGE_PIXEL_PACK); + + if (!packStorage) + { + // Out-of-memory + return NULL; + } + + ASSERT(HAS_DYNAMIC_TYPE(PackStorage11*, packStorage)); + return static_cast<PackStorage11*>(packStorage); +} + +bool Buffer11::supportsDirectBinding() const +{ + // Do not support direct buffers for dynamic data. The streaming buffer + // offers better performance for data which changes every frame. + // Check for absence of static buffer interfaces to detect dynamic data. + return (mStaticVertexBuffer && mStaticIndexBuffer); +} + +Buffer11::BufferStorage11::BufferStorage11(Renderer11 *renderer, BufferUsage usage) + : mRenderer(renderer), + mUsage(usage), + mRevision(0), + mBufferSize(0) +{ +} + +Buffer11::NativeBuffer11::NativeBuffer11(Renderer11 *renderer, BufferUsage usage) + : BufferStorage11(renderer, usage), + mNativeBuffer(NULL) +{ +} + +Buffer11::NativeBuffer11::~NativeBuffer11() +{ + SafeRelease(mNativeBuffer); +} + +// Returns true if it recreates the direct buffer +bool Buffer11::NativeBuffer11::copyFromStorage(BufferStorage11 *source, size_t sourceOffset, + size_t size, size_t destOffset) +{ + ID3D11DeviceContext *context = mRenderer->getDeviceContext(); + + size_t requiredSize = sourceOffset + size; + bool createBuffer = !mNativeBuffer || mBufferSize < requiredSize; + + // (Re)initialize D3D buffer if needed + if (createBuffer) + { + bool preserveData = (destOffset > 0); + resize(source->getSize(), preserveData); + } + + if (source->getUsage() == BUFFER_USAGE_PIXEL_PACK) + { + ASSERT(HAS_DYNAMIC_TYPE(PackStorage11*, source)); + + void *sourcePointer = source->map(sourceOffset, size, GL_MAP_READ_BIT); + + D3D11_MAPPED_SUBRESOURCE mappedResource; + HRESULT hr = context->Map(mNativeBuffer, 0, D3D11_MAP_WRITE, 0, &mappedResource); + UNUSED_ASSERTION_VARIABLE(hr); + ASSERT(SUCCEEDED(hr)); + + unsigned char *destPointer = static_cast<unsigned char *>(mappedResource.pData) + destOffset; + + // Offset bounds are validated at the API layer + ASSERT(sourceOffset + size <= destOffset + mBufferSize); + memcpy(destPointer, sourcePointer, size); + } + else + { + ASSERT(HAS_DYNAMIC_TYPE(NativeBuffer11*, source)); + + D3D11_BOX srcBox; + srcBox.left = sourceOffset; + srcBox.right = sourceOffset + size; + srcBox.top = 0; + srcBox.bottom = 1; + srcBox.front = 0; + srcBox.back = 1; + + ASSERT(HAS_DYNAMIC_TYPE(NativeBuffer11*, source)); + ID3D11Buffer *sourceBuffer = static_cast<NativeBuffer11*>(source)->getNativeBuffer(); + + context->CopySubresourceRegion(mNativeBuffer, 0, destOffset, 0, 0, sourceBuffer, 0, &srcBox); + } + + return createBuffer; +} + +bool Buffer11::NativeBuffer11::resize(size_t size, bool preserveData) +{ + ID3D11Device *device = mRenderer->getDevice(); + ID3D11DeviceContext *context = mRenderer->getDeviceContext(); + + D3D11_BUFFER_DESC bufferDesc; + fillBufferDesc(&bufferDesc, mRenderer, mUsage, size); + + ID3D11Buffer *newBuffer; + HRESULT result = device->CreateBuffer(&bufferDesc, NULL, &newBuffer); + + if (FAILED(result)) + { + return gl::error(GL_OUT_OF_MEMORY, false); + } + + if (mNativeBuffer && preserveData) + { + // We don't call resize if the buffer is big enough already. + ASSERT(mBufferSize <= size); + + D3D11_BOX srcBox; + srcBox.left = 0; + srcBox.right = mBufferSize; + srcBox.top = 0; + srcBox.bottom = 1; + srcBox.front = 0; + srcBox.back = 1; + + context->CopySubresourceRegion(newBuffer, 0, 0, 0, 0, mNativeBuffer, 0, &srcBox); + } + + // No longer need the old buffer + SafeRelease(mNativeBuffer); + mNativeBuffer = newBuffer; + + mBufferSize = bufferDesc.ByteWidth; + + return true; +} + +void Buffer11::NativeBuffer11::fillBufferDesc(D3D11_BUFFER_DESC* bufferDesc, Renderer *renderer, + BufferUsage usage, unsigned int bufferSize) +{ + bufferDesc->ByteWidth = bufferSize; + bufferDesc->MiscFlags = 0; + bufferDesc->StructureByteStride = 0; + + switch (usage) + { + case BUFFER_USAGE_STAGING: + bufferDesc->Usage = D3D11_USAGE_STAGING; + bufferDesc->BindFlags = 0; + bufferDesc->CPUAccessFlags = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; + break; + + case BUFFER_USAGE_VERTEX_OR_TRANSFORM_FEEDBACK: + bufferDesc->Usage = D3D11_USAGE_DEFAULT; + bufferDesc->BindFlags = D3D11_BIND_VERTEX_BUFFER; + if (!static_cast<Renderer11 *>(renderer)->isLevel9()) + bufferDesc->BindFlags |= D3D11_BIND_STREAM_OUTPUT; + bufferDesc->CPUAccessFlags = 0; + break; + + case BUFFER_USAGE_INDEX: + bufferDesc->Usage = D3D11_USAGE_DEFAULT; + bufferDesc->BindFlags = D3D11_BIND_INDEX_BUFFER; + bufferDesc->CPUAccessFlags = 0; + break; + + case BUFFER_USAGE_PIXEL_UNPACK: + bufferDesc->Usage = D3D11_USAGE_DEFAULT; + bufferDesc->BindFlags = D3D11_BIND_SHADER_RESOURCE; + bufferDesc->CPUAccessFlags = 0; + break; + + case BUFFER_USAGE_UNIFORM: + bufferDesc->Usage = D3D11_USAGE_DYNAMIC; + bufferDesc->BindFlags = D3D11_BIND_CONSTANT_BUFFER; + bufferDesc->CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; + + // Constant buffers must be of a limited size, and aligned to 16 byte boundaries + // For our purposes we ignore any buffer data past the maximum constant buffer size + bufferDesc->ByteWidth = roundUp(bufferDesc->ByteWidth, 16u); + bufferDesc->ByteWidth = std::min<UINT>(bufferDesc->ByteWidth, renderer->getRendererCaps().maxUniformBlockSize); + break; + + default: + UNREACHABLE(); + } +} + +void *Buffer11::NativeBuffer11::map(size_t offset, size_t length, GLbitfield access) +{ + ASSERT(mUsage == BUFFER_USAGE_STAGING); + + D3D11_MAPPED_SUBRESOURCE mappedResource; + ID3D11DeviceContext *context = mRenderer->getDeviceContext(); + D3D11_MAP d3dMapType = gl_d3d11::GetD3DMapTypeFromBits(access); + UINT d3dMapFlag = ((access & GL_MAP_UNSYNCHRONIZED_BIT) != 0 ? D3D11_MAP_FLAG_DO_NOT_WAIT : 0); + + HRESULT result = context->Map(mNativeBuffer, 0, d3dMapType, d3dMapFlag, &mappedResource); + UNUSED_ASSERTION_VARIABLE(result); + ASSERT(SUCCEEDED(result)); + + return static_cast<GLubyte*>(mappedResource.pData) + offset; +} + +bool Buffer11::NativeBuffer11::setData(D3D11_MAP mapMode, const uint8_t *data, size_t size, size_t offset) +{ + ID3D11DeviceContext *context = mRenderer->getDeviceContext(); + + D3D11_MAPPED_SUBRESOURCE mappedResource; + HRESULT result = context->Map(mNativeBuffer, 0, mapMode, 0, &mappedResource); + if (FAILED(result)) + { + return gl::error(GL_OUT_OF_MEMORY, false); + } + + uint8_t *offsetBufferPointer = reinterpret_cast<uint8_t *>(mappedResource.pData) + offset; + memcpy(offsetBufferPointer, data, size); + + context->Unmap(mNativeBuffer, 0); + + return true; +} + +void Buffer11::NativeBuffer11::unmap() +{ + ASSERT(mUsage == BUFFER_USAGE_STAGING); + ID3D11DeviceContext *context = mRenderer->getDeviceContext(); + context->Unmap(mNativeBuffer, 0); +} + +Buffer11::PackStorage11::PackStorage11(Renderer11 *renderer) + : BufferStorage11(renderer, BUFFER_USAGE_PIXEL_PACK), + mStagingTexture(NULL), + mTextureFormat(DXGI_FORMAT_UNKNOWN), + mQueuedPackCommand(NULL), + mDataModified(false) +{ +} + +Buffer11::PackStorage11::~PackStorage11() +{ + SafeRelease(mStagingTexture); + SafeDelete(mQueuedPackCommand); +} + +bool Buffer11::PackStorage11::copyFromStorage(BufferStorage11 *source, size_t sourceOffset, + size_t size, size_t destOffset) +{ + // We copy through a staging buffer when drawing with a pack buffer, + // or for other cases where we access the pack buffer + UNREACHABLE(); + return false; +} + +bool Buffer11::PackStorage11::resize(size_t size, bool preserveData) +{ + if (size != mBufferSize) + { + if (!mMemoryBuffer.resize(size)) + { + return false; + } + mBufferSize = size; + } + + return true; +} + +void *Buffer11::PackStorage11::map(size_t offset, size_t length, GLbitfield access) +{ + ASSERT(offset + length <= getSize()); + // TODO: fast path + // We might be able to optimize out one or more memcpy calls by detecting when + // and if D3D packs the staging texture memory identically to how we would fill + // the pack buffer according to the current pack state. + + flushQueuedPackCommand(); + mDataModified = (mDataModified || (access & GL_MAP_WRITE_BIT) != 0); + + return mMemoryBuffer.data() + offset; +} + +void Buffer11::PackStorage11::unmap() +{ + // No-op +} + +gl::Error Buffer11::PackStorage11::packPixels(ID3D11Texture2D *srcTexure, UINT srcSubresource, const PackPixelsParams ¶ms) +{ + flushQueuedPackCommand(); + mQueuedPackCommand = new PackPixelsParams(params); + + D3D11_TEXTURE2D_DESC textureDesc; + srcTexure->GetDesc(&textureDesc); + + if (mStagingTexture != NULL && + (mTextureFormat != textureDesc.Format || + mTextureSize.width != params.area.width || + mTextureSize.height != params.area.height)) + { + SafeRelease(mStagingTexture); + mTextureSize.width = 0; + mTextureSize.height = 0; + mTextureFormat = DXGI_FORMAT_UNKNOWN; + } + + if (mStagingTexture == NULL) + { + ID3D11Device *device = mRenderer->getDevice(); + HRESULT hr; + + mTextureSize.width = params.area.width; + mTextureSize.height = params.area.height; + mTextureFormat = textureDesc.Format; + + D3D11_TEXTURE2D_DESC stagingDesc; + stagingDesc.Width = params.area.width; + stagingDesc.Height = params.area.height; + stagingDesc.MipLevels = 1; + stagingDesc.ArraySize = 1; + stagingDesc.Format = mTextureFormat; + stagingDesc.SampleDesc.Count = 1; + stagingDesc.SampleDesc.Quality = 0; + stagingDesc.Usage = D3D11_USAGE_STAGING; + stagingDesc.BindFlags = 0; + stagingDesc.CPUAccessFlags = D3D11_CPU_ACCESS_READ; + stagingDesc.MiscFlags = 0; + + hr = device->CreateTexture2D(&stagingDesc, NULL, &mStagingTexture); + if (FAILED(hr)) + { + ASSERT(hr == E_OUTOFMEMORY); + return gl::Error(GL_OUT_OF_MEMORY, "Failed to allocate internal staging texture."); + } + } + + // ReadPixels from multisampled FBOs isn't supported in current GL + ASSERT(textureDesc.SampleDesc.Count <= 1); + + ID3D11DeviceContext *immediateContext = mRenderer->getDeviceContext(); + D3D11_BOX srcBox; + srcBox.left = params.area.x; + srcBox.right = params.area.x + params.area.width; + srcBox.top = params.area.y; + srcBox.bottom = params.area.y + params.area.height; + srcBox.front = 0; + srcBox.back = 1; + + // Asynchronous copy + immediateContext->CopySubresourceRegion(mStagingTexture, 0, 0, 0, 0, srcTexure, srcSubresource, &srcBox); + + return gl::Error(GL_NO_ERROR); +} + +void Buffer11::PackStorage11::flushQueuedPackCommand() +{ + ASSERT(mMemoryBuffer.size() > 0); + + if (mQueuedPackCommand) + { + mRenderer->packPixels(mStagingTexture, *mQueuedPackCommand, mMemoryBuffer.data()); + SafeDelete(mQueuedPackCommand); + } +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Buffer11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Buffer11.h new file mode 100644 index 0000000000..5f24fb4e2d --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Buffer11.h @@ -0,0 +1,105 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Buffer11.h: Defines the rx::Buffer11 class which implements rx::BufferImpl via rx::BufferD3D. + +#ifndef LIBGLESV2_RENDERER_BUFFER11_H_ +#define LIBGLESV2_RENDERER_BUFFER11_H_ + +#include "libGLESv2/renderer/d3d/BufferD3D.h" +#include "libGLESv2/renderer/d3d/MemoryBuffer.h" +#include "libGLESv2/angletypes.h" + +namespace rx +{ +class Renderer11; + +enum BufferUsage +{ + BUFFER_USAGE_STAGING, + BUFFER_USAGE_VERTEX_OR_TRANSFORM_FEEDBACK, + BUFFER_USAGE_INDEX, + BUFFER_USAGE_PIXEL_UNPACK, + BUFFER_USAGE_PIXEL_PACK, + BUFFER_USAGE_UNIFORM, +}; + +struct PackPixelsParams +{ + PackPixelsParams(); + PackPixelsParams(const gl::Rectangle &area, GLenum format, GLenum type, GLuint outputPitch, + const gl::PixelPackState &pack, ptrdiff_t offset); + + gl::Rectangle area; + GLenum format; + GLenum type; + GLuint outputPitch; + gl::Buffer *packBuffer; + gl::PixelPackState pack; + ptrdiff_t offset; +}; + +typedef size_t DataRevision; + +class Buffer11 : public BufferD3D +{ + public: + Buffer11(rx::Renderer11 *renderer); + virtual ~Buffer11(); + + static Buffer11 *makeBuffer11(BufferImpl *buffer); + + ID3D11Buffer *getBuffer(BufferUsage usage); + ID3D11ShaderResourceView *getSRV(DXGI_FORMAT srvFormat); + bool isMapped() const { return mMappedStorage != NULL; } + gl::Error packPixels(ID3D11Texture2D *srcTexure, UINT srcSubresource, const PackPixelsParams ¶ms); + + // BufferD3D implementation + virtual size_t getSize() const { return mSize; } + virtual bool supportsDirectBinding() const; + virtual Renderer* getRenderer(); + + // BufferImpl implementation + virtual gl::Error setData(const void* data, size_t size, GLenum usage); + virtual void *getData(); + virtual gl::Error setSubData(const void* data, size_t size, size_t offset); + virtual gl::Error copySubData(BufferImpl* source, GLintptr sourceOffset, GLintptr destOffset, GLsizeiptr size); + virtual gl::Error map(size_t offset, size_t length, GLbitfield access, GLvoid **mapPtr); + virtual gl::Error unmap(); + virtual void markTransformFeedbackUsage(); + + private: + DISALLOW_COPY_AND_ASSIGN(Buffer11); + + class BufferStorage11; + class NativeBuffer11; + class PackStorage11; + + rx::Renderer11 *mRenderer; + size_t mSize; + + BufferStorage11 *mMappedStorage; + + std::map<BufferUsage, BufferStorage11*> mBufferStorages; + + typedef std::pair<ID3D11Buffer *, ID3D11ShaderResourceView *> BufferSRVPair; + std::map<DXGI_FORMAT, BufferSRVPair> mBufferResourceViews; + + MemoryBuffer mResolvedData; + DataRevision mResolvedDataRevision; + unsigned int mReadUsageCount; + + void markBufferUsage(); + NativeBuffer11 *getStagingBuffer(); + PackStorage11 *getPackStorage(); + + BufferStorage11 *getBufferStorage(BufferUsage usage); + BufferStorage11 *getLatestBufferStorage() const; +}; + +} + +#endif // LIBGLESV2_RENDERER_BUFFER11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Clear11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Clear11.cpp new file mode 100644 index 0000000000..765d34fd3f --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Clear11.cpp @@ -0,0 +1,555 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Clear11.cpp: Framebuffer clear utility class. + +#include "libGLESv2/renderer/d3d/d3d11/Clear11.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" +#include "libGLESv2/renderer/d3d/d3d11/renderer11_utils.h" +#include "libGLESv2/renderer/d3d/d3d11/RenderTarget11.h" +#include "libGLESv2/formatutils.h" +#include "libGLESv2/Framebuffer.h" +#include "libGLESv2/FramebufferAttachment.h" + +// Precompiled shaders +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/clearfloatvs.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/clearfloatps.h" + +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/clearuintvs.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/clearuintps.h" + +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/clearsintvs.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/clearsintps.h" + +namespace rx +{ + +template <typename T> +static void ApplyVertices(const gl::Extents &framebufferSize, const gl::Rectangle *scissor, const gl::Color<T> &color, float depth, void *buffer) +{ + d3d11::PositionDepthColorVertex<T> *vertices = reinterpret_cast<d3d11::PositionDepthColorVertex<T>*>(buffer); + + float depthClear = gl::clamp01(depth); + float left = -1.0f; + float right = 1.0f; + float top = -1.0f; + float bottom = 1.0f; + + // Clip the quad coordinates to the scissor if needed + if (scissor != NULL) + { + left = std::max(left, (scissor->x / float(framebufferSize.width)) * 2.0f - 1.0f); + right = std::min(right, ((scissor->x + scissor->width) / float(framebufferSize.width)) * 2.0f - 1.0f); + top = std::max(top, ((framebufferSize.height - scissor->y - scissor->height) / float(framebufferSize.height)) * 2.0f - 1.0f); + bottom = std::min(bottom, ((framebufferSize.height - scissor->y) / float(framebufferSize.height)) * 2.0f - 1.0f); + } + + d3d11::SetPositionDepthColorVertex<T>(vertices + 0, left, bottom, depthClear, color); + d3d11::SetPositionDepthColorVertex<T>(vertices + 1, left, top, depthClear, color); + d3d11::SetPositionDepthColorVertex<T>(vertices + 2, right, bottom, depthClear, color); + d3d11::SetPositionDepthColorVertex<T>(vertices + 3, right, top, depthClear, color); +} + +template <unsigned int vsSize, unsigned int psSize> +Clear11::ClearShader Clear11::CreateClearShader(ID3D11Device *device, DXGI_FORMAT colorType, const BYTE (&vsByteCode)[vsSize], const BYTE (&psByteCode)[psSize]) +{ + HRESULT result; + + ClearShader shader = { 0 }; + + D3D11_INPUT_ELEMENT_DESC quadLayout[] = + { + { "POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0 }, + { "COLOR", 0, colorType, 0, 12, D3D11_INPUT_PER_VERTEX_DATA, 0 }, + }; + + result = device->CreateInputLayout(quadLayout, ArraySize(quadLayout), vsByteCode, vsSize, &shader.inputLayout); + ASSERT(SUCCEEDED(result)); + + result = device->CreateVertexShader(vsByteCode, vsSize, NULL, &shader.vertexShader); + ASSERT(SUCCEEDED(result)); + + result = device->CreatePixelShader(psByteCode, psSize, NULL, &shader.pixelShader); + ASSERT(SUCCEEDED(result)); + + return shader; +} + +Clear11::Clear11(Renderer11 *renderer) + : mRenderer(renderer), mClearBlendStates(StructLessThan<ClearBlendInfo>), mClearDepthStencilStates(StructLessThan<ClearDepthStencilInfo>), + mVertexBuffer(NULL), mRasterizerState(NULL) +{ + HRESULT result; + ID3D11Device *device = renderer->getDevice(); + + D3D11_BUFFER_DESC vbDesc; + vbDesc.ByteWidth = sizeof(d3d11::PositionDepthColorVertex<float>) * 4; + vbDesc.Usage = D3D11_USAGE_DYNAMIC; + vbDesc.BindFlags = D3D11_BIND_VERTEX_BUFFER; + vbDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; + vbDesc.MiscFlags = 0; + vbDesc.StructureByteStride = 0; + + result = device->CreateBuffer(&vbDesc, NULL, &mVertexBuffer); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mVertexBuffer, "Clear11 masked clear vertex buffer"); + + D3D11_RASTERIZER_DESC rsDesc; + rsDesc.FillMode = D3D11_FILL_SOLID; + rsDesc.CullMode = D3D11_CULL_NONE; + rsDesc.FrontCounterClockwise = FALSE; + rsDesc.DepthBias = 0; + rsDesc.DepthBiasClamp = 0.0f; + rsDesc.SlopeScaledDepthBias = 0.0f; + rsDesc.DepthClipEnable = mRenderer->isLevel9(); + rsDesc.ScissorEnable = FALSE; + rsDesc.MultisampleEnable = FALSE; + rsDesc.AntialiasedLineEnable = FALSE; + + result = device->CreateRasterizerState(&rsDesc, &mRasterizerState); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mRasterizerState, "Clear11 masked clear rasterizer state"); + + mFloatClearShader = CreateClearShader(device, DXGI_FORMAT_R32G32B32A32_FLOAT, g_VS_ClearFloat, g_PS_ClearFloat); + if (mRenderer->isLevel9()) { + memset(&mUintClearShader, 0, sizeof(ClearShader)); + memset(&mIntClearShader, 0, sizeof(ClearShader)); + return; + } + + mUintClearShader = CreateClearShader(device, DXGI_FORMAT_R32G32B32A32_UINT, g_VS_ClearUint, g_PS_ClearUint ); + mIntClearShader = CreateClearShader(device, DXGI_FORMAT_R32G32B32A32_SINT, g_VS_ClearSint, g_PS_ClearSint ); +} + +Clear11::~Clear11() +{ + for (ClearBlendStateMap::iterator i = mClearBlendStates.begin(); i != mClearBlendStates.end(); i++) + { + SafeRelease(i->second); + } + mClearBlendStates.clear(); + + SafeRelease(mFloatClearShader.inputLayout); + SafeRelease(mFloatClearShader.vertexShader); + SafeRelease(mFloatClearShader.pixelShader); + + SafeRelease(mUintClearShader.inputLayout); + SafeRelease(mUintClearShader.vertexShader); + SafeRelease(mUintClearShader.pixelShader); + + SafeRelease(mIntClearShader.inputLayout); + SafeRelease(mIntClearShader.vertexShader); + SafeRelease(mIntClearShader.pixelShader); + + for (ClearDepthStencilStateMap::iterator i = mClearDepthStencilStates.begin(); i != mClearDepthStencilStates.end(); i++) + { + SafeRelease(i->second); + } + mClearDepthStencilStates.clear(); + + SafeRelease(mVertexBuffer); + SafeRelease(mRasterizerState); +} + +gl::Error Clear11::clearFramebuffer(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer) +{ + // First determine if a scissored clear is needed, this will always require drawing a quad. + // + // Otherwise, iterate over the color buffers which require clearing and determine if they can be + // cleared with ID3D11DeviceContext::ClearRenderTargetView... This requires: + // 1) The render target is being cleared to a float value (will be cast to integer when clearing integer + // render targets as expected but does not work the other way around) + // 2) The format of the render target has no color channels that are currently masked out. + // Clear the easy-to-clear buffers on the spot and accumulate the ones that require special work. + // + // Also determine if the depth stencil can be cleared with ID3D11DeviceContext::ClearDepthStencilView + // by checking if the stencil write mask covers the entire stencil. + // + // To clear the remaining buffers, quads must be drawn containing an int, uint or float vertex color + // attribute. + + gl::Extents framebufferSize; + if (frameBuffer->getFirstColorbuffer() != NULL) + { + gl::FramebufferAttachment *attachment = frameBuffer->getFirstColorbuffer(); + framebufferSize.width = attachment->getWidth(); + framebufferSize.height = attachment->getHeight(); + framebufferSize.depth = 1; + } + else if (frameBuffer->getDepthOrStencilbuffer() != NULL) + { + gl::FramebufferAttachment *attachment = frameBuffer->getDepthOrStencilbuffer(); + framebufferSize.width = attachment->getWidth(); + framebufferSize.height = attachment->getHeight(); + framebufferSize.depth = 1; + } + else + { + UNREACHABLE(); + return gl::Error(GL_INVALID_OPERATION); + } + + if (clearParams.scissorEnabled && (clearParams.scissor.x >= framebufferSize.width || + clearParams.scissor.y >= framebufferSize.height || + clearParams.scissor.x + clearParams.scissor.width <= 0 || + clearParams.scissor.y + clearParams.scissor.height <= 0)) + { + // Scissor is enabled and the scissor rectangle is outside the renderbuffer + return gl::Error(GL_NO_ERROR); + } + + bool needScissoredClear = clearParams.scissorEnabled && (clearParams.scissor.x > 0 || clearParams.scissor.y > 0 || + clearParams.scissor.x + clearParams.scissor.width < framebufferSize.width || + clearParams.scissor.y + clearParams.scissor.height < framebufferSize.height); + + std::vector<MaskedRenderTarget> maskedClearRenderTargets; + RenderTarget11* maskedClearDepthStencil = NULL; + + ID3D11DeviceContext *deviceContext = mRenderer->getDeviceContext(); + + for (unsigned int colorAttachment = 0; colorAttachment < gl::IMPLEMENTATION_MAX_DRAW_BUFFERS; colorAttachment++) + { + if (clearParams.clearColor[colorAttachment] && frameBuffer->isEnabledColorAttachment(colorAttachment)) + { + gl::FramebufferAttachment *attachment = frameBuffer->getColorbuffer(colorAttachment); + if (attachment) + { + RenderTarget11 *renderTarget = d3d11::GetAttachmentRenderTarget(attachment); + if (!renderTarget) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal render target view pointer unexpectedly null."); + } + + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(attachment->getInternalFormat()); + + if (clearParams.colorClearType == GL_FLOAT && + !(formatInfo.componentType == GL_FLOAT || formatInfo.componentType == GL_UNSIGNED_NORMALIZED || formatInfo.componentType == GL_SIGNED_NORMALIZED)) + { + ERR("It is undefined behaviour to clear a render buffer which is not normalized fixed point or floating-" + "point to floating point values (color attachment %u has internal format 0x%X).", colorAttachment, + attachment->getInternalFormat()); + } + + if ((formatInfo.redBits == 0 || !clearParams.colorMaskRed) && + (formatInfo.greenBits == 0 || !clearParams.colorMaskGreen) && + (formatInfo.blueBits == 0 || !clearParams.colorMaskBlue) && + (formatInfo.alphaBits == 0 || !clearParams.colorMaskAlpha)) + { + // Every channel either does not exist in the render target or is masked out + continue; + } + else if (needScissoredClear || clearParams.colorClearType != GL_FLOAT || + (formatInfo.redBits > 0 && !clearParams.colorMaskRed) || + (formatInfo.greenBits > 0 && !clearParams.colorMaskGreen) || + (formatInfo.blueBits > 0 && !clearParams.colorMaskBlue) || + (formatInfo.alphaBits > 0 && !clearParams.colorMaskAlpha)) + { + // A scissored or masked clear is required + MaskedRenderTarget maskAndRt; + bool clearColor = clearParams.clearColor[colorAttachment]; + maskAndRt.colorMask[0] = (clearColor && clearParams.colorMaskRed); + maskAndRt.colorMask[1] = (clearColor && clearParams.colorMaskGreen); + maskAndRt.colorMask[2] = (clearColor && clearParams.colorMaskBlue); + maskAndRt.colorMask[3] = (clearColor && clearParams.colorMaskAlpha); + maskAndRt.renderTarget = renderTarget; + maskedClearRenderTargets.push_back(maskAndRt); + } + else + { + // ID3D11DeviceContext::ClearRenderTargetView is possible + + ID3D11RenderTargetView *framebufferRTV = renderTarget->getRenderTargetView(); + if (!framebufferRTV) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal render target view pointer unexpectedly null."); + } + + const gl::InternalFormat &actualFormatInfo = gl::GetInternalFormatInfo(attachment->getActualFormat()); + + // Check if the actual format has a channel that the internal format does not and set them to the + // default values + const float clearValues[4] = + { + ((formatInfo.redBits == 0 && actualFormatInfo.redBits > 0) ? 0.0f : clearParams.colorFClearValue.red), + ((formatInfo.greenBits == 0 && actualFormatInfo.greenBits > 0) ? 0.0f : clearParams.colorFClearValue.green), + ((formatInfo.blueBits == 0 && actualFormatInfo.blueBits > 0) ? 0.0f : clearParams.colorFClearValue.blue), + ((formatInfo.alphaBits == 0 && actualFormatInfo.alphaBits > 0) ? 1.0f : clearParams.colorFClearValue.alpha), + }; + + deviceContext->ClearRenderTargetView(framebufferRTV, clearValues); + } + } + } + } + + if (clearParams.clearDepth || clearParams.clearStencil) + { + gl::FramebufferAttachment *attachment = frameBuffer->getDepthOrStencilbuffer(); + if (attachment) + { + RenderTarget11 *renderTarget = d3d11::GetAttachmentRenderTarget(attachment); + if (!renderTarget) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal depth stencil view pointer unexpectedly null."); + } + + const gl::InternalFormat &actualFormatInfo = gl::GetInternalFormatInfo(attachment->getActualFormat()); + + unsigned int stencilUnmasked = frameBuffer->hasStencil() ? (1 << actualFormatInfo.stencilBits) - 1 : 0; + bool needMaskedStencilClear = clearParams.clearStencil && (clearParams.stencilWriteMask & stencilUnmasked) != stencilUnmasked; + + if (needScissoredClear || needMaskedStencilClear) + { + maskedClearDepthStencil = renderTarget; + } + else + { + ID3D11DepthStencilView *framebufferDSV = renderTarget->getDepthStencilView(); + if (!framebufferDSV) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal depth stencil view pointer unexpectedly null."); + } + + UINT clearFlags = (clearParams.clearDepth ? D3D11_CLEAR_DEPTH : 0) | + (clearParams.clearStencil ? D3D11_CLEAR_STENCIL : 0); + FLOAT depthClear = gl::clamp01(clearParams.depthClearValue); + UINT8 stencilClear = clearParams.stencilClearValue & 0xFF; + + deviceContext->ClearDepthStencilView(framebufferDSV, clearFlags, depthClear, stencilClear); + } + } + } + + if (maskedClearRenderTargets.size() > 0 || maskedClearDepthStencil) + { + // To clear the render targets and depth stencil in one pass: + // + // Render a quad clipped to the scissor rectangle which draws the clear color and a blend + // state that will perform the required color masking. + // + // The quad's depth is equal to the depth clear value with a depth stencil state that + // will enable or disable depth test/writes if the depth buffer should be cleared or not. + // + // The rasterizer state's stencil is set to always pass or fail based on if the stencil + // should be cleared or not with a stencil write mask of the stencil clear value. + // + // ====================================================================================== + // + // Luckily, the gl spec (ES 3.0.2 pg 183) states that the results of clearing a render- + // buffer that is not normalized fixed point or floating point with floating point values + // are undefined so we can just write floats to them and D3D11 will bit cast them to + // integers. + // + // Also, we don't have to worry about attempting to clear a normalized fixed/floating point + // buffer with integer values because there is no gl API call which would allow it, + // glClearBuffer* calls only clear a single renderbuffer at a time which is verified to + // be a compatible clear type. + + // Bind all the render targets which need clearing + ASSERT(maskedClearRenderTargets.size() <= mRenderer->getRendererCaps().maxDrawBuffers); + std::vector<ID3D11RenderTargetView*> rtvs(maskedClearRenderTargets.size()); + for (unsigned int i = 0; i < maskedClearRenderTargets.size(); i++) + { + RenderTarget11 *renderTarget = maskedClearRenderTargets[i].renderTarget; + ID3D11RenderTargetView *rtv = renderTarget->getRenderTargetView(); + if (!rtv) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal render target view pointer unexpectedly null."); + } + + rtvs[i] = rtv; + } + ID3D11DepthStencilView *dsv = maskedClearDepthStencil ? maskedClearDepthStencil->getDepthStencilView() : NULL; + + ID3D11BlendState *blendState = getBlendState(maskedClearRenderTargets); + const FLOAT blendFactors[4] = { 1.0f, 1.0f, 1.0f, 1.0f }; + const UINT sampleMask = 0xFFFFFFFF; + + ID3D11DepthStencilState *dsState = getDepthStencilState(clearParams); + const UINT stencilClear = clearParams.stencilClearValue & 0xFF; + + // Set the vertices + UINT vertexStride = 0; + const UINT startIdx = 0; + const ClearShader* shader = NULL; + D3D11_MAPPED_SUBRESOURCE mappedResource; + HRESULT result = deviceContext->Map(mVertexBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to map internal masked clear vertex buffer, HRESULT: 0x%X.", result); + } + + const gl::Rectangle *scissorPtr = clearParams.scissorEnabled ? &clearParams.scissor : NULL; + switch (clearParams.colorClearType) + { + case GL_FLOAT: + ApplyVertices(framebufferSize, scissorPtr, clearParams.colorFClearValue, clearParams.depthClearValue, mappedResource.pData); + vertexStride = sizeof(d3d11::PositionDepthColorVertex<float>); + shader = &mFloatClearShader; + break; + + case GL_UNSIGNED_INT: + ApplyVertices(framebufferSize, scissorPtr, clearParams.colorUIClearValue, clearParams.depthClearValue, mappedResource.pData); + vertexStride = sizeof(d3d11::PositionDepthColorVertex<unsigned int>); + shader = &mUintClearShader; + break; + + case GL_INT: + ApplyVertices(framebufferSize, scissorPtr, clearParams.colorIClearValue, clearParams.depthClearValue, mappedResource.pData); + vertexStride = sizeof(d3d11::PositionDepthColorVertex<int>); + shader = &mIntClearShader; + break; + + default: + UNREACHABLE(); + break; + } + + deviceContext->Unmap(mVertexBuffer, 0); + + // Set the viewport to be the same size as the framebuffer + D3D11_VIEWPORT viewport; + viewport.TopLeftX = 0; + viewport.TopLeftY = 0; + viewport.Width = framebufferSize.width; + viewport.Height = framebufferSize.height; + viewport.MinDepth = 0; + viewport.MaxDepth = 1; + deviceContext->RSSetViewports(1, &viewport); + + // Apply state + deviceContext->OMSetBlendState(blendState, blendFactors, sampleMask); + deviceContext->OMSetDepthStencilState(dsState, stencilClear); + deviceContext->RSSetState(mRasterizerState); + + // Apply shaders + deviceContext->IASetInputLayout(shader->inputLayout); + deviceContext->VSSetShader(shader->vertexShader, NULL, 0); + deviceContext->PSSetShader(shader->pixelShader, NULL, 0); + deviceContext->GSSetShader(NULL, NULL, 0); + + // Apply vertex buffer + deviceContext->IASetVertexBuffers(0, 1, &mVertexBuffer, &vertexStride, &startIdx); + deviceContext->IASetPrimitiveTopology(D3D_PRIMITIVE_TOPOLOGY_TRIANGLESTRIP); + + // Apply render targets + deviceContext->OMSetRenderTargets(rtvs.size(), (rtvs.empty() ? NULL : &rtvs[0]), dsv); + + // Draw the clear quad + deviceContext->Draw(4, 0); + + // Clean up + mRenderer->markAllStateDirty(); + } + + return gl::Error(GL_NO_ERROR); +} + +ID3D11BlendState *Clear11::getBlendState(const std::vector<MaskedRenderTarget>& rts) +{ + ClearBlendInfo blendKey = { 0 }; + for (unsigned int i = 0; i < D3D11_SIMULTANEOUS_RENDER_TARGET_COUNT; i++) + { + if (i < rts.size()) + { + RenderTarget11 *rt = rts[i].renderTarget; + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(rt->getInternalFormat()); + + blendKey.maskChannels[i][0] = (rts[i].colorMask[0] && formatInfo.redBits > 0); + blendKey.maskChannels[i][1] = (rts[i].colorMask[1] && formatInfo.greenBits > 0); + blendKey.maskChannels[i][2] = (rts[i].colorMask[2] && formatInfo.blueBits > 0); + blendKey.maskChannels[i][3] = (rts[i].colorMask[3] && formatInfo.alphaBits > 0); + } + else + { + blendKey.maskChannels[i][0] = false; + blendKey.maskChannels[i][1] = false; + blendKey.maskChannels[i][2] = false; + blendKey.maskChannels[i][3] = false; + } + } + + ClearBlendStateMap::const_iterator i = mClearBlendStates.find(blendKey); + if (i != mClearBlendStates.end()) + { + return i->second; + } + else + { + D3D11_BLEND_DESC blendDesc = { 0 }; + blendDesc.AlphaToCoverageEnable = FALSE; + blendDesc.IndependentBlendEnable = (rts.size() > 1) ? TRUE : FALSE; + + for (unsigned int i = 0; i < D3D11_SIMULTANEOUS_RENDER_TARGET_COUNT; i++) + { + blendDesc.RenderTarget[i].BlendEnable = FALSE; + blendDesc.RenderTarget[i].RenderTargetWriteMask = gl_d3d11::ConvertColorMask(blendKey.maskChannels[i][0], + blendKey.maskChannels[i][1], + blendKey.maskChannels[i][2], + blendKey.maskChannels[i][3]); + } + + ID3D11Device *device = mRenderer->getDevice(); + ID3D11BlendState* blendState = NULL; + HRESULT result = device->CreateBlendState(&blendDesc, &blendState); + if (FAILED(result) || !blendState) + { + ERR("Unable to create a ID3D11BlendState, HRESULT: 0x%X.", result); + return NULL; + } + + mClearBlendStates[blendKey] = blendState; + + return blendState; + } +} + +ID3D11DepthStencilState *Clear11::getDepthStencilState(const gl::ClearParameters &clearParams) +{ + ClearDepthStencilInfo dsKey = { 0 }; + dsKey.clearDepth = clearParams.clearDepth; + dsKey.clearStencil = clearParams.clearStencil; + dsKey.stencilWriteMask = clearParams.stencilWriteMask & 0xFF; + + ClearDepthStencilStateMap::const_iterator i = mClearDepthStencilStates.find(dsKey); + if (i != mClearDepthStencilStates.end()) + { + return i->second; + } + else + { + D3D11_DEPTH_STENCIL_DESC dsDesc = { 0 }; + dsDesc.DepthEnable = dsKey.clearDepth ? TRUE : FALSE; + dsDesc.DepthWriteMask = dsKey.clearDepth ? D3D11_DEPTH_WRITE_MASK_ALL : D3D11_DEPTH_WRITE_MASK_ZERO; + dsDesc.DepthFunc = D3D11_COMPARISON_ALWAYS; + dsDesc.StencilEnable = dsKey.clearStencil ? TRUE : FALSE; + dsDesc.StencilReadMask = 0; + dsDesc.StencilWriteMask = dsKey.stencilWriteMask; + dsDesc.FrontFace.StencilFailOp = D3D11_STENCIL_OP_REPLACE; + dsDesc.FrontFace.StencilDepthFailOp = D3D11_STENCIL_OP_REPLACE; + dsDesc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_REPLACE; + dsDesc.FrontFace.StencilFunc = D3D11_COMPARISON_ALWAYS; + dsDesc.BackFace.StencilFailOp = D3D11_STENCIL_OP_REPLACE; + dsDesc.BackFace.StencilDepthFailOp = D3D11_STENCIL_OP_REPLACE; + dsDesc.BackFace.StencilPassOp = D3D11_STENCIL_OP_REPLACE; + dsDesc.BackFace.StencilFunc = D3D11_COMPARISON_ALWAYS; + + ID3D11Device *device = mRenderer->getDevice(); + ID3D11DepthStencilState* dsState = NULL; + HRESULT result = device->CreateDepthStencilState(&dsDesc, &dsState); + if (FAILED(result) || !dsState) + { + ERR("Unable to create a ID3D11DepthStencilState, HRESULT: 0x%X.", result); + return NULL; + } + + mClearDepthStencilStates[dsKey] = dsState; + + return dsState; + } +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Clear11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Clear11.h new file mode 100644 index 0000000000..be8e187c40 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Clear11.h @@ -0,0 +1,87 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Clear11.h: Framebuffer clear utility class. + +#ifndef LIBGLESV2_RENDERER_CLEAR11_H_ +#define LIBGLESV2_RENDERER_CLEAR11_H_ + +#include "libGLESv2/angletypes.h" +#include "libGLESv2/Error.h" + +#include <map> +#include <vector> + +namespace gl +{ +class Framebuffer; +} + +namespace rx +{ +class Renderer11; +class RenderTarget11; + +class Clear11 +{ + public: + explicit Clear11(Renderer11 *renderer); + ~Clear11(); + + // Clears the framebuffer with the supplied clear parameters, assumes that the framebuffer is currently applied. + gl::Error clearFramebuffer(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer); + + private: + Renderer11 *mRenderer; + + struct ClearBlendInfo + { + bool maskChannels[D3D11_SIMULTANEOUS_RENDER_TARGET_COUNT][4]; + }; + typedef bool (*ClearBlendInfoComparisonFunction)(const ClearBlendInfo&, const ClearBlendInfo &); + typedef std::map<ClearBlendInfo, ID3D11BlendState*, ClearBlendInfoComparisonFunction> ClearBlendStateMap; + ClearBlendStateMap mClearBlendStates; + + struct MaskedRenderTarget + { + bool colorMask[4]; + RenderTarget11 *renderTarget; + }; + + ID3D11BlendState *getBlendState(const std::vector<MaskedRenderTarget> &rts); + + struct ClearShader + { + ID3D11InputLayout *inputLayout; + ID3D11VertexShader *vertexShader; + ID3D11PixelShader *pixelShader; + }; + ClearShader mFloatClearShader; + ClearShader mUintClearShader; + ClearShader mIntClearShader; + + template <unsigned int vsSize, unsigned int psSize> + static ClearShader CreateClearShader(ID3D11Device *device, DXGI_FORMAT colorType, const BYTE (&vsByteCode)[vsSize], const BYTE (&psByteCode)[psSize]); + + struct ClearDepthStencilInfo + { + bool clearDepth; + bool clearStencil; + UINT8 stencilWriteMask; + }; + typedef bool (*ClearDepthStencilInfoComparisonFunction)(const ClearDepthStencilInfo&, const ClearDepthStencilInfo &); + typedef std::map<ClearDepthStencilInfo, ID3D11DepthStencilState*, ClearDepthStencilInfoComparisonFunction> ClearDepthStencilStateMap; + ClearDepthStencilStateMap mClearDepthStencilStates; + + ID3D11DepthStencilState *getDepthStencilState(const gl::ClearParameters &clearParams); + + ID3D11Buffer *mVertexBuffer; + ID3D11RasterizerState *mRasterizerState; +}; + +} + +#endif // LIBGLESV2_RENDERER_CLEAR11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Fence11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Fence11.cpp new file mode 100644 index 0000000000..a841b52862 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Fence11.cpp @@ -0,0 +1,70 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Fence11.cpp: Defines the rx::Fence11 class which implements rx::FenceImpl. + +#include "libGLESv2/renderer/d3d/d3d11/Fence11.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" +#include "libGLESv2/main.h" + +namespace rx +{ + +Fence11::Fence11(rx::Renderer11 *renderer) +{ + mRenderer = renderer; + mQuery = NULL; +} + +Fence11::~Fence11() +{ + SafeRelease(mQuery); +} + +bool Fence11::isSet() const +{ + return mQuery != NULL; +} + +void Fence11::set() +{ + if (!mQuery) + { + D3D11_QUERY_DESC queryDesc; + queryDesc.Query = D3D11_QUERY_EVENT; + queryDesc.MiscFlags = 0; + + if (FAILED(mRenderer->getDevice()->CreateQuery(&queryDesc, &mQuery))) + { + return gl::error(GL_OUT_OF_MEMORY); + } + } + + mRenderer->getDeviceContext()->End(mQuery); +} + +bool Fence11::test(bool flushCommandBuffer) +{ + ASSERT(mQuery); + + UINT getDataFlags = (flushCommandBuffer ? 0 : D3D11_ASYNC_GETDATA_DONOTFLUSH); + HRESULT result = mRenderer->getDeviceContext()->GetData(mQuery, NULL, 0, getDataFlags); + + if (mRenderer->isDeviceLost()) + { + return gl::error(GL_OUT_OF_MEMORY, true); + } + + ASSERT(result == S_OK || result == S_FALSE); + return (result == S_OK); +} + +bool Fence11::hasError() const +{ + return mRenderer->isDeviceLost(); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Fence11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Fence11.h index a5398bca14..50c7621776 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Fence11.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Fence11.h @@ -21,11 +21,10 @@ class Fence11 : public FenceImpl explicit Fence11(rx::Renderer11 *renderer); virtual ~Fence11(); - GLboolean isFence(); - void setFence(GLenum condition); - GLboolean testFence(); - void finishFence(); - void getFenceiv(GLenum pname, GLint *params); + bool isSet() const; + void set(); + bool test(bool flushCommandBuffer); + bool hasError() const; private: DISALLOW_COPY_AND_ASSIGN(Fence11); diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Image11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Image11.cpp new file mode 100644 index 0000000000..7536713af4 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Image11.cpp @@ -0,0 +1,569 @@ +// +// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Image11.h: Implements the rx::Image11 class, which acts as the interface to +// the actual underlying resources of a Texture + +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" +#include "libGLESv2/renderer/d3d/d3d11/Image11.h" +#include "libGLESv2/renderer/d3d/d3d11/TextureStorage11.h" +#include "libGLESv2/renderer/d3d/d3d11/formatutils11.h" +#include "libGLESv2/renderer/d3d/d3d11/renderer11_utils.h" +#include "libGLESv2/Framebuffer.h" +#include "libGLESv2/FramebufferAttachment.h" +#include "libGLESv2/main.h" + +#include "common/utilities.h" + +namespace rx +{ + +Image11::Image11() +{ + mStagingTexture = NULL; + mRenderer = NULL; + mDXGIFormat = DXGI_FORMAT_UNKNOWN; + mRecoverFromStorage = false; + mAssociatedStorage = NULL; + mAssociatedStorageLevel = 0; + mAssociatedStorageLayerTarget = 0; + mRecoveredFromStorageCount = 0; +} + +Image11::~Image11() +{ + disassociateStorage(); + releaseStagingTexture(); +} + +Image11 *Image11::makeImage11(Image *img) +{ + ASSERT(HAS_DYNAMIC_TYPE(rx::Image11*, img)); + return static_cast<rx::Image11*>(img); +} + +void Image11::generateMipmap(Image11 *dest, Image11 *src) +{ + ASSERT(src->getDXGIFormat() == dest->getDXGIFormat()); + ASSERT(src->getWidth() == 1 || src->getWidth() / 2 == dest->getWidth()); + ASSERT(src->getHeight() == 1 || src->getHeight() / 2 == dest->getHeight()); + + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(src->getDXGIFormat()); + ASSERT(dxgiFormatInfo.mipGenerationFunction != NULL); + + D3D11_MAPPED_SUBRESOURCE destMapped; + HRESULT destMapResult = dest->map(D3D11_MAP_WRITE, &destMapped); + if (FAILED(destMapResult)) + { + ERR("Failed to map destination image for mip map generation. HRESULT:0x%X", destMapResult); + return; + } + + D3D11_MAPPED_SUBRESOURCE srcMapped; + HRESULT srcMapResult = src->map(D3D11_MAP_READ, &srcMapped); + if (FAILED(srcMapResult)) + { + ERR("Failed to map source image for mip map generation. HRESULT:0x%X", srcMapResult); + + dest->unmap(); + return; + } + + const uint8_t *sourceData = reinterpret_cast<const uint8_t*>(srcMapped.pData); + uint8_t *destData = reinterpret_cast<uint8_t*>(destMapped.pData); + + dxgiFormatInfo.mipGenerationFunction(src->getWidth(), src->getHeight(), src->getDepth(), + sourceData, srcMapped.RowPitch, srcMapped.DepthPitch, + destData, destMapped.RowPitch, destMapped.DepthPitch); + + dest->unmap(); + src->unmap(); + + dest->markDirty(); +} + +bool Image11::isDirty() const +{ + // If mDirty is true + // AND mStagingTexture doesn't exist AND mStagingTexture doesn't need to be recovered from TextureStorage + // AND the texture doesn't require init data (i.e. a blank new texture will suffice) + // then isDirty should still return false. + if (mDirty && !mStagingTexture && !mRecoverFromStorage && !(d3d11::GetTextureFormatInfo(mInternalFormat).dataInitializerFunction != NULL)) + { + return false; + } + + return mDirty; +} + +bool Image11::copyToStorage2D(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) +{ + TextureStorage11_2D *storage11 = TextureStorage11_2D::makeTextureStorage11_2D(storage); + return copyToStorageImpl(storage11, level, 0, xoffset, yoffset, width, height); +} + +bool Image11::copyToStorageCube(TextureStorage *storage, int face, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) +{ + TextureStorage11_Cube *storage11 = TextureStorage11_Cube::makeTextureStorage11_Cube(storage); + return copyToStorageImpl(storage11, level, face, xoffset, yoffset, width, height); +} + +bool Image11::copyToStorage3D(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth) +{ + TextureStorage11_3D *storage11 = TextureStorage11_3D::makeTextureStorage11_3D(storage); + return copyToStorageImpl(storage11, level, 0, xoffset, yoffset, width, height); +} + +bool Image11::copyToStorage2DArray(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLint arrayLayer, GLsizei width, GLsizei height) +{ + TextureStorage11_2DArray *storage11 = TextureStorage11_2DArray::makeTextureStorage11_2DArray(storage); + return copyToStorageImpl(storage11, level, arrayLayer, xoffset, yoffset, width, height); +} + +bool Image11::copyToStorageImpl(TextureStorage11 *storage11, int level, int layerTarget, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) +{ + // If an app's behavior results in an Image11 copying its data to/from to a TextureStorage multiple times, + // then we should just keep the staging texture around to prevent the copying from impacting perf. + // We allow the Image11 to copy its data to/from TextureStorage once. + // This accounts for an app making a late call to glGenerateMipmap. + bool attemptToReleaseStagingTexture = (mRecoveredFromStorageCount < 2); + + if (attemptToReleaseStagingTexture) + { + // If another image is relying on this Storage for its data, then we must let it recover its data before we overwrite it. + storage11->releaseAssociatedImage(level, layerTarget, this); + } + + bool updateSubresourceSuccess = storage11->updateSubresourceLevel(getStagingTexture(), getStagingSubresource(), level, layerTarget, xoffset, yoffset, 0, width, height, 1); + + // Once the image data has been copied into the Storage, we can release it locally. + if (attemptToReleaseStagingTexture && updateSubresourceSuccess) + { + storage11->associateImage(this, level, layerTarget); + releaseStagingTexture(); + mRecoverFromStorage = true; + mAssociatedStorage = storage11; + mAssociatedStorageLevel = level; + mAssociatedStorageLayerTarget = layerTarget; + } + + return updateSubresourceSuccess; +} + +bool Image11::isAssociatedStorageValid(TextureStorage11* textureStorage) const +{ + return (mAssociatedStorage == textureStorage); +} + +bool Image11::recoverFromAssociatedStorage() +{ + if (mRecoverFromStorage) + { + createStagingTexture(); + + bool textureStorageCorrect = mAssociatedStorage->isAssociatedImageValid(mAssociatedStorageLevel, mAssociatedStorageLayerTarget, this); + + // This means that the cached TextureStorage has been modified after this Image11 released its copy of its data. + // This should not have happened. The TextureStorage should have told this Image11 to recover its data before it was overwritten. + ASSERT(textureStorageCorrect); + + if (textureStorageCorrect) + { + // CopySubResource from the Storage to the Staging texture + mAssociatedStorage->copySubresourceLevel(mStagingTexture, mStagingSubresource, mAssociatedStorageLevel, mAssociatedStorageLayerTarget, 0, 0, 0, mWidth, mHeight, mDepth); + mRecoveredFromStorageCount += 1; + } + + // Reset all the recovery parameters, even if the texture storage association is broken. + disassociateStorage(); + + return textureStorageCorrect; + } + + return false; +} + +void Image11::disassociateStorage() +{ + if (mRecoverFromStorage) + { + // Make the texturestorage release the Image11 too + mAssociatedStorage->disassociateImage(mAssociatedStorageLevel, mAssociatedStorageLayerTarget, this); + + mRecoverFromStorage = false; + mAssociatedStorage = NULL; + mAssociatedStorageLevel = 0; + mAssociatedStorageLayerTarget = 0; + } +} + +bool Image11::redefine(Renderer *renderer, GLenum target, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, bool forceRelease) +{ + if (mWidth != width || + mHeight != height || + mInternalFormat != internalformat || + forceRelease) + { + // End the association with the TextureStorage, since that data will be out of date. + // Also reset mRecoveredFromStorageCount since this Image is getting completely redefined. + disassociateStorage(); + mRecoveredFromStorageCount = 0; + + mRenderer = Renderer11::makeRenderer11(renderer); + + mWidth = width; + mHeight = height; + mDepth = depth; + mInternalFormat = internalformat; + mTarget = target; + + // compute the d3d format that will be used + const d3d11::TextureFormat &formatInfo = d3d11::GetTextureFormatInfo(internalformat); + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(formatInfo.texFormat); + mDXGIFormat = formatInfo.texFormat; + mActualFormat = dxgiFormatInfo.internalFormat; + mRenderable = (formatInfo.rtvFormat != DXGI_FORMAT_UNKNOWN); + + SafeRelease(mStagingTexture); + mDirty = (formatInfo.dataInitializerFunction != NULL); + + return true; + } + + return false; +} + +DXGI_FORMAT Image11::getDXGIFormat() const +{ + // this should only happen if the image hasn't been redefined first + // which would be a bug by the caller + ASSERT(mDXGIFormat != DXGI_FORMAT_UNKNOWN); + + return mDXGIFormat; +} + +// Store the pixel rectangle designated by xoffset,yoffset,width,height with pixels stored as format/type at input +// into the target pixel rectangle. +void Image11::loadData(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, + GLint unpackAlignment, GLenum type, const void *input) +{ + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(mInternalFormat); + GLsizei inputRowPitch = formatInfo.computeRowPitch(type, width, unpackAlignment); + GLsizei inputDepthPitch = formatInfo.computeDepthPitch(type, width, height, unpackAlignment); + + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(mDXGIFormat); + GLuint outputPixelSize = dxgiFormatInfo.pixelBytes; + + const d3d11::TextureFormat &d3dFormatInfo = d3d11::GetTextureFormatInfo(mInternalFormat); + LoadImageFunction loadFunction = d3dFormatInfo.loadFunctions.at(type); + + D3D11_MAPPED_SUBRESOURCE mappedImage; + HRESULT result = map(D3D11_MAP_WRITE, &mappedImage); + if (FAILED(result)) + { + ERR("Could not map image for loading."); + return; + } + + uint8_t* offsetMappedData = (reinterpret_cast<uint8_t*>(mappedImage.pData) + (yoffset * mappedImage.RowPitch + xoffset * outputPixelSize + zoffset * mappedImage.DepthPitch)); + loadFunction(width, height, depth, + reinterpret_cast<const uint8_t*>(input), inputRowPitch, inputDepthPitch, + offsetMappedData, mappedImage.RowPitch, mappedImage.DepthPitch); + + unmap(); +} + +void Image11::loadCompressedData(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, + const void *input) +{ + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(mInternalFormat); + GLsizei inputRowPitch = formatInfo.computeRowPitch(GL_UNSIGNED_BYTE, width, 1); + GLsizei inputDepthPitch = formatInfo.computeDepthPitch(GL_UNSIGNED_BYTE, width, height, 1); + + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(mDXGIFormat); + GLuint outputPixelSize = dxgiFormatInfo.pixelBytes; + GLuint outputBlockWidth = dxgiFormatInfo.blockWidth; + GLuint outputBlockHeight = dxgiFormatInfo.blockHeight; + + ASSERT(xoffset % outputBlockWidth == 0); + ASSERT(yoffset % outputBlockHeight == 0); + + const d3d11::TextureFormat &d3dFormatInfo = d3d11::GetTextureFormatInfo(mInternalFormat); + LoadImageFunction loadFunction = d3dFormatInfo.loadFunctions.at(GL_UNSIGNED_BYTE); + + D3D11_MAPPED_SUBRESOURCE mappedImage; + HRESULT result = map(D3D11_MAP_WRITE, &mappedImage); + if (FAILED(result)) + { + ERR("Could not map image for loading."); + return; + } + + uint8_t* offsetMappedData = reinterpret_cast<uint8_t*>(mappedImage.pData) + ((yoffset / outputBlockHeight) * mappedImage.RowPitch + + (xoffset / outputBlockWidth) * outputPixelSize + + zoffset * mappedImage.DepthPitch); + + loadFunction(width, height, depth, + reinterpret_cast<const uint8_t*>(input), inputRowPitch, inputDepthPitch, + offsetMappedData, mappedImage.RowPitch, mappedImage.DepthPitch); + + unmap(); +} + +void Image11::copy(GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) +{ + gl::FramebufferAttachment *colorbuffer = source->getReadColorbuffer(); + + if (colorbuffer && colorbuffer->getActualFormat() == mActualFormat) + { + // No conversion needed-- use copyback fastpath + ID3D11Texture2D *colorBufferTexture = NULL; + unsigned int subresourceIndex = 0; + + if (mRenderer->getRenderTargetResource(colorbuffer, &subresourceIndex, &colorBufferTexture)) + { + D3D11_TEXTURE2D_DESC textureDesc; + colorBufferTexture->GetDesc(&textureDesc); + + ID3D11Device *device = mRenderer->getDevice(); + ID3D11DeviceContext *deviceContext = mRenderer->getDeviceContext(); + + ID3D11Texture2D* srcTex = NULL; + if (textureDesc.SampleDesc.Count > 1) + { + D3D11_TEXTURE2D_DESC resolveDesc; + resolveDesc.Width = textureDesc.Width; + resolveDesc.Height = textureDesc.Height; + resolveDesc.MipLevels = 1; + resolveDesc.ArraySize = 1; + resolveDesc.Format = textureDesc.Format; + resolveDesc.SampleDesc.Count = 1; + resolveDesc.SampleDesc.Quality = 0; + resolveDesc.Usage = D3D11_USAGE_DEFAULT; + resolveDesc.BindFlags = 0; + resolveDesc.CPUAccessFlags = 0; + resolveDesc.MiscFlags = 0; + + HRESULT result = device->CreateTexture2D(&resolveDesc, NULL, &srcTex); + if (FAILED(result)) + { + ERR("Failed to create resolve texture for Image11::copy, HRESULT: 0x%X.", result); + return; + } + + deviceContext->ResolveSubresource(srcTex, 0, colorBufferTexture, subresourceIndex, textureDesc.Format); + subresourceIndex = 0; + } + else + { + srcTex = colorBufferTexture; + srcTex->AddRef(); + } + + D3D11_BOX srcBox; + srcBox.left = x; + srcBox.right = x + width; + srcBox.top = y; + srcBox.bottom = y + height; + srcBox.front = 0; + srcBox.back = 1; + + deviceContext->CopySubresourceRegion(mStagingTexture, 0, xoffset, yoffset, zoffset, srcTex, subresourceIndex, &srcBox); + + SafeRelease(srcTex); + SafeRelease(colorBufferTexture); + } + } + else + { + // This format requires conversion, so we must copy the texture to staging and manually convert via readPixels + D3D11_MAPPED_SUBRESOURCE mappedImage; + HRESULT result = map(D3D11_MAP_WRITE, &mappedImage); + if (FAILED(result)) + { + ERR("Failed to map texture for Image11::copy, HRESULT: 0x%X.", result); + return; + } + + // determine the offset coordinate into the destination buffer + GLsizei rowOffset = gl::GetInternalFormatInfo(mActualFormat).pixelBytes * xoffset; + uint8_t *dataOffset = static_cast<uint8_t*>(mappedImage.pData) + mappedImage.RowPitch * yoffset + rowOffset + zoffset * mappedImage.DepthPitch; + + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(mInternalFormat); + + mRenderer->readPixels(source, x, y, width, height, formatInfo.format, formatInfo.type, mappedImage.RowPitch, gl::PixelPackState(), dataOffset); + + unmap(); + } +} + +ID3D11Resource *Image11::getStagingTexture() +{ + createStagingTexture(); + + return mStagingTexture; +} + +void Image11::releaseStagingTexture() +{ + SafeRelease(mStagingTexture); +} + +unsigned int Image11::getStagingSubresource() +{ + createStagingTexture(); + + return mStagingSubresource; +} + +void Image11::createStagingTexture() +{ + if (mStagingTexture) + { + return; + } + + const DXGI_FORMAT dxgiFormat = getDXGIFormat(); + + if (mWidth > 0 && mHeight > 0 && mDepth > 0) + { + ID3D11Device *device = mRenderer->getDevice(); + HRESULT result; + + int lodOffset = 1; + GLsizei width = mWidth; + GLsizei height = mHeight; + + // adjust size if needed for compressed textures + d3d11::MakeValidSize(false, dxgiFormat, &width, &height, &lodOffset); + + if (mTarget == GL_TEXTURE_3D) + { + ID3D11Texture3D *newTexture = NULL; + + D3D11_TEXTURE3D_DESC desc; + desc.Width = width; + desc.Height = height; + desc.Depth = mDepth; + desc.MipLevels = lodOffset + 1; + desc.Format = dxgiFormat; + desc.Usage = D3D11_USAGE_STAGING; + desc.BindFlags = 0; + desc.CPUAccessFlags = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; + desc.MiscFlags = 0; + + if (d3d11::GetTextureFormatInfo(mInternalFormat).dataInitializerFunction != NULL) + { + std::vector<D3D11_SUBRESOURCE_DATA> initialData; + std::vector< std::vector<BYTE> > textureData; + d3d11::GenerateInitialTextureData(mInternalFormat, width, height, mDepth, + lodOffset + 1, &initialData, &textureData); + + result = device->CreateTexture3D(&desc, initialData.data(), &newTexture); + } + else + { + result = device->CreateTexture3D(&desc, NULL, &newTexture); + } + + if (FAILED(result)) + { + ASSERT(result == E_OUTOFMEMORY); + ERR("Creating image failed."); + return gl::error(GL_OUT_OF_MEMORY); + } + + mStagingTexture = newTexture; + mStagingSubresource = D3D11CalcSubresource(lodOffset, 0, lodOffset + 1); + } + else if (mTarget == GL_TEXTURE_2D || mTarget == GL_TEXTURE_2D_ARRAY || mTarget == GL_TEXTURE_CUBE_MAP) + { + ID3D11Texture2D *newTexture = NULL; + + D3D11_TEXTURE2D_DESC desc; + desc.Width = width; + desc.Height = height; + desc.MipLevels = lodOffset + 1; + desc.ArraySize = 1; + desc.Format = dxgiFormat; + desc.SampleDesc.Count = 1; + desc.SampleDesc.Quality = 0; + desc.Usage = D3D11_USAGE_STAGING; + desc.BindFlags = 0; + desc.CPUAccessFlags = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; + desc.MiscFlags = 0; + + if (d3d11::GetTextureFormatInfo(mInternalFormat).dataInitializerFunction != NULL) + { + std::vector<D3D11_SUBRESOURCE_DATA> initialData; + std::vector< std::vector<BYTE> > textureData; + d3d11::GenerateInitialTextureData(mInternalFormat, width, height, 1, + lodOffset + 1, &initialData, &textureData); + + result = device->CreateTexture2D(&desc, initialData.data(), &newTexture); + } + else + { + result = device->CreateTexture2D(&desc, NULL, &newTexture); + } + + if (FAILED(result)) + { + ASSERT(result == E_OUTOFMEMORY); + ERR("Creating image failed."); + return gl::error(GL_OUT_OF_MEMORY); + } + + mStagingTexture = newTexture; + mStagingSubresource = D3D11CalcSubresource(lodOffset, 0, lodOffset + 1); + } + else + { + UNREACHABLE(); + } + } + + mDirty = false; +} + +HRESULT Image11::map(D3D11_MAP mapType, D3D11_MAPPED_SUBRESOURCE *map) +{ + createStagingTexture(); + + // We must recover from the TextureStorage if necessary, even for D3D11_MAP_WRITE. + recoverFromAssociatedStorage(); + + HRESULT result = E_FAIL; + + if (mStagingTexture) + { + ID3D11DeviceContext *deviceContext = mRenderer->getDeviceContext(); + result = deviceContext->Map(mStagingTexture, mStagingSubresource, mapType, 0, map); + + // this can fail if the device is removed (from TDR) + if (d3d11::isDeviceLostError(result)) + { + mRenderer->notifyDeviceLost(); + } + else if (SUCCEEDED(result)) + { + mDirty = true; + } + } + + return result; +} + +void Image11::unmap() +{ + if (mStagingTexture) + { + ID3D11DeviceContext *deviceContext = mRenderer->getDeviceContext(); + deviceContext->Unmap(mStagingTexture, mStagingSubresource); + } +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Image11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Image11.h new file mode 100644 index 0000000000..a76a61f036 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Image11.h @@ -0,0 +1,89 @@ +// +// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Image11.h: Defines the rx::Image11 class, which acts as the interface to +// the actual underlying resources of a Texture + +#ifndef LIBGLESV2_RENDERER_IMAGE11_H_ +#define LIBGLESV2_RENDERER_IMAGE11_H_ + +#include "libGLESv2/renderer/d3d/ImageD3D.h" + +#include "common/debug.h" + +namespace gl +{ +class Framebuffer; +} + +namespace rx +{ +class Renderer; +class Renderer11; +class TextureStorage11; + +class Image11 : public ImageD3D +{ + public: + Image11(); + virtual ~Image11(); + + static Image11 *makeImage11(Image *img); + + static void generateMipmap(Image11 *dest, Image11 *src); + + virtual bool isDirty() const; + + virtual bool copyToStorage2D(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); + virtual bool copyToStorageCube(TextureStorage *storage, int face, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); + virtual bool copyToStorage3D(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth); + virtual bool copyToStorage2DArray(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLint arrayLayer, GLsizei width, GLsizei height); + + virtual bool redefine(Renderer *renderer, GLenum target, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, bool forceRelease); + + DXGI_FORMAT getDXGIFormat() const; + + virtual void loadData(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, + GLint unpackAlignment, GLenum type, const void *input); + virtual void loadCompressedData(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, + const void *input); + + virtual void copy(GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source); + + bool recoverFromAssociatedStorage(); + bool isAssociatedStorageValid(TextureStorage11* textureStorage) const; + void disassociateStorage(); + + protected: + HRESULT map(D3D11_MAP mapType, D3D11_MAPPED_SUBRESOURCE *map); + void unmap(); + + private: + DISALLOW_COPY_AND_ASSIGN(Image11); + + bool copyToStorageImpl(TextureStorage11 *storage11, int level, int layerTarget, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); + + ID3D11Resource *getStagingTexture(); + unsigned int getStagingSubresource(); + void createStagingTexture(); + void releaseStagingTexture(); + + Renderer11 *mRenderer; + + DXGI_FORMAT mDXGIFormat; + ID3D11Resource *mStagingTexture; + unsigned int mStagingSubresource; + + bool mRecoverFromStorage; + TextureStorage11 *mAssociatedStorage; + int mAssociatedStorageLevel; + int mAssociatedStorageLayerTarget; + unsigned int mRecoveredFromStorageCount; +}; + +} + +#endif // LIBGLESV2_RENDERER_IMAGE11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/IndexBuffer11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/IndexBuffer11.cpp new file mode 100644 index 0000000000..9a61182ee9 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/IndexBuffer11.cpp @@ -0,0 +1,162 @@ +// +// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// IndexBuffer11.cpp: Defines the D3D11 IndexBuffer implementation. + +#include "libGLESv2/renderer/d3d/d3d11/IndexBuffer11.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" + +namespace rx +{ + +IndexBuffer11::IndexBuffer11(Renderer11 *const renderer) : mRenderer(renderer) +{ + mBuffer = NULL; + mBufferSize = 0; + mDynamicUsage = false; +} + +IndexBuffer11::~IndexBuffer11() +{ + SafeRelease(mBuffer); +} + +gl::Error IndexBuffer11::initialize(unsigned int bufferSize, GLenum indexType, bool dynamic) +{ + SafeRelease(mBuffer); + + updateSerial(); + + if (bufferSize > 0) + { + ID3D11Device* dxDevice = mRenderer->getDevice(); + + D3D11_BUFFER_DESC bufferDesc; + bufferDesc.ByteWidth = bufferSize; + bufferDesc.Usage = D3D11_USAGE_DYNAMIC; + bufferDesc.BindFlags = D3D11_BIND_INDEX_BUFFER; + bufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; + bufferDesc.MiscFlags = 0; + bufferDesc.StructureByteStride = 0; + + HRESULT result = dxDevice->CreateBuffer(&bufferDesc, NULL, &mBuffer); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to allocate internal index buffer of size, %lu.", bufferSize); + } + } + + mBufferSize = bufferSize; + mIndexType = indexType; + mDynamicUsage = dynamic; + + return gl::Error(GL_NO_ERROR); +} + +IndexBuffer11 *IndexBuffer11::makeIndexBuffer11(IndexBuffer *indexBuffer) +{ + ASSERT(HAS_DYNAMIC_TYPE(IndexBuffer11*, indexBuffer)); + return static_cast<IndexBuffer11*>(indexBuffer); +} + +gl::Error IndexBuffer11::mapBuffer(unsigned int offset, unsigned int size, void** outMappedMemory) +{ + if (!mBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal index buffer is not initialized."); + } + + // Check for integer overflows and out-out-bounds map requests + if (offset + size < offset || offset + size > mBufferSize) + { + return gl::Error(GL_OUT_OF_MEMORY, "Index buffer map range is not inside the buffer."); + } + + ID3D11DeviceContext *dxContext = mRenderer->getDeviceContext(); + + D3D11_MAPPED_SUBRESOURCE mappedResource; + HRESULT result = dxContext->Map(mBuffer, 0, D3D11_MAP_WRITE_NO_OVERWRITE, 0, &mappedResource); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to map internal index buffer, HRESULT: 0x%08x.", result); + } + + *outMappedMemory = reinterpret_cast<char*>(mappedResource.pData) + offset; + return gl::Error(GL_NO_ERROR); +} + +gl::Error IndexBuffer11::unmapBuffer() +{ + if (!mBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal index buffer is not initialized."); + } + + ID3D11DeviceContext *dxContext = mRenderer->getDeviceContext(); + dxContext->Unmap(mBuffer, 0); + return gl::Error(GL_NO_ERROR); +} + +GLenum IndexBuffer11::getIndexType() const +{ + return mIndexType; +} + +unsigned int IndexBuffer11::getBufferSize() const +{ + return mBufferSize; +} + +gl::Error IndexBuffer11::setSize(unsigned int bufferSize, GLenum indexType) +{ + if (bufferSize > mBufferSize || indexType != mIndexType) + { + return initialize(bufferSize, indexType, mDynamicUsage); + } + else + { + return gl::Error(GL_NO_ERROR); + } +} + +gl::Error IndexBuffer11::discard() +{ + if (!mBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal index buffer is not initialized."); + } + + ID3D11DeviceContext *dxContext = mRenderer->getDeviceContext(); + + D3D11_MAPPED_SUBRESOURCE mappedResource; + HRESULT result = dxContext->Map(mBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to map internal index buffer, HRESULT: 0x%08x.", result); + } + + dxContext->Unmap(mBuffer, 0); + + return gl::Error(GL_NO_ERROR); +} + +DXGI_FORMAT IndexBuffer11::getIndexFormat() const +{ + switch (mIndexType) + { + case GL_UNSIGNED_BYTE: return DXGI_FORMAT_R16_UINT; + case GL_UNSIGNED_SHORT: return DXGI_FORMAT_R16_UINT; + case GL_UNSIGNED_INT: return DXGI_FORMAT_R32_UINT; + default: UNREACHABLE(); return DXGI_FORMAT_UNKNOWN; + } +} + +ID3D11Buffer *IndexBuffer11::getBuffer() const +{ + return mBuffer; +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/IndexBuffer11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/IndexBuffer11.h index 39a61946ad..f7c2b38e7e 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/IndexBuffer11.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/IndexBuffer11.h @@ -9,7 +9,7 @@ #ifndef LIBGLESV2_RENDERER_INDEXBUFFER11_H_ #define LIBGLESV2_RENDERER_INDEXBUFFER11_H_ -#include "libGLESv2/renderer/IndexBuffer.h" +#include "libGLESv2/renderer/d3d/IndexBuffer.h" namespace rx { @@ -21,18 +21,18 @@ class IndexBuffer11 : public IndexBuffer explicit IndexBuffer11(Renderer11 *const renderer); virtual ~IndexBuffer11(); - virtual bool initialize(unsigned int bufferSize, GLenum indexType, bool dynamic); + virtual gl::Error initialize(unsigned int bufferSize, GLenum indexType, bool dynamic); static IndexBuffer11 *makeIndexBuffer11(IndexBuffer *indexBuffer); - virtual bool mapBuffer(unsigned int offset, unsigned int size, void** outMappedMemory); - virtual bool unmapBuffer(); + virtual gl::Error mapBuffer(unsigned int offset, unsigned int size, void** outMappedMemory); + virtual gl::Error unmapBuffer(); virtual GLenum getIndexType() const; virtual unsigned int getBufferSize() const; - virtual bool setSize(unsigned int bufferSize, GLenum indexType); + virtual gl::Error setSize(unsigned int bufferSize, GLenum indexType); - virtual bool discard(); + virtual gl::Error discard(); DXGI_FORMAT getIndexFormat() const; ID3D11Buffer *getBuffer() const; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/InputLayoutCache.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/InputLayoutCache.cpp index 4940b8c638..d835e4fa68 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/InputLayoutCache.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/InputLayoutCache.cpp @@ -1,4 +1,3 @@ -#include "precompiled.h" // // Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be @@ -8,19 +7,35 @@ // InputLayoutCache.cpp: Defines InputLayoutCache, a class that builds and caches // D3D11 input layouts. -#include "libGLESv2/renderer/d3d11/InputLayoutCache.h" -#include "libGLESv2/renderer/d3d11/VertexBuffer11.h" -#include "libGLESv2/renderer/d3d11/BufferStorage11.h" -#include "libGLESv2/renderer/d3d11/ShaderExecutable11.h" +#include "libGLESv2/renderer/d3d/d3d11/InputLayoutCache.h" +#include "libGLESv2/renderer/d3d/d3d11/VertexBuffer11.h" +#include "libGLESv2/renderer/d3d/d3d11/Buffer11.h" +#include "libGLESv2/renderer/d3d/d3d11/ShaderExecutable11.h" +#include "libGLESv2/renderer/d3d/d3d11/formatutils11.h" +#include "libGLESv2/renderer/d3d/VertexDataManager.h" #include "libGLESv2/ProgramBinary.h" -#include "libGLESv2/Context.h" -#include "libGLESv2/renderer/VertexDataManager.h" +#include "libGLESv2/VertexAttribute.h" #include "third_party/murmurhash/MurmurHash3.h" namespace rx { +static void GetInputLayout(const TranslatedAttribute translatedAttributes[gl::MAX_VERTEX_ATTRIBS], + gl::VertexFormat inputLayout[gl::MAX_VERTEX_ATTRIBS]) +{ + for (unsigned int attributeIndex = 0; attributeIndex < gl::MAX_VERTEX_ATTRIBS; attributeIndex++) + { + const TranslatedAttribute &translatedAttribute = translatedAttributes[attributeIndex]; + + if (translatedAttributes[attributeIndex].active) + { + inputLayout[attributeIndex] = gl::VertexFormat(*translatedAttribute.attribute, + translatedAttribute.currentValueType); + } + } +} + const unsigned int InputLayoutCache::kMaxInputLayouts = 1024; InputLayoutCache::InputLayoutCache() : mInputLayoutMap(kMaxInputLayouts, hashInputLayout, compareInputLayouts) @@ -31,7 +46,7 @@ InputLayoutCache::InputLayoutCache() : mInputLayoutMap(kMaxInputLayouts, hashInp mCurrentIL = NULL; for (unsigned int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) { - mCurrentBuffers[i] = -1; + mCurrentBuffers[i] = NULL; mCurrentVertexStrides[i] = -1; mCurrentVertexOffsets[i] = -1; } @@ -53,7 +68,7 @@ void InputLayoutCache::clear() { for (InputLayoutMap::iterator i = mInputLayoutMap.begin(); i != mInputLayoutMap.end(); i++) { - i->second.inputLayout->Release(); + SafeRelease(i->second.inputLayout); } mInputLayoutMap.clear(); markDirty(); @@ -64,42 +79,36 @@ void InputLayoutCache::markDirty() mCurrentIL = NULL; for (unsigned int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) { - mCurrentBuffers[i] = -1; + mCurrentBuffers[i] = NULL; mCurrentVertexStrides[i] = -1; mCurrentVertexOffsets[i] = -1; } } -GLenum InputLayoutCache::applyVertexBuffers(TranslatedAttribute attributes[gl::MAX_VERTEX_ATTRIBS], - gl::ProgramBinary *programBinary) +gl::Error InputLayoutCache::applyVertexBuffers(TranslatedAttribute attributes[gl::MAX_VERTEX_ATTRIBS], + gl::ProgramBinary *programBinary) { int sortedSemanticIndices[gl::MAX_VERTEX_ATTRIBS]; programBinary->sortAttributesByLayout(attributes, sortedSemanticIndices); if (!mDevice || !mDeviceContext) { - ERR("InputLayoutCache is not initialized."); - return GL_INVALID_OPERATION; + return gl::Error(GL_OUT_OF_MEMORY, "Internal input layout cache is not initialized."); } InputLayoutKey ilKey = { 0 }; - ID3D11Buffer *vertexBuffers[gl::MAX_VERTEX_ATTRIBS] = { NULL }; - unsigned int vertexBufferSerials[gl::MAX_VERTEX_ATTRIBS] = { 0 }; - UINT vertexStrides[gl::MAX_VERTEX_ATTRIBS] = { 0 }; - UINT vertexOffsets[gl::MAX_VERTEX_ATTRIBS] = { 0 }; - static const char* semanticName = "TEXCOORD"; for (unsigned int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) { if (attributes[i].active) { - VertexBuffer11 *vertexBuffer = VertexBuffer11::makeVertexBuffer11(attributes[i].vertexBuffer); - BufferStorage11 *bufferStorage = attributes[i].storage ? BufferStorage11::makeBufferStorage11(attributes[i].storage) : NULL; - D3D11_INPUT_CLASSIFICATION inputClass = attributes[i].divisor > 0 ? D3D11_INPUT_PER_INSTANCE_DATA : D3D11_INPUT_PER_VERTEX_DATA; + gl::VertexFormat vertexFormat(*attributes[i].attribute, attributes[i].currentValueType); + const d3d11::VertexFormat &vertexFormatInfo = d3d11::GetVertexFormatInfo(vertexFormat); + // Record the type of the associated vertex shader vector in our key // This will prevent mismatched vertex shaders from using the same input layout GLint attributeSize; @@ -107,31 +116,28 @@ GLenum InputLayoutCache::applyVertexBuffers(TranslatedAttribute attributes[gl::M ilKey.elements[ilKey.elementCount].desc.SemanticName = semanticName; ilKey.elements[ilKey.elementCount].desc.SemanticIndex = i; - ilKey.elements[ilKey.elementCount].desc.Format = attributes[i].attribute->mArrayEnabled ? vertexBuffer->getDXGIFormat(*attributes[i].attribute) : DXGI_FORMAT_R32G32B32A32_FLOAT; + ilKey.elements[ilKey.elementCount].desc.Format = vertexFormatInfo.nativeFormat; ilKey.elements[ilKey.elementCount].desc.InputSlot = i; ilKey.elements[ilKey.elementCount].desc.AlignedByteOffset = 0; ilKey.elements[ilKey.elementCount].desc.InputSlotClass = inputClass; ilKey.elements[ilKey.elementCount].desc.InstanceDataStepRate = attributes[i].divisor; ilKey.elementCount++; - - vertexBuffers[i] = bufferStorage ? bufferStorage->getBuffer(BUFFER_USAGE_VERTEX) : vertexBuffer->getBuffer(); - vertexBufferSerials[i] = bufferStorage ? bufferStorage->getSerial() : vertexBuffer->getSerial(); - vertexStrides[i] = attributes[i].stride; - vertexOffsets[i] = attributes[i].offset; } } ID3D11InputLayout *inputLayout = NULL; - InputLayoutMap::iterator i = mInputLayoutMap.find(ilKey); - if (i != mInputLayoutMap.end()) + InputLayoutMap::iterator keyIter = mInputLayoutMap.find(ilKey); + if (keyIter != mInputLayoutMap.end()) { - inputLayout = i->second.inputLayout; - i->second.lastUsedTime = mCounter++; + inputLayout = keyIter->second.inputLayout; + keyIter->second.lastUsedTime = mCounter++; } else { - ShaderExecutable11 *shader = ShaderExecutable11::makeShaderExecutable11(programBinary->getVertexExecutable()); + gl::VertexFormat shaderInputLayout[gl::MAX_VERTEX_ATTRIBS]; + GetInputLayout(attributes, shaderInputLayout); + ShaderExecutable11 *shader = ShaderExecutable11::makeShaderExecutable11(programBinary->getVertexExecutableForInputLayout(shaderInputLayout)); D3D11_INPUT_ELEMENT_DESC descs[gl::MAX_VERTEX_ATTRIBS]; for (unsigned int j = 0; j < ilKey.elementCount; ++j) @@ -142,8 +148,7 @@ GLenum InputLayoutCache::applyVertexBuffers(TranslatedAttribute attributes[gl::M HRESULT result = mDevice->CreateInputLayout(descs, ilKey.elementCount, shader->getFunction(), shader->getLength(), &inputLayout); if (FAILED(result)) { - ERR("Failed to crate input layout, result: 0x%08x", result); - return GL_INVALID_OPERATION; + return gl::Error(GL_OUT_OF_MEMORY, "Failed to create internal input layout, HRESULT: 0x%08x", result); } if (mInputLayoutMap.size() >= kMaxInputLayouts) @@ -159,7 +164,7 @@ GLenum InputLayoutCache::applyVertexBuffers(TranslatedAttribute attributes[gl::M leastRecentlyUsed = i; } } - leastRecentlyUsed->second.inputLayout->Release(); + SafeRelease(leastRecentlyUsed->second.inputLayout); mInputLayoutMap.erase(leastRecentlyUsed); } @@ -176,19 +181,46 @@ GLenum InputLayoutCache::applyVertexBuffers(TranslatedAttribute attributes[gl::M mCurrentIL = inputLayout; } + bool dirtyBuffers = false; + size_t minDiff = gl::MAX_VERTEX_ATTRIBS; + size_t maxDiff = 0; for (unsigned int i = 0; i < gl::MAX_VERTEX_ATTRIBS; i++) { - if (vertexBufferSerials[i] != mCurrentBuffers[i] || vertexStrides[i] != mCurrentVertexStrides[i] || - vertexOffsets[i] != mCurrentVertexOffsets[i]) + ID3D11Buffer *buffer = NULL; + + if (attributes[i].active) + { + VertexBuffer11 *vertexBuffer = VertexBuffer11::makeVertexBuffer11(attributes[i].vertexBuffer); + Buffer11 *bufferStorage = attributes[i].storage ? Buffer11::makeBuffer11(attributes[i].storage) : NULL; + + buffer = bufferStorage ? bufferStorage->getBuffer(BUFFER_USAGE_VERTEX_OR_TRANSFORM_FEEDBACK) + : vertexBuffer->getBuffer(); + } + + UINT vertexStride = attributes[i].stride; + UINT vertexOffset = attributes[i].offset; + + if (buffer != mCurrentBuffers[i] || vertexStride != mCurrentVertexStrides[i] || + vertexOffset != mCurrentVertexOffsets[i]) { - mDeviceContext->IASetVertexBuffers(i, 1, &vertexBuffers[i], &vertexStrides[i], &vertexOffsets[i]); - mCurrentBuffers[i] = vertexBufferSerials[i]; - mCurrentVertexStrides[i] = vertexStrides[i]; - mCurrentVertexOffsets[i] = vertexOffsets[i]; + dirtyBuffers = true; + minDiff = std::min(minDiff, static_cast<size_t>(i)); + maxDiff = std::max(maxDiff, static_cast<size_t>(i)); + + mCurrentBuffers[i] = buffer; + mCurrentVertexStrides[i] = vertexStride; + mCurrentVertexOffsets[i] = vertexOffset; } } - return GL_NO_ERROR; + if (dirtyBuffers) + { + ASSERT(minDiff <= maxDiff && maxDiff < gl::MAX_VERTEX_ATTRIBS); + mDeviceContext->IASetVertexBuffers(minDiff, maxDiff - minDiff + 1, mCurrentBuffers + minDiff, + mCurrentVertexStrides + minDiff, mCurrentVertexOffsets + minDiff); + } + + return gl::Error(GL_NO_ERROR); } std::size_t InputLayoutCache::hashInputLayout(const InputLayoutKey &inputLayout) diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/InputLayoutCache.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/InputLayoutCache.h index bb1a8eebcf..cc71ac3f6f 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/InputLayoutCache.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/InputLayoutCache.h @@ -11,8 +11,14 @@ #define LIBGLESV2_RENDERER_INPUTLAYOUTCACHE_H_ #include "libGLESv2/Constants.h" +#include "libGLESv2/Error.h" #include "common/angleutils.h" +#include <GLES2/gl2.h> + +#include <cstddef> +#include <unordered_map> + namespace gl { class ProgramBinary; @@ -32,8 +38,8 @@ class InputLayoutCache void clear(); void markDirty(); - GLenum applyVertexBuffers(TranslatedAttribute attributes[gl::MAX_VERTEX_ATTRIBS], - gl::ProgramBinary *programBinary); + gl::Error applyVertexBuffers(TranslatedAttribute attributes[gl::MAX_VERTEX_ATTRIBS], + gl::ProgramBinary *programBinary); private: DISALLOW_COPY_AND_ASSIGN(InputLayoutCache); @@ -67,7 +73,7 @@ class InputLayoutCache }; ID3D11InputLayout *mCurrentIL; - unsigned int mCurrentBuffers[gl::MAX_VERTEX_ATTRIBS]; + ID3D11Buffer *mCurrentBuffers[gl::MAX_VERTEX_ATTRIBS]; UINT mCurrentVertexStrides[gl::MAX_VERTEX_ATTRIBS]; UINT mCurrentVertexOffsets[gl::MAX_VERTEX_ATTRIBS]; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/PixelTransfer11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/PixelTransfer11.cpp new file mode 100644 index 0000000000..a4e84f91c2 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/PixelTransfer11.cpp @@ -0,0 +1,252 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// PixelTransfer11.cpp: +// Implementation for buffer-to-texture and texture-to-buffer copies. +// Used to implement pixel transfers from unpack and to pack buffers. +// + +#include "libGLESv2/renderer/d3d/d3d11/PixelTransfer11.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" +#include "libGLESv2/renderer/d3d/d3d11/renderer11_utils.h" +#include "libGLESv2/renderer/d3d/d3d11/formatutils11.h" +#include "libGLESv2/renderer/d3d/d3d11/Buffer11.h" +#include "libGLESv2/renderer/d3d/d3d11/TextureStorage11.h" +#include "libGLESv2/renderer/d3d/d3d11/RenderTarget11.h" +#include "libGLESv2/formatutils.h" +#include "libGLESv2/Texture.h" +#include "libGLESv2/Buffer.h" +#include "libGLESv2/Context.h" + +// Precompiled shaders +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/buffertotexturevs.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/buffertotexturegs.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/buffertotexture_4fps.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/buffertotexture_4ips.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/buffertotexture_4uips.h" + +namespace rx +{ + +PixelTransfer11::PixelTransfer11(Renderer11 *renderer) + : mRenderer(renderer), + mBufferToTextureVS(NULL), + mBufferToTextureGS(NULL), + mParamsConstantBuffer(NULL), + mCopyRasterizerState(NULL), + mCopyDepthStencilState(NULL) +{ + HRESULT result = S_OK; + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_RASTERIZER_DESC rasterDesc; + rasterDesc.FillMode = D3D11_FILL_SOLID; + rasterDesc.CullMode = D3D11_CULL_NONE; + rasterDesc.FrontCounterClockwise = FALSE; + rasterDesc.DepthBias = 0; + rasterDesc.SlopeScaledDepthBias = 0.0f; + rasterDesc.DepthBiasClamp = 0.0f; + rasterDesc.DepthClipEnable = TRUE; + rasterDesc.ScissorEnable = FALSE; + rasterDesc.MultisampleEnable = FALSE; + rasterDesc.AntialiasedLineEnable = FALSE; + + result = device->CreateRasterizerState(&rasterDesc, &mCopyRasterizerState); + ASSERT(SUCCEEDED(result)); + + D3D11_DEPTH_STENCIL_DESC depthStencilDesc; + depthStencilDesc.DepthEnable = true; + depthStencilDesc.DepthWriteMask = D3D11_DEPTH_WRITE_MASK_ALL; + depthStencilDesc.DepthFunc = D3D11_COMPARISON_ALWAYS; + depthStencilDesc.StencilEnable = FALSE; + depthStencilDesc.StencilReadMask = D3D11_DEFAULT_STENCIL_READ_MASK; + depthStencilDesc.StencilWriteMask = D3D11_DEFAULT_STENCIL_WRITE_MASK; + depthStencilDesc.FrontFace.StencilFailOp = D3D11_STENCIL_OP_KEEP; + depthStencilDesc.FrontFace.StencilDepthFailOp = D3D11_STENCIL_OP_KEEP; + depthStencilDesc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_KEEP; + depthStencilDesc.FrontFace.StencilFunc = D3D11_COMPARISON_ALWAYS; + depthStencilDesc.BackFace.StencilFailOp = D3D11_STENCIL_OP_KEEP; + depthStencilDesc.BackFace.StencilDepthFailOp = D3D11_STENCIL_OP_KEEP; + depthStencilDesc.BackFace.StencilPassOp = D3D11_STENCIL_OP_KEEP; + depthStencilDesc.BackFace.StencilFunc = D3D11_COMPARISON_ALWAYS; + + result = device->CreateDepthStencilState(&depthStencilDesc, &mCopyDepthStencilState); + ASSERT(SUCCEEDED(result)); + + D3D11_BUFFER_DESC constantBufferDesc = { 0 }; + constantBufferDesc.ByteWidth = rx::roundUp<UINT>(sizeof(CopyShaderParams), 32u); + constantBufferDesc.Usage = D3D11_USAGE_DYNAMIC; + constantBufferDesc.BindFlags = D3D11_BIND_CONSTANT_BUFFER; + constantBufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; + constantBufferDesc.MiscFlags = 0; + constantBufferDesc.StructureByteStride = 0; + + result = device->CreateBuffer(&constantBufferDesc, NULL, &mParamsConstantBuffer); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mParamsConstantBuffer, "PixelTransfer11 constant buffer"); + + StructZero(&mParamsData); + + // init shaders + if (mRenderer->isLevel9()) + return; + + mBufferToTextureVS = d3d11::CompileVS(device, g_VS_BufferToTexture, "BufferToTexture VS"); + mBufferToTextureGS = d3d11::CompileGS(device, g_GS_BufferToTexture, "BufferToTexture GS"); + + buildShaderMap(); +} + +PixelTransfer11::~PixelTransfer11() +{ + for (auto shaderMapIt = mBufferToTexturePSMap.begin(); shaderMapIt != mBufferToTexturePSMap.end(); shaderMapIt++) + { + SafeRelease(shaderMapIt->second); + } + + mBufferToTexturePSMap.clear(); + + SafeRelease(mBufferToTextureVS); + SafeRelease(mBufferToTextureGS); + SafeRelease(mParamsConstantBuffer); + SafeRelease(mCopyRasterizerState); + SafeRelease(mCopyDepthStencilState); +} + +void PixelTransfer11::setBufferToTextureCopyParams(const gl::Box &destArea, const gl::Extents &destSize, GLenum internalFormat, + const gl::PixelUnpackState &unpack, unsigned int offset, CopyShaderParams *parametersOut) +{ + StructZero(parametersOut); + + float texelCenterX = 0.5f / static_cast<float>(destSize.width - 1); + float texelCenterY = 0.5f / static_cast<float>(destSize.height - 1); + + unsigned int bytesPerPixel = gl::GetInternalFormatInfo(internalFormat).pixelBytes; + unsigned int alignmentBytes = static_cast<unsigned int>(unpack.alignment); + unsigned int alignmentPixels = (alignmentBytes <= bytesPerPixel ? 1 : alignmentBytes / bytesPerPixel); + + parametersOut->FirstPixelOffset = offset; + parametersOut->PixelsPerRow = static_cast<unsigned int>(destArea.width); + parametersOut->RowStride = roundUp(parametersOut->PixelsPerRow, alignmentPixels); + parametersOut->RowsPerSlice = static_cast<unsigned int>(destArea.height); + parametersOut->PositionOffset[0] = texelCenterX + (destArea.x / float(destSize.width)) * 2.0f - 1.0f; + parametersOut->PositionOffset[1] = texelCenterY + ((destSize.height - destArea.y - 1) / float(destSize.height)) * 2.0f - 1.0f; + parametersOut->PositionScale[0] = 2.0f / static_cast<float>(destSize.width); + parametersOut->PositionScale[1] = -2.0f / static_cast<float>(destSize.height); +} + +bool PixelTransfer11::copyBufferToTexture(const gl::PixelUnpackState &unpack, unsigned int offset, RenderTarget *destRenderTarget, + GLenum destinationFormat, GLenum sourcePixelsType, const gl::Box &destArea) +{ + gl::Extents destSize = destRenderTarget->getExtents(); + + if (destArea.x < 0 || destArea.x + destArea.width > destSize.width || + destArea.y < 0 || destArea.y + destArea.height > destSize.height || + destArea.z < 0 || destArea.z + destArea.depth > destSize.depth ) + { + return false; + } + + const gl::Buffer &sourceBuffer = *unpack.pixelBuffer.get(); + + ASSERT(mRenderer->supportsFastCopyBufferToTexture(destinationFormat)); + + ID3D11PixelShader *pixelShader = findBufferToTexturePS(destinationFormat); + ASSERT(pixelShader); + + // The SRV must be in the proper read format, which may be different from the destination format + // EG: for half float data, we can load full precision floats with implicit conversion + GLenum unsizedFormat = gl::GetInternalFormatInfo(destinationFormat).format; + GLenum sourceFormat = gl::GetFormatTypeInfo(unsizedFormat, sourcePixelsType).internalFormat; + + const d3d11::TextureFormat &sourceFormatInfo = d3d11::GetTextureFormatInfo(sourceFormat); + DXGI_FORMAT srvFormat = sourceFormatInfo.srvFormat; + ASSERT(srvFormat != DXGI_FORMAT_UNKNOWN); + Buffer11 *bufferStorage11 = Buffer11::makeBuffer11(sourceBuffer.getImplementation()); + ID3D11ShaderResourceView *bufferSRV = bufferStorage11->getSRV(srvFormat); + ASSERT(bufferSRV != NULL); + + ID3D11RenderTargetView *textureRTV = RenderTarget11::makeRenderTarget11(destRenderTarget)->getRenderTargetView(); + ASSERT(textureRTV != NULL); + + CopyShaderParams shaderParams; + setBufferToTextureCopyParams(destArea, destSize, sourceFormat, unpack, offset, &shaderParams); + + ID3D11DeviceContext *deviceContext = mRenderer->getDeviceContext(); + + ID3D11ShaderResourceView *nullSRV = NULL; + ID3D11Buffer *nullBuffer = NULL; + UINT zero = 0; + + // Are we doing a 2D or 3D copy? + ID3D11GeometryShader *geometryShader = ((destSize.depth > 1) ? mBufferToTextureGS : NULL); + + deviceContext->VSSetShader(mBufferToTextureVS, NULL, 0); + deviceContext->GSSetShader(geometryShader, NULL, 0); + deviceContext->PSSetShader(pixelShader, NULL, 0); + deviceContext->PSSetShaderResources(0, 1, &bufferSRV); + deviceContext->IASetInputLayout(NULL); + deviceContext->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_POINTLIST); + + deviceContext->IASetVertexBuffers(0, 1, &nullBuffer, &zero, &zero); + deviceContext->OMSetBlendState(NULL, NULL, 0xFFFFFFF); + deviceContext->OMSetDepthStencilState(mCopyDepthStencilState, 0xFFFFFFFF); + deviceContext->RSSetState(mCopyRasterizerState); + + mRenderer->setOneTimeRenderTarget(textureRTV); + + if (!StructEquals(mParamsData, shaderParams)) + { + d3d11::SetBufferData(deviceContext, mParamsConstantBuffer, shaderParams); + mParamsData = shaderParams; + } + + deviceContext->VSSetConstantBuffers(0, 1, &mParamsConstantBuffer); + + // Set the viewport + D3D11_VIEWPORT viewport; + viewport.TopLeftX = 0; + viewport.TopLeftY = 0; + viewport.Width = destSize.width; + viewport.Height = destSize.height; + viewport.MinDepth = 0.0f; + viewport.MaxDepth = 1.0f; + deviceContext->RSSetViewports(1, &viewport); + + UINT numPixels = (destArea.width * destArea.height * destArea.depth); + deviceContext->Draw(numPixels, 0); + + // Unbind textures and render targets and vertex buffer + deviceContext->PSSetShaderResources(0, 1, &nullSRV); + deviceContext->VSSetConstantBuffers(0, 1, &nullBuffer); + + mRenderer->markAllStateDirty(); + + return true; +} + +void PixelTransfer11::buildShaderMap() +{ + ID3D11Device *device = mRenderer->getDevice(); + + mBufferToTexturePSMap[GL_FLOAT] = d3d11::CompilePS(device, g_PS_BufferToTexture_4F, "BufferToTexture RGBA ps"); + mBufferToTexturePSMap[GL_INT] = d3d11::CompilePS(device, g_PS_BufferToTexture_4I, "BufferToTexture RGBA-I ps"); + mBufferToTexturePSMap[GL_UNSIGNED_INT] = d3d11::CompilePS(device, g_PS_BufferToTexture_4UI, "BufferToTexture RGBA-UI ps"); +} + +ID3D11PixelShader *PixelTransfer11::findBufferToTexturePS(GLenum internalFormat) const +{ + GLenum componentType = gl::GetInternalFormatInfo(internalFormat).componentType; + if (componentType == GL_SIGNED_NORMALIZED || componentType == GL_UNSIGNED_NORMALIZED) + { + componentType = GL_FLOAT; + } + + auto shaderMapIt = mBufferToTexturePSMap.find(componentType); + return (shaderMapIt == mBufferToTexturePSMap.end() ? NULL : shaderMapIt->second); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/PixelTransfer11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/PixelTransfer11.h new file mode 100644 index 0000000000..ed1a3ae1d0 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/PixelTransfer11.h @@ -0,0 +1,86 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// PixelTransfer11.h: +// Buffer-to-Texture and Texture-to-Buffer data transfers. +// Used to implement pixel unpack and pixel pack buffers in ES3. + +#ifndef LIBGLESV2_PIXELTRANSFER11_H_ +#define LIBGLESV2_PIXELTRANSFER11_H_ + +#include "common/platform.h" + +#include <GLES2/gl2.h> + +#include <map> + +namespace gl +{ + +class Buffer; +struct Box; +struct Extents; +struct PixelUnpackState; + +} + +namespace rx +{ +class Renderer11; +class RenderTarget; + +class PixelTransfer11 +{ + public: + explicit PixelTransfer11(Renderer11 *renderer); + ~PixelTransfer11(); + + static bool supportsBufferToTextureCopy(GLenum internalFormat); + + // unpack: the source buffer is stored in the unpack state, and buffer strides + // offset: the start of the data within the unpack buffer + // destRenderTarget: individual slice/layer of a target texture + // destinationFormat/sourcePixelsType: determines shaders + shader parameters + // destArea: the sub-section of destRenderTarget to copy to + bool copyBufferToTexture(const gl::PixelUnpackState &unpack, unsigned int offset, RenderTarget *destRenderTarget, + GLenum destinationFormat, GLenum sourcePixelsType, const gl::Box &destArea); + + private: + + struct CopyShaderParams + { + unsigned int FirstPixelOffset; + unsigned int PixelsPerRow; + unsigned int RowStride; + unsigned int RowsPerSlice; + float PositionOffset[2]; + float PositionScale[2]; + int TexLocationOffset[2]; + int TexLocationScale[2]; + }; + + static void setBufferToTextureCopyParams(const gl::Box &destArea, const gl::Extents &destSize, GLenum internalFormat, + const gl::PixelUnpackState &unpack, unsigned int offset, CopyShaderParams *parametersOut); + + void buildShaderMap(); + ID3D11PixelShader *findBufferToTexturePS(GLenum internalFormat) const; + + Renderer11 *mRenderer; + + std::map<GLenum, ID3D11PixelShader *> mBufferToTexturePSMap; + ID3D11VertexShader *mBufferToTextureVS; + ID3D11GeometryShader *mBufferToTextureGS; + ID3D11Buffer *mParamsConstantBuffer; + CopyShaderParams mParamsData; + + ID3D11RasterizerState *mCopyRasterizerState; + ID3D11DepthStencilState *mCopyDepthStencilState; + +}; + +} + +#endif // LIBGLESV2_PIXELTRANSFER11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Query11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Query11.cpp new file mode 100644 index 0000000000..7109be3e28 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Query11.cpp @@ -0,0 +1,156 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Query11.cpp: Defines the rx::Query11 class which implements rx::QueryImpl. + +#include "libGLESv2/renderer/d3d/d3d11/Query11.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" +#include "libGLESv2/renderer/d3d/d3d11/renderer11_utils.h" +#include "libGLESv2/main.h" + +#include <GLES2/gl2ext.h> + +namespace rx +{ + +Query11::Query11(rx::Renderer11 *renderer, GLenum type) + : QueryImpl(type), + mResult(0), + mQueryFinished(false), + mRenderer(renderer), + mQuery(NULL) +{ +} + +Query11::~Query11() +{ + SafeRelease(mQuery); +} + +gl::Error Query11::begin() +{ + if (mQuery == NULL) + { + D3D11_QUERY_DESC queryDesc; + queryDesc.Query = gl_d3d11::ConvertQueryType(getType()); + queryDesc.MiscFlags = 0; + + HRESULT result = mRenderer->getDevice()->CreateQuery(&queryDesc, &mQuery); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal query creation failed, result: 0x%X.", result); + } + } + + mRenderer->getDeviceContext()->Begin(mQuery); + return gl::Error(GL_NO_ERROR); +} + +gl::Error Query11::end() +{ + ASSERT(mQuery); + mRenderer->getDeviceContext()->End(mQuery); + + mQueryFinished = false; + mResult = GL_FALSE; + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Query11::getResult(GLuint *params) +{ + while (!mQueryFinished) + { + gl::Error error = testQuery(); + if (error.isError()) + { + return error; + } + + if (!mQueryFinished) + { + Sleep(0); + } + } + + ASSERT(mQueryFinished); + *params = mResult; + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Query11::isResultAvailable(GLuint *available) +{ + gl::Error error = testQuery(); + if (error.isError()) + { + return error; + } + + *available = (mQueryFinished ? GL_TRUE : GL_FALSE); + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Query11::testQuery() +{ + if (!mQueryFinished) + { + ASSERT(mQuery); + + ID3D11DeviceContext *context = mRenderer->getDeviceContext(); + switch (getType()) + { + case GL_ANY_SAMPLES_PASSED_EXT: + case GL_ANY_SAMPLES_PASSED_CONSERVATIVE_EXT: + { + UINT64 numPixels = 0; + HRESULT result = context->GetData(mQuery, &numPixels, sizeof(numPixels), 0); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to get the data of an internal query, result: 0x%X.", result); + } + + if (result == S_OK) + { + mQueryFinished = true; + mResult = (numPixels > 0) ? GL_TRUE : GL_FALSE; + } + } + break; + + case GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN: + { + D3D11_QUERY_DATA_SO_STATISTICS soStats = { 0 }; + HRESULT result = context->GetData(mQuery, &soStats, sizeof(soStats), 0); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to get the data of an internal query, result: 0x%X.", result); + } + + if (result == S_OK) + { + mQueryFinished = true; + mResult = static_cast<GLuint>(soStats.NumPrimitivesWritten); + } + } + break; + + default: + UNREACHABLE(); + break; + } + + if (!mQueryFinished && mRenderer->testDeviceLost(true)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to test get query result, device is lost."); + } + } + + return gl::Error(GL_NO_ERROR); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Query11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Query11.h index 0a03de77ca..822f2542ee 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Query11.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Query11.h @@ -21,15 +21,19 @@ class Query11 : public QueryImpl Query11(rx::Renderer11 *renderer, GLenum type); virtual ~Query11(); - void begin(); - void end(); - GLuint getResult(); - GLboolean isResultAvailable(); + virtual gl::Error begin(); + virtual gl::Error end(); + virtual gl::Error getResult(GLuint *params); + virtual gl::Error isResultAvailable(GLuint *available); private: DISALLOW_COPY_AND_ASSIGN(Query11); - GLboolean testQuery(); + gl::Error testQuery(); + + GLuint mResult; + + bool mQueryFinished; rx::Renderer11 *mRenderer; ID3D11Query *mQuery; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/RenderStateCache.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/RenderStateCache.cpp index a1c324cd80..71b931f27e 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/RenderStateCache.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/RenderStateCache.cpp @@ -1,6 +1,5 @@ -#include "precompiled.h" // -// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // @@ -8,18 +7,29 @@ // RenderStateCache.cpp: Defines rx::RenderStateCache, a cache of Direct3D render // state objects. -#include "libGLESv2/renderer/d3d11/RenderStateCache.h" -#include "libGLESv2/renderer/d3d11/renderer11_utils.h" - +#include "libGLESv2/renderer/d3d/d3d11/RenderStateCache.h" +#include "libGLESv2/renderer/d3d/d3d11/renderer11_utils.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" #include "libGLESv2/Framebuffer.h" -#include "libGLESv2/Renderbuffer.h" -#include "libGLESv2/utilities.h" +#include "libGLESv2/FramebufferAttachment.h" + #include "common/debug.h" + #include "third_party/murmurhash/MurmurHash3.h" namespace rx { +template <typename mapType> +static void ClearStateMap(mapType &map) +{ + for (typename mapType::iterator i = map.begin(); i != map.end(); i++) + { + SafeRelease(i->second.first); + } + map.clear(); +} + // MSDN's documentation of ID3D11Device::CreateBlendState, ID3D11Device::CreateRasterizerState, // ID3D11Device::CreateDepthStencilState and ID3D11Device::CreateSamplerState claims the maximum // number of unique states of each type an application can create is 4096 @@ -49,29 +59,10 @@ void RenderStateCache::initialize(ID3D11Device *device) void RenderStateCache::clear() { - for (BlendStateMap::iterator i = mBlendStateCache.begin(); i != mBlendStateCache.end(); i++) - { - i->second.first->Release(); - } - mBlendStateCache.clear(); - - for (RasterizerStateMap::iterator i = mRasterizerStateCache.begin(); i != mRasterizerStateCache.end(); i++) - { - i->second.first->Release(); - } - mRasterizerStateCache.clear(); - - for (DepthStencilStateMap::iterator i = mDepthStencilStateCache.begin(); i != mDepthStencilStateCache.end(); i++) - { - i->second.first->Release(); - } - mDepthStencilStateCache.clear(); - - for (SamplerStateMap::iterator i = mSamplerStateCache.begin(); i != mSamplerStateCache.end(); i++) - { - i->second.first->Release(); - } - mSamplerStateCache.clear(); + ClearStateMap(mBlendStateCache); + ClearStateMap(mRasterizerStateCache); + ClearStateMap(mDepthStencilStateCache); + ClearStateMap(mSamplerStateCache); } std::size_t RenderStateCache::hashBlendState(const BlendStateKey &blendState) @@ -79,58 +70,56 @@ std::size_t RenderStateCache::hashBlendState(const BlendStateKey &blendState) static const unsigned int seed = 0xABCDEF98; std::size_t hash = 0; - MurmurHash3_x86_32(&blendState, sizeof(BlendStateKey), seed, &hash); + MurmurHash3_x86_32(&blendState, sizeof(gl::BlendState), seed, &hash); return hash; } bool RenderStateCache::compareBlendStates(const BlendStateKey &a, const BlendStateKey &b) { - return memcmp(&a, &b, sizeof(gl::BlendState)) == 0; + return memcmp(&a, &b, sizeof(BlendStateKey)) == 0; } -ID3D11BlendState *RenderStateCache::getBlendState(gl::Framebuffer *framebuffer, const gl::BlendState &blendState) +gl::Error RenderStateCache::getBlendState(const gl::Framebuffer *framebuffer, const gl::BlendState &blendState, + ID3D11BlendState **outBlendState) { if (!mDevice) { - ERR("RenderStateCache is not initialized."); - return NULL; + return gl::Error(GL_OUT_OF_MEMORY, "Internal error, RenderStateCache is not initialized."); } bool mrt = false; + const gl::ColorbufferInfo &colorbuffers = framebuffer->getColorbuffersForRender(); + BlendStateKey key = { 0 }; key.blendState = blendState; - for (unsigned int i = 0; i < D3D11_SIMULTANEOUS_RENDER_TARGET_COUNT; i++) + for (size_t colorAttachment = 0; colorAttachment < colorbuffers.size(); ++colorAttachment) { - gl::Renderbuffer *renderBuffer = framebuffer->getColorbuffer(i); - if (renderBuffer) + const gl::FramebufferAttachment *attachment = colorbuffers[colorAttachment]; + + auto rtChannels = key.rtChannels[colorAttachment]; + + if (attachment) { - if (i > 0) + if (colorAttachment > 0) { mrt = true; } - GLenum internalFormat = renderBuffer->getInternalFormat(); - key.rtChannels[i][0] = gl::GetRedSize(internalFormat) > 0; - key.rtChannels[i][1] = gl::GetGreenSize(internalFormat) > 0; - key.rtChannels[i][2] = gl::GetBlueSize(internalFormat) > 0;; - key.rtChannels[i][3] = gl::GetAlphaSize(internalFormat) > 0; - } - else - { - key.rtChannels[i][0] = false; - key.rtChannels[i][1] = false; - key.rtChannels[i][2] = false; - key.rtChannels[i][3] = false; + rtChannels[0] = attachment->getRedSize() > 0; + rtChannels[1] = attachment->getGreenSize() > 0; + rtChannels[2] = attachment->getBlueSize() > 0; + rtChannels[3] = attachment->getAlphaSize() > 0; } } - BlendStateMap::iterator i = mBlendStateCache.find(key); - if (i != mBlendStateCache.end()) + BlendStateMap::iterator keyIter = mBlendStateCache.find(key); + if (keyIter != mBlendStateCache.end()) { - BlendStateCounterPair &state = i->second; + BlendStateCounterPair &state = keyIter->second; state.second = mCounter++; - return state.first; + *outBlendState = state.first; + return gl::Error(GL_NO_ERROR); } else { @@ -147,7 +136,7 @@ ID3D11BlendState *RenderStateCache::getBlendState(gl::Framebuffer *framebuffer, leastRecentlyUsed = i; } } - leastRecentlyUsed->second.first->Release(); + SafeRelease(leastRecentlyUsed->second.first); mBlendStateCache.erase(leastRecentlyUsed); } @@ -182,13 +171,13 @@ ID3D11BlendState *RenderStateCache::getBlendState(gl::Framebuffer *framebuffer, HRESULT result = mDevice->CreateBlendState(&blendDesc, &dx11BlendState); if (FAILED(result) || !dx11BlendState) { - ERR("Unable to create a ID3D11BlendState, HRESULT: 0x%X.", result); - return NULL; + return gl::Error(GL_OUT_OF_MEMORY, "Unable to create a ID3D11BlendState, HRESULT: 0x%X.", result); } mBlendStateCache.insert(std::make_pair(key, std::make_pair(dx11BlendState, mCounter++))); - return dx11BlendState; + *outBlendState = dx11BlendState; + return gl::Error(GL_NO_ERROR); } } @@ -206,26 +195,25 @@ bool RenderStateCache::compareRasterizerStates(const RasterizerStateKey &a, cons return memcmp(&a, &b, sizeof(RasterizerStateKey)) == 0; } -ID3D11RasterizerState *RenderStateCache::getRasterizerState(const gl::RasterizerState &rasterState, - bool scissorEnabled, unsigned int depthSize) +gl::Error RenderStateCache::getRasterizerState(const gl::RasterizerState &rasterState, bool scissorEnabled, + ID3D11RasterizerState **outRasterizerState) { if (!mDevice) { - ERR("RenderStateCache is not initialized."); - return NULL; + return gl::Error(GL_OUT_OF_MEMORY, "Internal error, RenderStateCache is not initialized."); } - RasterizerStateKey key; + RasterizerStateKey key = { 0 }; key.rasterizerState = rasterState; key.scissorEnabled = scissorEnabled; - key.depthSize = depthSize; - RasterizerStateMap::iterator i = mRasterizerStateCache.find(key); - if (i != mRasterizerStateCache.end()) + RasterizerStateMap::iterator keyIter = mRasterizerStateCache.find(key); + if (keyIter != mRasterizerStateCache.end()) { - RasterizerStateCounterPair &state = i->second; + RasterizerStateCounterPair &state = keyIter->second; state.second = mCounter++; - return state.first; + *outRasterizerState = state.first; + return gl::Error(GL_NO_ERROR); } else { @@ -242,7 +230,7 @@ ID3D11RasterizerState *RenderStateCache::getRasterizerState(const gl::Rasterizer leastRecentlyUsed = i; } } - leastRecentlyUsed->second.first->Release(); + SafeRelease(leastRecentlyUsed->second.first); mRasterizerStateCache.erase(leastRecentlyUsed); } @@ -258,25 +246,34 @@ ID3D11RasterizerState *RenderStateCache::getRasterizerState(const gl::Rasterizer rasterDesc.FillMode = D3D11_FILL_SOLID; rasterDesc.CullMode = cullMode; rasterDesc.FrontCounterClockwise = (rasterState.frontFace == GL_CCW) ? FALSE: TRUE; - rasterDesc.DepthBias = ldexp(rasterState.polygonOffsetUnits, -static_cast<int>(depthSize)); rasterDesc.DepthBiasClamp = 0.0f; // MSDN documentation of DepthBiasClamp implies a value of zero will preform no clamping, must be tested though. - rasterDesc.SlopeScaledDepthBias = rasterState.polygonOffsetFactor; rasterDesc.DepthClipEnable = TRUE; rasterDesc.ScissorEnable = scissorEnabled ? TRUE : FALSE; rasterDesc.MultisampleEnable = rasterState.multiSample; rasterDesc.AntialiasedLineEnable = FALSE; + if (rasterState.polygonOffsetFill) + { + rasterDesc.SlopeScaledDepthBias = rasterState.polygonOffsetFactor; + rasterDesc.DepthBias = (INT)rasterState.polygonOffsetUnits; + } + else + { + rasterDesc.SlopeScaledDepthBias = 0.0f; + rasterDesc.DepthBias = 0; + } + ID3D11RasterizerState *dx11RasterizerState = NULL; HRESULT result = mDevice->CreateRasterizerState(&rasterDesc, &dx11RasterizerState); if (FAILED(result) || !dx11RasterizerState) { - ERR("Unable to create a ID3D11RasterizerState, HRESULT: 0x%X.", result); - return NULL; + return gl::Error(GL_OUT_OF_MEMORY, "Unable to create a ID3D11RasterizerState, HRESULT: 0x%X.", result); } mRasterizerStateCache.insert(std::make_pair(key, std::make_pair(dx11RasterizerState, mCounter++))); - return dx11RasterizerState; + *outRasterizerState = dx11RasterizerState; + return gl::Error(GL_NO_ERROR); } } @@ -294,20 +291,20 @@ bool RenderStateCache::compareDepthStencilStates(const gl::DepthStencilState &a, return memcmp(&a, &b, sizeof(gl::DepthStencilState)) == 0; } -ID3D11DepthStencilState *RenderStateCache::getDepthStencilState(const gl::DepthStencilState &dsState) +gl::Error RenderStateCache::getDepthStencilState(const gl::DepthStencilState &dsState, ID3D11DepthStencilState **outDSState) { if (!mDevice) { - ERR("RenderStateCache is not initialized."); - return NULL; + return gl::Error(GL_OUT_OF_MEMORY, "Internal error, RenderStateCache is not initialized."); } - DepthStencilStateMap::iterator i = mDepthStencilStateCache.find(dsState); - if (i != mDepthStencilStateCache.end()) + DepthStencilStateMap::iterator keyIter = mDepthStencilStateCache.find(dsState); + if (keyIter != mDepthStencilStateCache.end()) { - DepthStencilStateCounterPair &state = i->second; + DepthStencilStateCounterPair &state = keyIter->second; state.second = mCounter++; - return state.first; + *outDSState = state.first; + return gl::Error(GL_NO_ERROR); } else { @@ -324,7 +321,7 @@ ID3D11DepthStencilState *RenderStateCache::getDepthStencilState(const gl::DepthS leastRecentlyUsed = i; } } - leastRecentlyUsed->second.first->Release(); + SafeRelease(leastRecentlyUsed->second.first); mDepthStencilStateCache.erase(leastRecentlyUsed); } @@ -348,13 +345,13 @@ ID3D11DepthStencilState *RenderStateCache::getDepthStencilState(const gl::DepthS HRESULT result = mDevice->CreateDepthStencilState(&dsDesc, &dx11DepthStencilState); if (FAILED(result) || !dx11DepthStencilState) { - ERR("Unable to create a ID3D11DepthStencilState, HRESULT: 0x%X.", result); - return NULL; + return gl::Error(GL_OUT_OF_MEMORY, "Unable to create a ID3D11DepthStencilState, HRESULT: 0x%X.", result); } mDepthStencilStateCache.insert(std::make_pair(dsState, std::make_pair(dx11DepthStencilState, mCounter++))); - return dx11DepthStencilState; + *outDSState = dx11DepthStencilState; + return gl::Error(GL_NO_ERROR); } } @@ -372,20 +369,20 @@ bool RenderStateCache::compareSamplerStates(const gl::SamplerState &a, const gl: return memcmp(&a, &b, sizeof(gl::SamplerState)) == 0; } -ID3D11SamplerState *RenderStateCache::getSamplerState(const gl::SamplerState &samplerState) +gl::Error RenderStateCache::getSamplerState(const gl::SamplerState &samplerState, ID3D11SamplerState **outSamplerState) { if (!mDevice) { - ERR("RenderStateCache is not initialized."); - return NULL; + return gl::Error(GL_OUT_OF_MEMORY, "Internal error, RenderStateCache is not initialized."); } - SamplerStateMap::iterator i = mSamplerStateCache.find(samplerState); - if (i != mSamplerStateCache.end()) + SamplerStateMap::iterator keyIter = mSamplerStateCache.find(samplerState); + if (keyIter != mSamplerStateCache.end()) { - SamplerStateCounterPair &state = i->second; + SamplerStateCounterPair &state = keyIter->second; state.second = mCounter++; - return state.first; + *outSamplerState = state.first; + return gl::Error(GL_NO_ERROR); } else { @@ -402,37 +399,37 @@ ID3D11SamplerState *RenderStateCache::getSamplerState(const gl::SamplerState &sa leastRecentlyUsed = i; } } - leastRecentlyUsed->second.first->Release(); + SafeRelease(leastRecentlyUsed->second.first); mSamplerStateCache.erase(leastRecentlyUsed); } D3D11_SAMPLER_DESC samplerDesc; - samplerDesc.Filter = gl_d3d11::ConvertFilter(samplerState.minFilter, samplerState.magFilter, samplerState.maxAnisotropy); + samplerDesc.Filter = gl_d3d11::ConvertFilter(samplerState.minFilter, samplerState.magFilter, + samplerState.maxAnisotropy, samplerState.compareMode); samplerDesc.AddressU = gl_d3d11::ConvertTextureWrap(samplerState.wrapS); samplerDesc.AddressV = gl_d3d11::ConvertTextureWrap(samplerState.wrapT); - samplerDesc.AddressW = D3D11_TEXTURE_ADDRESS_CLAMP; - samplerDesc.MipLODBias = static_cast<float>(samplerState.lodOffset); + samplerDesc.AddressW = gl_d3d11::ConvertTextureWrap(samplerState.wrapR); + samplerDesc.MipLODBias = 0; samplerDesc.MaxAnisotropy = samplerState.maxAnisotropy; - samplerDesc.ComparisonFunc = D3D11_COMPARISON_NEVER; + samplerDesc.ComparisonFunc = gl_d3d11::ConvertComparison(samplerState.compareFunc); samplerDesc.BorderColor[0] = 0.0f; samplerDesc.BorderColor[1] = 0.0f; samplerDesc.BorderColor[2] = 0.0f; samplerDesc.BorderColor[3] = 0.0f; - samplerDesc.MinLOD = gl_d3d11::ConvertMinLOD(samplerState.minFilter, samplerState.lodOffset); - samplerDesc.MaxLOD = mDevice->GetFeatureLevel() >= D3D_FEATURE_LEVEL_10_0 - ? gl_d3d11::ConvertMaxLOD(samplerState.minFilter, samplerState.lodOffset) : FLT_MAX; + samplerDesc.MinLOD = samplerState.minLod; + samplerDesc.MaxLOD = samplerState.maxLod; ID3D11SamplerState *dx11SamplerState = NULL; HRESULT result = mDevice->CreateSamplerState(&samplerDesc, &dx11SamplerState); if (FAILED(result) || !dx11SamplerState) { - ERR("Unable to create a ID3D11DepthStencilState, HRESULT: 0x%X.", result); - return NULL; + return gl::Error(GL_OUT_OF_MEMORY, "Unable to create a ID3D11SamplerState, HRESULT: 0x%X.", result); } mSamplerStateCache.insert(std::make_pair(samplerState, std::make_pair(dx11SamplerState, mCounter++))); - return dx11SamplerState; + *outSamplerState = dx11SamplerState; + return gl::Error(GL_NO_ERROR); } } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/RenderStateCache.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/RenderStateCache.h index b4b871a4bd..d5471a3061 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/RenderStateCache.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/RenderStateCache.h @@ -1,5 +1,5 @@ // -// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // @@ -11,8 +11,11 @@ #define LIBGLESV2_RENDERER_RENDERSTATECACHE_H_ #include "libGLESv2/angletypes.h" +#include "libGLESv2/Error.h" #include "common/angleutils.h" +#include <unordered_map> + namespace gl { class Framebuffer; @@ -20,6 +23,7 @@ class Framebuffer; namespace rx { +class Renderer11; class RenderStateCache { @@ -30,12 +34,10 @@ class RenderStateCache void initialize(ID3D11Device *device); void clear(); - // Increments refcount on the returned blend state, Release() must be called. - ID3D11BlendState *getBlendState(gl::Framebuffer *framebuffer, const gl::BlendState &blendState); - ID3D11RasterizerState *getRasterizerState(const gl::RasterizerState &rasterState, - bool scissorEnabled, unsigned int depthSize); - ID3D11DepthStencilState *getDepthStencilState(const gl::DepthStencilState &dsState); - ID3D11SamplerState *getSamplerState(const gl::SamplerState &samplerState); + gl::Error getBlendState(const gl::Framebuffer *framebuffer, const gl::BlendState &blendState, ID3D11BlendState **outBlendState); + gl::Error getRasterizerState(const gl::RasterizerState &rasterState, bool scissorEnabled, ID3D11RasterizerState **outRasterizerState); + gl::Error getDepthStencilState(const gl::DepthStencilState &dsState, ID3D11DepthStencilState **outDSState); + gl::Error getSamplerState(const gl::SamplerState &samplerState, ID3D11SamplerState **outSamplerState); private: DISALLOW_COPY_AND_ASSIGN(RenderStateCache); @@ -63,7 +65,6 @@ class RenderStateCache { gl::RasterizerState rasterizerState; bool scissorEnabled; - unsigned int depthSize; }; static std::size_t hashRasterizerState(const RasterizerStateKey &rasterState); static bool compareRasterizerStates(const RasterizerStateKey &a, const RasterizerStateKey &b); diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/RenderTarget11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/RenderTarget11.cpp index 3707097aa4..3041f21faa 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/RenderTarget11.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/RenderTarget11.cpp @@ -1,4 +1,3 @@ -#include "precompiled.h" // // Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be @@ -8,26 +7,66 @@ // RenderTarget11.cpp: Implements a DX11-specific wrapper for ID3D11View pointers // retained by Renderbuffers. -#include "libGLESv2/renderer/d3d11/RenderTarget11.h" -#include "libGLESv2/renderer/d3d11/Renderer11.h" - -#include "libGLESv2/renderer/d3d11/renderer11_utils.h" +#include "libGLESv2/renderer/d3d/d3d11/RenderTarget11.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" +#include "libGLESv2/renderer/d3d/d3d11/renderer11_utils.h" +#include "libGLESv2/renderer/d3d/d3d11/formatutils11.h" #include "libGLESv2/main.h" namespace rx { -static unsigned int getRTVSubresourceIndex(ID3D11Texture2D *texture, ID3D11RenderTargetView *view) +static bool getTextureProperties(ID3D11Resource *resource, unsigned int *mipLevels, unsigned int *samples) +{ + ID3D11Texture1D *texture1D = d3d11::DynamicCastComObject<ID3D11Texture1D>(resource); + if (texture1D) + { + D3D11_TEXTURE1D_DESC texDesc; + texture1D->GetDesc(&texDesc); + SafeRelease(texture1D); + + *mipLevels = texDesc.MipLevels; + *samples = 0; + + return true; + } + + ID3D11Texture2D *texture2D = d3d11::DynamicCastComObject<ID3D11Texture2D>(resource); + if (texture2D) + { + D3D11_TEXTURE2D_DESC texDesc; + texture2D->GetDesc(&texDesc); + SafeRelease(texture2D); + + *mipLevels = texDesc.MipLevels; + *samples = texDesc.SampleDesc.Count > 1 ? texDesc.SampleDesc.Count : 0; + + return true; + } + + ID3D11Texture3D *texture3D = d3d11::DynamicCastComObject<ID3D11Texture3D>(resource); + if (texture3D) + { + D3D11_TEXTURE3D_DESC texDesc; + texture3D->GetDesc(&texDesc); + SafeRelease(texture3D); + + *mipLevels = texDesc.MipLevels; + *samples = 0; + + return true; + } + + return false; +} + +static unsigned int getRTVSubresourceIndex(ID3D11Resource *resource, ID3D11RenderTargetView *view) { D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; view->GetDesc(&rtvDesc); - D3D11_TEXTURE2D_DESC texDesc; - texture->GetDesc(&texDesc); - unsigned int mipSlice = 0; unsigned int arraySlice = 0; - unsigned int mipLevels = texDesc.MipLevels; switch (rtvDesc.ViewDimension) { @@ -76,20 +115,19 @@ static unsigned int getRTVSubresourceIndex(ID3D11Texture2D *texture, ID3D11Rende break; } + unsigned int mipLevels, samples; + getTextureProperties(resource, &mipLevels, &samples); + return D3D11CalcSubresource(mipSlice, arraySlice, mipLevels); } -static unsigned int getDSVSubresourceIndex(ID3D11Texture2D *texture, ID3D11DepthStencilView *view) +static unsigned int getDSVSubresourceIndex(ID3D11Resource *resource, ID3D11DepthStencilView *view) { D3D11_DEPTH_STENCIL_VIEW_DESC dsvDesc; view->GetDesc(&dsvDesc); - D3D11_TEXTURE2D_DESC texDesc; - texture->GetDesc(&texDesc); - unsigned int mipSlice = 0; unsigned int arraySlice = 0; - unsigned int mipLevels = texDesc.MipLevels; switch (dsvDesc.ViewDimension) { @@ -123,7 +161,7 @@ static unsigned int getDSVSubresourceIndex(ID3D11Texture2D *texture, ID3D11Depth arraySlice = dsvDesc.Texture2DMSArray.FirstArraySlice; break; - case D3D11_RTV_DIMENSION_UNKNOWN: + case D3D11_DSV_DIMENSION_UNKNOWN: UNIMPLEMENTED(); break; @@ -132,16 +170,37 @@ static unsigned int getDSVSubresourceIndex(ID3D11Texture2D *texture, ID3D11Depth break; } + unsigned int mipLevels, samples; + getTextureProperties(resource, &mipLevels, &samples); + return D3D11CalcSubresource(mipSlice, arraySlice, mipLevels); } -RenderTarget11::RenderTarget11(Renderer *renderer, ID3D11RenderTargetView *rtv, ID3D11Texture2D *tex, ID3D11ShaderResourceView *srv, GLsizei width, GLsizei height) +RenderTarget11::RenderTarget11(Renderer *renderer, ID3D11RenderTargetView *rtv, ID3D11Resource *resource, + ID3D11ShaderResourceView *srv, GLsizei width, GLsizei height, GLsizei depth) { mRenderer = Renderer11::makeRenderer11(renderer); - mTexture = tex; + + mTexture = resource; + if (mTexture) + { + mTexture->AddRef(); + } + mRenderTarget = rtv; + if (mRenderTarget) + { + mRenderTarget->AddRef(); + } + mDepthStencil = NULL; + mShaderResource = srv; + if (mShaderResource) + { + mShaderResource->AddRef(); + } + mSubresourceIndex = 0; if (mRenderTarget && mTexture) @@ -149,26 +208,46 @@ RenderTarget11::RenderTarget11(Renderer *renderer, ID3D11RenderTargetView *rtv, D3D11_RENDER_TARGET_VIEW_DESC desc; mRenderTarget->GetDesc(&desc); - D3D11_TEXTURE2D_DESC texDesc; - mTexture->GetDesc(&texDesc); + unsigned int mipLevels, samples; + getTextureProperties(mTexture, &mipLevels, &samples); mSubresourceIndex = getRTVSubresourceIndex(mTexture, mRenderTarget); mWidth = width; mHeight = height; - mSamples = (texDesc.SampleDesc.Count > 1) ? texDesc.SampleDesc.Count : 0; + mDepth = depth; + mSamples = samples; - mInternalFormat = d3d11_gl::ConvertTextureInternalFormat(desc.Format); - mActualFormat = d3d11_gl::ConvertTextureInternalFormat(desc.Format); + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(desc.Format); + mInternalFormat = dxgiFormatInfo.internalFormat; + mActualFormat = dxgiFormatInfo.internalFormat; } } -RenderTarget11::RenderTarget11(Renderer *renderer, ID3D11DepthStencilView *dsv, ID3D11Texture2D *tex, ID3D11ShaderResourceView *srv, GLsizei width, GLsizei height) +RenderTarget11::RenderTarget11(Renderer *renderer, ID3D11DepthStencilView *dsv, ID3D11Resource *resource, + ID3D11ShaderResourceView *srv, GLsizei width, GLsizei height, GLsizei depth) { mRenderer = Renderer11::makeRenderer11(renderer); - mTexture = tex; + + mTexture = resource; + if (mTexture) + { + mTexture->AddRef(); + } + mRenderTarget = NULL; + mDepthStencil = dsv; + if (mDepthStencil) + { + mDepthStencil->AddRef(); + } + mShaderResource = srv; + if (mShaderResource) + { + mShaderResource->AddRef(); + } + mSubresourceIndex = 0; if (mDepthStencil && mTexture) @@ -176,20 +255,22 @@ RenderTarget11::RenderTarget11(Renderer *renderer, ID3D11DepthStencilView *dsv, D3D11_DEPTH_STENCIL_VIEW_DESC desc; mDepthStencil->GetDesc(&desc); - D3D11_TEXTURE2D_DESC texDesc; - mTexture->GetDesc(&texDesc); + unsigned int mipLevels, samples; + getTextureProperties(mTexture, &mipLevels, &samples); mSubresourceIndex = getDSVSubresourceIndex(mTexture, mDepthStencil); mWidth = width; mHeight = height; - mSamples = (texDesc.SampleDesc.Count > 1) ? texDesc.SampleDesc.Count : 0; + mDepth = depth; + mSamples = samples; - mInternalFormat = d3d11_gl::ConvertTextureInternalFormat(desc.Format); - mActualFormat = d3d11_gl::ConvertTextureInternalFormat(desc.Format); + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(desc.Format); + mInternalFormat = dxgiFormatInfo.internalFormat; + mActualFormat = dxgiFormatInfo.internalFormat; } } -RenderTarget11::RenderTarget11(Renderer *renderer, GLsizei width, GLsizei height, GLenum format, GLsizei samples, bool depth) +RenderTarget11::RenderTarget11(Renderer *renderer, GLsizei width, GLsizei height, GLenum internalFormat, GLsizei samples) { mRenderer = Renderer11::makeRenderer11(renderer); mTexture = NULL; @@ -197,14 +278,11 @@ RenderTarget11::RenderTarget11(Renderer *renderer, GLsizei width, GLsizei height mDepthStencil = NULL; mShaderResource = NULL; - DXGI_FORMAT requestedFormat = gl_d3d11::ConvertRenderbufferFormat(format); + const d3d11::TextureFormat &formatInfo = d3d11::GetTextureFormatInfo(internalFormat); + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(formatInfo.texFormat); - int supportedSamples = mRenderer->getNearestSupportedSamples(requestedFormat, samples); - if (supportedSamples < 0) - { - gl::error(GL_OUT_OF_MEMORY); - return; - } + const gl::TextureCaps &textureCaps = mRenderer->getRendererTextureCaps().get(internalFormat); + GLuint supportedSamples = textureCaps.getNearestSamples(samples); if (width > 0 && height > 0) { @@ -214,16 +292,35 @@ RenderTarget11::RenderTarget11(Renderer *renderer, GLsizei width, GLsizei height desc.Height = height; desc.MipLevels = 1; desc.ArraySize = 1; - desc.Format = requestedFormat; + desc.Format = formatInfo.texFormat; desc.SampleDesc.Count = (supportedSamples == 0) ? 1 : supportedSamples; desc.SampleDesc.Quality = 0; desc.Usage = D3D11_USAGE_DEFAULT; desc.CPUAccessFlags = 0; desc.MiscFlags = 0; - desc.BindFlags = (depth ? D3D11_BIND_DEPTH_STENCIL : (D3D11_BIND_RENDER_TARGET | D3D11_BIND_SHADER_RESOURCE)); + + // If a rendertarget or depthstencil format exists for this texture format, + // we'll flag it to allow binding that way. Shader resource views are a little + // more complicated. + bool bindRTV = false, bindDSV = false, bindSRV = false; + bindRTV = (formatInfo.rtvFormat != DXGI_FORMAT_UNKNOWN); + bindDSV = (formatInfo.dsvFormat != DXGI_FORMAT_UNKNOWN); + if (formatInfo.srvFormat != DXGI_FORMAT_UNKNOWN) + { + // Multisample targets flagged for binding as depth stencil cannot also be + // flagged for binding as SRV, so make certain not to add the SRV flag for + // these targets. + bindSRV = !(formatInfo.dsvFormat != DXGI_FORMAT_UNKNOWN && desc.SampleDesc.Count > 1); + } + + desc.BindFlags = (bindRTV ? D3D11_BIND_RENDER_TARGET : 0) | + (bindDSV ? D3D11_BIND_DEPTH_STENCIL : 0) | + (bindSRV ? D3D11_BIND_SHADER_RESOURCE : 0); ID3D11Device *device = mRenderer->getDevice(); - HRESULT result = device->CreateTexture2D(&desc, NULL, &mTexture); + ID3D11Texture2D *texture = NULL; + HRESULT result = device->CreateTexture2D(&desc, NULL, &texture); + mTexture = texture; if (result == E_OUTOFMEMORY) { @@ -232,10 +329,28 @@ RenderTarget11::RenderTarget11(Renderer *renderer, GLsizei width, GLsizei height } ASSERT(SUCCEEDED(result)); - if (depth) + if (bindSRV) + { + D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; + srvDesc.Format = formatInfo.srvFormat; + srvDesc.ViewDimension = (supportedSamples == 0) ? D3D11_SRV_DIMENSION_TEXTURE2D : D3D11_SRV_DIMENSION_TEXTURE2DMS; + srvDesc.Texture2D.MostDetailedMip = 0; + srvDesc.Texture2D.MipLevels = 1; + result = device->CreateShaderResourceView(mTexture, &srvDesc, &mShaderResource); + + if (result == E_OUTOFMEMORY) + { + SafeRelease(mTexture); + gl::error(GL_OUT_OF_MEMORY); + return; + } + ASSERT(SUCCEEDED(result)); + } + + if (bindDSV) { D3D11_DEPTH_STENCIL_VIEW_DESC dsvDesc; - dsvDesc.Format = requestedFormat; + dsvDesc.Format = formatInfo.dsvFormat; dsvDesc.ViewDimension = (supportedSamples == 0) ? D3D11_DSV_DIMENSION_TEXTURE2D : D3D11_DSV_DIMENSION_TEXTURE2DMS; dsvDesc.Texture2D.MipSlice = 0; dsvDesc.Flags = 0; @@ -243,82 +358,58 @@ RenderTarget11::RenderTarget11(Renderer *renderer, GLsizei width, GLsizei height if (result == E_OUTOFMEMORY) { - mTexture->Release(); - mTexture = NULL; + SafeRelease(mTexture); + SafeRelease(mShaderResource); gl::error(GL_OUT_OF_MEMORY); + return; } ASSERT(SUCCEEDED(result)); } - else + + if (bindRTV) { D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; - rtvDesc.Format = requestedFormat; + rtvDesc.Format = formatInfo.rtvFormat; rtvDesc.ViewDimension = (supportedSamples == 0) ? D3D11_RTV_DIMENSION_TEXTURE2D : D3D11_RTV_DIMENSION_TEXTURE2DMS; rtvDesc.Texture2D.MipSlice = 0; result = device->CreateRenderTargetView(mTexture, &rtvDesc, &mRenderTarget); if (result == E_OUTOFMEMORY) { - mTexture->Release(); - mTexture = NULL; + SafeRelease(mTexture); + SafeRelease(mShaderResource); + SafeRelease(mDepthStencil); gl::error(GL_OUT_OF_MEMORY); return; } ASSERT(SUCCEEDED(result)); - D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; - srvDesc.Format = requestedFormat; - srvDesc.ViewDimension = (supportedSamples == 0) ? D3D11_SRV_DIMENSION_TEXTURE2D : D3D11_SRV_DIMENSION_TEXTURE2DMS; - srvDesc.Texture2D.MostDetailedMip = 0; - srvDesc.Texture2D.MipLevels = 1; - result = device->CreateShaderResourceView(mTexture, &srvDesc, &mShaderResource); - - if (result == E_OUTOFMEMORY) + if (formatInfo.dataInitializerFunction != NULL) { - mTexture->Release(); - mTexture = NULL; - mRenderTarget->Release(); - mRenderTarget = NULL; - gl::error(GL_OUT_OF_MEMORY); - return; + ID3D11DeviceContext *context = mRenderer->getDeviceContext(); + + const float clearValues[4] = { 0.0f, 0.0f, 0.0f, 1.0f }; + context->ClearRenderTargetView(mRenderTarget, clearValues); } - ASSERT(SUCCEEDED(result)); } } + mWidth = width; mHeight = height; - mInternalFormat = format; + mDepth = 1; + mInternalFormat = internalFormat; mSamples = supportedSamples; - mActualFormat = d3d11_gl::ConvertTextureInternalFormat(requestedFormat); + mActualFormat = dxgiFormatInfo.internalFormat; mSubresourceIndex = D3D11CalcSubresource(0, 0, 1); } RenderTarget11::~RenderTarget11() { - if (mTexture) - { - mTexture->Release(); - mTexture = NULL; - } - - if (mRenderTarget) - { - mRenderTarget->Release(); - mRenderTarget = NULL; - } - - if (mDepthStencil) - { - mDepthStencil->Release(); - mDepthStencil = NULL; - } - - if (mShaderResource) - { - mShaderResource->Release(); - mShaderResource = NULL; - } + SafeRelease(mTexture); + SafeRelease(mRenderTarget); + SafeRelease(mDepthStencil); + SafeRelease(mShaderResource); } RenderTarget11 *RenderTarget11::makeRenderTarget11(RenderTarget *target) @@ -327,7 +418,12 @@ RenderTarget11 *RenderTarget11::makeRenderTarget11(RenderTarget *target) return static_cast<rx::RenderTarget11*>(target); } -ID3D11Texture2D *RenderTarget11::getTexture() const +void RenderTarget11::invalidate(GLint x, GLint y, GLsizei width, GLsizei height) +{ + // Currently a no-op +} + +ID3D11Resource *RenderTarget11::getTexture() const { return mTexture; } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/RenderTarget11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/RenderTarget11.h index 97827f2639..82182957af 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/RenderTarget11.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/RenderTarget11.h @@ -20,14 +20,17 @@ class Renderer11; class RenderTarget11 : public RenderTarget { public: - RenderTarget11(Renderer *renderer, ID3D11RenderTargetView *rtv, ID3D11Texture2D *tex, ID3D11ShaderResourceView *srv, GLsizei width, GLsizei height); - RenderTarget11(Renderer *renderer, ID3D11DepthStencilView *dsv, ID3D11Texture2D *tex, ID3D11ShaderResourceView *srv, GLsizei width, GLsizei height); - RenderTarget11(Renderer *renderer, GLsizei width, GLsizei height, GLenum format, GLsizei samples, bool depth); + // RenderTarget11 takes ownership of any D3D11 resources it is given and will AddRef them + RenderTarget11(Renderer *renderer, ID3D11RenderTargetView *rtv, ID3D11Resource *resource, ID3D11ShaderResourceView *srv, GLsizei width, GLsizei height, GLsizei depth); + RenderTarget11(Renderer *renderer, ID3D11DepthStencilView *dsv, ID3D11Resource *resource, ID3D11ShaderResourceView *srv, GLsizei width, GLsizei height, GLsizei depth); + RenderTarget11(Renderer *renderer, GLsizei width, GLsizei height, GLenum internalFormat, GLsizei samples); virtual ~RenderTarget11(); static RenderTarget11 *makeRenderTarget11(RenderTarget *renderTarget); - ID3D11Texture2D *getTexture() const; + virtual void invalidate(GLint x, GLint y, GLsizei width, GLsizei height); + + ID3D11Resource *getTexture() const; ID3D11RenderTargetView *getRenderTargetView() const; ID3D11DepthStencilView *getDepthStencilView() const; ID3D11ShaderResourceView *getShaderResourceView() const; @@ -38,7 +41,7 @@ class RenderTarget11 : public RenderTarget DISALLOW_COPY_AND_ASSIGN(RenderTarget11); unsigned int mSubresourceIndex; - ID3D11Texture2D *mTexture; + ID3D11Resource *mTexture; ID3D11RenderTargetView *mRenderTarget; ID3D11DepthStencilView *mDepthStencil; ID3D11ShaderResourceView *mShaderResource; @@ -48,4 +51,4 @@ class RenderTarget11 : public RenderTarget } -#endif LIBGLESV2_RENDERER_RENDERTARGET11_H_
\ No newline at end of file +#endif // LIBGLESV2_RENDERER_RENDERTARGET11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Renderer11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Renderer11.cpp new file mode 100644 index 0000000000..b29b2ef910 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Renderer11.cpp @@ -0,0 +1,3266 @@ +// +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Renderer11.cpp: Implements a back-end specific class for the D3D11 renderer. + +#include "common/platform.h" +#include "libGLESv2/main.h" +#include "libGLESv2/Buffer.h" +#include "libGLESv2/FramebufferAttachment.h" +#include "libGLESv2/ProgramBinary.h" +#include "libGLESv2/Framebuffer.h" +#include "libGLESv2/renderer/d3d/ProgramD3D.h" +#include "libGLESv2/renderer/d3d/ShaderD3D.h" +#include "libGLESv2/renderer/d3d/TextureD3D.h" +#include "libGLESv2/renderer/d3d/TransformFeedbackD3D.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" +#include "libGLESv2/renderer/d3d/d3d11/RenderTarget11.h" +#include "libGLESv2/renderer/d3d/d3d11/renderer11_utils.h" +#include "libGLESv2/renderer/d3d/d3d11/formatutils11.h" +#include "libGLESv2/renderer/d3d/d3d11/ShaderExecutable11.h" +#include "libGLESv2/renderer/d3d/d3d11/SwapChain11.h" +#include "libGLESv2/renderer/d3d/d3d11/Image11.h" +#include "libGLESv2/renderer/d3d/d3d11/VertexBuffer11.h" +#include "libGLESv2/renderer/d3d/d3d11/IndexBuffer11.h" +#include "libGLESv2/renderer/d3d/d3d11/Buffer11.h" +#include "libGLESv2/renderer/d3d/VertexDataManager.h" +#include "libGLESv2/renderer/d3d/IndexDataManager.h" +#include "libGLESv2/renderer/d3d/d3d11/TextureStorage11.h" +#include "libGLESv2/renderer/d3d/d3d11/Query11.h" +#include "libGLESv2/renderer/d3d/d3d11/Fence11.h" +#include "libGLESv2/renderer/d3d/d3d11/Blit11.h" +#include "libGLESv2/renderer/d3d/d3d11/Clear11.h" +#include "libGLESv2/renderer/d3d/d3d11/PixelTransfer11.h" +#include "libGLESv2/renderer/d3d/d3d11/VertexArray11.h" +#include "libGLESv2/renderer/d3d/d3d11/Buffer11.h" + +#include "libEGL/Display.h" + +#include "common/utilities.h" + +#include <EGL/eglext.h> + +#include <sstream> + +// Enable ANGLE_SKIP_DXGI_1_2_CHECK if there is not a possibility of using cross-process +// HWNDs or the Windows 7 Platform Update (KB2670838) is expected to be installed. +#ifndef ANGLE_SKIP_DXGI_1_2_CHECK +#define ANGLE_SKIP_DXGI_1_2_CHECK 0 +#endif + +#ifdef _DEBUG +// this flag enables suppressing some spurious warnings that pop up in certain WebGL samples +// and conformance tests. to enable all warnings, remove this define. +#define ANGLE_SUPPRESS_D3D11_HAZARD_WARNINGS 1 +#endif + +namespace rx +{ +static const DXGI_FORMAT RenderTargetFormats[] = + { + DXGI_FORMAT_B8G8R8A8_UNORM, + DXGI_FORMAT_R8G8B8A8_UNORM + }; + +static const DXGI_FORMAT DepthStencilFormats[] = + { + DXGI_FORMAT_UNKNOWN, + DXGI_FORMAT_D24_UNORM_S8_UINT, + DXGI_FORMAT_D16_UNORM + }; + +enum +{ + MAX_TEXTURE_IMAGE_UNITS_VTF_SM4 = 16 +}; + +Renderer11::Renderer11(egl::Display *display, EGLNativeDisplayType hDc, EGLint requestedDisplay) + : Renderer(display), + mDc(hDc), + mRequestedDisplay(requestedDisplay) +{ + mVertexDataManager = NULL; + mIndexDataManager = NULL; + + mLineLoopIB = NULL; + mTriangleFanIB = NULL; + + mBlit = NULL; + mPixelTransfer = NULL; + + mClear = NULL; + + mSyncQuery = NULL; + + mD3d11Module = NULL; + mDxgiModule = NULL; + + mDeviceLost = false; + + mDevice = NULL; + mDeviceContext = NULL; + mDxgiAdapter = NULL; + mDxgiFactory = NULL; + + mDriverConstantBufferVS = NULL; + mDriverConstantBufferPS = NULL; + + mAppliedVertexShader = NULL; + mAppliedGeometryShader = NULL; + mCurPointGeometryShader = NULL; + mAppliedPixelShader = NULL; +} + +Renderer11::~Renderer11() +{ + release(); +} + +Renderer11 *Renderer11::makeRenderer11(Renderer *renderer) +{ + ASSERT(HAS_DYNAMIC_TYPE(rx::Renderer11*, renderer)); + return static_cast<rx::Renderer11*>(renderer); +} + +#ifndef __d3d11_1_h__ +#define D3D11_MESSAGE_ID_DEVICE_DRAW_RENDERTARGETVIEW_NOT_SET ((D3D11_MESSAGE_ID)3146081) +#endif + +EGLint Renderer11::initialize() +{ + if (!mCompiler.initialize()) + { + return EGL_NOT_INITIALIZED; + } + +#if !defined(ANGLE_PLATFORM_WINRT) + mDxgiModule = LoadLibrary(TEXT("dxgi.dll")); + mD3d11Module = LoadLibrary(TEXT("d3d11.dll")); + + if (mD3d11Module == NULL || mDxgiModule == NULL) + { + ERR("Could not load D3D11 or DXGI library - aborting!\n"); + return EGL_NOT_INITIALIZED; + } + + // create the D3D11 device + ASSERT(mDevice == NULL); + PFN_D3D11_CREATE_DEVICE D3D11CreateDevice = (PFN_D3D11_CREATE_DEVICE)GetProcAddress(mD3d11Module, "D3D11CreateDevice"); + + if (D3D11CreateDevice == NULL) + { + ERR("Could not retrieve D3D11CreateDevice address - aborting!\n"); + return EGL_NOT_INITIALIZED; + } +#endif + + D3D_FEATURE_LEVEL featureLevels[] = + { + D3D_FEATURE_LEVEL_11_0, + D3D_FEATURE_LEVEL_10_1, + D3D_FEATURE_LEVEL_10_0, +#if !defined(ANGLE_ENABLE_D3D9) + D3D_FEATURE_LEVEL_9_3, + D3D_FEATURE_LEVEL_9_2, + D3D_FEATURE_LEVEL_9_1, +#endif + }; + + D3D_DRIVER_TYPE driverType = D3D_DRIVER_TYPE_HARDWARE; + if (mRequestedDisplay == EGL_PLATFORM_ANGLE_TYPE_D3D11_WARP_ANGLE) + { + driverType = D3D_DRIVER_TYPE_WARP; + } + + HRESULT result = S_OK; + +#ifdef _DEBUG + result = D3D11CreateDevice(NULL, + driverType, + NULL, + D3D11_CREATE_DEVICE_DEBUG, + featureLevels, + ArraySize(featureLevels), + D3D11_SDK_VERSION, + &mDevice, + &mFeatureLevel, + &mDeviceContext); + + if (!mDevice || FAILED(result)) + { + ERR("Failed creating Debug D3D11 device - falling back to release runtime.\n"); + } + + if (!mDevice || FAILED(result)) +#endif + { + result = D3D11CreateDevice(NULL, + driverType, + NULL, + 0, + featureLevels, + ArraySize(featureLevels), + D3D11_SDK_VERSION, + &mDevice, + &mFeatureLevel, + &mDeviceContext); + + if (!mDevice || FAILED(result)) + { + ERR("Could not create D3D11 device - aborting!\n"); + return EGL_NOT_INITIALIZED; // Cleanup done by destructor through glDestroyRenderer + } + } + +#if !ANGLE_SKIP_DXGI_1_2_CHECK && !defined(ANGLE_PLATFORM_WINRT) + // In order to create a swap chain for an HWND owned by another process, DXGI 1.2 is required. + // The easiest way to check is to query for a IDXGIDevice2. + bool requireDXGI1_2 = false; + HWND hwnd = WindowFromDC(mDc); + if (hwnd) + { + DWORD currentProcessId = GetCurrentProcessId(); + DWORD wndProcessId; + GetWindowThreadProcessId(hwnd, &wndProcessId); + requireDXGI1_2 = (currentProcessId != wndProcessId); + } + else + { + requireDXGI1_2 = true; + } + + if (requireDXGI1_2) + { + IDXGIDevice2 *dxgiDevice2 = NULL; + result = mDevice->QueryInterface(__uuidof(IDXGIDevice2), (void**)&dxgiDevice2); + if (FAILED(result)) + { + ERR("DXGI 1.2 required to present to HWNDs owned by another process.\n"); + return EGL_NOT_INITIALIZED; + } + SafeRelease(dxgiDevice2); + } +#endif + +#if !defined(ANGLE_PLATFORM_WINRT) + IDXGIDevice *dxgiDevice = NULL; +#else + IDXGIDevice1 *dxgiDevice = NULL; +#endif + result = mDevice->QueryInterface(IID_PPV_ARGS(&dxgiDevice)); + + if (FAILED(result)) + { + ERR("Could not query DXGI device - aborting!\n"); + return EGL_NOT_INITIALIZED; + } + + result = dxgiDevice->GetParent(__uuidof(IDXGIAdapter), (void**)&mDxgiAdapter); + + if (FAILED(result)) + { + ERR("Could not retrieve DXGI adapter - aborting!\n"); + return EGL_NOT_INITIALIZED; + } + + SafeRelease(dxgiDevice); + + mDxgiAdapter->GetDesc(&mAdapterDescription); + memset(mDescription, 0, sizeof(mDescription)); + wcstombs(mDescription, mAdapterDescription.Description, sizeof(mDescription) - 1); + + result = mDxgiAdapter->GetParent(__uuidof(IDXGIFactory), (void**)&mDxgiFactory); + + if (!mDxgiFactory || FAILED(result)) + { + ERR("Could not create DXGI factory - aborting!\n"); + return EGL_NOT_INITIALIZED; + } + + // Disable some spurious D3D11 debug warnings to prevent them from flooding the output log +#if !defined(__MINGW32__) && defined(ANGLE_SUPPRESS_D3D11_HAZARD_WARNINGS) && defined(_DEBUG) + ID3D11InfoQueue *infoQueue; + result = mDevice->QueryInterface(__uuidof(ID3D11InfoQueue), (void **)&infoQueue); + + if (SUCCEEDED(result)) + { + D3D11_MESSAGE_ID hideMessages[] = + { + D3D11_MESSAGE_ID_DEVICE_DRAW_RENDERTARGETVIEW_NOT_SET + }; + + D3D11_INFO_QUEUE_FILTER filter = {0}; + filter.DenyList.NumIDs = ArraySize(hideMessages); + filter.DenyList.pIDList = hideMessages; + + infoQueue->AddStorageFilterEntries(&filter); + SafeRelease(infoQueue); + } +#endif + +#if !defined(ANGLE_PLATFORM_WINRT) + static wchar_t *qt_d3dcreate_multihreaded_var = _wgetenv(L"QT_D3DCREATE_MULTITHREADED"); + if (qt_d3dcreate_multihreaded_var && wcsstr(qt_d3dcreate_multihreaded_var, L"1")) + { + ID3D10Multithread *multithread; + result = mDevice->QueryInterface(IID_PPV_ARGS(&multithread)); + ASSERT(SUCCEEDED(result)); + result = multithread->SetMultithreadProtected(true); + ASSERT(SUCCEEDED(result)); + multithread->Release(); + } +#endif + + initializeDevice(); + + return EGL_SUCCESS; +} + +// do any one-time device initialization +// NOTE: this is also needed after a device lost/reset +// to reset the scene status and ensure the default states are reset. +void Renderer11::initializeDevice() +{ + mStateCache.initialize(mDevice); + mInputLayoutCache.initialize(mDevice, mDeviceContext); + + ASSERT(!mVertexDataManager && !mIndexDataManager); + mVertexDataManager = new VertexDataManager(this); + mIndexDataManager = new IndexDataManager(this); + + ASSERT(!mBlit); + mBlit = new Blit11(this); + + ASSERT(!mClear); + mClear = new Clear11(this); + + ASSERT(!mPixelTransfer); + mPixelTransfer = new PixelTransfer11(this); + + const gl::Caps &rendererCaps = getRendererCaps(); + + mForceSetVertexSamplerStates.resize(rendererCaps.maxVertexTextureImageUnits); + mCurVertexSamplerStates.resize(rendererCaps.maxVertexTextureImageUnits); + + mForceSetPixelSamplerStates.resize(rendererCaps.maxTextureImageUnits); + mCurPixelSamplerStates.resize(rendererCaps.maxTextureImageUnits); + + mCurVertexSRVs.resize(rendererCaps.maxVertexTextureImageUnits); + mCurPixelSRVs.resize(rendererCaps.maxTextureImageUnits); + + markAllStateDirty(); +} + +int Renderer11::generateConfigs(ConfigDesc **configDescList) +{ + unsigned int numRenderFormats = ArraySize(RenderTargetFormats); + unsigned int numDepthFormats = ArraySize(DepthStencilFormats); + (*configDescList) = new ConfigDesc[numRenderFormats * numDepthFormats]; + int numConfigs = 0; + + for (unsigned int formatIndex = 0; formatIndex < numRenderFormats; formatIndex++) + { + const d3d11::DXGIFormat &renderTargetFormatInfo = d3d11::GetDXGIFormatInfo(RenderTargetFormats[formatIndex]); + const gl::TextureCaps &renderTargetFormatCaps = getRendererTextureCaps().get(renderTargetFormatInfo.internalFormat); + if (renderTargetFormatCaps.renderable) + { + for (unsigned int depthStencilIndex = 0; depthStencilIndex < numDepthFormats; depthStencilIndex++) + { + const d3d11::DXGIFormat &depthStencilFormatInfo = d3d11::GetDXGIFormatInfo(DepthStencilFormats[depthStencilIndex]); + const gl::TextureCaps &depthStencilFormatCaps = getRendererTextureCaps().get(depthStencilFormatInfo.internalFormat); + if (depthStencilFormatCaps.renderable || DepthStencilFormats[depthStencilIndex] == DXGI_FORMAT_UNKNOWN) + { + ConfigDesc newConfig; + newConfig.renderTargetFormat = renderTargetFormatInfo.internalFormat; + newConfig.depthStencilFormat = depthStencilFormatInfo.internalFormat; + newConfig.multiSample = 0; // FIXME: enumerate multi-sampling + newConfig.fastConfig = true; // Assume all DX11 format conversions to be fast + newConfig.es3Capable = true; + + (*configDescList)[numConfigs++] = newConfig; + } + } + } + } + + return numConfigs; +} + +void Renderer11::deleteConfigs(ConfigDesc *configDescList) +{ + delete [] (configDescList); +} + +void Renderer11::sync(bool block) +{ + if (block) + { + HRESULT result; + + if (!mSyncQuery) + { + D3D11_QUERY_DESC queryDesc; + queryDesc.Query = D3D11_QUERY_EVENT; + queryDesc.MiscFlags = 0; + + result = mDevice->CreateQuery(&queryDesc, &mSyncQuery); + ASSERT(SUCCEEDED(result)); + } + + mDeviceContext->End(mSyncQuery); + mDeviceContext->Flush(); + + do + { + result = mDeviceContext->GetData(mSyncQuery, NULL, 0, D3D11_ASYNC_GETDATA_DONOTFLUSH); + + // Keep polling, but allow other threads to do something useful first + Sleep(0); + + if (testDeviceLost(true)) + { + return; + } + } + while (result == S_FALSE); + } + else + { + mDeviceContext->Flush(); + } +} + +SwapChain *Renderer11::createSwapChain(EGLNativeWindowType window, HANDLE shareHandle, GLenum backBufferFormat, GLenum depthBufferFormat) +{ + return new rx::SwapChain11(this, window, shareHandle, backBufferFormat, depthBufferFormat); +} + +gl::Error Renderer11::generateSwizzle(gl::Texture *texture) +{ + if (texture) + { + TextureStorage *texStorage = texture->getNativeTexture(); + if (texStorage) + { + TextureStorage11 *storage11 = TextureStorage11::makeTextureStorage11(texStorage); + + gl::Error error = storage11->generateSwizzles(texture->getSamplerState().swizzleRed, + texture->getSamplerState().swizzleGreen, + texture->getSamplerState().swizzleBlue, + texture->getSamplerState().swizzleAlpha); + if (error.isError()) + { + return error; + } + } + } + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer11::setSamplerState(gl::SamplerType type, int index, const gl::SamplerState &samplerState) +{ + if (type == gl::SAMPLER_PIXEL) + { + ASSERT(static_cast<unsigned int>(index) < getRendererCaps().maxTextureImageUnits); + + if (mForceSetPixelSamplerStates[index] || memcmp(&samplerState, &mCurPixelSamplerStates[index], sizeof(gl::SamplerState)) != 0) + { + ID3D11SamplerState *dxSamplerState = NULL; + gl::Error error = mStateCache.getSamplerState(samplerState, &dxSamplerState); + if (error.isError()) + { + return error; + } + + ASSERT(dxSamplerState != NULL); + mDeviceContext->PSSetSamplers(index, 1, &dxSamplerState); + + mCurPixelSamplerStates[index] = samplerState; + } + + mForceSetPixelSamplerStates[index] = false; + } + else if (type == gl::SAMPLER_VERTEX) + { + ASSERT(static_cast<unsigned int>(index) < getRendererCaps().maxVertexTextureImageUnits); + + if (mForceSetVertexSamplerStates[index] || memcmp(&samplerState, &mCurVertexSamplerStates[index], sizeof(gl::SamplerState)) != 0) + { + ID3D11SamplerState *dxSamplerState = NULL; + gl::Error error = mStateCache.getSamplerState(samplerState, &dxSamplerState); + if (error.isError()) + { + return error; + } + + ASSERT(dxSamplerState != NULL); + mDeviceContext->VSSetSamplers(index, 1, &dxSamplerState); + + mCurVertexSamplerStates[index] = samplerState; + } + + mForceSetVertexSamplerStates[index] = false; + } + else UNREACHABLE(); + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer11::setTexture(gl::SamplerType type, int index, gl::Texture *texture) +{ + ID3D11ShaderResourceView *textureSRV = NULL; + bool forceSetTexture = false; + + if (texture) + { + TextureD3D* textureImpl = TextureD3D::makeTextureD3D(texture->getImplementation()); + TextureStorage *texStorage = textureImpl->getNativeTexture(); + ASSERT(texStorage != NULL); + + TextureStorage11 *storage11 = TextureStorage11::makeTextureStorage11(texStorage); + gl::SamplerState samplerState; + texture->getSamplerStateWithNativeOffset(&samplerState); + textureSRV = storage11->getSRV(samplerState); + + // If we get NULL back from getSRV here, something went wrong in the texture class and we're unexpectedly + // missing the shader resource view + ASSERT(textureSRV != NULL); + + forceSetTexture = textureImpl->hasDirtyImages(); + textureImpl->resetDirty(); + } + + if (type == gl::SAMPLER_PIXEL) + { + ASSERT(static_cast<unsigned int>(index) < getRendererCaps().maxTextureImageUnits); + + if (forceSetTexture || mCurPixelSRVs[index] != textureSRV) + { + mDeviceContext->PSSetShaderResources(index, 1, &textureSRV); + } + + mCurPixelSRVs[index] = textureSRV; + } + else if (type == gl::SAMPLER_VERTEX) + { + ASSERT(static_cast<unsigned int>(index) < getRendererCaps().maxVertexTextureImageUnits); + + if (forceSetTexture || mCurVertexSRVs[index] != textureSRV) + { + mDeviceContext->VSSetShaderResources(index, 1, &textureSRV); + } + + mCurVertexSRVs[index] = textureSRV; + } + else UNREACHABLE(); + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer11::setUniformBuffers(const gl::Buffer *vertexUniformBuffers[], const gl::Buffer *fragmentUniformBuffers[]) +{ + for (unsigned int uniformBufferIndex = 0; uniformBufferIndex < gl::IMPLEMENTATION_MAX_VERTEX_SHADER_UNIFORM_BUFFERS; uniformBufferIndex++) + { + const gl::Buffer *uniformBuffer = vertexUniformBuffers[uniformBufferIndex]; + if (uniformBuffer) + { + Buffer11 *bufferStorage = Buffer11::makeBuffer11(uniformBuffer->getImplementation()); + ID3D11Buffer *constantBuffer = bufferStorage->getBuffer(BUFFER_USAGE_UNIFORM); + + if (!constantBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY); + } + + if (mCurrentConstantBufferVS[uniformBufferIndex] != bufferStorage->getSerial()) + { + mDeviceContext->VSSetConstantBuffers(getReservedVertexUniformBuffers() + uniformBufferIndex, + 1, &constantBuffer); + mCurrentConstantBufferVS[uniformBufferIndex] = bufferStorage->getSerial(); + } + } + } + + for (unsigned int uniformBufferIndex = 0; uniformBufferIndex < gl::IMPLEMENTATION_MAX_FRAGMENT_SHADER_UNIFORM_BUFFERS; uniformBufferIndex++) + { + const gl::Buffer *uniformBuffer = fragmentUniformBuffers[uniformBufferIndex]; + if (uniformBuffer) + { + Buffer11 *bufferStorage = Buffer11::makeBuffer11(uniformBuffer->getImplementation()); + ID3D11Buffer *constantBuffer = bufferStorage->getBuffer(BUFFER_USAGE_UNIFORM); + + if (!constantBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY); + } + + if (mCurrentConstantBufferPS[uniformBufferIndex] != bufferStorage->getSerial()) + { + mDeviceContext->PSSetConstantBuffers(getReservedFragmentUniformBuffers() + uniformBufferIndex, + 1, &constantBuffer); + mCurrentConstantBufferPS[uniformBufferIndex] = bufferStorage->getSerial(); + } + } + } + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer11::setRasterizerState(const gl::RasterizerState &rasterState) +{ + if (mForceSetRasterState || memcmp(&rasterState, &mCurRasterState, sizeof(gl::RasterizerState)) != 0) + { + ID3D11RasterizerState *dxRasterState = NULL; + gl::Error error = mStateCache.getRasterizerState(rasterState, mScissorEnabled, &dxRasterState); + if (error.isError()) + { + return error; + } + + mDeviceContext->RSSetState(dxRasterState); + + mCurRasterState = rasterState; + } + + mForceSetRasterState = false; + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer11::setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState &blendState, const gl::ColorF &blendColor, + unsigned int sampleMask) +{ + if (mForceSetBlendState || + memcmp(&blendState, &mCurBlendState, sizeof(gl::BlendState)) != 0 || + memcmp(&blendColor, &mCurBlendColor, sizeof(gl::ColorF)) != 0 || + sampleMask != mCurSampleMask) + { + ID3D11BlendState *dxBlendState = NULL; + gl::Error error = mStateCache.getBlendState(framebuffer, blendState, &dxBlendState); + if (error.isError()) + { + return error; + } + + ASSERT(dxBlendState != NULL); + + float blendColors[4] = {0.0f}; + if (blendState.sourceBlendRGB != GL_CONSTANT_ALPHA && blendState.sourceBlendRGB != GL_ONE_MINUS_CONSTANT_ALPHA && + blendState.destBlendRGB != GL_CONSTANT_ALPHA && blendState.destBlendRGB != GL_ONE_MINUS_CONSTANT_ALPHA) + { + blendColors[0] = blendColor.red; + blendColors[1] = blendColor.green; + blendColors[2] = blendColor.blue; + blendColors[3] = blendColor.alpha; + } + else + { + blendColors[0] = blendColor.alpha; + blendColors[1] = blendColor.alpha; + blendColors[2] = blendColor.alpha; + blendColors[3] = blendColor.alpha; + } + + mDeviceContext->OMSetBlendState(dxBlendState, blendColors, sampleMask); + + mCurBlendState = blendState; + mCurBlendColor = blendColor; + mCurSampleMask = sampleMask; + } + + mForceSetBlendState = false; + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer11::setDepthStencilState(const gl::DepthStencilState &depthStencilState, int stencilRef, + int stencilBackRef, bool frontFaceCCW) +{ + if (mForceSetDepthStencilState || + memcmp(&depthStencilState, &mCurDepthStencilState, sizeof(gl::DepthStencilState)) != 0 || + stencilRef != mCurStencilRef || stencilBackRef != mCurStencilBackRef) + { + ASSERT(depthStencilState.stencilWritemask == depthStencilState.stencilBackWritemask); + ASSERT(stencilRef == stencilBackRef); + ASSERT(depthStencilState.stencilMask == depthStencilState.stencilBackMask); + + ID3D11DepthStencilState *dxDepthStencilState = NULL; + gl::Error error = mStateCache.getDepthStencilState(depthStencilState, &dxDepthStencilState); + if (error.isError()) + { + return error; + } + + ASSERT(dxDepthStencilState); + + // Max D3D11 stencil reference value is 0xFF, corresponding to the max 8 bits in a stencil buffer + // GL specifies we should clamp the ref value to the nearest bit depth when doing stencil ops + META_ASSERT(D3D11_DEFAULT_STENCIL_READ_MASK == 0xFF); + META_ASSERT(D3D11_DEFAULT_STENCIL_WRITE_MASK == 0xFF); + UINT dxStencilRef = std::min<UINT>(stencilRef, 0xFFu); + + mDeviceContext->OMSetDepthStencilState(dxDepthStencilState, dxStencilRef); + + mCurDepthStencilState = depthStencilState; + mCurStencilRef = stencilRef; + mCurStencilBackRef = stencilBackRef; + } + + mForceSetDepthStencilState = false; + + return gl::Error(GL_NO_ERROR); +} + +void Renderer11::setScissorRectangle(const gl::Rectangle &scissor, bool enabled) +{ + if (mForceSetScissor || memcmp(&scissor, &mCurScissor, sizeof(gl::Rectangle)) != 0 || + enabled != mScissorEnabled) + { + if (enabled) + { + D3D11_RECT rect; + rect.left = std::max(0, scissor.x); + rect.top = std::max(0, scissor.y); + rect.right = scissor.x + std::max(0, scissor.width); + rect.bottom = scissor.y + std::max(0, scissor.height); + + mDeviceContext->RSSetScissorRects(1, &rect); + } + + if (enabled != mScissorEnabled) + { + mForceSetRasterState = true; + } + + mCurScissor = scissor; + mScissorEnabled = enabled; + } + + mForceSetScissor = false; +} + +void Renderer11::setViewport(const gl::Rectangle &viewport, float zNear, float zFar, GLenum drawMode, GLenum frontFace, + bool ignoreViewport) +{ + gl::Rectangle actualViewport = viewport; + float actualZNear = gl::clamp01(zNear); + float actualZFar = gl::clamp01(zFar); + if (ignoreViewport) + { + actualViewport.x = 0; + actualViewport.y = 0; + actualViewport.width = mRenderTargetDesc.width; + actualViewport.height = mRenderTargetDesc.height; + actualZNear = 0.0f; + actualZFar = 1.0f; + } + + const gl::Caps& caps = getRendererCaps(); + + // Clamp width and height first to the gl maximum, then clamp further if we extend past the D3D maximum bounds + D3D11_VIEWPORT dxViewport; + dxViewport.TopLeftX = gl::clamp(actualViewport.x, -static_cast<int>(caps.maxViewportWidth), static_cast<int>(caps.maxViewportWidth)); + dxViewport.TopLeftY = gl::clamp(actualViewport.y, -static_cast<int>(caps.maxViewportHeight), static_cast<int>(caps.maxViewportHeight)); + dxViewport.Width = gl::clamp(actualViewport.width, 0, static_cast<int>(caps.maxViewportWidth - dxViewport.TopLeftX)); + dxViewport.Height = gl::clamp(actualViewport.height, 0, static_cast<int>(caps.maxViewportHeight - dxViewport.TopLeftY)); + dxViewport.MinDepth = actualZNear; + dxViewport.MaxDepth = actualZFar; + + bool viewportChanged = mForceSetViewport || memcmp(&actualViewport, &mCurViewport, sizeof(gl::Rectangle)) != 0 || + actualZNear != mCurNear || actualZFar != mCurFar; + + if (viewportChanged) + { + mDeviceContext->RSSetViewports(1, &dxViewport); + + mCurViewport = actualViewport; + mCurNear = actualZNear; + mCurFar = actualZFar; + + mPixelConstants.viewCoords[0] = actualViewport.width * 0.5f; + mPixelConstants.viewCoords[1] = actualViewport.height * 0.5f; + mPixelConstants.viewCoords[2] = actualViewport.x + (actualViewport.width * 0.5f); + mPixelConstants.viewCoords[3] = actualViewport.y + (actualViewport.height * 0.5f); + + mPixelConstants.depthFront[0] = (actualZFar - actualZNear) * 0.5f; + mPixelConstants.depthFront[1] = (actualZNear + actualZFar) * 0.5f; + + mVertexConstants.depthRange[0] = actualZNear; + mVertexConstants.depthRange[1] = actualZFar; + mVertexConstants.depthRange[2] = actualZFar - actualZNear; + + mPixelConstants.depthRange[0] = actualZNear; + mPixelConstants.depthRange[1] = actualZFar; + mPixelConstants.depthRange[2] = actualZFar - actualZNear; + } + + mForceSetViewport = false; +} + +bool Renderer11::applyPrimitiveType(GLenum mode, GLsizei count) +{ + D3D11_PRIMITIVE_TOPOLOGY primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_UNDEFINED; + + GLsizei minCount = 0; + + switch (mode) + { + case GL_POINTS: primitiveTopology = D3D11_PRIMITIVE_TOPOLOGY_POINTLIST; minCount = 1; break; + case GL_LINES: primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_LINELIST; minCount = 2; break; + case GL_LINE_LOOP: primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_LINESTRIP; minCount = 2; break; + case GL_LINE_STRIP: primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_LINESTRIP; minCount = 2; break; + case GL_TRIANGLES: primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_TRIANGLELIST; minCount = 3; break; + case GL_TRIANGLE_STRIP: primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_TRIANGLESTRIP; minCount = 3; break; + // emulate fans via rewriting index buffer + case GL_TRIANGLE_FAN: primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_TRIANGLELIST; minCount = 3; break; + default: + UNREACHABLE(); + return false; + } + + if (primitiveTopology != mCurrentPrimitiveTopology) + { + mDeviceContext->IASetPrimitiveTopology(primitiveTopology); + mCurrentPrimitiveTopology = primitiveTopology; + } + + return count >= minCount; +} + +gl::Error Renderer11::applyRenderTarget(gl::Framebuffer *framebuffer) +{ + // Get the color render buffer and serial + // Also extract the render target dimensions and view + unsigned int renderTargetWidth = 0; + unsigned int renderTargetHeight = 0; + GLenum renderTargetFormat = 0; + unsigned int renderTargetSerials[gl::IMPLEMENTATION_MAX_DRAW_BUFFERS] = {0}; + ID3D11RenderTargetView* framebufferRTVs[gl::IMPLEMENTATION_MAX_DRAW_BUFFERS] = {NULL}; + bool missingColorRenderTarget = true; + + const gl::ColorbufferInfo &colorbuffers = framebuffer->getColorbuffersForRender(); + + for (size_t colorAttachment = 0; colorAttachment < colorbuffers.size(); ++colorAttachment) + { + gl::FramebufferAttachment *colorbuffer = colorbuffers[colorAttachment]; + + if (colorbuffer) + { + // the draw buffer must be either "none", "back" for the default buffer or the same index as this color (in order) + + // check for zero-sized default framebuffer, which is a special case. + // in this case we do not wish to modify any state and just silently return false. + // this will not report any gl error but will cause the calling method to return. + if (colorbuffer->getWidth() == 0 || colorbuffer->getHeight() == 0) + { + return gl::Error(GL_NO_ERROR); + } + + renderTargetSerials[colorAttachment] = GetAttachmentSerial(colorbuffer); + + // Extract the render target dimensions and view + RenderTarget11 *renderTarget = d3d11::GetAttachmentRenderTarget(colorbuffer); + if (!renderTarget) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal render target pointer unexpectedly null."); + } + + framebufferRTVs[colorAttachment] = renderTarget->getRenderTargetView(); + if (!framebufferRTVs[colorAttachment]) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal render target view pointer unexpectedly null."); + } + + if (missingColorRenderTarget) + { + renderTargetWidth = colorbuffer->getWidth(); + renderTargetHeight = colorbuffer->getHeight(); + renderTargetFormat = colorbuffer->getActualFormat(); + missingColorRenderTarget = false; + } + + // TODO: Detect if this color buffer is already bound as a texture and unbind it first to prevent + // D3D11 warnings. + } + } + + // Get the depth stencil render buffter and serials + gl::FramebufferAttachment *depthStencil = framebuffer->getDepthbuffer(); + unsigned int depthbufferSerial = 0; + unsigned int stencilbufferSerial = 0; + if (depthStencil) + { + depthbufferSerial = GetAttachmentSerial(depthStencil); + } + else if (framebuffer->getStencilbuffer()) + { + depthStencil = framebuffer->getStencilbuffer(); + stencilbufferSerial = GetAttachmentSerial(depthStencil); + } + + ID3D11DepthStencilView* framebufferDSV = NULL; + if (depthStencil) + { + RenderTarget11 *depthStencilRenderTarget = d3d11::GetAttachmentRenderTarget(depthStencil); + if (!depthStencilRenderTarget) + { + SafeRelease(framebufferRTVs); + return gl::Error(GL_OUT_OF_MEMORY, "Internal render target pointer unexpectedly null."); + } + + framebufferDSV = depthStencilRenderTarget->getDepthStencilView(); + if (!framebufferDSV) + { + SafeRelease(framebufferRTVs); + return gl::Error(GL_OUT_OF_MEMORY, "Internal depth stencil view pointer unexpectedly null."); + } + + // If there is no render buffer, the width, height and format values come from + // the depth stencil + if (missingColorRenderTarget) + { + renderTargetWidth = depthStencil->getWidth(); + renderTargetHeight = depthStencil->getHeight(); + renderTargetFormat = depthStencil->getActualFormat(); + } + } + + // Apply the render target and depth stencil + if (!mRenderTargetDescInitialized || !mDepthStencilInitialized || + memcmp(renderTargetSerials, mAppliedRenderTargetSerials, sizeof(renderTargetSerials)) != 0 || + depthbufferSerial != mAppliedDepthbufferSerial || + stencilbufferSerial != mAppliedStencilbufferSerial) + { + mDeviceContext->OMSetRenderTargets(getRendererCaps().maxDrawBuffers, framebufferRTVs, framebufferDSV); + + mRenderTargetDesc.width = renderTargetWidth; + mRenderTargetDesc.height = renderTargetHeight; + mRenderTargetDesc.format = renderTargetFormat; + mForceSetViewport = true; + mForceSetScissor = true; + mForceSetBlendState = true; + + if (!mDepthStencilInitialized) + { + mForceSetRasterState = true; + } + + for (unsigned int rtIndex = 0; rtIndex < gl::IMPLEMENTATION_MAX_DRAW_BUFFERS; rtIndex++) + { + mAppliedRenderTargetSerials[rtIndex] = renderTargetSerials[rtIndex]; + } + mAppliedDepthbufferSerial = depthbufferSerial; + mAppliedStencilbufferSerial = stencilbufferSerial; + mRenderTargetDescInitialized = true; + mDepthStencilInitialized = true; + } + + invalidateFramebufferSwizzles(framebuffer); + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer11::applyVertexBuffer(gl::ProgramBinary *programBinary, const gl::VertexAttribute vertexAttributes[], const gl::VertexAttribCurrentValueData currentValues[], + GLint first, GLsizei count, GLsizei instances) +{ + TranslatedAttribute attributes[gl::MAX_VERTEX_ATTRIBS]; + gl::Error error = mVertexDataManager->prepareVertexData(vertexAttributes, currentValues, programBinary, first, count, attributes, instances); + if (error.isError()) + { + return error; + } + + return mInputLayoutCache.applyVertexBuffers(attributes, programBinary); +} + +gl::Error Renderer11::applyIndexBuffer(const GLvoid *indices, gl::Buffer *elementArrayBuffer, GLsizei count, GLenum mode, GLenum type, TranslatedIndexData *indexInfo) +{ + gl::Error error = mIndexDataManager->prepareIndexData(type, count, elementArrayBuffer, indices, indexInfo); + if (error.isError()) + { + return error; + } + + ID3D11Buffer *buffer = NULL; + DXGI_FORMAT bufferFormat = (indexInfo->indexType == GL_UNSIGNED_INT) ? DXGI_FORMAT_R32_UINT : DXGI_FORMAT_R16_UINT; + + if (indexInfo->storage) + { + Buffer11 *storage = Buffer11::makeBuffer11(indexInfo->storage); + buffer = storage->getBuffer(BUFFER_USAGE_INDEX); + } + else + { + IndexBuffer11* indexBuffer = IndexBuffer11::makeIndexBuffer11(indexInfo->indexBuffer); + buffer = indexBuffer->getBuffer(); + } + + if (buffer != mAppliedIB || bufferFormat != mAppliedIBFormat || indexInfo->startOffset != mAppliedIBOffset) + { + mDeviceContext->IASetIndexBuffer(buffer, bufferFormat, indexInfo->startOffset); + + mAppliedIB = buffer; + mAppliedIBFormat = bufferFormat; + mAppliedIBOffset = indexInfo->startOffset; + } + + return gl::Error(GL_NO_ERROR); +} + +void Renderer11::applyTransformFeedbackBuffers(gl::Buffer *transformFeedbackBuffers[], GLintptr offsets[]) +{ + ID3D11Buffer* d3dBuffers[gl::IMPLEMENTATION_MAX_TRANSFORM_FEEDBACK_BUFFERS]; + UINT d3dOffsets[gl::IMPLEMENTATION_MAX_TRANSFORM_FEEDBACK_BUFFERS]; + bool requiresUpdate = false; + for (size_t i = 0; i < gl::IMPLEMENTATION_MAX_TRANSFORM_FEEDBACK_BUFFERS; i++) + { + if (transformFeedbackBuffers[i]) + { + Buffer11 *storage = Buffer11::makeBuffer11(transformFeedbackBuffers[i]->getImplementation()); + ID3D11Buffer *buffer = storage->getBuffer(BUFFER_USAGE_VERTEX_OR_TRANSFORM_FEEDBACK); + + d3dBuffers[i] = buffer; + d3dOffsets[i] = (mAppliedTFBuffers[i] != buffer) ? static_cast<UINT>(offsets[i]) : -1; + } + else + { + d3dBuffers[i] = NULL; + d3dOffsets[i] = 0; + } + + if (d3dBuffers[i] != mAppliedTFBuffers[i] || offsets[i] != mAppliedTFOffsets[i]) + { + requiresUpdate = true; + } + } + + if (requiresUpdate) + { + mDeviceContext->SOSetTargets(ArraySize(d3dBuffers), d3dBuffers, d3dOffsets); + for (size_t i = 0; i < gl::IMPLEMENTATION_MAX_TRANSFORM_FEEDBACK_BUFFERS; i++) + { + mAppliedTFBuffers[i] = d3dBuffers[i]; + mAppliedTFOffsets[i] = offsets[i]; + } + } +} + +gl::Error Renderer11::drawArrays(GLenum mode, GLsizei count, GLsizei instances, bool transformFeedbackActive) +{ + if (mode == GL_POINTS && transformFeedbackActive) + { + // Since point sprites are generated with a geometry shader, too many vertices will + // be written if transform feedback is active. To work around this, draw only the points + // with the stream out shader and no pixel shader to feed the stream out buffers and then + // draw again with the point sprite geometry shader to rasterize the point sprites. + + mDeviceContext->PSSetShader(NULL, NULL, 0); + + if (instances > 0) + { + mDeviceContext->DrawInstanced(count, instances, 0, 0); + } + else + { + mDeviceContext->Draw(count, 0); + } + + mDeviceContext->GSSetShader(mCurPointGeometryShader, NULL, 0); + mDeviceContext->PSSetShader(mAppliedPixelShader, NULL, 0); + + if (instances > 0) + { + mDeviceContext->DrawInstanced(count, instances, 0, 0); + } + else + { + mDeviceContext->Draw(count, 0); + } + + mDeviceContext->GSSetShader(mAppliedGeometryShader, NULL, 0); + + return gl::Error(GL_NO_ERROR); + } + else if (mode == GL_LINE_LOOP) + { + return drawLineLoop(count, GL_NONE, NULL, 0, NULL); + } + else if (mode == GL_TRIANGLE_FAN) + { + return drawTriangleFan(count, GL_NONE, NULL, 0, NULL, instances); + } + else if (instances > 0) + { + mDeviceContext->DrawInstanced(count, instances, 0, 0); + return gl::Error(GL_NO_ERROR); + } + else + { + mDeviceContext->Draw(count, 0); + return gl::Error(GL_NO_ERROR); + } +} + +gl::Error Renderer11::drawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, + gl::Buffer *elementArrayBuffer, const TranslatedIndexData &indexInfo, GLsizei instances) +{ + int minIndex = static_cast<int>(indexInfo.indexRange.start); + + if (mode == GL_LINE_LOOP) + { + return drawLineLoop(count, type, indices, minIndex, elementArrayBuffer); + } + else if (mode == GL_TRIANGLE_FAN) + { + return drawTriangleFan(count, type, indices, minIndex, elementArrayBuffer, instances); + } + else if (instances > 0) + { + mDeviceContext->DrawIndexedInstanced(count, instances, 0, -minIndex, 0); + return gl::Error(GL_NO_ERROR); + } + else + { + mDeviceContext->DrawIndexed(count, 0, -minIndex); + return gl::Error(GL_NO_ERROR); + } +} + +template<typename T> +static void fillLineLoopIndices(GLenum type, GLsizei count, const GLvoid *indices, T *data) +{ + switch (type) + { + case GL_NONE: // Non-indexed draw + for (int i = 0; i < count; i++) + { + data[i] = i; + } + data[count] = 0; + break; + case GL_UNSIGNED_BYTE: + for (int i = 0; i < count; i++) + { + data[i] = static_cast<const GLubyte*>(indices)[i]; + } + data[count] = static_cast<const GLubyte*>(indices)[0]; + break; + case GL_UNSIGNED_SHORT: + for (int i = 0; i < count; i++) + { + data[i] = static_cast<const GLushort*>(indices)[i]; + } + data[count] = static_cast<const GLushort*>(indices)[0]; + break; + case GL_UNSIGNED_INT: + for (int i = 0; i < count; i++) + { + data[i] = static_cast<const GLuint*>(indices)[i]; + } + data[count] = static_cast<const GLuint*>(indices)[0]; + break; + default: UNREACHABLE(); + } +} + +template<typename T> +static void fillTriangleFanIndices(GLenum type, unsigned int numTris, const GLvoid *indices, T *data) +{ + switch (type) + { + case GL_NONE: // Non-indexed draw + for (unsigned int i = 0; i < numTris; i++) + { + data[i*3 + 0] = 0; + data[i*3 + 1] = i + 1; + data[i*3 + 2] = i + 2; + } + break; + case GL_UNSIGNED_BYTE: + for (unsigned int i = 0; i < numTris; i++) + { + data[i*3 + 0] = static_cast<const GLubyte*>(indices)[0]; + data[i*3 + 1] = static_cast<const GLubyte*>(indices)[i + 1]; + data[i*3 + 2] = static_cast<const GLubyte*>(indices)[i + 2]; + } + break; + case GL_UNSIGNED_SHORT: + for (unsigned int i = 0; i < numTris; i++) + { + data[i*3 + 0] = static_cast<const GLushort*>(indices)[0]; + data[i*3 + 1] = static_cast<const GLushort*>(indices)[i + 1]; + data[i*3 + 2] = static_cast<const GLushort*>(indices)[i + 2]; + } + break; + case GL_UNSIGNED_INT: + for (unsigned int i = 0; i < numTris; i++) + { + data[i*3 + 0] = static_cast<const GLuint*>(indices)[0]; + data[i*3 + 1] = static_cast<const GLuint*>(indices)[i + 1]; + data[i*3 + 2] = static_cast<const GLuint*>(indices)[i + 2]; + } + break; + default: UNREACHABLE(); + } +} + +gl::Error Renderer11::drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer) +{ + // Get the raw indices for an indexed draw + if (type != GL_NONE && elementArrayBuffer) + { + gl::Buffer *indexBuffer = elementArrayBuffer; + BufferImpl *storage = indexBuffer->getImplementation(); + intptr_t offset = reinterpret_cast<intptr_t>(indices); + indices = static_cast<const GLubyte*>(storage->getData()) + offset; + } + + // TODO: some level 9 hardware supports 32-bit indices; test and store support instead + const int indexType = isLevel9() ? GL_UNSIGNED_SHORT : GL_UNSIGNED_INT; + + if (!mLineLoopIB) + { + mLineLoopIB = new StreamingIndexBufferInterface(this); + gl::Error error = mLineLoopIB->reserveBufferSpace(INITIAL_INDEX_BUFFER_SIZE, indexType); + if (error.isError()) + { + SafeDelete(mLineLoopIB); + return error; + } + } + + // Checked by Renderer11::applyPrimitiveType + ASSERT(count >= 0); + + int indexTypeSize = indexType == GL_UNSIGNED_SHORT ? sizeof(unsigned short) : sizeof(unsigned int); + if (static_cast<unsigned int>(count) + 1 > (std::numeric_limits<unsigned int>::max() / indexTypeSize)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to create a 32-bit looping index buffer for GL_LINE_LOOP, too many indices required."); + } + + const unsigned int spaceNeeded = (static_cast<unsigned int>(count) + 1) * sizeof(unsigned int); + gl::Error error = mLineLoopIB->reserveBufferSpace(spaceNeeded, GL_UNSIGNED_INT); + if (error.isError()) + { + return error; + } + + void* mappedMemory = NULL; + unsigned int offset; + error = mLineLoopIB->mapBuffer(spaceNeeded, &mappedMemory, &offset); + if (error.isError()) + { + return error; + } + + if (indexType == GL_UNSIGNED_SHORT) + fillLineLoopIndices(type, count, indices, reinterpret_cast<unsigned short*>(mappedMemory)); + else + fillLineLoopIndices(type, count, indices, reinterpret_cast<unsigned int*>(mappedMemory)); + unsigned int indexBufferOffset = offset; + + error = mLineLoopIB->unmapBuffer(); + if (error.isError()) + { + return error; + } + + IndexBuffer11 *indexBuffer = IndexBuffer11::makeIndexBuffer11(mLineLoopIB->getIndexBuffer()); + ID3D11Buffer *d3dIndexBuffer = indexBuffer->getBuffer(); + DXGI_FORMAT indexFormat = indexBuffer->getIndexFormat(); + + if (mAppliedIB != d3dIndexBuffer || mAppliedIBFormat != indexFormat || mAppliedIBOffset != indexBufferOffset) + { + mDeviceContext->IASetIndexBuffer(d3dIndexBuffer, indexFormat, indexBufferOffset); + mAppliedIB = d3dIndexBuffer; + mAppliedIBFormat = indexFormat; + mAppliedIBOffset = indexBufferOffset; + } + + mDeviceContext->DrawIndexed(count + 1, 0, -minIndex); + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer11::drawTriangleFan(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer, int instances) +{ + // Get the raw indices for an indexed draw + if (type != GL_NONE && elementArrayBuffer) + { + gl::Buffer *indexBuffer = elementArrayBuffer; + BufferImpl *storage = indexBuffer->getImplementation(); + intptr_t offset = reinterpret_cast<intptr_t>(indices); + indices = static_cast<const GLubyte*>(storage->getData()) + offset; + } + + const int indexType = isLevel9() ? GL_UNSIGNED_SHORT : GL_UNSIGNED_INT; + + if (!mTriangleFanIB) + { + mTriangleFanIB = new StreamingIndexBufferInterface(this); + gl::Error error = mTriangleFanIB->reserveBufferSpace(INITIAL_INDEX_BUFFER_SIZE, indexType); + if (error.isError()) + { + SafeDelete(mTriangleFanIB); + return error; + } + } + + // Checked by Renderer11::applyPrimitiveType + ASSERT(count >= 3); + + const unsigned int numTris = count - 2; + + int indexTypeSize = indexType == GL_UNSIGNED_SHORT ? sizeof(unsigned short) : sizeof(unsigned int); + if (numTris > (std::numeric_limits<unsigned int>::max() / (indexTypeSize * 3))) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to create a scratch index buffer for GL_TRIANGLE_FAN, too many indices required."); + } + + const unsigned int spaceNeeded = (numTris * 3) * indexTypeSize; + gl::Error error = mTriangleFanIB->reserveBufferSpace(spaceNeeded, indexType); + if (error.isError()) + { + return error; + } + + void* mappedMemory = NULL; + unsigned int offset; + error = mTriangleFanIB->mapBuffer(spaceNeeded, &mappedMemory, &offset); + if (error.isError()) + { + return error; + } + + if (indexType == GL_UNSIGNED_SHORT) + fillTriangleFanIndices(type, numTris, indices, reinterpret_cast<unsigned short*>(mappedMemory)); + else + fillTriangleFanIndices(type, numTris, indices, reinterpret_cast<unsigned int*>(mappedMemory)); + + unsigned int indexBufferOffset = offset; + + error = mTriangleFanIB->unmapBuffer(); + if (error.isError()) + { + return error; + } + + IndexBuffer11 *indexBuffer = IndexBuffer11::makeIndexBuffer11(mTriangleFanIB->getIndexBuffer()); + ID3D11Buffer *d3dIndexBuffer = indexBuffer->getBuffer(); + DXGI_FORMAT indexFormat = indexBuffer->getIndexFormat(); + + if (mAppliedIB != d3dIndexBuffer || mAppliedIBFormat != indexFormat || mAppliedIBOffset != indexBufferOffset) + { + mDeviceContext->IASetIndexBuffer(d3dIndexBuffer, indexFormat, indexBufferOffset); + mAppliedIB = d3dIndexBuffer; + mAppliedIBFormat = indexFormat; + mAppliedIBOffset = indexBufferOffset; + } + + if (instances > 0) + { + mDeviceContext->DrawIndexedInstanced(numTris * 3, instances, 0, -minIndex, 0); + } + else + { + mDeviceContext->DrawIndexed(numTris * 3, 0, -minIndex); + } + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer11::applyShaders(gl::ProgramBinary *programBinary, const gl::VertexFormat inputLayout[], const gl::Framebuffer *framebuffer, + bool rasterizerDiscard, bool transformFeedbackActive) +{ + ShaderExecutable *vertexExe = programBinary->getVertexExecutableForInputLayout(inputLayout); + ShaderExecutable *pixelExe = programBinary->getPixelExecutableForFramebuffer(framebuffer); + ShaderExecutable *geometryExe = programBinary->getGeometryExecutable(); + + ID3D11VertexShader *vertexShader = (vertexExe ? ShaderExecutable11::makeShaderExecutable11(vertexExe)->getVertexShader() : NULL); + + ID3D11PixelShader *pixelShader = NULL; + // Skip pixel shader if we're doing rasterizer discard. + if (!rasterizerDiscard) + { + pixelShader = (pixelExe ? ShaderExecutable11::makeShaderExecutable11(pixelExe)->getPixelShader() : NULL); + } + + ID3D11GeometryShader *geometryShader = NULL; + if (transformFeedbackActive) + { + geometryShader = (vertexExe ? ShaderExecutable11::makeShaderExecutable11(vertexExe)->getStreamOutShader() : NULL); + } + else if (mCurRasterState.pointDrawMode) + { + geometryShader = (geometryExe ? ShaderExecutable11::makeShaderExecutable11(geometryExe)->getGeometryShader() : NULL); + } + + bool dirtyUniforms = false; + + if (vertexShader != mAppliedVertexShader) + { + mDeviceContext->VSSetShader(vertexShader, NULL, 0); + mAppliedVertexShader = vertexShader; + dirtyUniforms = true; + } + + if (geometryShader != mAppliedGeometryShader) + { + mDeviceContext->GSSetShader(geometryShader, NULL, 0); + mAppliedGeometryShader = geometryShader; + dirtyUniforms = true; + } + + if (geometryExe && mCurRasterState.pointDrawMode) + { + mCurPointGeometryShader = ShaderExecutable11::makeShaderExecutable11(geometryExe)->getGeometryShader(); + } + else + { + mCurPointGeometryShader = NULL; + } + + if (pixelShader != mAppliedPixelShader) + { + mDeviceContext->PSSetShader(pixelShader, NULL, 0); + mAppliedPixelShader = pixelShader; + dirtyUniforms = true; + } + + if (dirtyUniforms) + { + programBinary->dirtyAllUniforms(); + } + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer11::applyUniforms(const gl::ProgramBinary &programBinary) +{ + const std::vector<gl::LinkedUniform*> &uniformArray = programBinary.getUniforms(); + + unsigned int totalRegisterCountVS = 0; + unsigned int totalRegisterCountPS = 0; + + bool vertexUniformsDirty = false; + bool pixelUniformsDirty = false; + + for (size_t uniformIndex = 0; uniformIndex < uniformArray.size(); uniformIndex++) + { + const gl::LinkedUniform &uniform = *uniformArray[uniformIndex]; + + if (uniform.isReferencedByVertexShader() && !uniform.isSampler()) + { + totalRegisterCountVS += uniform.registerCount; + vertexUniformsDirty = (vertexUniformsDirty || uniform.dirty); + } + + if (uniform.isReferencedByFragmentShader() && !uniform.isSampler()) + { + totalRegisterCountPS += uniform.registerCount; + pixelUniformsDirty = (pixelUniformsDirty || uniform.dirty); + } + } + + const ProgramD3D *programD3D = ProgramD3D::makeProgramD3D(programBinary.getImplementation()); + const UniformStorage11 *vertexUniformStorage = UniformStorage11::makeUniformStorage11(&programD3D->getVertexUniformStorage()); + const UniformStorage11 *fragmentUniformStorage = UniformStorage11::makeUniformStorage11(&programD3D->getFragmentUniformStorage()); + ASSERT(vertexUniformStorage); + ASSERT(fragmentUniformStorage); + + ID3D11Buffer *vertexConstantBuffer = vertexUniformStorage->getConstantBuffer(); + ID3D11Buffer *pixelConstantBuffer = fragmentUniformStorage->getConstantBuffer(); + + float (*mapVS)[4] = NULL; + float (*mapPS)[4] = NULL; + + if (totalRegisterCountVS > 0 && vertexUniformsDirty) + { + D3D11_MAPPED_SUBRESOURCE map = {0}; + HRESULT result = mDeviceContext->Map(vertexConstantBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &map); + UNUSED_ASSERTION_VARIABLE(result); + ASSERT(SUCCEEDED(result)); + mapVS = (float(*)[4])map.pData; + } + + if (totalRegisterCountPS > 0 && pixelUniformsDirty) + { + D3D11_MAPPED_SUBRESOURCE map = {0}; + HRESULT result = mDeviceContext->Map(pixelConstantBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &map); + UNUSED_ASSERTION_VARIABLE(result); + ASSERT(SUCCEEDED(result)); + mapPS = (float(*)[4])map.pData; + } + + for (size_t uniformIndex = 0; uniformIndex < uniformArray.size(); uniformIndex++) + { + gl::LinkedUniform *uniform = uniformArray[uniformIndex]; + + if (!uniform->isSampler()) + { + unsigned int componentCount = (4 - uniform->registerElement); + + // we assume that uniforms from structs are arranged in struct order in our uniforms list. otherwise we would + // overwrite previously written regions of memory. + + if (uniform->isReferencedByVertexShader() && mapVS) + { + memcpy(&mapVS[uniform->vsRegisterIndex][uniform->registerElement], uniform->data, uniform->registerCount * sizeof(float) * componentCount); + } + + if (uniform->isReferencedByFragmentShader() && mapPS) + { + memcpy(&mapPS[uniform->psRegisterIndex][uniform->registerElement], uniform->data, uniform->registerCount * sizeof(float) * componentCount); + } + } + } + + if (mapVS) + { + mDeviceContext->Unmap(vertexConstantBuffer, 0); + } + + if (mapPS) + { + mDeviceContext->Unmap(pixelConstantBuffer, 0); + } + + if (mCurrentVertexConstantBuffer != vertexConstantBuffer) + { + mDeviceContext->VSSetConstantBuffers(0, 1, &vertexConstantBuffer); + mCurrentVertexConstantBuffer = vertexConstantBuffer; + } + + if (mCurrentPixelConstantBuffer != pixelConstantBuffer) + { + mDeviceContext->PSSetConstantBuffers(0, 1, &pixelConstantBuffer); + mCurrentPixelConstantBuffer = pixelConstantBuffer; + } + + // Driver uniforms + if (!mDriverConstantBufferVS) + { + D3D11_BUFFER_DESC constantBufferDescription = {0}; + constantBufferDescription.ByteWidth = sizeof(dx_VertexConstants); + constantBufferDescription.Usage = D3D11_USAGE_DEFAULT; + constantBufferDescription.BindFlags = D3D11_BIND_CONSTANT_BUFFER; + constantBufferDescription.CPUAccessFlags = 0; + constantBufferDescription.MiscFlags = 0; + constantBufferDescription.StructureByteStride = 0; + + HRESULT result = mDevice->CreateBuffer(&constantBufferDescription, NULL, &mDriverConstantBufferVS); + UNUSED_ASSERTION_VARIABLE(result); + ASSERT(SUCCEEDED(result)); + + mDeviceContext->VSSetConstantBuffers(1, 1, &mDriverConstantBufferVS); + } + + if (!mDriverConstantBufferPS) + { + D3D11_BUFFER_DESC constantBufferDescription = {0}; + constantBufferDescription.ByteWidth = sizeof(dx_PixelConstants); + constantBufferDescription.Usage = D3D11_USAGE_DEFAULT; + constantBufferDescription.BindFlags = D3D11_BIND_CONSTANT_BUFFER; + constantBufferDescription.CPUAccessFlags = 0; + constantBufferDescription.MiscFlags = 0; + constantBufferDescription.StructureByteStride = 0; + + HRESULT result = mDevice->CreateBuffer(&constantBufferDescription, NULL, &mDriverConstantBufferPS); + UNUSED_ASSERTION_VARIABLE(result); + ASSERT(SUCCEEDED(result)); + + mDeviceContext->PSSetConstantBuffers(1, 1, &mDriverConstantBufferPS); + } + + if (memcmp(&mVertexConstants, &mAppliedVertexConstants, sizeof(dx_VertexConstants)) != 0) + { + mDeviceContext->UpdateSubresource(mDriverConstantBufferVS, 0, NULL, &mVertexConstants, 16, 0); + memcpy(&mAppliedVertexConstants, &mVertexConstants, sizeof(dx_VertexConstants)); + } + + if (memcmp(&mPixelConstants, &mAppliedPixelConstants, sizeof(dx_PixelConstants)) != 0) + { + mDeviceContext->UpdateSubresource(mDriverConstantBufferPS, 0, NULL, &mPixelConstants, 16, 0); + memcpy(&mAppliedPixelConstants, &mPixelConstants, sizeof(dx_PixelConstants)); + } + + // needed for the point sprite geometry shader + if (mFeatureLevel >= D3D_FEATURE_LEVEL_10_0 && mCurrentGeometryConstantBuffer != mDriverConstantBufferPS) + { + mDeviceContext->GSSetConstantBuffers(0, 1, &mDriverConstantBufferPS); + mCurrentGeometryConstantBuffer = mDriverConstantBufferPS; + } + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer11::clear(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer) +{ + gl::Error error = mClear->clearFramebuffer(clearParams, frameBuffer); + if (error.isError()) + { + return error; + } + + invalidateFramebufferSwizzles(frameBuffer); + + return gl::Error(GL_NO_ERROR); +} + +void Renderer11::markAllStateDirty() +{ + for (unsigned int rtIndex = 0; rtIndex < gl::IMPLEMENTATION_MAX_DRAW_BUFFERS; rtIndex++) + { + mAppliedRenderTargetSerials[rtIndex] = 0; + } + mAppliedDepthbufferSerial = 0; + mAppliedStencilbufferSerial = 0; + mDepthStencilInitialized = false; + mRenderTargetDescInitialized = false; + + ASSERT(mForceSetVertexSamplerStates.size() == mCurVertexSRVs.size()); + for (size_t vsamplerId = 0; vsamplerId < mForceSetVertexSamplerStates.size(); ++vsamplerId) + { + mForceSetVertexSamplerStates[vsamplerId] = true; + mCurVertexSRVs[vsamplerId] = NULL; + } + + ASSERT(mForceSetPixelSamplerStates.size() == mCurPixelSRVs.size()); + for (size_t fsamplerId = 0; fsamplerId < mForceSetPixelSamplerStates.size(); ++fsamplerId) + { + mForceSetPixelSamplerStates[fsamplerId] = true; + mCurPixelSRVs[fsamplerId] = NULL; + } + + mForceSetBlendState = true; + mForceSetRasterState = true; + mForceSetDepthStencilState = true; + mForceSetScissor = true; + mForceSetViewport = true; + + mAppliedIB = NULL; + mAppliedIBFormat = DXGI_FORMAT_UNKNOWN; + mAppliedIBOffset = 0; + + mAppliedVertexShader = NULL; + mAppliedGeometryShader = NULL; + mCurPointGeometryShader = NULL; + mAppliedPixelShader = NULL; + + for (size_t i = 0; i < gl::IMPLEMENTATION_MAX_TRANSFORM_FEEDBACK_BUFFERS; i++) + { + mAppliedTFBuffers[i] = NULL; + mAppliedTFOffsets[i] = 0; + } + + memset(&mAppliedVertexConstants, 0, sizeof(dx_VertexConstants)); + memset(&mAppliedPixelConstants, 0, sizeof(dx_PixelConstants)); + + mInputLayoutCache.markDirty(); + + for (unsigned int i = 0; i < gl::IMPLEMENTATION_MAX_VERTEX_SHADER_UNIFORM_BUFFERS; i++) + { + mCurrentConstantBufferVS[i] = -1; + mCurrentConstantBufferPS[i] = -1; + } + + mCurrentVertexConstantBuffer = NULL; + mCurrentPixelConstantBuffer = NULL; + mCurrentGeometryConstantBuffer = NULL; + + mCurrentPrimitiveTopology = D3D_PRIMITIVE_TOPOLOGY_UNDEFINED; +} + +void Renderer11::releaseDeviceResources() +{ + mStateCache.clear(); + mInputLayoutCache.clear(); + + SafeDelete(mVertexDataManager); + SafeDelete(mIndexDataManager); + SafeDelete(mLineLoopIB); + SafeDelete(mTriangleFanIB); + SafeDelete(mBlit); + SafeDelete(mClear); + SafeDelete(mPixelTransfer); + + SafeRelease(mDriverConstantBufferVS); + SafeRelease(mDriverConstantBufferPS); + SafeRelease(mSyncQuery); +} + +void Renderer11::notifyDeviceLost() +{ + mDeviceLost = true; + mDisplay->notifyDeviceLost(); +} + +bool Renderer11::isDeviceLost() +{ + return mDeviceLost; +} + +// set notify to true to broadcast a message to all contexts of the device loss +bool Renderer11::testDeviceLost(bool notify) +{ + bool isLost = false; + + // GetRemovedReason is used to test if the device is removed + HRESULT result = mDevice->GetDeviceRemovedReason(); + isLost = d3d11::isDeviceLostError(result); + + if (isLost) + { + // Log error if this is a new device lost event + if (mDeviceLost == false) + { + ERR("The D3D11 device was removed: 0x%08X", result); + } + + // ensure we note the device loss -- + // we'll probably get this done again by notifyDeviceLost + // but best to remember it! + // Note that we don't want to clear the device loss status here + // -- this needs to be done by resetDevice + mDeviceLost = true; + if (notify) + { + notifyDeviceLost(); + } + } + + return isLost; +} + +bool Renderer11::testDeviceResettable() +{ + // determine if the device is resettable by creating a dummy device + PFN_D3D11_CREATE_DEVICE D3D11CreateDevice = (PFN_D3D11_CREATE_DEVICE)GetProcAddress(mD3d11Module, "D3D11CreateDevice"); + + if (D3D11CreateDevice == NULL) + { + return false; + } + + D3D_FEATURE_LEVEL featureLevels[] = + { + D3D_FEATURE_LEVEL_11_0, + D3D_FEATURE_LEVEL_10_1, + D3D_FEATURE_LEVEL_10_0, +#if !defined(ANGLE_ENABLE_D3D9) + D3D_FEATURE_LEVEL_9_3, + D3D_FEATURE_LEVEL_9_2, + D3D_FEATURE_LEVEL_9_1, +#endif + }; + + ID3D11Device* dummyDevice; + D3D_FEATURE_LEVEL dummyFeatureLevel; + ID3D11DeviceContext* dummyContext; + + HRESULT result = D3D11CreateDevice(NULL, + D3D_DRIVER_TYPE_HARDWARE, + NULL, + #if defined(_DEBUG) + D3D11_CREATE_DEVICE_DEBUG, + #else + 0, + #endif + featureLevels, + ArraySize(featureLevels), + D3D11_SDK_VERSION, + &dummyDevice, + &dummyFeatureLevel, + &dummyContext); + + if (!mDevice || FAILED(result)) + { + return false; + } + + SafeRelease(dummyContext); + SafeRelease(dummyDevice); + + return true; +} + +void Renderer11::release() +{ + releaseShaderCompiler(); + releaseDeviceResources(); + + SafeRelease(mDxgiFactory); + SafeRelease(mDxgiAdapter); + + if (mDeviceContext) + { + mDeviceContext->ClearState(); + mDeviceContext->Flush(); + SafeRelease(mDeviceContext); + } + + SafeRelease(mDevice); + + if (mD3d11Module) + { + FreeLibrary(mD3d11Module); + mD3d11Module = NULL; + } + + if (mDxgiModule) + { + FreeLibrary(mDxgiModule); + mDxgiModule = NULL; + } + + mCompiler.release(); +} + +bool Renderer11::resetDevice() +{ + // recreate everything + release(); + EGLint result = initialize(); + + if (result != EGL_SUCCESS) + { + ERR("Could not reinitialize D3D11 device: %08X", result); + return false; + } + + mDeviceLost = false; + + return true; +} + +DWORD Renderer11::getAdapterVendor() const +{ + return mAdapterDescription.VendorId; +} + +std::string Renderer11::getRendererDescription() const +{ + std::ostringstream rendererString; + + rendererString << mDescription; + rendererString << " Direct3D11"; + + rendererString << " vs_" << getMajorShaderModel() << "_" << getMinorShaderModel(); + rendererString << " ps_" << getMajorShaderModel() << "_" << getMinorShaderModel(); + + return rendererString.str(); +} + +GUID Renderer11::getAdapterIdentifier() const +{ + // Use the adapter LUID as our adapter ID + // This number is local to a machine is only guaranteed to be unique between restarts + META_ASSERT(sizeof(LUID) <= sizeof(GUID)); + GUID adapterId = {0}; + memcpy(&adapterId, &mAdapterDescription.AdapterLuid, sizeof(LUID)); + return adapterId; +} + +unsigned int Renderer11::getReservedVertexUniformVectors() const +{ + return 0; // Driver uniforms are stored in a separate constant buffer +} + +unsigned int Renderer11::getReservedFragmentUniformVectors() const +{ + return 0; // Driver uniforms are stored in a separate constant buffer +} + +unsigned int Renderer11::getReservedVertexUniformBuffers() const +{ + // we reserve one buffer for the application uniforms, and one for driver uniforms + return 2; +} + +unsigned int Renderer11::getReservedFragmentUniformBuffers() const +{ + // we reserve one buffer for the application uniforms, and one for driver uniforms + return 2; +} + +bool Renderer11::getShareHandleSupport() const +{ + // We only currently support share handles with BGRA surfaces, because + // chrome needs BGRA. Once chrome fixes this, we should always support them. + // PIX doesn't seem to support using share handles, so disable them. + return getRendererExtensions().textureFormatBGRA8888 && !gl::perfActive(); +} + +bool Renderer11::getPostSubBufferSupport() const +{ + // D3D11 does not support present with dirty rectangles until D3D11.1 and DXGI 1.2. + return false; +} + +int Renderer11::getMajorShaderModel() const +{ + switch (mFeatureLevel) + { + case D3D_FEATURE_LEVEL_11_0: return D3D11_SHADER_MAJOR_VERSION; // 5 + case D3D_FEATURE_LEVEL_10_1: return D3D10_1_SHADER_MAJOR_VERSION; // 4 + case D3D_FEATURE_LEVEL_10_0: + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return D3D10_SHADER_MAJOR_VERSION; // 4 + default: UNREACHABLE(); return 0; + } +} + +int Renderer11::getMinorShaderModel() const +{ + switch (mFeatureLevel) + { + case D3D_FEATURE_LEVEL_11_0: return D3D11_SHADER_MINOR_VERSION; // 0 + case D3D_FEATURE_LEVEL_10_1: return D3D10_1_SHADER_MINOR_VERSION; // 1 + case D3D_FEATURE_LEVEL_10_0: + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return D3D10_SHADER_MINOR_VERSION; // 0 + default: UNREACHABLE(); return 0; + } +} + +int Renderer11::getMinSwapInterval() const +{ + return 0; +} + +int Renderer11::getMaxSwapInterval() const +{ + return 4; +} + +bool Renderer11::copyToRenderTarget2D(TextureStorage *dest, TextureStorage *source) +{ + if (source && dest) + { + TextureStorage11_2D *source11 = TextureStorage11_2D::makeTextureStorage11_2D(source); + TextureStorage11_2D *dest11 = TextureStorage11_2D::makeTextureStorage11_2D(dest); + + mDeviceContext->CopyResource(dest11->getResource(), source11->getResource()); + + dest11->invalidateSwizzleCache(); + + return true; + } + + return false; +} + +bool Renderer11::copyToRenderTargetCube(TextureStorage *dest, TextureStorage *source) +{ + if (source && dest) + { + TextureStorage11_Cube *source11 = TextureStorage11_Cube::makeTextureStorage11_Cube(source); + TextureStorage11_Cube *dest11 = TextureStorage11_Cube::makeTextureStorage11_Cube(dest); + + mDeviceContext->CopyResource(dest11->getResource(), source11->getResource()); + + dest11->invalidateSwizzleCache(); + + return true; + } + + return false; +} + +bool Renderer11::copyToRenderTarget3D(TextureStorage *dest, TextureStorage *source) +{ + if (source && dest) + { + TextureStorage11_3D *source11 = TextureStorage11_3D::makeTextureStorage11_3D(source); + TextureStorage11_3D *dest11 = TextureStorage11_3D::makeTextureStorage11_3D(dest); + + mDeviceContext->CopyResource(dest11->getResource(), source11->getResource()); + + dest11->invalidateSwizzleCache(); + + return true; + } + + return false; +} + +bool Renderer11::copyToRenderTarget2DArray(TextureStorage *dest, TextureStorage *source) +{ + if (source && dest) + { + TextureStorage11_2DArray *source11 = TextureStorage11_2DArray::makeTextureStorage11_2DArray(source); + TextureStorage11_2DArray *dest11 = TextureStorage11_2DArray::makeTextureStorage11_2DArray(dest); + + mDeviceContext->CopyResource(dest11->getResource(), source11->getResource()); + + dest11->invalidateSwizzleCache(); + + return true; + } + + return false; +} + +bool Renderer11::copyImage2D(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, TextureStorage *storage, GLint level) +{ + gl::FramebufferAttachment *colorbuffer = framebuffer->getReadColorbuffer(); + if (!colorbuffer) + { + ERR("Failed to retrieve the color buffer from the frame buffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + RenderTarget11 *sourceRenderTarget = d3d11::GetAttachmentRenderTarget(colorbuffer); + if (!sourceRenderTarget) + { + ERR("Failed to retrieve the render target from the frame buffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + ID3D11ShaderResourceView *source = sourceRenderTarget->getShaderResourceView(); + if (!source) + { + ERR("Failed to retrieve the render target view from the render target."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + TextureStorage11_2D *storage11 = TextureStorage11_2D::makeTextureStorage11_2D(storage); + if (!storage11) + { + ERR("Failed to retrieve the texture storage from the destination."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + gl::ImageIndex index = gl::ImageIndex::Make2D(level); + RenderTarget11 *destRenderTarget = RenderTarget11::makeRenderTarget11(storage11->getRenderTarget(index)); + if (!destRenderTarget) + { + ERR("Failed to retrieve the render target from the destination storage."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + ID3D11RenderTargetView *dest = destRenderTarget->getRenderTargetView(); + if (!dest) + { + ERR("Failed to retrieve the render target view from the destination render target."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + gl::Box sourceArea(sourceRect.x, sourceRect.y, 0, sourceRect.width, sourceRect.height, 1); + gl::Extents sourceSize(sourceRenderTarget->getWidth(), sourceRenderTarget->getHeight(), 1); + + gl::Box destArea(xoffset, yoffset, 0, sourceRect.width, sourceRect.height, 1); + gl::Extents destSize(destRenderTarget->getWidth(), destRenderTarget->getHeight(), 1); + + // Use nearest filtering because source and destination are the same size for the direct + // copy + bool ret = mBlit->copyTexture(source, sourceArea, sourceSize, dest, destArea, destSize, NULL, + destFormat, GL_NEAREST); + + storage11->invalidateSwizzleCacheLevel(level); + + return ret; +} + +bool Renderer11::copyImageCube(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, TextureStorage *storage, GLenum target, GLint level) +{ + gl::FramebufferAttachment *colorbuffer = framebuffer->getReadColorbuffer(); + if (!colorbuffer) + { + ERR("Failed to retrieve the color buffer from the frame buffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + RenderTarget11 *sourceRenderTarget = d3d11::GetAttachmentRenderTarget(colorbuffer); + if (!sourceRenderTarget) + { + ERR("Failed to retrieve the render target from the frame buffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + ID3D11ShaderResourceView *source = sourceRenderTarget->getShaderResourceView(); + if (!source) + { + ERR("Failed to retrieve the render target view from the render target."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + TextureStorage11_Cube *storage11 = TextureStorage11_Cube::makeTextureStorage11_Cube(storage); + if (!storage11) + { + ERR("Failed to retrieve the texture storage from the destination."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + gl::ImageIndex index = gl::ImageIndex::MakeCube(target, level); + RenderTarget11 *destRenderTarget = RenderTarget11::makeRenderTarget11(storage11->getRenderTarget(index)); + if (!destRenderTarget) + { + ERR("Failed to retrieve the render target from the destination storage."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + ID3D11RenderTargetView *dest = destRenderTarget->getRenderTargetView(); + if (!dest) + { + ERR("Failed to retrieve the render target view from the destination render target."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + gl::Box sourceArea(sourceRect.x, sourceRect.y, 0, sourceRect.width, sourceRect.height, 1); + gl::Extents sourceSize(sourceRenderTarget->getWidth(), sourceRenderTarget->getHeight(), 1); + + gl::Box destArea(xoffset, yoffset, 0, sourceRect.width, sourceRect.height, 1); + gl::Extents destSize(destRenderTarget->getWidth(), destRenderTarget->getHeight(), 1); + + // Use nearest filtering because source and destination are the same size for the direct + // copy + bool ret = mBlit->copyTexture(source, sourceArea, sourceSize, dest, destArea, destSize, NULL, + destFormat, GL_NEAREST); + + storage11->invalidateSwizzleCacheLevel(level); + + return ret; +} + +bool Renderer11::copyImage3D(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, GLint zOffset, TextureStorage *storage, GLint level) +{ + gl::FramebufferAttachment *colorbuffer = framebuffer->getReadColorbuffer(); + if (!colorbuffer) + { + ERR("Failed to retrieve the color buffer from the frame buffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + RenderTarget11 *sourceRenderTarget = d3d11::GetAttachmentRenderTarget(colorbuffer); + if (!sourceRenderTarget) + { + ERR("Failed to retrieve the render target from the frame buffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + ID3D11ShaderResourceView *source = sourceRenderTarget->getShaderResourceView(); + if (!source) + { + ERR("Failed to retrieve the render target view from the render target."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + TextureStorage11_3D *storage11 = TextureStorage11_3D::makeTextureStorage11_3D(storage); + if (!storage11) + { + ERR("Failed to retrieve the texture storage from the destination."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + gl::ImageIndex index = gl::ImageIndex::Make3D(level, zOffset); + RenderTarget11 *destRenderTarget = RenderTarget11::makeRenderTarget11(storage11->getRenderTarget(index)); + if (!destRenderTarget) + { + ERR("Failed to retrieve the render target from the destination storage."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + ID3D11RenderTargetView *dest = destRenderTarget->getRenderTargetView(); + if (!dest) + { + ERR("Failed to retrieve the render target view from the destination render target."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + gl::Box sourceArea(sourceRect.x, sourceRect.y, 0, sourceRect.width, sourceRect.height, 1); + gl::Extents sourceSize(sourceRenderTarget->getWidth(), sourceRenderTarget->getHeight(), 1); + + gl::Box destArea(xoffset, yoffset, 0, sourceRect.width, sourceRect.height, 1); + gl::Extents destSize(destRenderTarget->getWidth(), destRenderTarget->getHeight(), 1); + + // Use nearest filtering because source and destination are the same size for the direct + // copy + bool ret = mBlit->copyTexture(source, sourceArea, sourceSize, dest, destArea, destSize, NULL, + destFormat, GL_NEAREST); + + storage11->invalidateSwizzleCacheLevel(level); + + return ret; +} + +bool Renderer11::copyImage2DArray(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, GLint zOffset, TextureStorage *storage, GLint level) +{ + gl::FramebufferAttachment *colorbuffer = framebuffer->getReadColorbuffer(); + if (!colorbuffer) + { + ERR("Failed to retrieve the color buffer from the frame buffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + RenderTarget11 *sourceRenderTarget = d3d11::GetAttachmentRenderTarget(colorbuffer); + if (!sourceRenderTarget) + { + ERR("Failed to retrieve the render target from the frame buffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + ID3D11ShaderResourceView *source = sourceRenderTarget->getShaderResourceView(); + if (!source) + { + ERR("Failed to retrieve the render target view from the render target."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + TextureStorage11_2DArray *storage11 = TextureStorage11_2DArray::makeTextureStorage11_2DArray(storage); + if (!storage11) + { + SafeRelease(source); + ERR("Failed to retrieve the texture storage from the destination."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + gl::ImageIndex index = gl::ImageIndex::Make2DArray(level, zOffset); + RenderTarget11 *destRenderTarget = RenderTarget11::makeRenderTarget11(storage11->getRenderTarget(index)); + if (!destRenderTarget) + { + SafeRelease(source); + ERR("Failed to retrieve the render target from the destination storage."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + ID3D11RenderTargetView *dest = destRenderTarget->getRenderTargetView(); + if (!dest) + { + ERR("Failed to retrieve the render target view from the destination render target."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + gl::Box sourceArea(sourceRect.x, sourceRect.y, 0, sourceRect.width, sourceRect.height, 1); + gl::Extents sourceSize(sourceRenderTarget->getWidth(), sourceRenderTarget->getHeight(), 1); + + gl::Box destArea(xoffset, yoffset, 0, sourceRect.width, sourceRect.height, 1); + gl::Extents destSize(destRenderTarget->getWidth(), destRenderTarget->getHeight(), 1); + + // Use nearest filtering because source and destination are the same size for the direct + // copy + bool ret = mBlit->copyTexture(source, sourceArea, sourceSize, dest, destArea, destSize, NULL, + destFormat, GL_NEAREST); + + storage11->invalidateSwizzleCacheLevel(level); + + return ret; +} + +void Renderer11::unapplyRenderTargets() +{ + setOneTimeRenderTarget(NULL); +} + +void Renderer11::setOneTimeRenderTarget(ID3D11RenderTargetView *renderTargetView) +{ + ID3D11RenderTargetView *rtvArray[gl::IMPLEMENTATION_MAX_DRAW_BUFFERS] = {NULL}; + + rtvArray[0] = renderTargetView; + + mDeviceContext->OMSetRenderTargets(getRendererCaps().maxDrawBuffers, rtvArray, NULL); + + // Do not preserve the serial for this one-time-use render target + for (unsigned int rtIndex = 0; rtIndex < gl::IMPLEMENTATION_MAX_DRAW_BUFFERS; rtIndex++) + { + mAppliedRenderTargetSerials[rtIndex] = 0; + } +} + +RenderTarget *Renderer11::createRenderTarget(SwapChain *swapChain, bool depth) +{ + SwapChain11 *swapChain11 = SwapChain11::makeSwapChain11(swapChain); + RenderTarget11 *renderTarget = NULL; + + if (depth) + { + // Note: depth stencil may be NULL for 0 sized surfaces + renderTarget = new RenderTarget11(this, swapChain11->getDepthStencil(), + swapChain11->getDepthStencilTexture(), + swapChain11->getDepthStencilShaderResource(), + swapChain11->getWidth(), swapChain11->getHeight(), 1); + } + else + { + // Note: render target may be NULL for 0 sized surfaces + renderTarget = new RenderTarget11(this, swapChain11->getRenderTarget(), + swapChain11->getOffscreenTexture(), + swapChain11->getRenderTargetShaderResource(), + swapChain11->getWidth(), swapChain11->getHeight(), 1); + } + return renderTarget; +} + +RenderTarget *Renderer11::createRenderTarget(int width, int height, GLenum format, GLsizei samples) +{ + RenderTarget11 *renderTarget = new RenderTarget11(this, width, height, format, samples); + return renderTarget; +} + +ShaderImpl *Renderer11::createShader(GLenum type) +{ + return new ShaderD3D(type, this); +} + +ProgramImpl *Renderer11::createProgram() +{ + return new ProgramD3D(this); +} + +void Renderer11::releaseShaderCompiler() +{ + ShaderD3D::releaseCompiler(); +} + +ShaderExecutable *Renderer11::loadExecutable(const void *function, size_t length, rx::ShaderType type, + const std::vector<gl::LinkedVarying> &transformFeedbackVaryings, + bool separatedOutputBuffers) +{ + ShaderExecutable11 *executable = NULL; + HRESULT result; + + switch (type) + { + case rx::SHADER_VERTEX: + { + ID3D11VertexShader *vertexShader = NULL; + ID3D11GeometryShader *streamOutShader = NULL; + + result = mDevice->CreateVertexShader(function, length, NULL, &vertexShader); + ASSERT(SUCCEEDED(result)); + + if (transformFeedbackVaryings.size() > 0) + { + std::vector<D3D11_SO_DECLARATION_ENTRY> soDeclaration; + for (size_t i = 0; i < transformFeedbackVaryings.size(); i++) + { + const gl::LinkedVarying &varying = transformFeedbackVaryings[i]; + GLenum transposedType = gl::TransposeMatrixType(varying.type); + + for (size_t j = 0; j < varying.semanticIndexCount; j++) + { + D3D11_SO_DECLARATION_ENTRY entry = { 0 }; + entry.Stream = 0; + entry.SemanticName = varying.semanticName.c_str(); + entry.SemanticIndex = varying.semanticIndex + j; + entry.StartComponent = 0; + entry.ComponentCount = gl::VariableColumnCount(transposedType); + entry.OutputSlot = (separatedOutputBuffers ? i : 0); + soDeclaration.push_back(entry); + } + } + + result = mDevice->CreateGeometryShaderWithStreamOutput(function, length, soDeclaration.data(), soDeclaration.size(), + NULL, 0, 0, NULL, &streamOutShader); + ASSERT(SUCCEEDED(result)); + } + + if (vertexShader) + { + executable = new ShaderExecutable11(function, length, vertexShader, streamOutShader); + } + } + break; + case rx::SHADER_PIXEL: + { + ID3D11PixelShader *pixelShader = NULL; + + result = mDevice->CreatePixelShader(function, length, NULL, &pixelShader); + ASSERT(SUCCEEDED(result)); + + if (pixelShader) + { + executable = new ShaderExecutable11(function, length, pixelShader); + } + } + break; + case rx::SHADER_GEOMETRY: + { + ID3D11GeometryShader *geometryShader = NULL; + + result = mDevice->CreateGeometryShader(function, length, NULL, &geometryShader); + ASSERT(SUCCEEDED(result)); + + if (geometryShader) + { + executable = new ShaderExecutable11(function, length, geometryShader); + } + } + break; + default: + UNREACHABLE(); + break; + } + + return executable; +} + +ShaderExecutable *Renderer11::compileToExecutable(gl::InfoLog &infoLog, const char *shaderHLSL, rx::ShaderType type, + const std::vector<gl::LinkedVarying> &transformFeedbackVaryings, + bool separatedOutputBuffers, D3DWorkaroundType workaround) +{ + const char *profileType = NULL; + switch (type) + { + case rx::SHADER_VERTEX: + profileType = "vs"; + break; + case rx::SHADER_PIXEL: + profileType = "ps"; + break; + case rx::SHADER_GEOMETRY: + profileType = "gs"; + break; + default: + UNREACHABLE(); + return NULL; + } + + const char *profileVersion = NULL; + switch (mFeatureLevel) + { + case D3D_FEATURE_LEVEL_11_0: + profileVersion = "5_0"; + break; + case D3D_FEATURE_LEVEL_10_1: + profileVersion = "4_1"; + break; + case D3D_FEATURE_LEVEL_10_0: + profileVersion = "4_0"; + break; + case D3D_FEATURE_LEVEL_9_3: + profileVersion = "4_0_level_9_3"; + break; + case D3D_FEATURE_LEVEL_9_2: + profileVersion = "4_0_level_9_2"; + break; + case D3D_FEATURE_LEVEL_9_1: + profileVersion = "4_0_level_9_1"; + break; + default: + UNREACHABLE(); + return NULL; + } + + char profile[32]; + snprintf(profile, ArraySize(profile), "%s_%s", profileType, profileVersion); + + UINT flags = D3DCOMPILE_OPTIMIZATION_LEVEL0; + + if (gl::perfActive()) + { +#ifndef NDEBUG + flags = D3DCOMPILE_SKIP_OPTIMIZATION; +#endif + + flags |= D3DCOMPILE_DEBUG; + + std::string sourcePath = getTempPath(); + std::string sourceText = std::string("#line 2 \"") + sourcePath + std::string("\"\n\n") + std::string(shaderHLSL); + writeFile(sourcePath.c_str(), sourceText.c_str(), sourceText.size()); + } + + // Sometimes D3DCompile will fail with the default compilation flags for complicated shaders when it would otherwise pass with alternative options. + // Try the default flags first and if compilation fails, try some alternatives. + const UINT extraFlags[] = + { + flags, + flags | D3DCOMPILE_SKIP_VALIDATION, + flags | D3DCOMPILE_SKIP_OPTIMIZATION + }; + + const static char *extraFlagNames[] = + { + "default", + "skip validation", + "skip optimization" + }; + + int attempts = ArraySize(extraFlags); + + ID3DBlob *binary = (ID3DBlob*)mCompiler.compileToBinary(infoLog, shaderHLSL, profile, extraFlags, extraFlagNames, attempts); + if (!binary) + { + return NULL; + } + + ShaderExecutable *executable = loadExecutable((DWORD *)binary->GetBufferPointer(), binary->GetBufferSize(), type, + transformFeedbackVaryings, separatedOutputBuffers); + SafeRelease(binary); + + return executable; +} + +rx::UniformStorage *Renderer11::createUniformStorage(size_t storageSize) +{ + return new UniformStorage11(this, storageSize); +} + +VertexBuffer *Renderer11::createVertexBuffer() +{ + return new VertexBuffer11(this); +} + +IndexBuffer *Renderer11::createIndexBuffer() +{ + return new IndexBuffer11(this); +} + +BufferImpl *Renderer11::createBuffer() +{ + return new Buffer11(this); +} + +VertexArrayImpl *Renderer11::createVertexArray() +{ + return new VertexArray11(this); +} + +QueryImpl *Renderer11::createQuery(GLenum type) +{ + return new Query11(this, type); +} + +FenceImpl *Renderer11::createFence() +{ + return new Fence11(this); +} + +TransformFeedbackImpl* Renderer11::createTransformFeedback() +{ + return new TransformFeedbackD3D(); +} + +bool Renderer11::supportsFastCopyBufferToTexture(GLenum internalFormat) const +{ + ASSERT(getRendererExtensions().pixelBufferObject); + + const gl::InternalFormat &internalFormatInfo = gl::GetInternalFormatInfo(internalFormat); + const d3d11::TextureFormat &d3d11FormatInfo = d3d11::GetTextureFormatInfo(internalFormat); + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(d3d11FormatInfo.texFormat); + + // sRGB formats do not work with D3D11 buffer SRVs + if (internalFormatInfo.colorEncoding == GL_SRGB) + { + return false; + } + + // We cannot support direct copies to non-color-renderable formats + if (d3d11FormatInfo.rtvFormat != DXGI_FORMAT_UNKNOWN) + { + return false; + } + + // We skip all 3-channel formats since sometimes format support is missing + if (internalFormatInfo.componentCount == 3) + { + return false; + } + + // We don't support formats which we can't represent without conversion + if (dxgiFormatInfo.internalFormat != internalFormat) + { + return false; + } + + return true; +} + +bool Renderer11::fastCopyBufferToTexture(const gl::PixelUnpackState &unpack, unsigned int offset, RenderTarget *destRenderTarget, + GLenum destinationFormat, GLenum sourcePixelsType, const gl::Box &destArea) +{ + ASSERT(supportsFastCopyBufferToTexture(destinationFormat)); + return mPixelTransfer->copyBufferToTexture(unpack, offset, destRenderTarget, destinationFormat, sourcePixelsType, destArea); +} + +bool Renderer11::getRenderTargetResource(gl::FramebufferAttachment *colorbuffer, unsigned int *subresourceIndex, ID3D11Texture2D **resource) +{ + ASSERT(colorbuffer != NULL); + + RenderTarget11 *renderTarget = d3d11::GetAttachmentRenderTarget(colorbuffer); + if (renderTarget) + { + *subresourceIndex = renderTarget->getSubresourceIndex(); + + ID3D11RenderTargetView *colorBufferRTV = renderTarget->getRenderTargetView(); + if (colorBufferRTV) + { + ID3D11Resource *textureResource = NULL; + colorBufferRTV->GetResource(&textureResource); + + if (textureResource) + { + HRESULT result = textureResource->QueryInterface(__uuidof(ID3D11Texture2D), (void**)resource); + SafeRelease(textureResource); + + if (SUCCEEDED(result)) + { + return true; + } + else + { + ERR("Failed to extract the ID3D11Texture2D from the render target resource, " + "HRESULT: 0x%X.", result); + } + } + } + } + + return false; +} + +bool Renderer11::blitRect(gl::Framebuffer *readTarget, const gl::Rectangle &readRect, gl::Framebuffer *drawTarget, const gl::Rectangle &drawRect, + const gl::Rectangle *scissor, bool blitRenderTarget, bool blitDepth, bool blitStencil, GLenum filter) +{ + if (blitRenderTarget) + { + gl::FramebufferAttachment *readBuffer = readTarget->getReadColorbuffer(); + + if (!readBuffer) + { + ERR("Failed to retrieve the read buffer from the read framebuffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + RenderTarget *readRenderTarget = GetAttachmentRenderTarget(readBuffer); + + for (unsigned int colorAttachment = 0; colorAttachment < gl::IMPLEMENTATION_MAX_DRAW_BUFFERS; colorAttachment++) + { + if (drawTarget->isEnabledColorAttachment(colorAttachment)) + { + gl::FramebufferAttachment *drawBuffer = drawTarget->getColorbuffer(colorAttachment); + + if (!drawBuffer) + { + ERR("Failed to retrieve the draw buffer from the draw framebuffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + RenderTarget *drawRenderTarget = GetAttachmentRenderTarget(drawBuffer); + + if (!blitRenderbufferRect(readRect, drawRect, readRenderTarget, drawRenderTarget, filter, scissor, + blitRenderTarget, false, false)) + { + return false; + } + } + } + } + + if (blitDepth || blitStencil) + { + gl::FramebufferAttachment *readBuffer = readTarget->getDepthOrStencilbuffer(); + gl::FramebufferAttachment *drawBuffer = drawTarget->getDepthOrStencilbuffer(); + + if (!readBuffer) + { + ERR("Failed to retrieve the read depth-stencil buffer from the read framebuffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + if (!drawBuffer) + { + ERR("Failed to retrieve the draw depth-stencil buffer from the draw framebuffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + RenderTarget *readRenderTarget = GetAttachmentRenderTarget(readBuffer); + RenderTarget *drawRenderTarget = GetAttachmentRenderTarget(drawBuffer); + ASSERT(readRenderTarget && drawRenderTarget); + + if (!blitRenderbufferRect(readRect, drawRect, readRenderTarget, drawRenderTarget, filter, scissor, + false, blitDepth, blitStencil)) + { + return false; + } + } + + invalidateFramebufferSwizzles(drawTarget); + + return true; +} + +gl::Error Renderer11::readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, + GLenum type, GLuint outputPitch, const gl::PixelPackState &pack, uint8_t *pixels) +{ + ID3D11Texture2D *colorBufferTexture = NULL; + unsigned int subresourceIndex = 0; + + gl::FramebufferAttachment *colorbuffer = framebuffer->getReadColorbuffer(); + + if (colorbuffer && getRenderTargetResource(colorbuffer, &subresourceIndex, &colorBufferTexture)) + { + gl::Rectangle area; + area.x = x; + area.y = y; + area.width = width; + area.height = height; + + gl::Buffer *packBuffer = pack.pixelBuffer.get(); + if (packBuffer != NULL) + { + rx::Buffer11 *packBufferStorage = Buffer11::makeBuffer11(packBuffer->getImplementation()); + PackPixelsParams packParams(area, format, type, outputPitch, pack, reinterpret_cast<ptrdiff_t>(pixels)); + + gl::Error error = packBufferStorage->packPixels(colorBufferTexture, subresourceIndex, packParams); + if (error.isError()) + { + return error; + } + + packBuffer->getIndexRangeCache()->clear(); + } + else + { + gl::Error error = readTextureData(colorBufferTexture, subresourceIndex, area, format, type, outputPitch, pack, pixels); + if (error.isError()) + { + return error; + } + } + + SafeRelease(colorBufferTexture); + } + + return gl::Error(GL_NO_ERROR); +} + +Image *Renderer11::createImage() +{ + return new Image11(); +} + +void Renderer11::generateMipmap(Image *dest, Image *src) +{ + Image11 *dest11 = Image11::makeImage11(dest); + Image11 *src11 = Image11::makeImage11(src); + Image11::generateMipmap(dest11, src11); +} + +TextureStorage *Renderer11::createTextureStorage2D(SwapChain *swapChain) +{ + SwapChain11 *swapChain11 = SwapChain11::makeSwapChain11(swapChain); + return new TextureStorage11_2D(this, swapChain11); +} + +TextureStorage *Renderer11::createTextureStorage2D(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, int levels) +{ + return new TextureStorage11_2D(this, internalformat, renderTarget, width, height, levels); +} + +TextureStorage *Renderer11::createTextureStorageCube(GLenum internalformat, bool renderTarget, int size, int levels) +{ + return new TextureStorage11_Cube(this, internalformat, renderTarget, size, levels); +} + +TextureStorage *Renderer11::createTextureStorage3D(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, GLsizei depth, int levels) +{ + return new TextureStorage11_3D(this, internalformat, renderTarget, width, height, depth, levels); +} + +TextureStorage *Renderer11::createTextureStorage2DArray(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, GLsizei depth, int levels) +{ + return new TextureStorage11_2DArray(this, internalformat, renderTarget, width, height, depth, levels); +} + +TextureImpl *Renderer11::createTexture(GLenum target) +{ + switch(target) + { + case GL_TEXTURE_2D: return new TextureD3D_2D(this); + case GL_TEXTURE_CUBE_MAP: return new TextureD3D_Cube(this); + case GL_TEXTURE_3D: return new TextureD3D_3D(this); + case GL_TEXTURE_2D_ARRAY: return new TextureD3D_2DArray(this); + default: + UNREACHABLE(); + } + + return NULL; +} + +gl::Error Renderer11::readTextureData(ID3D11Texture2D *texture, unsigned int subResource, const gl::Rectangle &area, GLenum format, + GLenum type, GLuint outputPitch, const gl::PixelPackState &pack, uint8_t *pixels) +{ + ASSERT(area.width >= 0); + ASSERT(area.height >= 0); + + D3D11_TEXTURE2D_DESC textureDesc; + texture->GetDesc(&textureDesc); + + // Clamp read region to the defined texture boundaries, preventing out of bounds reads + // and reads of uninitialized data. + gl::Rectangle safeArea; + safeArea.x = gl::clamp(area.x, 0, static_cast<int>(textureDesc.Width)); + safeArea.y = gl::clamp(area.y, 0, static_cast<int>(textureDesc.Height)); + safeArea.width = gl::clamp(area.width + std::min(area.x, 0), 0, + static_cast<int>(textureDesc.Width) - safeArea.x); + safeArea.height = gl::clamp(area.height + std::min(area.y, 0), 0, + static_cast<int>(textureDesc.Height) - safeArea.y); + + ASSERT(safeArea.x >= 0 && safeArea.y >= 0); + ASSERT(safeArea.x + safeArea.width <= static_cast<int>(textureDesc.Width)); + ASSERT(safeArea.y + safeArea.height <= static_cast<int>(textureDesc.Height)); + + if (safeArea.width == 0 || safeArea.height == 0) + { + // no work to do + return gl::Error(GL_NO_ERROR); + } + + D3D11_TEXTURE2D_DESC stagingDesc; + stagingDesc.Width = safeArea.width; + stagingDesc.Height = safeArea.height; + stagingDesc.MipLevels = 1; + stagingDesc.ArraySize = 1; + stagingDesc.Format = textureDesc.Format; + stagingDesc.SampleDesc.Count = 1; + stagingDesc.SampleDesc.Quality = 0; + stagingDesc.Usage = D3D11_USAGE_STAGING; + stagingDesc.BindFlags = 0; + stagingDesc.CPUAccessFlags = D3D11_CPU_ACCESS_READ; + stagingDesc.MiscFlags = 0; + + ID3D11Texture2D* stagingTex = NULL; + HRESULT result = mDevice->CreateTexture2D(&stagingDesc, NULL, &stagingTex); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to create internal staging texture for ReadPixels, HRESULT: 0x%X.", result); + } + + ID3D11Texture2D* srcTex = NULL; + if (textureDesc.SampleDesc.Count > 1) + { + D3D11_TEXTURE2D_DESC resolveDesc; + resolveDesc.Width = textureDesc.Width; + resolveDesc.Height = textureDesc.Height; + resolveDesc.MipLevels = 1; + resolveDesc.ArraySize = 1; + resolveDesc.Format = textureDesc.Format; + resolveDesc.SampleDesc.Count = 1; + resolveDesc.SampleDesc.Quality = 0; + resolveDesc.Usage = D3D11_USAGE_DEFAULT; + resolveDesc.BindFlags = 0; + resolveDesc.CPUAccessFlags = 0; + resolveDesc.MiscFlags = 0; + + result = mDevice->CreateTexture2D(&resolveDesc, NULL, &srcTex); + if (FAILED(result)) + { + SafeRelease(stagingTex); + return gl::Error(GL_OUT_OF_MEMORY, "Failed to create internal resolve texture for ReadPixels, HRESULT: 0x%X.", result); + } + + mDeviceContext->ResolveSubresource(srcTex, 0, texture, subResource, textureDesc.Format); + subResource = 0; + } + else + { + srcTex = texture; + srcTex->AddRef(); + } + + D3D11_BOX srcBox; + srcBox.left = static_cast<UINT>(safeArea.x); + srcBox.right = static_cast<UINT>(safeArea.x + safeArea.width); + srcBox.top = static_cast<UINT>(safeArea.y); + srcBox.bottom = static_cast<UINT>(safeArea.y + safeArea.height); + srcBox.front = 0; + srcBox.back = 1; + + mDeviceContext->CopySubresourceRegion(stagingTex, 0, 0, 0, 0, srcTex, subResource, &srcBox); + + SafeRelease(srcTex); + + PackPixelsParams packParams(safeArea, format, type, outputPitch, pack, 0); + packPixels(stagingTex, packParams, pixels); + + SafeRelease(stagingTex); + + return gl::Error(GL_NO_ERROR); +} + +void Renderer11::packPixels(ID3D11Texture2D *readTexture, const PackPixelsParams ¶ms, uint8_t *pixelsOut) +{ + D3D11_TEXTURE2D_DESC textureDesc; + readTexture->GetDesc(&textureDesc); + + D3D11_MAPPED_SUBRESOURCE mapping; + HRESULT hr = mDeviceContext->Map(readTexture, 0, D3D11_MAP_READ, 0, &mapping); + UNUSED_ASSERTION_VARIABLE(hr); + ASSERT(SUCCEEDED(hr)); + + uint8_t *source; + int inputPitch; + if (params.pack.reverseRowOrder) + { + source = static_cast<uint8_t*>(mapping.pData) + mapping.RowPitch * (params.area.height - 1); + inputPitch = -static_cast<int>(mapping.RowPitch); + } + else + { + source = static_cast<uint8_t*>(mapping.pData); + inputPitch = static_cast<int>(mapping.RowPitch); + } + + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(textureDesc.Format); + const gl::InternalFormat &sourceFormatInfo = gl::GetInternalFormatInfo(dxgiFormatInfo.internalFormat); + if (sourceFormatInfo.format == params.format && sourceFormatInfo.type == params.type) + { + uint8_t *dest = pixelsOut + params.offset; + for (int y = 0; y < params.area.height; y++) + { + memcpy(dest + y * params.outputPitch, source + y * inputPitch, params.area.width * sourceFormatInfo.pixelBytes); + } + } + else + { + const d3d11::DXGIFormat &sourceDXGIFormatInfo = d3d11::GetDXGIFormatInfo(textureDesc.Format); + ColorCopyFunction fastCopyFunc = sourceDXGIFormatInfo.getFastCopyFunction(params.format, params.type); + + const gl::FormatType &destFormatTypeInfo = gl::GetFormatTypeInfo(params.format, params.type); + const gl::InternalFormat &destFormatInfo = gl::GetInternalFormatInfo(destFormatTypeInfo.internalFormat); + + if (fastCopyFunc) + { + // Fast copy is possible through some special function + for (int y = 0; y < params.area.height; y++) + { + for (int x = 0; x < params.area.width; x++) + { + uint8_t *dest = pixelsOut + params.offset + y * params.outputPitch + x * destFormatInfo.pixelBytes; + const uint8_t *src = source + y * inputPitch + x * sourceFormatInfo.pixelBytes; + + fastCopyFunc(src, dest); + } + } + } + else + { + uint8_t temp[16]; // Maximum size of any Color<T> type used. + META_ASSERT(sizeof(temp) >= sizeof(gl::ColorF) && + sizeof(temp) >= sizeof(gl::ColorUI) && + sizeof(temp) >= sizeof(gl::ColorI)); + + for (int y = 0; y < params.area.height; y++) + { + for (int x = 0; x < params.area.width; x++) + { + uint8_t *dest = pixelsOut + params.offset + y * params.outputPitch + x * destFormatInfo.pixelBytes; + const uint8_t *src = source + y * inputPitch + x * sourceFormatInfo.pixelBytes; + + // readFunc and writeFunc will be using the same type of color, CopyTexImage + // will not allow the copy otherwise. + sourceDXGIFormatInfo.colorReadFunction(src, temp); + destFormatTypeInfo.colorWriteFunction(temp, dest); + } + } + } + } + + mDeviceContext->Unmap(readTexture, 0); +} + +bool Renderer11::blitRenderbufferRect(const gl::Rectangle &readRect, const gl::Rectangle &drawRect, RenderTarget *readRenderTarget, + RenderTarget *drawRenderTarget, GLenum filter, const gl::Rectangle *scissor, + bool colorBlit, bool depthBlit, bool stencilBlit) +{ + // Since blitRenderbufferRect is called for each render buffer that needs to be blitted, + // it should never be the case that both color and depth/stencil need to be blitted at + // at the same time. + ASSERT(colorBlit != (depthBlit || stencilBlit)); + + bool result = true; + + RenderTarget11 *drawRenderTarget11 = RenderTarget11::makeRenderTarget11(drawRenderTarget); + if (!drawRenderTarget) + { + ERR("Failed to retrieve the draw render target from the draw framebuffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + ID3D11Resource *drawTexture = drawRenderTarget11->getTexture(); + unsigned int drawSubresource = drawRenderTarget11->getSubresourceIndex(); + ID3D11RenderTargetView *drawRTV = drawRenderTarget11->getRenderTargetView(); + ID3D11DepthStencilView *drawDSV = drawRenderTarget11->getDepthStencilView(); + + RenderTarget11 *readRenderTarget11 = RenderTarget11::makeRenderTarget11(readRenderTarget); + if (!readRenderTarget) + { + ERR("Failed to retrieve the read render target from the read framebuffer."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + ID3D11Resource *readTexture = NULL; + ID3D11ShaderResourceView *readSRV = NULL; + unsigned int readSubresource = 0; + if (readRenderTarget->getSamples() > 0) + { + ID3D11Resource *unresolvedResource = readRenderTarget11->getTexture(); + ID3D11Texture2D *unresolvedTexture = d3d11::DynamicCastComObject<ID3D11Texture2D>(unresolvedResource); + + if (unresolvedTexture) + { + readTexture = resolveMultisampledTexture(unresolvedTexture, readRenderTarget11->getSubresourceIndex()); + readSubresource = 0; + + SafeRelease(unresolvedTexture); + + HRESULT hresult = mDevice->CreateShaderResourceView(readTexture, NULL, &readSRV); + if (FAILED(hresult)) + { + SafeRelease(readTexture); + return gl::error(GL_OUT_OF_MEMORY, false); + } + } + } + else + { + readTexture = readRenderTarget11->getTexture(); + readTexture->AddRef(); + readSubresource = readRenderTarget11->getSubresourceIndex(); + readSRV = readRenderTarget11->getShaderResourceView(); + readSRV->AddRef(); + } + + if (!readTexture || !readSRV) + { + SafeRelease(readTexture); + SafeRelease(readSRV); + ERR("Failed to retrieve the read render target view from the read render target."); + return gl::error(GL_OUT_OF_MEMORY, false); + } + + gl::Extents readSize(readRenderTarget->getWidth(), readRenderTarget->getHeight(), 1); + gl::Extents drawSize(drawRenderTarget->getWidth(), drawRenderTarget->getHeight(), 1); + + bool scissorNeeded = scissor && gl::ClipRectangle(drawRect, *scissor, NULL); + + bool wholeBufferCopy = !scissorNeeded && + readRect.x == 0 && readRect.width == readSize.width && + readRect.y == 0 && readRect.height == readSize.height && + drawRect.x == 0 && drawRect.width == drawSize.width && + drawRect.y == 0 && drawRect.height == drawSize.height; + + bool stretchRequired = readRect.width != drawRect.width || readRect.height != drawRect.height; + + bool flipRequired = readRect.width < 0 || readRect.height < 0 || drawRect.width < 0 || drawRect.height < 0; + + bool outOfBounds = readRect.x < 0 || readRect.x + readRect.width > readSize.width || + readRect.y < 0 || readRect.y + readRect.height > readSize.height || + drawRect.x < 0 || drawRect.x + drawRect.width > drawSize.width || + drawRect.y < 0 || drawRect.y + drawRect.height > drawSize.height; + + const gl::InternalFormat &actualFormatInfo = gl::GetInternalFormatInfo(drawRenderTarget->getActualFormat()); + bool partialDSBlit = (actualFormatInfo.depthBits > 0 && depthBlit) != (actualFormatInfo.stencilBits > 0 && stencilBlit); + + if (readRenderTarget11->getActualFormat() == drawRenderTarget->getActualFormat() && + !stretchRequired && !outOfBounds && !flipRequired && !partialDSBlit && + (!(depthBlit || stencilBlit) || wholeBufferCopy)) + { + UINT dstX = drawRect.x; + UINT dstY = drawRect.y; + + D3D11_BOX readBox; + readBox.left = readRect.x; + readBox.right = readRect.x + readRect.width; + readBox.top = readRect.y; + readBox.bottom = readRect.y + readRect.height; + readBox.front = 0; + readBox.back = 1; + + if (scissorNeeded) + { + // drawRect is guaranteed to have positive width and height because stretchRequired is false. + ASSERT(drawRect.width >= 0 || drawRect.height >= 0); + + if (drawRect.x < scissor->x) + { + dstX = scissor->x; + readBox.left += (scissor->x - drawRect.x); + } + if (drawRect.y < scissor->y) + { + dstY = scissor->y; + readBox.top += (scissor->y - drawRect.y); + } + if (drawRect.x + drawRect.width > scissor->x + scissor->width) + { + readBox.right -= ((drawRect.x + drawRect.width) - (scissor->x + scissor->width)); + } + if (drawRect.y + drawRect.height > scissor->y + scissor->height) + { + readBox.bottom -= ((drawRect.y + drawRect.height) - (scissor->y + scissor->height)); + } + } + + // D3D11 needs depth-stencil CopySubresourceRegions to have a NULL pSrcBox + // We also require complete framebuffer copies for depth-stencil blit. + D3D11_BOX *pSrcBox = wholeBufferCopy ? NULL : &readBox; + + mDeviceContext->CopySubresourceRegion(drawTexture, drawSubresource, dstX, dstY, 0, + readTexture, readSubresource, pSrcBox); + result = true; + } + else + { + gl::Box readArea(readRect.x, readRect.y, 0, readRect.width, readRect.height, 1); + gl::Box drawArea(drawRect.x, drawRect.y, 0, drawRect.width, drawRect.height, 1); + + if (depthBlit && stencilBlit) + { + result = mBlit->copyDepthStencil(readTexture, readSubresource, readArea, readSize, + drawTexture, drawSubresource, drawArea, drawSize, + scissor); + } + else if (depthBlit) + { + result = mBlit->copyDepth(readSRV, readArea, readSize, drawDSV, drawArea, drawSize, + scissor); + } + else if (stencilBlit) + { + result = mBlit->copyStencil(readTexture, readSubresource, readArea, readSize, + drawTexture, drawSubresource, drawArea, drawSize, + scissor); + } + else + { + GLenum format = gl::GetInternalFormatInfo(drawRenderTarget->getInternalFormat()).format; + result = mBlit->copyTexture(readSRV, readArea, readSize, drawRTV, drawArea, drawSize, + scissor, format, filter); + } + } + + SafeRelease(readTexture); + SafeRelease(readSRV); + + return result; +} + +ID3D11Texture2D *Renderer11::resolveMultisampledTexture(ID3D11Texture2D *source, unsigned int subresource) +{ + D3D11_TEXTURE2D_DESC textureDesc; + source->GetDesc(&textureDesc); + + if (textureDesc.SampleDesc.Count > 1) + { + D3D11_TEXTURE2D_DESC resolveDesc; + resolveDesc.Width = textureDesc.Width; + resolveDesc.Height = textureDesc.Height; + resolveDesc.MipLevels = 1; + resolveDesc.ArraySize = 1; + resolveDesc.Format = textureDesc.Format; + resolveDesc.SampleDesc.Count = 1; + resolveDesc.SampleDesc.Quality = 0; + resolveDesc.Usage = textureDesc.Usage; + resolveDesc.BindFlags = textureDesc.BindFlags; + resolveDesc.CPUAccessFlags = 0; + resolveDesc.MiscFlags = 0; + + ID3D11Texture2D *resolveTexture = NULL; + HRESULT result = mDevice->CreateTexture2D(&resolveDesc, NULL, &resolveTexture); + if (FAILED(result)) + { + ERR("Failed to create a multisample resolve texture, HRESULT: 0x%X.", result); + return NULL; + } + + mDeviceContext->ResolveSubresource(resolveTexture, 0, source, subresource, textureDesc.Format); + return resolveTexture; + } + else + { + source->AddRef(); + return source; + } +} + +void Renderer11::invalidateFBOAttachmentSwizzles(gl::FramebufferAttachment *attachment, int mipLevel) +{ + ASSERT(attachment->isTexture()); + gl::Texture *texture = attachment->getTexture(); + + TextureStorage *texStorage = texture->getNativeTexture(); + if (texStorage) + { + TextureStorage11 *texStorage11 = TextureStorage11::makeTextureStorage11(texStorage); + if (!texStorage11) + { + ERR("texture storage pointer unexpectedly null."); + return; + } + + texStorage11->invalidateSwizzleCacheLevel(mipLevel); + } +} + +void Renderer11::invalidateFramebufferSwizzles(gl::Framebuffer *framebuffer) +{ + for (unsigned int colorAttachment = 0; colorAttachment < gl::IMPLEMENTATION_MAX_DRAW_BUFFERS; colorAttachment++) + { + gl::FramebufferAttachment *attachment = framebuffer->getColorbuffer(colorAttachment); + if (attachment && attachment->isTexture()) + { + invalidateFBOAttachmentSwizzles(attachment, attachment->mipLevel()); + } + } + + gl::FramebufferAttachment *depthAttachment = framebuffer->getDepthbuffer(); + if (depthAttachment && depthAttachment->isTexture()) + { + invalidateFBOAttachmentSwizzles(depthAttachment, depthAttachment->mipLevel()); + } + + gl::FramebufferAttachment *stencilAttachment = framebuffer->getStencilbuffer(); + if (stencilAttachment && stencilAttachment->isTexture()) + { + invalidateFBOAttachmentSwizzles(stencilAttachment, stencilAttachment->mipLevel()); + } +} + +bool Renderer11::getLUID(LUID *adapterLuid) const +{ + adapterLuid->HighPart = 0; + adapterLuid->LowPart = 0; + + if (!mDxgiAdapter) + { + return false; + } + + DXGI_ADAPTER_DESC adapterDesc; + if (FAILED(mDxgiAdapter->GetDesc(&adapterDesc))) + { + return false; + } + + *adapterLuid = adapterDesc.AdapterLuid; + return true; +} + +rx::VertexConversionType Renderer11::getVertexConversionType(const gl::VertexFormat &vertexFormat) const +{ + return d3d11::GetVertexFormatInfo(vertexFormat).conversionType; +} + +GLenum Renderer11::getVertexComponentType(const gl::VertexFormat &vertexFormat) const +{ + return d3d11::GetDXGIFormatInfo(d3d11::GetVertexFormatInfo(vertexFormat).nativeFormat).componentType; +} + +void Renderer11::generateCaps(gl::Caps *outCaps, gl::TextureCapsMap *outTextureCaps, gl::Extensions *outExtensions) const +{ + d3d11_gl::GenerateCaps(mDevice, outCaps, outTextureCaps, outExtensions); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Renderer11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Renderer11.h new file mode 100644 index 0000000000..2a53fa1672 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/Renderer11.h @@ -0,0 +1,346 @@ +// +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Renderer11.h: Defines a back-end specific class for the D3D11 renderer. + +#ifndef LIBGLESV2_RENDERER_RENDERER11_H_ +#define LIBGLESV2_RENDERER_RENDERER11_H_ + +#include "common/angleutils.h" +#include "libGLESv2/angletypes.h" +#include "common/mathutil.h" + +#include "libGLESv2/renderer/Renderer.h" +#include "libGLESv2/renderer/d3d/HLSLCompiler.h" +#include "libGLESv2/renderer/d3d/d3d11/RenderStateCache.h" +#include "libGLESv2/renderer/d3d/d3d11/InputLayoutCache.h" +#include "libGLESv2/renderer/RenderTarget.h" + +namespace gl +{ +class FramebufferAttachment; +} + +namespace rx +{ + +class VertexDataManager; +class IndexDataManager; +class StreamingIndexBufferInterface; +class Blit11; +class Clear11; +class PixelTransfer11; +struct PackPixelsParams; + +enum +{ + MAX_VERTEX_UNIFORM_VECTORS_D3D11 = 1024, + MAX_FRAGMENT_UNIFORM_VECTORS_D3D11 = 1024 +}; + +class Renderer11 : public Renderer +{ + public: + Renderer11(egl::Display *display, EGLNativeDisplayType hDc, EGLint requestedDisplay); + virtual ~Renderer11(); + + static Renderer11 *makeRenderer11(Renderer *renderer); + + virtual EGLint initialize(); + virtual bool resetDevice(); + + virtual int generateConfigs(ConfigDesc **configDescList); + virtual void deleteConfigs(ConfigDesc *configDescList); + + virtual void sync(bool block); + + virtual SwapChain *createSwapChain(EGLNativeWindowType window, HANDLE shareHandle, GLenum backBufferFormat, GLenum depthBufferFormat); + + virtual gl::Error generateSwizzle(gl::Texture *texture); + virtual gl::Error setSamplerState(gl::SamplerType type, int index, const gl::SamplerState &sampler); + virtual gl::Error setTexture(gl::SamplerType type, int index, gl::Texture *texture); + + virtual gl::Error setUniformBuffers(const gl::Buffer *vertexUniformBuffers[], const gl::Buffer *fragmentUniformBuffers[]); + + virtual gl::Error setRasterizerState(const gl::RasterizerState &rasterState); + virtual gl::Error setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState &blendState, const gl::ColorF &blendColor, + unsigned int sampleMask); + virtual gl::Error setDepthStencilState(const gl::DepthStencilState &depthStencilState, int stencilRef, + int stencilBackRef, bool frontFaceCCW); + + virtual void setScissorRectangle(const gl::Rectangle &scissor, bool enabled); + virtual void setViewport(const gl::Rectangle &viewport, float zNear, float zFar, GLenum drawMode, GLenum frontFace, + bool ignoreViewport); + + virtual bool applyPrimitiveType(GLenum mode, GLsizei count); + virtual gl::Error applyRenderTarget(gl::Framebuffer *frameBuffer); + virtual gl::Error applyShaders(gl::ProgramBinary *programBinary, const gl::VertexFormat inputLayout[], const gl::Framebuffer *framebuffer, + bool rasterizerDiscard, bool transformFeedbackActive); + virtual gl::Error applyUniforms(const gl::ProgramBinary &programBinary); + virtual gl::Error applyVertexBuffer(gl::ProgramBinary *programBinary, const gl::VertexAttribute vertexAttributes[], const gl::VertexAttribCurrentValueData currentValues[], + GLint first, GLsizei count, GLsizei instances); + virtual gl::Error applyIndexBuffer(const GLvoid *indices, gl::Buffer *elementArrayBuffer, GLsizei count, GLenum mode, GLenum type, TranslatedIndexData *indexInfo); + virtual void applyTransformFeedbackBuffers(gl::Buffer *transformFeedbackBuffers[], GLintptr offsets[]); + + virtual gl::Error drawArrays(GLenum mode, GLsizei count, GLsizei instances, bool transformFeedbackActive); + virtual gl::Error drawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, + gl::Buffer *elementArrayBuffer, const TranslatedIndexData &indexInfo, GLsizei instances); + + virtual gl::Error clear(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer); + + virtual void markAllStateDirty(); + + // lost device + void notifyDeviceLost(); + virtual bool isDeviceLost(); + virtual bool testDeviceLost(bool notify); + virtual bool testDeviceResettable(); + + virtual DWORD getAdapterVendor() const; + virtual std::string getRendererDescription() const; + virtual GUID getAdapterIdentifier() const; + + virtual unsigned int getReservedVertexUniformVectors() const; + virtual unsigned int getReservedFragmentUniformVectors() const; + virtual unsigned int getReservedVertexUniformBuffers() const; + virtual unsigned int getReservedFragmentUniformBuffers() const; + virtual bool getShareHandleSupport() const; + virtual bool getPostSubBufferSupport() const; + + virtual int getMajorShaderModel() const; + virtual int getMinSwapInterval() const; + virtual int getMaxSwapInterval() const; + + // Pixel operations + virtual bool copyToRenderTarget2D(TextureStorage *dest, TextureStorage *source); + virtual bool copyToRenderTargetCube(TextureStorage *dest, TextureStorage *source); + virtual bool copyToRenderTarget3D(TextureStorage *dest, TextureStorage *source); + virtual bool copyToRenderTarget2DArray(TextureStorage *dest, TextureStorage *source); + + virtual bool copyImage2D(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, TextureStorage *storage, GLint level); + virtual bool copyImageCube(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, TextureStorage *storage, GLenum target, GLint level); + virtual bool copyImage3D(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, GLint zOffset, TextureStorage *storage, GLint level); + virtual bool copyImage2DArray(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, GLint zOffset, TextureStorage *storage, GLint level); + + virtual bool blitRect(gl::Framebuffer *readTarget, const gl::Rectangle &readRect, gl::Framebuffer *drawTarget, const gl::Rectangle &drawRect, + const gl::Rectangle *scissor, bool blitRenderTarget, bool blitDepth, bool blitStencil, GLenum filter); + + virtual gl::Error readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, + GLenum type, GLuint outputPitch, const gl::PixelPackState &pack, uint8_t *pixels); + + // RenderTarget creation + virtual RenderTarget *createRenderTarget(SwapChain *swapChain, bool depth); + virtual RenderTarget *createRenderTarget(int width, int height, GLenum format, GLsizei samples); + + // Shader creation + virtual ShaderImpl *createShader(GLenum type); + virtual ProgramImpl *createProgram(); + + // Shader operations + virtual void releaseShaderCompiler(); + virtual ShaderExecutable *loadExecutable(const void *function, size_t length, rx::ShaderType type, + const std::vector<gl::LinkedVarying> &transformFeedbackVaryings, + bool separatedOutputBuffers); + virtual ShaderExecutable *compileToExecutable(gl::InfoLog &infoLog, const char *shaderHLSL, rx::ShaderType type, + const std::vector<gl::LinkedVarying> &transformFeedbackVaryings, + bool separatedOutputBuffers, D3DWorkaroundType workaround); + virtual UniformStorage *createUniformStorage(size_t storageSize); + + // Image operations + virtual Image *createImage(); + virtual void generateMipmap(Image *dest, Image *source); + virtual TextureStorage *createTextureStorage2D(SwapChain *swapChain); + virtual TextureStorage *createTextureStorage2D(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, int levels); + virtual TextureStorage *createTextureStorageCube(GLenum internalformat, bool renderTarget, int size, int levels); + virtual TextureStorage *createTextureStorage3D(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, GLsizei depth, int levels); + virtual TextureStorage *createTextureStorage2DArray(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, GLsizei depth, int levels); + + // Texture creation + virtual TextureImpl *createTexture(GLenum target); + + // Buffer creation + virtual BufferImpl *createBuffer(); + virtual VertexBuffer *createVertexBuffer(); + virtual IndexBuffer *createIndexBuffer(); + + // Vertex Array creation + virtual VertexArrayImpl *createVertexArray(); + + // Query and Fence creation + virtual QueryImpl *createQuery(GLenum type); + virtual FenceImpl *createFence(); + + // Transform Feedback creation + virtual TransformFeedbackImpl* createTransformFeedback(); + + // D3D11-renderer specific methods + ID3D11Device *getDevice() { return mDevice; } + ID3D11DeviceContext *getDeviceContext() { return mDeviceContext; }; + IDXGIFactory *getDxgiFactory() { return mDxgiFactory; }; + bool isLevel9() { return mFeatureLevel <= D3D_FEATURE_LEVEL_9_3; } + + Blit11 *getBlitter() { return mBlit; } + + // Buffer-to-texture and Texture-to-buffer copies + virtual bool supportsFastCopyBufferToTexture(GLenum internalFormat) const; + virtual bool fastCopyBufferToTexture(const gl::PixelUnpackState &unpack, unsigned int offset, RenderTarget *destRenderTarget, + GLenum destinationFormat, GLenum sourcePixelsType, const gl::Box &destArea); + + bool getRenderTargetResource(gl::FramebufferAttachment *colorbuffer, unsigned int *subresourceIndex, ID3D11Texture2D **resource); + void unapplyRenderTargets(); + void setOneTimeRenderTarget(ID3D11RenderTargetView *renderTargetView); + void packPixels(ID3D11Texture2D *readTexture, const PackPixelsParams ¶ms, uint8_t *pixelsOut); + + virtual bool getLUID(LUID *adapterLuid) const; + virtual rx::VertexConversionType getVertexConversionType(const gl::VertexFormat &vertexFormat) const; + virtual GLenum getVertexComponentType(const gl::VertexFormat &vertexFormat) const; + + private: + DISALLOW_COPY_AND_ASSIGN(Renderer11); + + virtual void generateCaps(gl::Caps *outCaps, gl::TextureCapsMap *outTextureCaps, gl::Extensions *outExtensions) const; + + gl::Error drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer); + gl::Error drawTriangleFan(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer, int instances); + + gl::Error readTextureData(ID3D11Texture2D *texture, unsigned int subResource, const gl::Rectangle &area, GLenum format, + GLenum type, GLuint outputPitch, const gl::PixelPackState &pack, uint8_t *pixels); + + bool blitRenderbufferRect(const gl::Rectangle &readRect, const gl::Rectangle &drawRect, RenderTarget *readRenderTarget, + RenderTarget *drawRenderTarget, GLenum filter, const gl::Rectangle *scissor, + bool colorBlit, bool depthBlit, bool stencilBlit); + ID3D11Texture2D *resolveMultisampledTexture(ID3D11Texture2D *source, unsigned int subresource); + + static void invalidateFBOAttachmentSwizzles(gl::FramebufferAttachment *attachment, int mipLevel); + static void invalidateFramebufferSwizzles(gl::Framebuffer *framebuffer); + + HMODULE mD3d11Module; + HMODULE mDxgiModule; + EGLNativeDisplayType mDc; + EGLint mRequestedDisplay; + + HLSLCompiler mCompiler; + + bool mDeviceLost; + + void initializeDevice(); + void releaseDeviceResources(); + int getMinorShaderModel() const; + void release(); + + RenderStateCache mStateCache; + + // current render target states + unsigned int mAppliedRenderTargetSerials[gl::IMPLEMENTATION_MAX_DRAW_BUFFERS]; + unsigned int mAppliedDepthbufferSerial; + unsigned int mAppliedStencilbufferSerial; + bool mDepthStencilInitialized; + bool mRenderTargetDescInitialized; + rx::RenderTarget::Desc mRenderTargetDesc; + + // Currently applied sampler states + std::vector<bool> mForceSetVertexSamplerStates; + std::vector<gl::SamplerState> mCurVertexSamplerStates; + + std::vector<bool> mForceSetPixelSamplerStates; + std::vector<gl::SamplerState> mCurPixelSamplerStates; + + // Currently applied textures + std::vector<ID3D11ShaderResourceView*> mCurVertexSRVs; + std::vector<ID3D11ShaderResourceView*> mCurPixelSRVs; + + // Currently applied blend state + bool mForceSetBlendState; + gl::BlendState mCurBlendState; + gl::ColorF mCurBlendColor; + unsigned int mCurSampleMask; + + // Currently applied rasterizer state + bool mForceSetRasterState; + gl::RasterizerState mCurRasterState; + + // Currently applied depth stencil state + bool mForceSetDepthStencilState; + gl::DepthStencilState mCurDepthStencilState; + int mCurStencilRef; + int mCurStencilBackRef; + + // Currently applied scissor rectangle + bool mForceSetScissor; + bool mScissorEnabled; + gl::Rectangle mCurScissor; + + // Currently applied viewport + bool mForceSetViewport; + gl::Rectangle mCurViewport; + float mCurNear; + float mCurFar; + + // Currently applied primitive topology + D3D11_PRIMITIVE_TOPOLOGY mCurrentPrimitiveTopology; + + // Currently applied index buffer + ID3D11Buffer *mAppliedIB; + DXGI_FORMAT mAppliedIBFormat; + unsigned int mAppliedIBOffset; + + // Currently applied transform feedback buffers + ID3D11Buffer *mAppliedTFBuffers[gl::IMPLEMENTATION_MAX_TRANSFORM_FEEDBACK_BUFFERS]; + GLintptr mAppliedTFOffsets[gl::IMPLEMENTATION_MAX_TRANSFORM_FEEDBACK_BUFFERS]; + + // Currently applied shaders + ID3D11VertexShader *mAppliedVertexShader; + ID3D11GeometryShader *mAppliedGeometryShader; + ID3D11GeometryShader *mCurPointGeometryShader; + ID3D11PixelShader *mAppliedPixelShader; + + dx_VertexConstants mVertexConstants; + dx_VertexConstants mAppliedVertexConstants; + ID3D11Buffer *mDriverConstantBufferVS; + ID3D11Buffer *mCurrentVertexConstantBuffer; + unsigned int mCurrentConstantBufferVS[gl::IMPLEMENTATION_MAX_VERTEX_SHADER_UNIFORM_BUFFERS]; + + dx_PixelConstants mPixelConstants; + dx_PixelConstants mAppliedPixelConstants; + ID3D11Buffer *mDriverConstantBufferPS; + ID3D11Buffer *mCurrentPixelConstantBuffer; + unsigned int mCurrentConstantBufferPS[gl::IMPLEMENTATION_MAX_VERTEX_SHADER_UNIFORM_BUFFERS]; + + ID3D11Buffer *mCurrentGeometryConstantBuffer; + + // Vertex, index and input layouts + VertexDataManager *mVertexDataManager; + IndexDataManager *mIndexDataManager; + InputLayoutCache mInputLayoutCache; + + StreamingIndexBufferInterface *mLineLoopIB; + StreamingIndexBufferInterface *mTriangleFanIB; + + // Texture copy resources + Blit11 *mBlit; + PixelTransfer11 *mPixelTransfer; + + // Masked clear resources + Clear11 *mClear; + + // Sync query + ID3D11Query *mSyncQuery; + + ID3D11Device *mDevice; + D3D_FEATURE_LEVEL mFeatureLevel; + ID3D11DeviceContext *mDeviceContext; + IDXGIAdapter *mDxgiAdapter; + DXGI_ADAPTER_DESC mAdapterDescription; + char mDescription[128]; + IDXGIFactory *mDxgiFactory; +}; + +} +#endif // LIBGLESV2_RENDERER_RENDERER11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/ShaderExecutable11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/ShaderExecutable11.cpp index 2e455e3af5..52f34887fb 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/ShaderExecutable11.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/ShaderExecutable11.cpp @@ -1,6 +1,5 @@ -#include "precompiled.h" // -// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // @@ -8,9 +7,8 @@ // ShaderExecutable11.cpp: Implements a D3D11-specific class to contain shader // executable implementation details. -#include "libGLESv2/renderer/d3d11/ShaderExecutable11.h" - -#include "common/debug.h" +#include "libGLESv2/renderer/d3d/d3d11/ShaderExecutable11.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" namespace rx { @@ -21,18 +19,16 @@ ShaderExecutable11::ShaderExecutable11(const void *function, size_t length, ID3D mPixelExecutable = executable; mVertexExecutable = NULL; mGeometryExecutable = NULL; - - mConstantBuffer = NULL; + mStreamOutExecutable = NULL; } -ShaderExecutable11::ShaderExecutable11(const void *function, size_t length, ID3D11VertexShader *executable) +ShaderExecutable11::ShaderExecutable11(const void *function, size_t length, ID3D11VertexShader *executable, ID3D11GeometryShader *streamOut) : ShaderExecutable(function, length) { mVertexExecutable = executable; mPixelExecutable = NULL; mGeometryExecutable = NULL; - - mConstantBuffer = NULL; + mStreamOutExecutable = streamOut; } ShaderExecutable11::ShaderExecutable11(const void *function, size_t length, ID3D11GeometryShader *executable) @@ -41,29 +37,15 @@ ShaderExecutable11::ShaderExecutable11(const void *function, size_t length, ID3D mGeometryExecutable = executable; mVertexExecutable = NULL; mPixelExecutable = NULL; - - mConstantBuffer = NULL; + mStreamOutExecutable = NULL; } ShaderExecutable11::~ShaderExecutable11() { - if (mVertexExecutable) - { - mVertexExecutable->Release(); - } - if (mPixelExecutable) - { - mPixelExecutable->Release(); - } - if (mGeometryExecutable) - { - mGeometryExecutable->Release(); - } - - if (mConstantBuffer) - { - mConstantBuffer->Release(); - } + SafeRelease(mVertexExecutable); + SafeRelease(mPixelExecutable); + SafeRelease(mGeometryExecutable); + SafeRelease(mStreamOutExecutable); } ShaderExecutable11 *ShaderExecutable11::makeShaderExecutable11(ShaderExecutable *executable) @@ -87,23 +69,42 @@ ID3D11GeometryShader *ShaderExecutable11::getGeometryShader() const return mGeometryExecutable; } -ID3D11Buffer *ShaderExecutable11::getConstantBuffer(ID3D11Device *device, unsigned int registerCount) +ID3D11GeometryShader *ShaderExecutable11::getStreamOutShader() const +{ + return mStreamOutExecutable; +} + +UniformStorage11::UniformStorage11(Renderer11 *renderer, size_t initialSize) + : UniformStorage(initialSize), + mConstantBuffer(NULL) { - if (!mConstantBuffer && registerCount > 0) + ID3D11Device *d3d11Device = renderer->getDevice(); + + if (initialSize > 0) { D3D11_BUFFER_DESC constantBufferDescription = {0}; - constantBufferDescription.ByteWidth = registerCount * sizeof(float[4]); + constantBufferDescription.ByteWidth = initialSize; constantBufferDescription.Usage = D3D11_USAGE_DYNAMIC; constantBufferDescription.BindFlags = D3D11_BIND_CONSTANT_BUFFER; constantBufferDescription.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; constantBufferDescription.MiscFlags = 0; constantBufferDescription.StructureByteStride = 0; - HRESULT result = device->CreateBuffer(&constantBufferDescription, NULL, &mConstantBuffer); + HRESULT result = d3d11Device->CreateBuffer(&constantBufferDescription, NULL, &mConstantBuffer); + UNUSED_ASSERTION_VARIABLE(result); ASSERT(SUCCEEDED(result)); } +} - return mConstantBuffer; +UniformStorage11::~UniformStorage11() +{ + SafeRelease(mConstantBuffer); } -}
\ No newline at end of file +const UniformStorage11 *UniformStorage11::makeUniformStorage11(const UniformStorage *uniformStorage) +{ + ASSERT(HAS_DYNAMIC_TYPE(const UniformStorage11*, uniformStorage)); + return static_cast<const UniformStorage11*>(uniformStorage); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/ShaderExecutable11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/ShaderExecutable11.h index c6ec1cf7d2..74a1e03915 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/ShaderExecutable11.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/ShaderExecutable11.h @@ -1,5 +1,5 @@ // -// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // @@ -14,12 +14,14 @@ namespace rx { +class Renderer11; +class UniformStorage11; class ShaderExecutable11 : public ShaderExecutable { public: ShaderExecutable11(const void *function, size_t length, ID3D11PixelShader *executable); - ShaderExecutable11(const void *function, size_t length, ID3D11VertexShader *executable); + ShaderExecutable11(const void *function, size_t length, ID3D11VertexShader *executable, ID3D11GeometryShader *streamOut); ShaderExecutable11(const void *function, size_t length, ID3D11GeometryShader *executable); virtual ~ShaderExecutable11(); @@ -29,8 +31,7 @@ class ShaderExecutable11 : public ShaderExecutable ID3D11PixelShader *getPixelShader() const; ID3D11VertexShader *getVertexShader() const; ID3D11GeometryShader *getGeometryShader() const; - - ID3D11Buffer *getConstantBuffer(ID3D11Device *device, unsigned int registerCount); + ID3D11GeometryShader *getStreamOutShader() const; private: DISALLOW_COPY_AND_ASSIGN(ShaderExecutable11); @@ -38,7 +39,20 @@ class ShaderExecutable11 : public ShaderExecutable ID3D11PixelShader *mPixelExecutable; ID3D11VertexShader *mVertexExecutable; ID3D11GeometryShader *mGeometryExecutable; + ID3D11GeometryShader *mStreamOutExecutable; +}; + +class UniformStorage11 : public UniformStorage +{ + public: + UniformStorage11(Renderer11 *renderer, size_t initialSize); + virtual ~UniformStorage11(); + + static const UniformStorage11 *makeUniformStorage11(const UniformStorage *uniformStorage); + ID3D11Buffer *getConstantBuffer() const { return mConstantBuffer; } + + private: ID3D11Buffer *mConstantBuffer; }; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/SwapChain11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/SwapChain11.cpp index bd97d5cff5..4b29be055d 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/SwapChain11.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/SwapChain11.cpp @@ -1,19 +1,20 @@ -#include "precompiled.h" // -// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // // SwapChain11.cpp: Implements a back-end specific class for the D3D11 swap chain. -#include "libGLESv2/renderer/d3d11/SwapChain11.h" +#include "common/platform.h" +#include "libGLESv2/renderer/d3d/d3d11/SwapChain11.h" +#include "libGLESv2/renderer/d3d/d3d11/renderer11_utils.h" +#include "libGLESv2/renderer/d3d/d3d11/formatutils11.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" -#include "libGLESv2/renderer/d3d11/renderer11_utils.h" -#include "libGLESv2/renderer/d3d11/Renderer11.h" - -#include "libGLESv2/renderer/d3d11/shaders/compiled/passthrough11vs.h" -#include "libGLESv2/renderer/d3d11/shaders/compiled/passthroughrgba11ps.h" +// Precompiled shaders +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthrough2dvs.h" +#include "libGLESv2/renderer/d3d/d3d11/shaders/compiled/passthroughrgba2dps.h" namespace rx { @@ -30,6 +31,7 @@ SwapChain11::SwapChain11(Renderer11 *renderer, EGLNativeWindowType window, HANDL mOffscreenSRView = NULL; mDepthStencilTexture = NULL; mDepthStencilDSView = NULL; + mDepthStencilSRView = NULL; mQuadVB = NULL; mPassThroughSampler = NULL; mPassThroughIL = NULL; @@ -37,6 +39,8 @@ SwapChain11::SwapChain11(Renderer11 *renderer, EGLNativeWindowType window, HANDL mPassThroughPS = NULL; mWidth = -1; mHeight = -1; + mViewportWidth = -1; + mViewportHeight = -1; mSwapInterval = 0; mAppCreatedShareHandle = mShareHandle != NULL; mPassThroughResourcesInit = false; @@ -57,6 +61,7 @@ void SwapChain11::release() SafeRelease(mOffscreenSRView); SafeRelease(mDepthStencilTexture); SafeRelease(mDepthStencilDSView); + SafeRelease(mDepthStencilSRView); SafeRelease(mQuadVB); SafeRelease(mPassThroughSampler); SafeRelease(mPassThroughIL); @@ -76,6 +81,7 @@ void SwapChain11::releaseOffscreenTexture() SafeRelease(mOffscreenSRView); SafeRelease(mDepthStencilTexture); SafeRelease(mDepthStencilDSView); + SafeRelease(mDepthStencilSRView); } EGLint SwapChain11::resetOffscreenTexture(int backbufferWidth, int backbufferHeight) @@ -89,6 +95,7 @@ EGLint SwapChain11::resetOffscreenTexture(int backbufferWidth, int backbufferHei ASSERT(backbufferHeight >= 1); // Preserve the render target content +#if !defined(ANGLE_PLATFORM_WINRT) ID3D11Texture2D *previousOffscreenTexture = mOffscreenTexture; if (previousOffscreenTexture) { @@ -96,9 +103,12 @@ EGLint SwapChain11::resetOffscreenTexture(int backbufferWidth, int backbufferHei } const int previousWidth = mWidth; const int previousHeight = mHeight; +#endif releaseOffscreenTexture(); + const d3d11::TextureFormat &backbufferFormatInfo = d3d11::GetTextureFormatInfo(mBackBufferFormat); + // If the app passed in a share handle, open the resource // See EGL_ANGLE_d3d_share_handle_client_buffer if (mAppCreatedShareHandle) @@ -114,7 +124,7 @@ EGLint SwapChain11::resetOffscreenTexture(int backbufferWidth, int backbufferHei } result = tempResource11->QueryInterface(__uuidof(ID3D11Texture2D), (void**)&mOffscreenTexture); - tempResource11->Release(); + SafeRelease(tempResource11); if (FAILED(result)) { @@ -127,11 +137,11 @@ EGLint SwapChain11::resetOffscreenTexture(int backbufferWidth, int backbufferHei D3D11_TEXTURE2D_DESC offscreenTextureDesc = {0}; mOffscreenTexture->GetDesc(&offscreenTextureDesc); - if (offscreenTextureDesc.Width != (UINT)backbufferWidth - || offscreenTextureDesc.Height != (UINT)backbufferHeight - || offscreenTextureDesc.Format != gl_d3d11::ConvertRenderbufferFormat(mBackBufferFormat) - || offscreenTextureDesc.MipLevels != 1 - || offscreenTextureDesc.ArraySize != 1) + if (offscreenTextureDesc.Width != (UINT)backbufferWidth || + offscreenTextureDesc.Height != (UINT)backbufferHeight || + offscreenTextureDesc.Format != backbufferFormatInfo.texFormat || + offscreenTextureDesc.MipLevels != 1 || + offscreenTextureDesc.ArraySize != 1) { ERR("Invalid texture parameters in the shared offscreen texture pbuffer"); release(); @@ -145,7 +155,7 @@ EGLint SwapChain11::resetOffscreenTexture(int backbufferWidth, int backbufferHei D3D11_TEXTURE2D_DESC offscreenTextureDesc = {0}; offscreenTextureDesc.Width = backbufferWidth; offscreenTextureDesc.Height = backbufferHeight; - offscreenTextureDesc.Format = gl_d3d11::ConvertRenderbufferFormat(mBackBufferFormat); + offscreenTextureDesc.Format = backbufferFormatInfo.texFormat; offscreenTextureDesc.MipLevels = 1; offscreenTextureDesc.ArraySize = 1; offscreenTextureDesc.SampleDesc.Count = 1; @@ -172,7 +182,7 @@ EGLint SwapChain11::resetOffscreenTexture(int backbufferWidth, int backbufferHei } } - d3d11::SetDebugName(mOffscreenTexture, "Offscreen texture"); + d3d11::SetDebugName(mOffscreenTexture, "Offscreen back buffer texture"); // EGL_ANGLE_surface_d3d_texture_2d_share_handle requires that we store a share handle for the client if (useSharedResource) @@ -188,7 +198,7 @@ EGLint SwapChain11::resetOffscreenTexture(int backbufferWidth, int backbufferHei else { result = offscreenTextureResource->GetSharedHandle(&mShareHandle); - offscreenTextureResource->Release(); + SafeRelease(offscreenTextureResource); if (FAILED(result)) { @@ -198,32 +208,45 @@ EGLint SwapChain11::resetOffscreenTexture(int backbufferWidth, int backbufferHei } } } - - HRESULT result = device->CreateRenderTargetView(mOffscreenTexture, NULL, &mOffscreenRTView); + + D3D11_RENDER_TARGET_VIEW_DESC offscreenRTVDesc; + offscreenRTVDesc.Format = backbufferFormatInfo.rtvFormat; + offscreenRTVDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2D; + offscreenRTVDesc.Texture2D.MipSlice = 0; + + HRESULT result = device->CreateRenderTargetView(mOffscreenTexture, &offscreenRTVDesc, &mOffscreenRTView); ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mOffscreenRTView, "Offscreen render target"); + d3d11::SetDebugName(mOffscreenRTView, "Offscreen back buffer render target"); - result = device->CreateShaderResourceView(mOffscreenTexture, NULL, &mOffscreenSRView); + D3D11_SHADER_RESOURCE_VIEW_DESC offscreenSRVDesc; + offscreenSRVDesc.Format = backbufferFormatInfo.srvFormat; + offscreenSRVDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2D; + offscreenSRVDesc.Texture2D.MostDetailedMip = 0; + offscreenSRVDesc.Texture2D.MipLevels = -1; + + result = device->CreateShaderResourceView(mOffscreenTexture, &offscreenSRVDesc, &mOffscreenSRView); ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mOffscreenSRView, "Offscreen shader resource"); + d3d11::SetDebugName(mOffscreenSRView, "Offscreen back buffer shader resource"); + + const d3d11::TextureFormat &depthBufferFormatInfo = d3d11::GetTextureFormatInfo(mDepthBufferFormat); if (mDepthBufferFormat != GL_NONE) { - D3D11_TEXTURE2D_DESC depthStencilDesc = {0}; - depthStencilDesc.Width = backbufferWidth; - depthStencilDesc.Height = backbufferHeight; - depthStencilDesc.Format = gl_d3d11::ConvertRenderbufferFormat(mDepthBufferFormat); - depthStencilDesc.MipLevels = 1; - depthStencilDesc.ArraySize = 1; - depthStencilDesc.SampleDesc.Count = 1; - depthStencilDesc.SampleDesc.Quality = 0; - depthStencilDesc.Usage = D3D11_USAGE_DEFAULT; - depthStencilDesc.BindFlags = D3D11_BIND_DEPTH_STENCIL; - depthStencilDesc.CPUAccessFlags = 0; - depthStencilDesc.MiscFlags = 0; - - result = device->CreateTexture2D(&depthStencilDesc, NULL, &mDepthStencilTexture); + D3D11_TEXTURE2D_DESC depthStencilTextureDesc; + depthStencilTextureDesc.Width = backbufferWidth; + depthStencilTextureDesc.Height = backbufferHeight; + depthStencilTextureDesc.Format = depthBufferFormatInfo.texFormat; + depthStencilTextureDesc.MipLevels = 1; + depthStencilTextureDesc.ArraySize = 1; + depthStencilTextureDesc.SampleDesc.Count = 1; + depthStencilTextureDesc.SampleDesc.Quality = 0; + depthStencilTextureDesc.Usage = D3D11_USAGE_DEFAULT; + depthStencilTextureDesc.BindFlags = D3D11_BIND_DEPTH_STENCIL | D3D11_BIND_SHADER_RESOURCE; + depthStencilTextureDesc.CPUAccessFlags = 0; + depthStencilTextureDesc.MiscFlags = 0; + + result = device->CreateTexture2D(&depthStencilTextureDesc, NULL, &mDepthStencilTexture); if (FAILED(result)) { ERR("Could not create depthstencil surface for new swap chain: 0x%08X", result); @@ -238,16 +261,37 @@ EGLint SwapChain11::resetOffscreenTexture(int backbufferWidth, int backbufferHei return EGL_BAD_ALLOC; } } - d3d11::SetDebugName(mDepthStencilTexture, "Depth stencil texture"); + d3d11::SetDebugName(mDepthStencilTexture, "Offscreen depth stencil texture"); + + D3D11_DEPTH_STENCIL_VIEW_DESC depthStencilDesc; + depthStencilDesc.Format = depthBufferFormatInfo.dsvFormat; + depthStencilDesc.ViewDimension = D3D11_DSV_DIMENSION_TEXTURE2D; + depthStencilDesc.Flags = 0; + depthStencilDesc.Texture2D.MipSlice = 0; - result = device->CreateDepthStencilView(mDepthStencilTexture, NULL, &mDepthStencilDSView); + result = device->CreateDepthStencilView(mDepthStencilTexture, &depthStencilDesc, &mDepthStencilDSView); ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mDepthStencilDSView, "Depth stencil view"); + d3d11::SetDebugName(mDepthStencilDSView, "Offscreen depth stencil view"); + + D3D11_SHADER_RESOURCE_VIEW_DESC depthStencilSRVDesc; + depthStencilSRVDesc.Format = depthBufferFormatInfo.srvFormat; + depthStencilSRVDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2D; + depthStencilSRVDesc.Texture2D.MostDetailedMip = 0; + depthStencilSRVDesc.Texture2D.MipLevels = -1; + + result = device->CreateShaderResourceView(mDepthStencilTexture, &depthStencilSRVDesc, &mDepthStencilSRView); + ASSERT(SUCCEEDED(result)); + d3d11::SetDebugName(mDepthStencilSRView, "Offscreen depth stencil shader resource"); } mWidth = backbufferWidth; mHeight = backbufferHeight; +#if !defined(ANGLE_PLATFORM_WINRT) || WINAPI_FAMILY==WINAPI_FAMILY_PC_APP + mViewportWidth = backbufferWidth; + mViewportHeight = backbufferHeight; +#endif +#if !defined(ANGLE_PLATFORM_WINRT) if (previousOffscreenTexture != NULL) { D3D11_BOX sourceBox = {0}; @@ -262,13 +306,14 @@ EGLint SwapChain11::resetOffscreenTexture(int backbufferWidth, int backbufferHei const int yoffset = std::max(mHeight - previousHeight, 0); deviceContext->CopySubresourceRegion(mOffscreenTexture, 0, 0, yoffset, 0, previousOffscreenTexture, 0, &sourceBox); - previousOffscreenTexture->Release(); + SafeRelease(previousOffscreenTexture); if (mSwapChain) { - swapRect(0, 0, mWidth, mHeight); + swapRect(0, 0, mWidth, mHeight, SWAP_NORMAL); } } +#endif return EGL_SUCCESS; } @@ -295,8 +340,15 @@ EGLint SwapChain11::resize(EGLint backbufferWidth, EGLint backbufferHeight) SafeRelease(mBackBufferRTView); // Resize swap chain - DXGI_FORMAT backbufferDXGIFormat = gl_d3d11::ConvertRenderbufferFormat(mBackBufferFormat); - HRESULT result = mSwapChain->ResizeBuffers(2, backbufferWidth, backbufferHeight, backbufferDXGIFormat, 0); + HRESULT result; +#if !defined(ANGLE_PLATFORM_WINRT) || WINAPI_FAMILY==WINAPI_FAMILY_PC_APP // Windows phone swap chain is never resized, only the texture is +#if !defined(ANGLE_PLATFORM_WINRT) + const int bufferCount = 1; +#else + const int bufferCount = 2; +#endif + const d3d11::TextureFormat &backbufferFormatInfo = d3d11::GetTextureFormatInfo(mBackBufferFormat); + result = mSwapChain->ResizeBuffers(bufferCount, backbufferWidth, backbufferHeight, backbufferFormatInfo.texFormat, 0); if (FAILED(result)) { @@ -312,6 +364,7 @@ EGLint SwapChain11::resize(EGLint backbufferWidth, EGLint backbufferHeight) return EGL_BAD_ALLOC; } } +#endif result = mSwapChain->GetBuffer(0, __uuidof(ID3D11Texture2D), (LPVOID*)&mBackBufferTexture); ASSERT(SUCCEEDED(result)); @@ -361,51 +414,60 @@ EGLint SwapChain11::reset(int backbufferWidth, int backbufferHeight, EGLint swap if (mWindow) { -#if !defined(ANGLE_OS_WINRT) + const d3d11::TextureFormat &backbufferFormatInfo = d3d11::GetTextureFormatInfo(mBackBufferFormat); + IDXGIFactory *factory = mRenderer->getDxgiFactory(); +#if !defined(ANGLE_PLATFORM_WINRT) DXGI_SWAP_CHAIN_DESC swapChainDesc = {0}; - swapChainDesc.BufferDesc.Format = gl_d3d11::ConvertRenderbufferFormat(mBackBufferFormat); swapChainDesc.BufferDesc.Width = backbufferWidth; swapChainDesc.BufferDesc.Height = backbufferHeight; - swapChainDesc.BufferCount = 2; - swapChainDesc.BufferDesc.Scaling = DXGI_MODE_SCALING_UNSPECIFIED; - swapChainDesc.BufferDesc.ScanlineOrdering = DXGI_MODE_SCANLINE_ORDER_UNSPECIFIED; swapChainDesc.BufferDesc.RefreshRate.Numerator = 0; swapChainDesc.BufferDesc.RefreshRate.Denominator = 1; - swapChainDesc.Windowed = TRUE; + swapChainDesc.BufferDesc.Format = backbufferFormatInfo.texFormat; + swapChainDesc.BufferDesc.ScanlineOrdering = DXGI_MODE_SCANLINE_ORDER_UNSPECIFIED; + swapChainDesc.BufferDesc.Scaling = DXGI_MODE_SCALING_UNSPECIFIED; + swapChainDesc.SampleDesc.Count = 1; + swapChainDesc.SampleDesc.Quality = 0; + swapChainDesc.BufferUsage = DXGI_USAGE_RENDER_TARGET_OUTPUT; + swapChainDesc.BufferCount = 1; swapChainDesc.OutputWindow = mWindow; + swapChainDesc.Windowed = TRUE; + swapChainDesc.SwapEffect = DXGI_SWAP_EFFECT_DISCARD; + swapChainDesc.Flags = 0; + + HRESULT result = factory->CreateSwapChain(device, &swapChainDesc, &mSwapChain); #else - IDXGIFactory2 *factory; - HRESULT result = mRenderer->getDxgiFactory()->QueryInterface(IID_PPV_ARGS(&factory)); + IDXGIFactory2 *factory2; + HRESULT result = factory->QueryInterface(IID_PPV_ARGS(&factory2)); ASSERT(SUCCEEDED(result)); DXGI_SWAP_CHAIN_DESC1 swapChainDesc = {0}; - swapChainDesc.Format = gl_d3d11::ConvertRenderbufferFormat(mBackBufferFormat); - swapChainDesc.Width = backbufferWidth; - swapChainDesc.Height = backbufferHeight; + swapChainDesc.Width = 0; + swapChainDesc.Height = 0; + swapChainDesc.Format = backbufferFormatInfo.texFormat; + swapChainDesc.SampleDesc.Count = 1; + swapChainDesc.SampleDesc.Quality = 0; + swapChainDesc.BufferUsage = DXGI_USAGE_RENDER_TARGET_OUTPUT; swapChainDesc.Stereo = FALSE; -#if !defined(ANGLE_OS_WINPHONE) + swapChainDesc.Flags = 0; +#if WINAPI_FAMILY==WINAPI_FAMILY_PC_APP + swapChainDesc.Scaling = DXGI_SCALING_NONE; swapChainDesc.BufferCount = 2; swapChainDesc.SwapEffect = DXGI_SWAP_EFFECT_FLIP_SEQUENTIAL; -#else +#elif WINAPI_FAMILY==WINAPI_FAMILY_PHONE_APP swapChainDesc.BufferCount = 1; swapChainDesc.SwapEffect = DXGI_SWAP_EFFECT_DISCARD; #endif -#endif - swapChainDesc.BufferUsage = DXGI_USAGE_RENDER_TARGET_OUTPUT; - swapChainDesc.Flags = 0; - swapChainDesc.SampleDesc.Count = 1; - swapChainDesc.SampleDesc.Quality = 0; -#if !defined(ANGLE_OS_WINRT) - HRESULT result = factory->CreateSwapChain(device, &swapChainDesc, &mSwapChain); -#else IDXGISwapChain1 *swapChain; - result = factory->CreateSwapChainForCoreWindow(device, mWindow, &swapChainDesc, NULL, &swapChain); + result = factory2->CreateSwapChainForCoreWindow(device, mWindow, &swapChainDesc, NULL, &swapChain); mSwapChain = swapChain; + HRESULT hr = swapChain->GetDesc1(&swapChainDesc); + ASSERT(SUCCEEDED(hr)); + mViewportWidth = swapChainDesc.Width; + mViewportHeight = swapChainDesc.Height; #endif - if (FAILED(result)) { ERR("Could not create additional swap chains or offscreen surfaces: %08lX", result); @@ -415,28 +477,10 @@ EGLint SwapChain11::reset(int backbufferWidth, int backbufferHeight, EGLint swap { return EGL_CONTEXT_LOST; } -#if !defined(ANGLE_OS_WINRT) else { - // We cannot create a swap chain for an HWND that is owned by a different process on some versions of - // windows - DWORD currentProcessId = GetCurrentProcessId(); - DWORD wndProcessId; - GetWindowThreadProcessId(mWindow, &wndProcessId); - - if (currentProcessId != wndProcessId) - { - ERR("Could not create swap chain, window owned by different process"); - return EGL_BAD_NATIVE_WINDOW; - } - else - { - return EGL_BAD_ALLOC; - } + return EGL_BAD_ALLOC; } -#else - return EGL_BAD_ALLOC; -#endif } result = mSwapChain->GetBuffer(0, __uuidof(ID3D11Texture2D), (LPVOID*)&mBackBufferTexture); @@ -493,7 +537,7 @@ void SwapChain11::initPassThroughResources() samplerDesc.BorderColor[2] = 0.0f; samplerDesc.BorderColor[3] = 0.0f; samplerDesc.MinLOD = 0; - samplerDesc.MaxLOD = D3D11_FLOAT32_MAX; + samplerDesc.MaxLOD = FLT_MAX; result = device->CreateSamplerState(&samplerDesc, &mPassThroughSampler); ASSERT(SUCCEEDED(result)); @@ -505,21 +549,21 @@ void SwapChain11::initPassThroughResources() { "TEXCOORD", 0, DXGI_FORMAT_R32G32_FLOAT, 0, 8, D3D11_INPUT_PER_VERTEX_DATA, 0 }, }; - result = device->CreateInputLayout(quadLayout, 2, g_VS_Passthrough, sizeof(g_VS_Passthrough), &mPassThroughIL); + result = device->CreateInputLayout(quadLayout, 2, g_VS_Passthrough2D, sizeof(g_VS_Passthrough2D), &mPassThroughIL); ASSERT(SUCCEEDED(result)); d3d11::SetDebugName(mPassThroughIL, "Swap chain pass through layout"); - result = device->CreateVertexShader(g_VS_Passthrough, sizeof(g_VS_Passthrough), NULL, &mPassThroughVS); + result = device->CreateVertexShader(g_VS_Passthrough2D, sizeof(g_VS_Passthrough2D), NULL, &mPassThroughVS); ASSERT(SUCCEEDED(result)); d3d11::SetDebugName(mPassThroughVS, "Swap chain pass through vertex shader"); - result = device->CreatePixelShader(g_PS_PassthroughRGBA, sizeof(g_PS_PassthroughRGBA), NULL, &mPassThroughPS); + result = device->CreatePixelShader(g_PS_PassthroughRGBA2D, sizeof(g_PS_PassthroughRGBA2D), NULL, &mPassThroughPS); ASSERT(SUCCEEDED(result)); d3d11::SetDebugName(mPassThroughPS, "Swap chain pass through pixel shader"); } // parameters should be validated/clamped by caller -EGLint SwapChain11::swapRect(EGLint x, EGLint y, EGLint width, EGLint height) +EGLint SwapChain11::swapRect(EGLint x, EGLint y, EGLint width, EGLint height, EGLint flags) { if (!mSwapChain) { @@ -550,10 +594,13 @@ EGLint SwapChain11::swapRect(EGLint x, EGLint y, EGLint width, EGLint height) float u2 = (x + width) / float(mWidth); float v2 = (y + height) / float(mHeight); - d3d11::SetPositionTexCoordVertex(&vertices[0], x1, y1, u1, v1); - d3d11::SetPositionTexCoordVertex(&vertices[1], x1, y2, u1, v2); - d3d11::SetPositionTexCoordVertex(&vertices[2], x2, y1, u2, v1); - d3d11::SetPositionTexCoordVertex(&vertices[3], x2, y2, u2, v2); + const int rotateL = flags & SWAP_ROTATE_90; + const int rotateR = flags & SWAP_ROTATE_270; + + d3d11::SetPositionTexCoordVertex(&vertices[0], x1, y1, rotateL ? u2 : u1, rotateR ? v2 : v1); + d3d11::SetPositionTexCoordVertex(&vertices[1], x1, y2, rotateR ? u2 : u1, rotateL ? v1 : v2); + d3d11::SetPositionTexCoordVertex(&vertices[2], x2, y1, rotateR ? u1 : u2, rotateL ? v2 : v1); + d3d11::SetPositionTexCoordVertex(&vertices[3], x2, y2, rotateL ? u1 : u2, rotateR ? v1 : v2); deviceContext->Unmap(mQuadVB, 0); @@ -583,8 +630,8 @@ EGLint SwapChain11::swapRect(EGLint x, EGLint y, EGLint width, EGLint height) D3D11_VIEWPORT viewport; viewport.TopLeftX = 0; viewport.TopLeftY = 0; - viewport.Width = mWidth; - viewport.Height = mHeight; + viewport.Width = mViewportWidth; + viewport.Height = mViewportHeight; viewport.MinDepth = 0.0f; viewport.MaxDepth = 1.0f; deviceContext->RSSetViewports(1, &viewport); @@ -596,7 +643,7 @@ EGLint SwapChain11::swapRect(EGLint x, EGLint y, EGLint width, EGLint height) // Draw deviceContext->Draw(4, 0); -#if ANGLE_FORCE_VSYNC_OFF +#ifdef ANGLE_FORCE_VSYNC_OFF result = mSwapChain->Present(0, 0); #else result = mSwapChain->Present(mSwapInterval, 0); @@ -605,6 +652,7 @@ EGLint SwapChain11::swapRect(EGLint x, EGLint y, EGLint width, EGLint height) if (result == DXGI_ERROR_DEVICE_REMOVED) { HRESULT removedReason = device->GetDeviceRemovedReason(); + UNUSED_TRACE_VARIABLE(removedReason); ERR("Present failed: the D3D11 device was removed: 0x%08X", removedReason); return EGL_CONTEXT_LOST; } @@ -628,61 +676,33 @@ EGLint SwapChain11::swapRect(EGLint x, EGLint y, EGLint width, EGLint height) return EGL_SUCCESS; } -// Increments refcount on texture. -// caller must Release() the returned texture ID3D11Texture2D *SwapChain11::getOffscreenTexture() { - if (mOffscreenTexture) - { - mOffscreenTexture->AddRef(); - } - return mOffscreenTexture; } -// Increments refcount on view. -// caller must Release() the returned view ID3D11RenderTargetView *SwapChain11::getRenderTarget() { - if (mOffscreenRTView) - { - mOffscreenRTView->AddRef(); - } - return mOffscreenRTView; } -// Increments refcount on view. -// caller must Release() the returned view ID3D11ShaderResourceView *SwapChain11::getRenderTargetShaderResource() { - if (mOffscreenSRView) - { - mOffscreenSRView->AddRef(); - } - return mOffscreenSRView; } -// Increments refcount on view. -// caller must Release() the returned view ID3D11DepthStencilView *SwapChain11::getDepthStencil() { - if (mDepthStencilDSView) - { - mDepthStencilDSView->AddRef(); - } - return mDepthStencilDSView; } -ID3D11Texture2D *SwapChain11::getDepthStencilTexture() +ID3D11ShaderResourceView * SwapChain11::getDepthStencilShaderResource() { - if (mDepthStencilTexture) - { - mDepthStencilTexture->AddRef(); - } + return mDepthStencilSRView; +} +ID3D11Texture2D *SwapChain11::getDepthStencilTexture() +{ return mDepthStencilTexture; } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/SwapChain11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/SwapChain11.h index 2a030c839d..b30b78568a 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/SwapChain11.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/SwapChain11.h @@ -25,7 +25,7 @@ class SwapChain11 : public SwapChain EGLint resize(EGLint backbufferWidth, EGLint backbufferHeight); virtual EGLint reset(EGLint backbufferWidth, EGLint backbufferHeight, EGLint swapInterval); - virtual EGLint swapRect(EGLint x, EGLint y, EGLint width, EGLint height); + virtual EGLint swapRect(EGLint x, EGLint y, EGLint width, EGLint height, EGLint flags); virtual void recreate(); virtual ID3D11Texture2D *getOffscreenTexture(); @@ -34,6 +34,7 @@ class SwapChain11 : public SwapChain virtual ID3D11Texture2D *getDepthStencilTexture(); virtual ID3D11DepthStencilView *getDepthStencil(); + virtual ID3D11ShaderResourceView *getDepthStencilShaderResource(); EGLint getWidth() const { return mWidth; } EGLint getHeight() const { return mHeight; } @@ -51,6 +52,8 @@ class SwapChain11 : public SwapChain Renderer11 *mRenderer; EGLint mHeight; EGLint mWidth; + EGLint mViewportWidth; + EGLint mViewportHeight; bool mAppCreatedShareHandle; unsigned int mSwapInterval; bool mPassThroughResourcesInit; @@ -66,6 +69,7 @@ class SwapChain11 : public SwapChain ID3D11Texture2D *mDepthStencilTexture; ID3D11DepthStencilView *mDepthStencilDSView; + ID3D11ShaderResourceView *mDepthStencilSRView; ID3D11Buffer *mQuadVB; ID3D11SamplerState *mPassThroughSampler; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/TextureStorage11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/TextureStorage11.cpp new file mode 100644 index 0000000000..91e7147da6 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/TextureStorage11.cpp @@ -0,0 +1,1937 @@ +// +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// TextureStorage11.cpp: Implements the abstract rx::TextureStorage11 class and its concrete derived +// classes TextureStorage11_2D and TextureStorage11_Cube, which act as the interface to the D3D11 texture. + +#include "libGLESv2/renderer/d3d/d3d11/TextureStorage11.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" +#include "libGLESv2/renderer/d3d/d3d11/RenderTarget11.h" +#include "libGLESv2/renderer/d3d/d3d11/SwapChain11.h" +#include "libGLESv2/renderer/d3d/d3d11/renderer11_utils.h" +#include "libGLESv2/renderer/d3d/d3d11/Blit11.h" +#include "libGLESv2/renderer/d3d/d3d11/formatutils11.h" +#include "libGLESv2/renderer/d3d/d3d11/Image11.h" +#include "libGLESv2/renderer/d3d/TextureD3D.h" +#include "libGLESv2/main.h" +#include "libGLESv2/ImageIndex.h" + +#include "common/utilities.h" + +namespace rx +{ + +TextureStorage11::SwizzleCacheValue::SwizzleCacheValue() + : swizzleRed(GL_NONE), swizzleGreen(GL_NONE), swizzleBlue(GL_NONE), swizzleAlpha(GL_NONE) +{ +} + +TextureStorage11::SwizzleCacheValue::SwizzleCacheValue(GLenum red, GLenum green, GLenum blue, GLenum alpha) + : swizzleRed(red), swizzleGreen(green), swizzleBlue(blue), swizzleAlpha(alpha) +{ +} + +bool TextureStorage11::SwizzleCacheValue::operator==(const SwizzleCacheValue &other) const +{ + return swizzleRed == other.swizzleRed && + swizzleGreen == other.swizzleGreen && + swizzleBlue == other.swizzleBlue && + swizzleAlpha == other.swizzleAlpha; +} + +bool TextureStorage11::SwizzleCacheValue::operator!=(const SwizzleCacheValue &other) const +{ + return !(*this == other); +} + +TextureStorage11::SRVKey::SRVKey(int baseLevel, int mipLevels, bool swizzle) + : baseLevel(baseLevel), mipLevels(mipLevels), swizzle(swizzle) +{ +} + +bool TextureStorage11::SRVKey::operator==(const SRVKey &rhs) const +{ + return baseLevel == rhs.baseLevel && + mipLevels == rhs.mipLevels && + swizzle == rhs.swizzle; +} + +TextureStorage11::SRVCache::~SRVCache() +{ + for (size_t i = 0; i < cache.size(); i++) + { + SafeRelease(cache[i].srv); + } +} + +ID3D11ShaderResourceView *TextureStorage11::SRVCache::find(const SRVKey &key) const +{ + for (size_t i = 0; i < cache.size(); i++) + { + if (cache[i].key == key) + { + return cache[i].srv; + } + } + + return NULL; +} + +ID3D11ShaderResourceView *TextureStorage11::SRVCache::add(const SRVKey &key, ID3D11ShaderResourceView *srv) +{ + SRVPair pair = {key, srv}; + cache.push_back(pair); + + return srv; +} + +TextureStorage11::TextureStorage11(Renderer *renderer, UINT bindFlags) + : mBindFlags(bindFlags), + mTopLevel(0), + mMipLevels(0), + mTextureFormat(DXGI_FORMAT_UNKNOWN), + mShaderResourceFormat(DXGI_FORMAT_UNKNOWN), + mRenderTargetFormat(DXGI_FORMAT_UNKNOWN), + mDepthStencilFormat(DXGI_FORMAT_UNKNOWN), + mTextureWidth(0), + mTextureHeight(0), + mTextureDepth(0) +{ + mRenderer = Renderer11::makeRenderer11(renderer); + + for (unsigned int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) + { + mLevelSRVs[i] = NULL; + } +} + +TextureStorage11::~TextureStorage11() +{ + for (unsigned int level = 0; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; level++) + { + SafeRelease(mLevelSRVs[level]); + } +} + +TextureStorage11 *TextureStorage11::makeTextureStorage11(TextureStorage *storage) +{ + ASSERT(HAS_DYNAMIC_TYPE(TextureStorage11*, storage)); + return static_cast<TextureStorage11*>(storage); +} + +DWORD TextureStorage11::GetTextureBindFlags(GLenum internalFormat, bool renderTarget) +{ + UINT bindFlags = 0; + + const d3d11::TextureFormat &formatInfo = d3d11::GetTextureFormatInfo(internalFormat); + if (formatInfo.srvFormat != DXGI_FORMAT_UNKNOWN) + { + bindFlags |= D3D11_BIND_SHADER_RESOURCE; + } + if (formatInfo.dsvFormat != DXGI_FORMAT_UNKNOWN) + { + bindFlags |= D3D11_BIND_DEPTH_STENCIL; + } + if (formatInfo.rtvFormat != DXGI_FORMAT_UNKNOWN && renderTarget) + { + bindFlags |= D3D11_BIND_RENDER_TARGET; + } + + return bindFlags; +} + +UINT TextureStorage11::getBindFlags() const +{ + return mBindFlags; +} + +int TextureStorage11::getTopLevel() const +{ + return mTopLevel; +} + +bool TextureStorage11::isRenderTarget() const +{ + return (mBindFlags & (D3D11_BIND_RENDER_TARGET | D3D11_BIND_DEPTH_STENCIL)) != 0; +} + +bool TextureStorage11::isManaged() const +{ + return false; +} + +int TextureStorage11::getLevelCount() const +{ + return mMipLevels - mTopLevel; +} + +int TextureStorage11::getLevelWidth(int mipLevel) const +{ + return std::max(static_cast<int>(mTextureWidth) >> mipLevel, 1); +} + +int TextureStorage11::getLevelHeight(int mipLevel) const +{ + return std::max(static_cast<int>(mTextureHeight) >> mipLevel, 1); +} + +int TextureStorage11::getLevelDepth(int mipLevel) const +{ + return std::max(static_cast<int>(mTextureDepth) >> mipLevel, 1); +} + +UINT TextureStorage11::getSubresourceIndex(int mipLevel, int layerTarget) const +{ + UINT index = 0; + if (getResource()) + { + index = D3D11CalcSubresource(mipLevel, layerTarget, mMipLevels); + } + return index; +} + +ID3D11ShaderResourceView *TextureStorage11::getSRV(const gl::SamplerState &samplerState) +{ + bool swizzleRequired = samplerState.swizzleRequired(); + bool mipmapping = gl::IsMipmapFiltered(samplerState); + unsigned int mipLevels = mipmapping ? (samplerState.maxLevel - samplerState.baseLevel) : 1; + + // Make sure there's 'mipLevels' mipmap levels below the base level (offset by the top level, which corresponds to GL level 0) + mipLevels = std::min(mipLevels, mMipLevels - mTopLevel - samplerState.baseLevel); + + if (swizzleRequired) + { + verifySwizzleExists(samplerState.swizzleRed, samplerState.swizzleGreen, samplerState.swizzleBlue, samplerState.swizzleAlpha); + } + + SRVKey key(samplerState.baseLevel, mipLevels, swizzleRequired); + ID3D11ShaderResourceView *srv = srvCache.find(key); + + if(srv) + { + return srv; + } + + DXGI_FORMAT format = (swizzleRequired ? mSwizzleShaderResourceFormat : mShaderResourceFormat); + ID3D11Resource *texture = swizzleRequired ? getSwizzleTexture() : getResource(); + + srv = createSRV(samplerState.baseLevel, mipLevels, format, texture); + + return srvCache.add(key, srv); +} + +ID3D11ShaderResourceView *TextureStorage11::getSRVLevel(int mipLevel) +{ + if (mipLevel >= 0 && mipLevel < getLevelCount()) + { + if (!mLevelSRVs[mipLevel]) + { + mLevelSRVs[mipLevel] = createSRV(mipLevel, 1, mShaderResourceFormat, getResource()); + } + + return mLevelSRVs[mipLevel]; + } + else + { + return NULL; + } +} + +gl::Error TextureStorage11::generateSwizzles(GLenum swizzleRed, GLenum swizzleGreen, GLenum swizzleBlue, GLenum swizzleAlpha) +{ + SwizzleCacheValue swizzleTarget(swizzleRed, swizzleGreen, swizzleBlue, swizzleAlpha); + for (int level = 0; level < getLevelCount(); level++) + { + // Check if the swizzle for this level is out of date + if (mSwizzleCache[level] != swizzleTarget) + { + // Need to re-render the swizzle for this level + ID3D11ShaderResourceView *sourceSRV = getSRVLevel(level); + ID3D11RenderTargetView *destRTV = getSwizzleRenderTarget(level); + + gl::Extents size(getLevelWidth(level), getLevelHeight(level), getLevelDepth(level)); + + Blit11 *blitter = mRenderer->getBlitter(); + + gl::Error error = blitter->swizzleTexture(sourceSRV, destRTV, size, swizzleRed, swizzleGreen, swizzleBlue, swizzleAlpha); + if (error.isError()) + { + return error; + } + + mSwizzleCache[level] = swizzleTarget; + } + } + + return gl::Error(GL_NO_ERROR); +} + +void TextureStorage11::invalidateSwizzleCacheLevel(int mipLevel) +{ + if (mipLevel >= 0 && static_cast<unsigned int>(mipLevel) < ArraySize(mSwizzleCache)) + { + // The default constructor of SwizzleCacheValue has GL_NONE for all channels which is not a + // valid swizzle combination + mSwizzleCache[mipLevel] = SwizzleCacheValue(); + } +} + +void TextureStorage11::invalidateSwizzleCache() +{ + for (unsigned int mipLevel = 0; mipLevel < ArraySize(mSwizzleCache); mipLevel++) + { + invalidateSwizzleCacheLevel(mipLevel); + } +} + +bool TextureStorage11::updateSubresourceLevel(ID3D11Resource *srcTexture, unsigned int sourceSubresource, + int level, int layerTarget, GLint xoffset, GLint yoffset, GLint zoffset, + GLsizei width, GLsizei height, GLsizei depth) +{ + if (srcTexture) + { + invalidateSwizzleCacheLevel(level); + + gl::Extents texSize(getLevelWidth(level), getLevelHeight(level), getLevelDepth(level)); + gl::Box copyArea(xoffset, yoffset, zoffset, width, height, depth); + + bool fullCopy = copyArea.x == 0 && + copyArea.y == 0 && + copyArea.z == 0 && + copyArea.width == texSize.width && + copyArea.height == texSize.height && + copyArea.depth == texSize.depth; + + ID3D11Resource *dstTexture = getResource(); + unsigned int dstSubresource = getSubresourceIndex(level + mTopLevel, layerTarget); + + ASSERT(dstTexture); + + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(mTextureFormat); + if (!fullCopy && (dxgiFormatInfo.depthBits > 0 || dxgiFormatInfo.stencilBits > 0)) + { + // CopySubresourceRegion cannot copy partial depth stencils, use the blitter instead + Blit11 *blitter = mRenderer->getBlitter(); + + return blitter->copyDepthStencil(srcTexture, sourceSubresource, copyArea, texSize, + dstTexture, dstSubresource, copyArea, texSize, + NULL); + } + else + { + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(mTextureFormat); + + D3D11_BOX srcBox; + srcBox.left = copyArea.x; + srcBox.top = copyArea.y; + srcBox.right = copyArea.x + roundUp((unsigned int)width, dxgiFormatInfo.blockWidth); + srcBox.bottom = copyArea.y + roundUp((unsigned int)height, dxgiFormatInfo.blockHeight); + srcBox.front = copyArea.z; + srcBox.back = copyArea.z + copyArea.depth; + + ID3D11DeviceContext *context = mRenderer->getDeviceContext(); + + context->CopySubresourceRegion(dstTexture, dstSubresource, copyArea.x, copyArea.y, copyArea.z, + srcTexture, sourceSubresource, fullCopy ? NULL : &srcBox); + return true; + } + } + + return false; +} + +bool TextureStorage11::copySubresourceLevel(ID3D11Resource* dstTexture, unsigned int dstSubresource, + int level, int layerTarget, GLint xoffset, GLint yoffset, GLint zoffset, + GLsizei width, GLsizei height, GLsizei depth) +{ + if (dstTexture) + { + ID3D11Resource *srcTexture = getResource(); + unsigned int srcSubresource = getSubresourceIndex(level + mTopLevel, layerTarget); + + ASSERT(srcTexture); + + ID3D11DeviceContext *context = mRenderer->getDeviceContext(); + + context->CopySubresourceRegion(dstTexture, dstSubresource, xoffset, yoffset, zoffset, + srcTexture, srcSubresource, NULL); + return true; + } + + return false; +} + +void TextureStorage11::generateMipmapLayer(RenderTarget11 *source, RenderTarget11 *dest) +{ + if (source && dest) + { + ID3D11ShaderResourceView *sourceSRV = source->getShaderResourceView(); + ID3D11RenderTargetView *destRTV = dest->getRenderTargetView(); + + if (sourceSRV && destRTV) + { + gl::Box sourceArea(0, 0, 0, source->getWidth(), source->getHeight(), source->getDepth()); + gl::Extents sourceSize(source->getWidth(), source->getHeight(), source->getDepth()); + + gl::Box destArea(0, 0, 0, dest->getWidth(), dest->getHeight(), dest->getDepth()); + gl::Extents destSize(dest->getWidth(), dest->getHeight(), dest->getDepth()); + + Blit11 *blitter = mRenderer->getBlitter(); + + blitter->copyTexture(sourceSRV, sourceArea, sourceSize, destRTV, destArea, destSize, NULL, + gl::GetInternalFormatInfo(source->getInternalFormat()).format, GL_LINEAR); + } + } +} + +void TextureStorage11::verifySwizzleExists(GLenum swizzleRed, GLenum swizzleGreen, GLenum swizzleBlue, GLenum swizzleAlpha) +{ + SwizzleCacheValue swizzleTarget(swizzleRed, swizzleGreen, swizzleBlue, swizzleAlpha); + for (unsigned int level = 0; level < mMipLevels; level++) + { + ASSERT(mSwizzleCache[level] == swizzleTarget); + } +} + +TextureStorage11_2D::TextureStorage11_2D(Renderer *renderer, SwapChain11 *swapchain) + : TextureStorage11(renderer, D3D11_BIND_RENDER_TARGET | D3D11_BIND_SHADER_RESOURCE), + mTexture(swapchain->getOffscreenTexture()), + mSwizzleTexture(NULL) +{ + mTexture->AddRef(); + + for (unsigned int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) + { + mAssociatedImages[i] = NULL; + mRenderTarget[i] = NULL; + mSwizzleRenderTargets[i] = NULL; + } + + D3D11_TEXTURE2D_DESC texDesc; + mTexture->GetDesc(&texDesc); + mMipLevels = texDesc.MipLevels; + mTextureFormat = texDesc.Format; + mTextureWidth = texDesc.Width; + mTextureHeight = texDesc.Height; + mTextureDepth = 1; + + ID3D11ShaderResourceView *srv = swapchain->getRenderTargetShaderResource(); + D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; + srv->GetDesc(&srvDesc); + mShaderResourceFormat = srvDesc.Format; + + ID3D11RenderTargetView* offscreenRTV = swapchain->getRenderTarget(); + D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; + offscreenRTV->GetDesc(&rtvDesc); + mRenderTargetFormat = rtvDesc.Format; + + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(mTextureFormat); + const d3d11::TextureFormat &formatInfo = d3d11::GetTextureFormatInfo(dxgiFormatInfo.internalFormat); + mSwizzleTextureFormat = formatInfo.swizzleTexFormat; + mSwizzleShaderResourceFormat = formatInfo.swizzleSRVFormat; + mSwizzleRenderTargetFormat = formatInfo.swizzleRTVFormat; + + mDepthStencilFormat = DXGI_FORMAT_UNKNOWN; + + initializeSerials(1, 1); +} + +TextureStorage11_2D::TextureStorage11_2D(Renderer *renderer, GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, int levels) + : TextureStorage11(renderer, GetTextureBindFlags(internalformat, renderTarget)), + mTexture(NULL), + mSwizzleTexture(NULL) +{ + for (unsigned int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) + { + mAssociatedImages[i] = NULL; + mRenderTarget[i] = NULL; + mSwizzleRenderTargets[i] = NULL; + } + + const d3d11::TextureFormat &formatInfo = d3d11::GetTextureFormatInfo(internalformat); + mTextureFormat = formatInfo.texFormat; + mShaderResourceFormat = formatInfo.srvFormat; + mDepthStencilFormat = formatInfo.dsvFormat; + mRenderTargetFormat = formatInfo.rtvFormat; + mSwizzleTextureFormat = formatInfo.swizzleTexFormat; + mSwizzleShaderResourceFormat = formatInfo.swizzleSRVFormat; + mSwizzleRenderTargetFormat = formatInfo.swizzleRTVFormat; + + // if the width or height is not positive this should be treated as an incomplete texture + // we handle that here by skipping the d3d texture creation + if (width > 0 && height > 0) + { + // adjust size if needed for compressed textures + d3d11::MakeValidSize(false, mTextureFormat, &width, &height, &mTopLevel); + + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_TEXTURE2D_DESC desc; + desc.Width = width; // Compressed texture size constraints? + desc.Height = height; + desc.MipLevels = desc.MipLevels = mRenderer->isLevel9() ? 1 : ((levels > 0) ? (mTopLevel + levels) : 0); + desc.ArraySize = 1; + desc.Format = mTextureFormat; + desc.SampleDesc.Count = 1; + desc.SampleDesc.Quality = 0; + desc.Usage = D3D11_USAGE_DEFAULT; + desc.BindFlags = getBindFlags(); + desc.CPUAccessFlags = 0; + desc.MiscFlags = 0; + + HRESULT result = device->CreateTexture2D(&desc, NULL, &mTexture); + + // this can happen from windows TDR + if (d3d11::isDeviceLostError(result)) + { + mRenderer->notifyDeviceLost(); + gl::error(GL_OUT_OF_MEMORY); + } + else if (FAILED(result)) + { + ASSERT(result == E_OUTOFMEMORY); + ERR("Creating image failed."); + gl::error(GL_OUT_OF_MEMORY); + } + else + { + mTexture->GetDesc(&desc); + mMipLevels = desc.MipLevels; + mTextureWidth = desc.Width; + mTextureHeight = desc.Height; + mTextureDepth = 1; + } + } + + initializeSerials(getLevelCount(), 1); +} + +TextureStorage11_2D::~TextureStorage11_2D() +{ + for (unsigned i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) + { + if (mAssociatedImages[i] != NULL) + { + bool imageAssociationCorrect = mAssociatedImages[i]->isAssociatedStorageValid(this); + ASSERT(imageAssociationCorrect); + + if (imageAssociationCorrect) + { + // We must let the Images recover their data before we delete it from the TextureStorage. + mAssociatedImages[i]->recoverFromAssociatedStorage(); + } + } + } + + SafeRelease(mTexture); + SafeRelease(mSwizzleTexture); + + for (unsigned int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) + { + SafeDelete(mRenderTarget[i]); + SafeRelease(mSwizzleRenderTargets[i]); + } +} + +TextureStorage11_2D *TextureStorage11_2D::makeTextureStorage11_2D(TextureStorage *storage) +{ + ASSERT(HAS_DYNAMIC_TYPE(TextureStorage11_2D*, storage)); + return static_cast<TextureStorage11_2D*>(storage); +} + +void TextureStorage11_2D::associateImage(Image11* image, int level, int layerTarget) +{ + ASSERT(0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + + if (0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + { + mAssociatedImages[level] = image; + } +} + +bool TextureStorage11_2D::isAssociatedImageValid(int level, int layerTarget, Image11* expectedImage) +{ + if (0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + { + // This validation check should never return false. It means the Image/TextureStorage association is broken. + bool retValue = (mAssociatedImages[level] == expectedImage); + ASSERT(retValue); + return retValue; + } + + return false; +} + +// disassociateImage allows an Image to end its association with a Storage. +void TextureStorage11_2D::disassociateImage(int level, int layerTarget, Image11* expectedImage) +{ + ASSERT(0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + + if (0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + { + ASSERT(mAssociatedImages[level] == expectedImage); + + if (mAssociatedImages[level] == expectedImage) + { + mAssociatedImages[level] = NULL; + } + } +} + +// releaseAssociatedImage prepares the Storage for a new Image association. It lets the old Image recover its data before ending the association. +void TextureStorage11_2D::releaseAssociatedImage(int level, int layerTarget, Image11* incomingImage) +{ + ASSERT(0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + + if (0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + { + // No need to let the old Image recover its data, if it is also the incoming Image. + if (mAssociatedImages[level] != NULL && mAssociatedImages[level] != incomingImage) + { + // Ensure that the Image is still associated with this TextureStorage. This should be true. + bool imageAssociationCorrect = mAssociatedImages[level]->isAssociatedStorageValid(this); + ASSERT(imageAssociationCorrect); + + if (imageAssociationCorrect) + { + // Force the image to recover from storage before its data is overwritten. + // This will reset mAssociatedImages[level] to NULL too. + mAssociatedImages[level]->recoverFromAssociatedStorage(); + } + } + } +} + +ID3D11Resource *TextureStorage11_2D::getResource() const +{ + return mTexture; +} + +RenderTarget *TextureStorage11_2D::getRenderTarget(const gl::ImageIndex &index) +{ + ASSERT(!index.hasLayer()); + + int level = index.mipIndex; + + if (level >= 0 && level < getLevelCount()) + { + if (!mRenderTarget[level]) + { + ID3D11ShaderResourceView *srv = getSRVLevel(level); + if (!srv) + { + return NULL; + } + + if (mRenderTargetFormat != DXGI_FORMAT_UNKNOWN) + { + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; + rtvDesc.Format = mRenderTargetFormat; + rtvDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2D; + rtvDesc.Texture2D.MipSlice = mTopLevel + level; + + ID3D11RenderTargetView *rtv; + HRESULT result = device->CreateRenderTargetView(mTexture, &rtvDesc, &rtv); + + if (result == E_OUTOFMEMORY) + { + return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + + mRenderTarget[level] = new RenderTarget11(mRenderer, rtv, mTexture, srv, getLevelWidth(level), getLevelHeight(level), 1); + + // RenderTarget will take ownership of these resources + SafeRelease(rtv); + } + else if (mDepthStencilFormat != DXGI_FORMAT_UNKNOWN) + { + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_DEPTH_STENCIL_VIEW_DESC dsvDesc; + dsvDesc.Format = mDepthStencilFormat; + dsvDesc.ViewDimension = D3D11_DSV_DIMENSION_TEXTURE2D; + dsvDesc.Texture2D.MipSlice = mTopLevel + level; + dsvDesc.Flags = 0; + + ID3D11DepthStencilView *dsv; + HRESULT result = device->CreateDepthStencilView(mTexture, &dsvDesc, &dsv); + + if (result == E_OUTOFMEMORY) + { + SafeRelease(srv); + return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + + mRenderTarget[level] = new RenderTarget11(mRenderer, dsv, mTexture, srv, getLevelWidth(level), getLevelHeight(level), 1); + + // RenderTarget will take ownership of these resources + SafeRelease(dsv); + } + else + { + UNREACHABLE(); + } + } + + return mRenderTarget[level]; + } + else + { + return NULL; + } +} + +ID3D11ShaderResourceView *TextureStorage11_2D::createSRV(int baseLevel, int mipLevels, DXGI_FORMAT format, ID3D11Resource *texture) +{ + D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; + srvDesc.Format = format; + srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2D; + srvDesc.Texture2D.MostDetailedMip = mTopLevel + baseLevel; + srvDesc.Texture2D.MipLevels = mipLevels; + + ID3D11ShaderResourceView *SRV = NULL; + + ID3D11Device *device = mRenderer->getDevice(); + HRESULT result = device->CreateShaderResourceView(texture, &srvDesc, &SRV); + + if (result == E_OUTOFMEMORY) + { + gl::error(GL_OUT_OF_MEMORY); + } + ASSERT(SUCCEEDED(result)); + + return SRV; +} + +void TextureStorage11_2D::generateMipmaps() +{ + // Base level must already be defined + + for (int level = 1; level < getLevelCount(); level++) + { + invalidateSwizzleCacheLevel(level); + + gl::ImageIndex srcIndex = gl::ImageIndex::Make2D(level - 1); + gl::ImageIndex destIndex = gl::ImageIndex::Make2D(level); + + RenderTarget11 *source = RenderTarget11::makeRenderTarget11(getRenderTarget(srcIndex)); + RenderTarget11 *dest = RenderTarget11::makeRenderTarget11(getRenderTarget(destIndex)); + + generateMipmapLayer(source, dest); + } +} + +ID3D11Resource *TextureStorage11_2D::getSwizzleTexture() +{ + if (!mSwizzleTexture) + { + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_TEXTURE2D_DESC desc; + desc.Width = mTextureWidth; + desc.Height = mTextureHeight; + desc.MipLevels = mMipLevels; + desc.ArraySize = 1; + desc.Format = mSwizzleTextureFormat; + desc.SampleDesc.Count = 1; + desc.SampleDesc.Quality = 0; + desc.Usage = D3D11_USAGE_DEFAULT; + desc.BindFlags = D3D11_BIND_SHADER_RESOURCE | D3D11_BIND_RENDER_TARGET; + desc.CPUAccessFlags = 0; + desc.MiscFlags = 0; + + HRESULT result = device->CreateTexture2D(&desc, NULL, &mSwizzleTexture); + + if (result == E_OUTOFMEMORY) + { + return gl::error(GL_OUT_OF_MEMORY, static_cast<ID3D11Texture2D*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + } + + return mSwizzleTexture; +} + +ID3D11RenderTargetView *TextureStorage11_2D::getSwizzleRenderTarget(int mipLevel) +{ + if (mipLevel >= 0 && mipLevel < getLevelCount()) + { + if (!mSwizzleRenderTargets[mipLevel]) + { + ID3D11Resource *swizzleTexture = getSwizzleTexture(); + if (!swizzleTexture) + { + return NULL; + } + + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; + rtvDesc.Format = mSwizzleRenderTargetFormat; + rtvDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2D; + rtvDesc.Texture2D.MipSlice = mTopLevel + mipLevel; + + HRESULT result = device->CreateRenderTargetView(mSwizzleTexture, &rtvDesc, &mSwizzleRenderTargets[mipLevel]); + if (result == E_OUTOFMEMORY) + { + return gl::error(GL_OUT_OF_MEMORY, static_cast<ID3D11RenderTargetView*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + } + + return mSwizzleRenderTargets[mipLevel]; + } + else + { + return NULL; + } +} + +TextureStorage11_Cube::TextureStorage11_Cube(Renderer *renderer, GLenum internalformat, bool renderTarget, int size, int levels) + : TextureStorage11(renderer, GetTextureBindFlags(internalformat, renderTarget)) +{ + mTexture = NULL; + mSwizzleTexture = NULL; + + for (unsigned int level = 0; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; level++) + { + mSwizzleRenderTargets[level] = NULL; + for (unsigned int face = 0; face < 6; face++) + { + mAssociatedImages[face][level] = NULL; + mRenderTarget[face][level] = NULL; + } + } + + const d3d11::TextureFormat &formatInfo = d3d11::GetTextureFormatInfo(internalformat); + mTextureFormat = formatInfo.texFormat; + mShaderResourceFormat = formatInfo.srvFormat; + mDepthStencilFormat = formatInfo.dsvFormat; + mRenderTargetFormat = formatInfo.rtvFormat; + mSwizzleTextureFormat = formatInfo.swizzleTexFormat; + mSwizzleShaderResourceFormat = formatInfo.swizzleSRVFormat; + mSwizzleRenderTargetFormat = formatInfo.swizzleRTVFormat; + + // if the size is not positive this should be treated as an incomplete texture + // we handle that here by skipping the d3d texture creation + if (size > 0) + { + // adjust size if needed for compressed textures + int height = size; + d3d11::MakeValidSize(false, mTextureFormat, &size, &height, &mTopLevel); + + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_TEXTURE2D_DESC desc; + desc.Width = size; + desc.Height = size; + desc.MipLevels = ((levels > 0) ? (mTopLevel + levels) : 0); + desc.ArraySize = 6; + desc.Format = mTextureFormat; + desc.SampleDesc.Count = 1; + desc.SampleDesc.Quality = 0; + desc.Usage = D3D11_USAGE_DEFAULT; + desc.BindFlags = getBindFlags(); + desc.CPUAccessFlags = 0; + desc.MiscFlags = D3D11_RESOURCE_MISC_TEXTURECUBE; + + HRESULT result = device->CreateTexture2D(&desc, NULL, &mTexture); + + if (FAILED(result)) + { + ASSERT(result == E_OUTOFMEMORY); + ERR("Creating image failed."); + gl::error(GL_OUT_OF_MEMORY); + } + else + { + mTexture->GetDesc(&desc); + mMipLevels = desc.MipLevels; + mTextureWidth = desc.Width; + mTextureHeight = desc.Height; + mTextureDepth = 1; + } + } + + initializeSerials(getLevelCount() * 6, 6); +} + + +TextureStorage11_Cube::~TextureStorage11_Cube() +{ + for (unsigned int level = 0; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; level++) + { + for (unsigned int face = 0; face < 6; face++) + { + if (mAssociatedImages[face][level] != NULL) + { + bool imageAssociationCorrect = mAssociatedImages[face][level]->isAssociatedStorageValid(this); + ASSERT(imageAssociationCorrect); + + if (imageAssociationCorrect) + { + // We must let the Images recover their data before we delete it from the TextureStorage. + mAssociatedImages[face][level]->recoverFromAssociatedStorage(); + } + } + } + } + + SafeRelease(mTexture); + SafeRelease(mSwizzleTexture); + + for (unsigned int level = 0; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; level++) + { + SafeRelease(mSwizzleRenderTargets[level]); + for (unsigned int face = 0; face < 6; face++) + { + SafeDelete(mRenderTarget[face][level]); + } + } +} + +TextureStorage11_Cube *TextureStorage11_Cube::makeTextureStorage11_Cube(TextureStorage *storage) +{ + ASSERT(HAS_DYNAMIC_TYPE(TextureStorage11_Cube*, storage)); + return static_cast<TextureStorage11_Cube*>(storage); +} + +void TextureStorage11_Cube::associateImage(Image11* image, int level, int layerTarget) +{ + ASSERT(0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + ASSERT(0 <= layerTarget && layerTarget < 6); + + if (0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + { + if (0 <= layerTarget && layerTarget < 6) + { + mAssociatedImages[layerTarget][level] = image; + } + } +} + +bool TextureStorage11_Cube::isAssociatedImageValid(int level, int layerTarget, Image11* expectedImage) +{ + if (0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + { + if (0 <= layerTarget && layerTarget < 6) + { + // This validation check should never return false. It means the Image/TextureStorage association is broken. + bool retValue = (mAssociatedImages[layerTarget][level] == expectedImage); + ASSERT(retValue); + return retValue; + } + } + + return false; +} + +// disassociateImage allows an Image to end its association with a Storage. +void TextureStorage11_Cube::disassociateImage(int level, int layerTarget, Image11* expectedImage) +{ + ASSERT(0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + ASSERT(0 <= layerTarget && layerTarget < 6); + + if (0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + { + if (0 <= layerTarget && layerTarget < 6) + { + ASSERT(mAssociatedImages[layerTarget][level] == expectedImage); + + if (mAssociatedImages[layerTarget][level] == expectedImage) + { + mAssociatedImages[layerTarget][level] = NULL; + } + } + } +} + +// releaseAssociatedImage prepares the Storage for a new Image association. It lets the old Image recover its data before ending the association. +void TextureStorage11_Cube::releaseAssociatedImage(int level, int layerTarget, Image11* incomingImage) +{ + ASSERT(0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + ASSERT(0 <= layerTarget && layerTarget < 6); + + if ((0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS)) + { + if (0 <= layerTarget && layerTarget < 6) + { + // No need to let the old Image recover its data, if it is also the incoming Image. + if (mAssociatedImages[layerTarget][level] != NULL && mAssociatedImages[layerTarget][level] != incomingImage) + { + // Ensure that the Image is still associated with this TextureStorage. This should be true. + bool imageAssociationCorrect = mAssociatedImages[layerTarget][level]->isAssociatedStorageValid(this); + ASSERT(imageAssociationCorrect); + + if (imageAssociationCorrect) + { + // Force the image to recover from storage before its data is overwritten. + // This will reset mAssociatedImages[level] to NULL too. + mAssociatedImages[layerTarget][level]->recoverFromAssociatedStorage(); + } + } + } + } +} + +ID3D11Resource *TextureStorage11_Cube::getResource() const +{ + return mTexture; +} + +RenderTarget *TextureStorage11_Cube::getRenderTarget(const gl::ImageIndex &index) +{ + int faceIndex = index.layerIndex; + int level = index.mipIndex; + + if (level >= 0 && level < getLevelCount()) + { + if (!mRenderTarget[faceIndex][level]) + { + ID3D11Device *device = mRenderer->getDevice(); + HRESULT result; + + D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; + srvDesc.Format = mShaderResourceFormat; + srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2DARRAY; // Will be used with Texture2D sampler, not TextureCube + srvDesc.Texture2DArray.MostDetailedMip = mTopLevel + level; + srvDesc.Texture2DArray.MipLevels = 1; + srvDesc.Texture2DArray.FirstArraySlice = faceIndex; + srvDesc.Texture2DArray.ArraySize = 1; + + ID3D11ShaderResourceView *srv; + result = device->CreateShaderResourceView(mTexture, &srvDesc, &srv); + + if (result == E_OUTOFMEMORY) + { + return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + + if (mRenderTargetFormat != DXGI_FORMAT_UNKNOWN) + { + D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; + rtvDesc.Format = mRenderTargetFormat; + rtvDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2DARRAY; + rtvDesc.Texture2DArray.MipSlice = mTopLevel + level; + rtvDesc.Texture2DArray.FirstArraySlice = faceIndex; + rtvDesc.Texture2DArray.ArraySize = 1; + + ID3D11RenderTargetView *rtv; + result = device->CreateRenderTargetView(mTexture, &rtvDesc, &rtv); + + if (result == E_OUTOFMEMORY) + { + SafeRelease(srv); + return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + + mRenderTarget[faceIndex][level] = new RenderTarget11(mRenderer, rtv, mTexture, srv, getLevelWidth(level), getLevelHeight(level), 1); + + // RenderTarget will take ownership of these resources + SafeRelease(rtv); + SafeRelease(srv); + } + else if (mDepthStencilFormat != DXGI_FORMAT_UNKNOWN) + { + D3D11_DEPTH_STENCIL_VIEW_DESC dsvDesc; + dsvDesc.Format = mDepthStencilFormat; + dsvDesc.ViewDimension = D3D11_DSV_DIMENSION_TEXTURE2DARRAY; + dsvDesc.Flags = 0; + dsvDesc.Texture2DArray.MipSlice = mTopLevel + level; + dsvDesc.Texture2DArray.FirstArraySlice = faceIndex; + dsvDesc.Texture2DArray.ArraySize = 1; + + ID3D11DepthStencilView *dsv; + result = device->CreateDepthStencilView(mTexture, &dsvDesc, &dsv); + + if (result == E_OUTOFMEMORY) + { + SafeRelease(srv); + return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + + mRenderTarget[faceIndex][level] = new RenderTarget11(mRenderer, dsv, mTexture, srv, getLevelWidth(level), getLevelHeight(level), 1); + + // RenderTarget will take ownership of these resources + SafeRelease(dsv); + SafeRelease(srv); + } + else + { + UNREACHABLE(); + } + } + + return mRenderTarget[faceIndex][level]; + } + else + { + return NULL; + } +} + +ID3D11ShaderResourceView *TextureStorage11_Cube::createSRV(int baseLevel, int mipLevels, DXGI_FORMAT format, ID3D11Resource *texture) +{ + D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; + srvDesc.Format = format; + + // Unnormalized integer cube maps are not supported by DX11; we emulate them as an array of six 2D textures + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(format); + if (dxgiFormatInfo.componentType == GL_INT || dxgiFormatInfo.componentType == GL_UNSIGNED_INT) + { + srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2DARRAY; + srvDesc.Texture2DArray.MostDetailedMip = mTopLevel + baseLevel; + srvDesc.Texture2DArray.MipLevels = 1; + srvDesc.Texture2DArray.FirstArraySlice = 0; + srvDesc.Texture2DArray.ArraySize = 6; + } + else + { + srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURECUBE; + srvDesc.TextureCube.MipLevels = mipLevels; + srvDesc.TextureCube.MostDetailedMip = mTopLevel + baseLevel; + } + + ID3D11ShaderResourceView *SRV = NULL; + + ID3D11Device *device = mRenderer->getDevice(); + HRESULT result = device->CreateShaderResourceView(texture, &srvDesc, &SRV); + + if (result == E_OUTOFMEMORY) + { + gl::error(GL_OUT_OF_MEMORY); + } + ASSERT(SUCCEEDED(result)); + + return SRV; +} + +void TextureStorage11_Cube::generateMipmaps() +{ + // Base level must already be defined + + for (int faceIndex = 0; faceIndex < 6; faceIndex++) + { + for (int level = 1; level < getLevelCount(); level++) + { + invalidateSwizzleCacheLevel(level); + + gl::ImageIndex srcIndex = gl::ImageIndex::MakeCube(GL_TEXTURE_CUBE_MAP_POSITIVE_X + faceIndex, level - 1); + gl::ImageIndex destIndex = gl::ImageIndex::MakeCube(GL_TEXTURE_CUBE_MAP_POSITIVE_X + faceIndex, level); + + RenderTarget11 *source = RenderTarget11::makeRenderTarget11(getRenderTarget(srcIndex)); + RenderTarget11 *dest = RenderTarget11::makeRenderTarget11(getRenderTarget(destIndex)); + + generateMipmapLayer(source, dest); + } + } +} + +ID3D11Resource *TextureStorage11_Cube::getSwizzleTexture() +{ + if (!mSwizzleTexture) + { + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_TEXTURE2D_DESC desc; + desc.Width = mTextureWidth; + desc.Height = mTextureHeight; + desc.MipLevels = mMipLevels; + desc.ArraySize = 6; + desc.Format = mSwizzleTextureFormat; + desc.SampleDesc.Count = 1; + desc.SampleDesc.Quality = 0; + desc.Usage = D3D11_USAGE_DEFAULT; + desc.BindFlags = D3D11_BIND_SHADER_RESOURCE | D3D11_BIND_RENDER_TARGET; + desc.CPUAccessFlags = 0; + desc.MiscFlags = D3D11_RESOURCE_MISC_TEXTURECUBE; + + HRESULT result = device->CreateTexture2D(&desc, NULL, &mSwizzleTexture); + + if (result == E_OUTOFMEMORY) + { + return gl::error(GL_OUT_OF_MEMORY, static_cast<ID3D11Texture2D*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + } + + return mSwizzleTexture; +} + +ID3D11RenderTargetView *TextureStorage11_Cube::getSwizzleRenderTarget(int mipLevel) +{ + if (mipLevel >= 0 && mipLevel < getLevelCount()) + { + if (!mSwizzleRenderTargets[mipLevel]) + { + ID3D11Resource *swizzleTexture = getSwizzleTexture(); + if (!swizzleTexture) + { + return NULL; + } + + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; + rtvDesc.Format = mSwizzleRenderTargetFormat; + rtvDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2DARRAY; + rtvDesc.Texture2DArray.MipSlice = mTopLevel + mipLevel; + rtvDesc.Texture2DArray.FirstArraySlice = 0; + rtvDesc.Texture2DArray.ArraySize = 6; + + HRESULT result = device->CreateRenderTargetView(mSwizzleTexture, &rtvDesc, &mSwizzleRenderTargets[mipLevel]); + + if (result == E_OUTOFMEMORY) + { + return gl::error(GL_OUT_OF_MEMORY, static_cast<ID3D11RenderTargetView*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + } + + return mSwizzleRenderTargets[mipLevel]; + } + else + { + return NULL; + } +} + +TextureStorage11_3D::TextureStorage11_3D(Renderer *renderer, GLenum internalformat, bool renderTarget, + GLsizei width, GLsizei height, GLsizei depth, int levels) + : TextureStorage11(renderer, GetTextureBindFlags(internalformat, renderTarget)) +{ + mTexture = NULL; + mSwizzleTexture = NULL; + + for (unsigned int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) + { + mAssociatedImages[i] = NULL; + mLevelRenderTargets[i] = NULL; + mSwizzleRenderTargets[i] = NULL; + } + + const d3d11::TextureFormat &formatInfo = d3d11::GetTextureFormatInfo(internalformat); + mTextureFormat = formatInfo.texFormat; + mShaderResourceFormat = formatInfo.srvFormat; + mDepthStencilFormat = formatInfo.dsvFormat; + mRenderTargetFormat = formatInfo.rtvFormat; + mSwizzleTextureFormat = formatInfo.swizzleTexFormat; + mSwizzleShaderResourceFormat = formatInfo.swizzleSRVFormat; + mSwizzleRenderTargetFormat = formatInfo.swizzleRTVFormat; + + // If the width, height or depth are not positive this should be treated as an incomplete texture + // we handle that here by skipping the d3d texture creation + if (width > 0 && height > 0 && depth > 0) + { + // adjust size if needed for compressed textures + d3d11::MakeValidSize(false, mTextureFormat, &width, &height, &mTopLevel); + + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_TEXTURE3D_DESC desc; + desc.Width = width; + desc.Height = height; + desc.Depth = depth; + desc.MipLevels = ((levels > 0) ? (mTopLevel + levels) : 0); + desc.Format = mTextureFormat; + desc.Usage = D3D11_USAGE_DEFAULT; + desc.BindFlags = getBindFlags(); + desc.CPUAccessFlags = 0; + desc.MiscFlags = 0; + + HRESULT result = device->CreateTexture3D(&desc, NULL, &mTexture); + + // this can happen from windows TDR + if (d3d11::isDeviceLostError(result)) + { + mRenderer->notifyDeviceLost(); + gl::error(GL_OUT_OF_MEMORY); + } + else if (FAILED(result)) + { + ASSERT(result == E_OUTOFMEMORY); + ERR("Creating image failed."); + gl::error(GL_OUT_OF_MEMORY); + } + else + { + mTexture->GetDesc(&desc); + mMipLevels = desc.MipLevels; + mTextureWidth = desc.Width; + mTextureHeight = desc.Height; + mTextureDepth = desc.Depth; + } + } + + initializeSerials(getLevelCount() * depth, depth); +} + +TextureStorage11_3D::~TextureStorage11_3D() +{ + for (unsigned i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) + { + if (mAssociatedImages[i] != NULL) + { + bool imageAssociationCorrect = mAssociatedImages[i]->isAssociatedStorageValid(this); + ASSERT(imageAssociationCorrect); + + if (imageAssociationCorrect) + { + // We must let the Images recover their data before we delete it from the TextureStorage. + mAssociatedImages[i]->recoverFromAssociatedStorage(); + } + } + } + + SafeRelease(mTexture); + SafeRelease(mSwizzleTexture); + + for (RenderTargetMap::iterator i = mLevelLayerRenderTargets.begin(); i != mLevelLayerRenderTargets.end(); i++) + { + SafeDelete(i->second); + } + mLevelLayerRenderTargets.clear(); + + for (unsigned int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) + { + SafeDelete(mLevelRenderTargets[i]); + SafeRelease(mSwizzleRenderTargets[i]); + } +} + +TextureStorage11_3D *TextureStorage11_3D::makeTextureStorage11_3D(TextureStorage *storage) +{ + ASSERT(HAS_DYNAMIC_TYPE(TextureStorage11_3D*, storage)); + return static_cast<TextureStorage11_3D*>(storage); +} + +void TextureStorage11_3D::associateImage(Image11* image, int level, int layerTarget) +{ + ASSERT(0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + + if (0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + { + mAssociatedImages[level] = image; + } +} + +bool TextureStorage11_3D::isAssociatedImageValid(int level, int layerTarget, Image11* expectedImage) +{ + if (0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + { + // This validation check should never return false. It means the Image/TextureStorage association is broken. + bool retValue = (mAssociatedImages[level] == expectedImage); + ASSERT(retValue); + return retValue; + } + + return false; +} + +// disassociateImage allows an Image to end its association with a Storage. +void TextureStorage11_3D::disassociateImage(int level, int layerTarget, Image11* expectedImage) +{ + ASSERT(0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS); + + if (0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + { + ASSERT(mAssociatedImages[level] == expectedImage); + + if (mAssociatedImages[level] == expectedImage) + { + mAssociatedImages[level] = NULL; + } + } +} + +// releaseAssociatedImage prepares the Storage for a new Image association. It lets the old Image recover its data before ending the association. +void TextureStorage11_3D::releaseAssociatedImage(int level, int layerTarget, Image11* incomingImage) +{ + ASSERT((0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS)); + + if (0 <= level && level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS) + { + // No need to let the old Image recover its data, if it is also the incoming Image. + if (mAssociatedImages[level] != NULL && mAssociatedImages[level] != incomingImage) + { + // Ensure that the Image is still associated with this TextureStorage. This should be true. + bool imageAssociationCorrect = mAssociatedImages[level]->isAssociatedStorageValid(this); + ASSERT(imageAssociationCorrect); + + if (imageAssociationCorrect) + { + // Force the image to recover from storage before its data is overwritten. + // This will reset mAssociatedImages[level] to NULL too. + mAssociatedImages[level]->recoverFromAssociatedStorage(); + } + } + } +} + +ID3D11Resource *TextureStorage11_3D::getResource() const +{ + return mTexture; +} + +ID3D11ShaderResourceView *TextureStorage11_3D::createSRV(int baseLevel, int mipLevels, DXGI_FORMAT format, ID3D11Resource *texture) +{ + D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; + srvDesc.Format = format; + srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE3D; + srvDesc.Texture3D.MostDetailedMip = baseLevel; + srvDesc.Texture3D.MipLevels = mipLevels; + + ID3D11ShaderResourceView *SRV = NULL; + + ID3D11Device *device = mRenderer->getDevice(); + HRESULT result = device->CreateShaderResourceView(texture, &srvDesc, &SRV); + + if (result == E_OUTOFMEMORY) + { + gl::error(GL_OUT_OF_MEMORY); + } + ASSERT(SUCCEEDED(result)); + + return SRV; +} + +RenderTarget *TextureStorage11_3D::getRenderTarget(const gl::ImageIndex &index) +{ + int mipLevel = index.mipIndex; + + if (mipLevel >= 0 && mipLevel < getLevelCount()) + { + ASSERT(mRenderTargetFormat != DXGI_FORMAT_UNKNOWN); + + if (!index.hasLayer()) + { + if (!mLevelRenderTargets[mipLevel]) + { + ID3D11ShaderResourceView *srv = getSRVLevel(mipLevel); + if (!srv) + { + return NULL; + } + + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; + rtvDesc.Format = mRenderTargetFormat; + rtvDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE3D; + rtvDesc.Texture3D.MipSlice = mTopLevel + mipLevel; + rtvDesc.Texture3D.FirstWSlice = 0; + rtvDesc.Texture3D.WSize = -1; + + ID3D11RenderTargetView *rtv; + HRESULT result = device->CreateRenderTargetView(mTexture, &rtvDesc, &rtv); + + if (result == E_OUTOFMEMORY) + { + SafeRelease(srv); + return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + + mLevelRenderTargets[mipLevel] = new RenderTarget11(mRenderer, rtv, mTexture, srv, getLevelWidth(mipLevel), getLevelHeight(mipLevel), getLevelDepth(mipLevel)); + + // RenderTarget will take ownership of these resources + SafeRelease(rtv); + } + + return mLevelRenderTargets[mipLevel]; + } + else + { + int layer = index.layerIndex; + + LevelLayerKey key(mipLevel, layer); + if (mLevelLayerRenderTargets.find(key) == mLevelLayerRenderTargets.end()) + { + ID3D11Device *device = mRenderer->getDevice(); + HRESULT result; + + // TODO, what kind of SRV is expected here? + ID3D11ShaderResourceView *srv = NULL; + + D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; + rtvDesc.Format = mRenderTargetFormat; + rtvDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE3D; + rtvDesc.Texture3D.MipSlice = mTopLevel + mipLevel; + rtvDesc.Texture3D.FirstWSlice = layer; + rtvDesc.Texture3D.WSize = 1; + + ID3D11RenderTargetView *rtv; + result = device->CreateRenderTargetView(mTexture, &rtvDesc, &rtv); + + if (result == E_OUTOFMEMORY) + { + SafeRelease(srv); + return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + + mLevelLayerRenderTargets[key] = new RenderTarget11(mRenderer, rtv, mTexture, srv, getLevelWidth(mipLevel), getLevelHeight(mipLevel), 1); + + // RenderTarget will take ownership of these resources + SafeRelease(rtv); + SafeRelease(srv); + } + + return mLevelLayerRenderTargets[key]; + } + } + + return NULL; +} + +void TextureStorage11_3D::generateMipmaps() +{ + // Base level must already be defined + + for (int level = 1; level < getLevelCount(); level++) + { + invalidateSwizzleCacheLevel(level); + + gl::ImageIndex srcIndex = gl::ImageIndex::Make3D(level - 1); + gl::ImageIndex destIndex = gl::ImageIndex::Make3D(level); + + RenderTarget11 *source = RenderTarget11::makeRenderTarget11(getRenderTarget(srcIndex)); + RenderTarget11 *dest = RenderTarget11::makeRenderTarget11(getRenderTarget(destIndex)); + + generateMipmapLayer(source, dest); + } +} + +ID3D11Resource *TextureStorage11_3D::getSwizzleTexture() +{ + if (!mSwizzleTexture) + { + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_TEXTURE3D_DESC desc; + desc.Width = mTextureWidth; + desc.Height = mTextureHeight; + desc.Depth = mTextureDepth; + desc.MipLevels = mMipLevels; + desc.Format = mSwizzleTextureFormat; + desc.Usage = D3D11_USAGE_DEFAULT; + desc.BindFlags = D3D11_BIND_SHADER_RESOURCE | D3D11_BIND_RENDER_TARGET; + desc.CPUAccessFlags = 0; + desc.MiscFlags = 0; + + HRESULT result = device->CreateTexture3D(&desc, NULL, &mSwizzleTexture); + + if (result == E_OUTOFMEMORY) + { + return gl::error(GL_OUT_OF_MEMORY, static_cast<ID3D11Texture3D*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + } + + return mSwizzleTexture; +} + +ID3D11RenderTargetView *TextureStorage11_3D::getSwizzleRenderTarget(int mipLevel) +{ + if (mipLevel >= 0 && mipLevel < getLevelCount()) + { + if (!mSwizzleRenderTargets[mipLevel]) + { + ID3D11Resource *swizzleTexture = getSwizzleTexture(); + if (!swizzleTexture) + { + return NULL; + } + + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; + rtvDesc.Format = mSwizzleRenderTargetFormat; + rtvDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE3D; + rtvDesc.Texture3D.MipSlice = mTopLevel + mipLevel; + rtvDesc.Texture3D.FirstWSlice = 0; + rtvDesc.Texture3D.WSize = -1; + + HRESULT result = device->CreateRenderTargetView(mSwizzleTexture, &rtvDesc, &mSwizzleRenderTargets[mipLevel]); + + if (result == E_OUTOFMEMORY) + { + return gl::error(GL_OUT_OF_MEMORY, static_cast<ID3D11RenderTargetView*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + } + + return mSwizzleRenderTargets[mipLevel]; + } + else + { + return NULL; + } +} + +TextureStorage11_2DArray::TextureStorage11_2DArray(Renderer *renderer, GLenum internalformat, bool renderTarget, + GLsizei width, GLsizei height, GLsizei depth, int levels) + : TextureStorage11(renderer, GetTextureBindFlags(internalformat, renderTarget)) +{ + mTexture = NULL; + mSwizzleTexture = NULL; + + for (unsigned int level = 0; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; level++) + { + mSwizzleRenderTargets[level] = NULL; + } + + const d3d11::TextureFormat &formatInfo = d3d11::GetTextureFormatInfo(internalformat); + mTextureFormat = formatInfo.texFormat; + mShaderResourceFormat = formatInfo.srvFormat; + mDepthStencilFormat = formatInfo.dsvFormat; + mRenderTargetFormat = formatInfo.rtvFormat; + mSwizzleTextureFormat = formatInfo.swizzleTexFormat; + mSwizzleShaderResourceFormat = formatInfo.swizzleSRVFormat; + mSwizzleRenderTargetFormat = formatInfo.swizzleRTVFormat; + + // if the width, height or depth is not positive this should be treated as an incomplete texture + // we handle that here by skipping the d3d texture creation + if (width > 0 && height > 0 && depth > 0) + { + // adjust size if needed for compressed textures + d3d11::MakeValidSize(false, mTextureFormat, &width, &height, &mTopLevel); + + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_TEXTURE2D_DESC desc; + desc.Width = width; + desc.Height = height; + desc.MipLevels = ((levels > 0) ? (mTopLevel + levels) : 0); + desc.ArraySize = depth; + desc.Format = mTextureFormat; + desc.SampleDesc.Count = 1; + desc.SampleDesc.Quality = 0; + desc.Usage = D3D11_USAGE_DEFAULT; + desc.BindFlags = getBindFlags(); + desc.CPUAccessFlags = 0; + desc.MiscFlags = 0; + + HRESULT result = device->CreateTexture2D(&desc, NULL, &mTexture); + + // this can happen from windows TDR + if (d3d11::isDeviceLostError(result)) + { + mRenderer->notifyDeviceLost(); + gl::error(GL_OUT_OF_MEMORY); + } + else if (FAILED(result)) + { + ASSERT(result == E_OUTOFMEMORY); + ERR("Creating image failed."); + gl::error(GL_OUT_OF_MEMORY); + } + else + { + mTexture->GetDesc(&desc); + mMipLevels = desc.MipLevels; + mTextureWidth = desc.Width; + mTextureHeight = desc.Height; + mTextureDepth = desc.ArraySize; + } + } + + initializeSerials(getLevelCount() * depth, depth); +} + +TextureStorage11_2DArray::~TextureStorage11_2DArray() +{ + for (ImageMap::iterator i = mAssociatedImages.begin(); i != mAssociatedImages.end(); i++) + { + bool imageAssociationCorrect = i->second->isAssociatedStorageValid(this); + ASSERT(imageAssociationCorrect); + + if (imageAssociationCorrect) + { + // We must let the Images recover their data before we delete it from the TextureStorage. + i->second->recoverFromAssociatedStorage(); + } + } + mAssociatedImages.clear(); + + SafeRelease(mTexture); + SafeRelease(mSwizzleTexture); + + for (unsigned int level = 0; level < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; level++) + { + SafeRelease(mSwizzleRenderTargets[level]); + } + + for (RenderTargetMap::iterator i = mRenderTargets.begin(); i != mRenderTargets.end(); i++) + { + SafeDelete(i->second); + } + mRenderTargets.clear(); +} + +TextureStorage11_2DArray *TextureStorage11_2DArray::makeTextureStorage11_2DArray(TextureStorage *storage) +{ + ASSERT(HAS_DYNAMIC_TYPE(TextureStorage11_2DArray*, storage)); + return static_cast<TextureStorage11_2DArray*>(storage); +} + +void TextureStorage11_2DArray::associateImage(Image11* image, int level, int layerTarget) +{ + ASSERT(0 <= level && level < getLevelCount()); + + if (0 <= level && level < getLevelCount()) + { + LevelLayerKey key(level, layerTarget); + mAssociatedImages[key] = image; + } +} + +bool TextureStorage11_2DArray::isAssociatedImageValid(int level, int layerTarget, Image11* expectedImage) +{ + LevelLayerKey key(level, layerTarget); + + // This validation check should never return false. It means the Image/TextureStorage association is broken. + bool retValue = (mAssociatedImages.find(key) != mAssociatedImages.end() && (mAssociatedImages[key] == expectedImage)); + ASSERT(retValue); + return retValue; +} + +// disassociateImage allows an Image to end its association with a Storage. +void TextureStorage11_2DArray::disassociateImage(int level, int layerTarget, Image11* expectedImage) +{ + LevelLayerKey key(level, layerTarget); + + bool imageAssociationCorrect = (mAssociatedImages.find(key) != mAssociatedImages.end() && (mAssociatedImages[key] == expectedImage)); + ASSERT(imageAssociationCorrect); + + if (imageAssociationCorrect) + { + mAssociatedImages[key] = NULL; + } +} + +// releaseAssociatedImage prepares the Storage for a new Image association. It lets the old Image recover its data before ending the association. +void TextureStorage11_2DArray::releaseAssociatedImage(int level, int layerTarget, Image11* incomingImage) +{ + LevelLayerKey key(level, layerTarget); + + ASSERT(mAssociatedImages.find(key) != mAssociatedImages.end()); + + if (mAssociatedImages.find(key) != mAssociatedImages.end()) + { + if (mAssociatedImages[key] != NULL && mAssociatedImages[key] != incomingImage) + { + // Ensure that the Image is still associated with this TextureStorage. This should be true. + bool imageAssociationCorrect = mAssociatedImages[key]->isAssociatedStorageValid(this); + ASSERT(imageAssociationCorrect); + + if (imageAssociationCorrect) + { + // Force the image to recover from storage before its data is overwritten. + // This will reset mAssociatedImages[level] to NULL too. + mAssociatedImages[key]->recoverFromAssociatedStorage(); + } + } + } +} + +ID3D11Resource *TextureStorage11_2DArray::getResource() const +{ + return mTexture; +} + +ID3D11ShaderResourceView *TextureStorage11_2DArray::createSRV(int baseLevel, int mipLevels, DXGI_FORMAT format, ID3D11Resource *texture) +{ + D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; + srvDesc.Format = format; + srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2DARRAY; + srvDesc.Texture2DArray.MostDetailedMip = mTopLevel + baseLevel; + srvDesc.Texture2DArray.MipLevels = mipLevels; + srvDesc.Texture2DArray.FirstArraySlice = 0; + srvDesc.Texture2DArray.ArraySize = mTextureDepth; + + ID3D11ShaderResourceView *SRV = NULL; + + ID3D11Device *device = mRenderer->getDevice(); + HRESULT result = device->CreateShaderResourceView(texture, &srvDesc, &SRV); + + if (result == E_OUTOFMEMORY) + { + gl::error(GL_OUT_OF_MEMORY); + } + ASSERT(SUCCEEDED(result)); + + return SRV; +} + +RenderTarget *TextureStorage11_2DArray::getRenderTarget(const gl::ImageIndex &index) +{ + ASSERT(index.hasLayer()); + + int mipLevel = index.mipIndex; + int layer = index.layerIndex; + + if (mipLevel >= 0 && mipLevel < getLevelCount()) + { + LevelLayerKey key(mipLevel, layer); + if (mRenderTargets.find(key) == mRenderTargets.end()) + { + ID3D11Device *device = mRenderer->getDevice(); + HRESULT result; + + D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; + srvDesc.Format = mShaderResourceFormat; + srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2DARRAY; + srvDesc.Texture2DArray.MostDetailedMip = mTopLevel + mipLevel; + srvDesc.Texture2DArray.MipLevels = 1; + srvDesc.Texture2DArray.FirstArraySlice = layer; + srvDesc.Texture2DArray.ArraySize = 1; + + ID3D11ShaderResourceView *srv; + result = device->CreateShaderResourceView(mTexture, &srvDesc, &srv); + + if (result == E_OUTOFMEMORY) + { + return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + + if (mRenderTargetFormat != DXGI_FORMAT_UNKNOWN) + { + D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; + rtvDesc.Format = mRenderTargetFormat; + rtvDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2DARRAY; + rtvDesc.Texture2DArray.MipSlice = mTopLevel + mipLevel; + rtvDesc.Texture2DArray.FirstArraySlice = layer; + rtvDesc.Texture2DArray.ArraySize = 1; + + ID3D11RenderTargetView *rtv; + result = device->CreateRenderTargetView(mTexture, &rtvDesc, &rtv); + + if (result == E_OUTOFMEMORY) + { + SafeRelease(srv); + return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + + mRenderTargets[key] = new RenderTarget11(mRenderer, rtv, mTexture, srv, getLevelWidth(mipLevel), getLevelHeight(mipLevel), 1); + + // RenderTarget will take ownership of these resources + SafeRelease(rtv); + SafeRelease(srv); + } + else + { + UNREACHABLE(); + } + } + + return mRenderTargets[key]; + } + else + { + return NULL; + } +} + +void TextureStorage11_2DArray::generateMipmaps() +{ + // Base level must already be defined + + for (int level = 0; level < getLevelCount(); level++) + { + invalidateSwizzleCacheLevel(level); + for (unsigned int layer = 0; layer < mTextureDepth; layer++) + { + gl::ImageIndex sourceIndex = gl::ImageIndex::Make2DArray(level - 1, layer); + gl::ImageIndex destIndex = gl::ImageIndex::Make2DArray(level, layer); + + RenderTarget11 *source = RenderTarget11::makeRenderTarget11(getRenderTarget(sourceIndex)); + RenderTarget11 *dest = RenderTarget11::makeRenderTarget11(getRenderTarget(destIndex)); + + generateMipmapLayer(source, dest); + } + } +} + +ID3D11Resource *TextureStorage11_2DArray::getSwizzleTexture() +{ + if (!mSwizzleTexture) + { + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_TEXTURE2D_DESC desc; + desc.Width = mTextureWidth; + desc.Height = mTextureHeight; + desc.MipLevels = mMipLevels; + desc.ArraySize = mTextureDepth; + desc.Format = mSwizzleTextureFormat; + desc.SampleDesc.Count = 1; + desc.SampleDesc.Quality = 0; + desc.Usage = D3D11_USAGE_DEFAULT; + desc.BindFlags = D3D11_BIND_SHADER_RESOURCE | D3D11_BIND_RENDER_TARGET; + desc.CPUAccessFlags = 0; + desc.MiscFlags = 0; + + HRESULT result = device->CreateTexture2D(&desc, NULL, &mSwizzleTexture); + + if (result == E_OUTOFMEMORY) + { + return gl::error(GL_OUT_OF_MEMORY, static_cast<ID3D11Texture2D*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + } + + return mSwizzleTexture; +} + +ID3D11RenderTargetView *TextureStorage11_2DArray::getSwizzleRenderTarget(int mipLevel) +{ + if (mipLevel >= 0 && mipLevel < getLevelCount()) + { + if (!mSwizzleRenderTargets[mipLevel]) + { + ID3D11Resource *swizzleTexture = getSwizzleTexture(); + if (!swizzleTexture) + { + return NULL; + } + + ID3D11Device *device = mRenderer->getDevice(); + + D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; + rtvDesc.Format = mSwizzleRenderTargetFormat; + rtvDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2DARRAY; + rtvDesc.Texture2DArray.MipSlice = mTopLevel + mipLevel; + rtvDesc.Texture2DArray.FirstArraySlice = 0; + rtvDesc.Texture2DArray.ArraySize = mTextureDepth; + + HRESULT result = device->CreateRenderTargetView(mSwizzleTexture, &rtvDesc, &mSwizzleRenderTargets[mipLevel]); + + if (result == E_OUTOFMEMORY) + { + return gl::error(GL_OUT_OF_MEMORY, static_cast<ID3D11RenderTargetView*>(NULL)); + } + ASSERT(SUCCEEDED(result)); + } + + return mSwizzleRenderTargets[mipLevel]; + } + else + { + return NULL; + } +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/TextureStorage11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/TextureStorage11.h new file mode 100644 index 0000000000..9d63b2699d --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/TextureStorage11.h @@ -0,0 +1,313 @@ +// +// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// TextureStorage11.h: Defines the abstract rx::TextureStorage11 class and its concrete derived +// classes TextureStorage11_2D and TextureStorage11_Cube, which act as the interface to the D3D11 texture. + +#ifndef LIBGLESV2_RENDERER_TEXTURESTORAGE11_H_ +#define LIBGLESV2_RENDERER_TEXTURESTORAGE11_H_ + +#include "libGLESv2/Texture.h" +#include "libGLESv2/Error.h" +#include "libGLESv2/renderer/d3d/TextureStorage.h" + +#include <map> + +namespace gl +{ +struct ImageIndex; +} + +namespace rx +{ +class RenderTarget; +class RenderTarget11; +class Renderer; +class Renderer11; +class SwapChain11; +class Image11; + +class TextureStorage11 : public TextureStorage +{ + public: + virtual ~TextureStorage11(); + + static TextureStorage11 *makeTextureStorage11(TextureStorage *storage); + + static DWORD GetTextureBindFlags(GLenum internalFormat, bool renderTarget); + + UINT getBindFlags() const; + + virtual ID3D11Resource *getResource() const = 0; + virtual ID3D11ShaderResourceView *getSRV(const gl::SamplerState &samplerState); + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index) = 0; + + virtual void generateMipmaps() = 0; + + virtual int getTopLevel() const; + virtual bool isRenderTarget() const; + virtual bool isManaged() const; + virtual int getLevelCount() const; + UINT getSubresourceIndex(int mipLevel, int layerTarget) const; + + gl::Error generateSwizzles(GLenum swizzleRed, GLenum swizzleGreen, GLenum swizzleBlue, GLenum swizzleAlpha); + void invalidateSwizzleCacheLevel(int mipLevel); + void invalidateSwizzleCache(); + + bool updateSubresourceLevel(ID3D11Resource *texture, unsigned int sourceSubresource, int level, + int layerTarget, GLint xoffset, GLint yoffset, GLint zoffset, + GLsizei width, GLsizei height, GLsizei depth); + + bool copySubresourceLevel(ID3D11Resource* dstTexture, unsigned int dstSubresource, int level, + int layerTarget, GLint xoffset, GLint yoffset, GLint zoffset, + GLsizei width, GLsizei height, GLsizei depth); + + virtual void associateImage(Image11* image, int level, int layerTarget) = 0; + virtual void disassociateImage(int level, int layerTarget, Image11* expectedImage) = 0; + virtual bool isAssociatedImageValid(int level, int layerTarget, Image11* expectedImage) = 0; + virtual void releaseAssociatedImage(int level, int layerTarget, Image11* incomingImage) = 0; + + protected: + TextureStorage11(Renderer *renderer, UINT bindFlags); + void generateMipmapLayer(RenderTarget11 *source, RenderTarget11 *dest); + int getLevelWidth(int mipLevel) const; + int getLevelHeight(int mipLevel) const; + int getLevelDepth(int mipLevel) const; + + virtual ID3D11Resource *getSwizzleTexture() = 0; + virtual ID3D11RenderTargetView *getSwizzleRenderTarget(int mipLevel) = 0; + ID3D11ShaderResourceView *getSRVLevel(int mipLevel); + + virtual ID3D11ShaderResourceView *createSRV(int baseLevel, int mipLevels, DXGI_FORMAT format, ID3D11Resource *texture) = 0; + + void verifySwizzleExists(GLenum swizzleRed, GLenum swizzleGreen, GLenum swizzleBlue, GLenum swizzleAlpha); + + Renderer11 *mRenderer; + int mTopLevel; + unsigned int mMipLevels; + + DXGI_FORMAT mTextureFormat; + DXGI_FORMAT mShaderResourceFormat; + DXGI_FORMAT mRenderTargetFormat; + DXGI_FORMAT mDepthStencilFormat; + DXGI_FORMAT mSwizzleTextureFormat; + DXGI_FORMAT mSwizzleShaderResourceFormat; + DXGI_FORMAT mSwizzleRenderTargetFormat; + unsigned int mTextureWidth; + unsigned int mTextureHeight; + unsigned int mTextureDepth; + + struct SwizzleCacheValue + { + GLenum swizzleRed; + GLenum swizzleGreen; + GLenum swizzleBlue; + GLenum swizzleAlpha; + + SwizzleCacheValue(); + SwizzleCacheValue(GLenum red, GLenum green, GLenum blue, GLenum alpha); + + bool operator ==(const SwizzleCacheValue &other) const; + bool operator !=(const SwizzleCacheValue &other) const; + }; + SwizzleCacheValue mSwizzleCache[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; + + struct SRVKey + { + SRVKey(int baseLevel = 0, int mipLevels = 0, bool swizzle = false); + + bool operator==(const SRVKey &rhs) const; + + int baseLevel; + int mipLevels; + bool swizzle; + }; + + struct SRVPair + { + SRVKey key; + ID3D11ShaderResourceView *srv; + }; + + struct SRVCache + { + ~SRVCache(); + + ID3D11ShaderResourceView *find(const SRVKey &key) const; + ID3D11ShaderResourceView *add(const SRVKey &key, ID3D11ShaderResourceView *srv); + + std::vector<SRVPair> cache; + }; + + private: + DISALLOW_COPY_AND_ASSIGN(TextureStorage11); + + const UINT mBindFlags; + + SRVCache srvCache; + ID3D11ShaderResourceView *mLevelSRVs[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; +}; + +class TextureStorage11_2D : public TextureStorage11 +{ + public: + TextureStorage11_2D(Renderer *renderer, SwapChain11 *swapchain); + TextureStorage11_2D(Renderer *renderer, GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, int levels); + virtual ~TextureStorage11_2D(); + + static TextureStorage11_2D *makeTextureStorage11_2D(TextureStorage *storage); + + virtual ID3D11Resource *getResource() const; + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index); + + virtual void generateMipmaps(); + + virtual void associateImage(Image11* image, int level, int layerTarget); + virtual void disassociateImage(int level, int layerTarget, Image11* expectedImage); + virtual bool isAssociatedImageValid(int level, int layerTarget, Image11* expectedImage); + virtual void releaseAssociatedImage(int level, int layerTarget, Image11* incomingImage); + + protected: + virtual ID3D11Resource *getSwizzleTexture(); + virtual ID3D11RenderTargetView *getSwizzleRenderTarget(int mipLevel); + + private: + DISALLOW_COPY_AND_ASSIGN(TextureStorage11_2D); + + virtual ID3D11ShaderResourceView *createSRV(int baseLevel, int mipLevels, DXGI_FORMAT format, ID3D11Resource *texture); + + ID3D11Texture2D *mTexture; + RenderTarget11 *mRenderTarget[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; + + ID3D11Texture2D *mSwizzleTexture; + ID3D11RenderTargetView *mSwizzleRenderTargets[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; + + Image11 *mAssociatedImages[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; +}; + +class TextureStorage11_Cube : public TextureStorage11 +{ + public: + TextureStorage11_Cube(Renderer *renderer, GLenum internalformat, bool renderTarget, int size, int levels); + virtual ~TextureStorage11_Cube(); + + static TextureStorage11_Cube *makeTextureStorage11_Cube(TextureStorage *storage); + + virtual ID3D11Resource *getResource() const; + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index); + + virtual void generateMipmaps(); + + virtual void associateImage(Image11* image, int level, int layerTarget); + virtual void disassociateImage(int level, int layerTarget, Image11* expectedImage); + virtual bool isAssociatedImageValid(int level, int layerTarget, Image11* expectedImage); + virtual void releaseAssociatedImage(int level, int layerTarget, Image11* incomingImage); + + protected: + virtual ID3D11Resource *getSwizzleTexture(); + virtual ID3D11RenderTargetView *getSwizzleRenderTarget(int mipLevel); + + private: + DISALLOW_COPY_AND_ASSIGN(TextureStorage11_Cube); + + virtual ID3D11ShaderResourceView *createSRV(int baseLevel, int mipLevels, DXGI_FORMAT format, ID3D11Resource *texture); + + ID3D11Texture2D *mTexture; + RenderTarget11 *mRenderTarget[6][gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; + + ID3D11Texture2D *mSwizzleTexture; + ID3D11RenderTargetView *mSwizzleRenderTargets[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; + + Image11 *mAssociatedImages[6][gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; +}; + +class TextureStorage11_3D : public TextureStorage11 +{ + public: + TextureStorage11_3D(Renderer *renderer, GLenum internalformat, bool renderTarget, + GLsizei width, GLsizei height, GLsizei depth, int levels); + virtual ~TextureStorage11_3D(); + + static TextureStorage11_3D *makeTextureStorage11_3D(TextureStorage *storage); + + virtual ID3D11Resource *getResource() const; + + // Handles both layer and non-layer RTs + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index); + + virtual void generateMipmaps(); + + virtual void associateImage(Image11* image, int level, int layerTarget); + virtual void disassociateImage(int level, int layerTarget, Image11* expectedImage); + virtual bool isAssociatedImageValid(int level, int layerTarget, Image11* expectedImage); + virtual void releaseAssociatedImage(int level, int layerTarget, Image11* incomingImage); + + protected: + virtual ID3D11Resource *getSwizzleTexture(); + virtual ID3D11RenderTargetView *getSwizzleRenderTarget(int mipLevel); + + private: + DISALLOW_COPY_AND_ASSIGN(TextureStorage11_3D); + + virtual ID3D11ShaderResourceView *createSRV(int baseLevel, int mipLevels, DXGI_FORMAT format, ID3D11Resource *texture); + + typedef std::pair<int, int> LevelLayerKey; + typedef std::map<LevelLayerKey, RenderTarget11*> RenderTargetMap; + RenderTargetMap mLevelLayerRenderTargets; + + RenderTarget11 *mLevelRenderTargets[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; + + ID3D11Texture3D *mTexture; + ID3D11Texture3D *mSwizzleTexture; + ID3D11RenderTargetView *mSwizzleRenderTargets[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; + + Image11 *mAssociatedImages[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; +}; + +class TextureStorage11_2DArray : public TextureStorage11 +{ + public: + TextureStorage11_2DArray(Renderer *renderer, GLenum internalformat, bool renderTarget, + GLsizei width, GLsizei height, GLsizei depth, int levels); + virtual ~TextureStorage11_2DArray(); + + static TextureStorage11_2DArray *makeTextureStorage11_2DArray(TextureStorage *storage); + + virtual ID3D11Resource *getResource() const; + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index); + + virtual void generateMipmaps(); + + virtual void associateImage(Image11* image, int level, int layerTarget); + virtual void disassociateImage(int level, int layerTarget, Image11* expectedImage); + virtual bool isAssociatedImageValid(int level, int layerTarget, Image11* expectedImage); + virtual void releaseAssociatedImage(int level, int layerTarget, Image11* incomingImage); + + protected: + virtual ID3D11Resource *getSwizzleTexture(); + virtual ID3D11RenderTargetView *getSwizzleRenderTarget(int mipLevel); + + private: + DISALLOW_COPY_AND_ASSIGN(TextureStorage11_2DArray); + + virtual ID3D11ShaderResourceView *createSRV(int baseLevel, int mipLevels, DXGI_FORMAT format, ID3D11Resource *texture); + + typedef std::pair<int, int> LevelLayerKey; + typedef std::map<LevelLayerKey, RenderTarget11*> RenderTargetMap; + RenderTargetMap mRenderTargets; + + ID3D11Texture2D *mTexture; + + ID3D11Texture2D *mSwizzleTexture; + ID3D11RenderTargetView *mSwizzleRenderTargets[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; + + typedef std::map<LevelLayerKey, Image11*> ImageMap; + ImageMap mAssociatedImages; +}; + +} + +#endif // LIBGLESV2_RENDERER_TEXTURESTORAGE11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/VertexArray11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/VertexArray11.h new file mode 100644 index 0000000000..590cb9f05a --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/VertexArray11.h @@ -0,0 +1,42 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// VertexArray11.h: Defines the rx::VertexArray11 class which implements rx::VertexArrayImpl. + +#ifndef LIBGLESV2_RENDERER_VERTEXARRAY11_H_ +#define LIBGLESV2_RENDERER_VERTEXARRAY11_H_ + +#include "libGLESv2/renderer/VertexArrayImpl.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" + +namespace rx +{ +class Renderer11; + +class VertexArray11 : public VertexArrayImpl +{ + public: + VertexArray11(rx::Renderer11 *renderer) + : VertexArrayImpl(), + mRenderer(renderer) + { + } + virtual ~VertexArray11() { } + + virtual void setElementArrayBuffer(const gl::Buffer *buffer) { } + virtual void setAttribute(size_t idx, const gl::VertexAttribute &attr) { } + virtual void setAttributeDivisor(size_t idx, GLuint divisor) { } + virtual void enableAttribute(size_t idx, bool enabledState) { } + + private: + DISALLOW_COPY_AND_ASSIGN(VertexArray11); + + rx::Renderer11 *mRenderer; +}; + +} + +#endif // LIBGLESV2_RENDERER_VERTEXARRAY11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/VertexBuffer11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/VertexBuffer11.cpp new file mode 100644 index 0000000000..9bc5b1d2d1 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/VertexBuffer11.cpp @@ -0,0 +1,209 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// VertexBuffer11.cpp: Defines the D3D11 VertexBuffer implementation. + +#include "libGLESv2/renderer/d3d/d3d11/VertexBuffer11.h" +#include "libGLESv2/renderer/d3d/d3d11/Buffer11.h" +#include "libGLESv2/renderer/d3d/d3d11/Renderer11.h" +#include "libGLESv2/renderer/d3d/d3d11/formatutils11.h" +#include "libGLESv2/Buffer.h" +#include "libGLESv2/VertexAttribute.h" + +namespace rx +{ + +VertexBuffer11::VertexBuffer11(rx::Renderer11 *const renderer) : mRenderer(renderer) +{ + mBuffer = NULL; + mBufferSize = 0; + mDynamicUsage = false; +} + +VertexBuffer11::~VertexBuffer11() +{ + SafeRelease(mBuffer); +} + +gl::Error VertexBuffer11::initialize(unsigned int size, bool dynamicUsage) +{ + SafeRelease(mBuffer); + + updateSerial(); + + if (size > 0) + { + ID3D11Device* dxDevice = mRenderer->getDevice(); + + D3D11_BUFFER_DESC bufferDesc; + bufferDesc.ByteWidth = size; + bufferDesc.Usage = D3D11_USAGE_DYNAMIC; + bufferDesc.BindFlags = D3D11_BIND_VERTEX_BUFFER; + bufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; + bufferDesc.MiscFlags = 0; + bufferDesc.StructureByteStride = 0; + + HRESULT result = dxDevice->CreateBuffer(&bufferDesc, NULL, &mBuffer); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to allocate internal vertex buffer of size, %lu.", size); + } + } + + mBufferSize = size; + mDynamicUsage = dynamicUsage; + + return gl::Error(GL_NO_ERROR); +} + +VertexBuffer11 *VertexBuffer11::makeVertexBuffer11(VertexBuffer *vetexBuffer) +{ + ASSERT(HAS_DYNAMIC_TYPE(VertexBuffer11*, vetexBuffer)); + return static_cast<VertexBuffer11*>(vetexBuffer); +} + +gl::Error VertexBuffer11::storeVertexAttributes(const gl::VertexAttribute &attrib, const gl::VertexAttribCurrentValueData ¤tValue, + GLint start, GLsizei count, GLsizei instances, unsigned int offset) +{ + if (!mBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal vertex buffer is not initialized."); + } + + gl::Buffer *buffer = attrib.buffer.get(); + int inputStride = ComputeVertexAttributeStride(attrib); + ID3D11DeviceContext *dxContext = mRenderer->getDeviceContext(); + + D3D11_MAPPED_SUBRESOURCE mappedResource; + HRESULT result = dxContext->Map(mBuffer, 0, D3D11_MAP_WRITE_NO_OVERWRITE, 0, &mappedResource); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to map internal vertex buffer, HRESULT: 0x%08x.", result); + } + + uint8_t *output = reinterpret_cast<uint8_t*>(mappedResource.pData) + offset; + + const uint8_t *input = NULL; + if (attrib.enabled) + { + if (buffer) + { + Buffer11 *storage = Buffer11::makeBuffer11(buffer->getImplementation()); + input = static_cast<const uint8_t*>(storage->getData()) + static_cast<int>(attrib.offset); + } + else + { + input = static_cast<const uint8_t*>(attrib.pointer); + } + } + else + { + input = reinterpret_cast<const uint8_t*>(currentValue.FloatValues); + } + + if (instances == 0 || attrib.divisor == 0) + { + input += inputStride * start; + } + + gl::VertexFormat vertexFormat(attrib, currentValue.Type); + const d3d11::VertexFormat &vertexFormatInfo = d3d11::GetVertexFormatInfo(vertexFormat); + ASSERT(vertexFormatInfo.copyFunction != NULL); + vertexFormatInfo.copyFunction(input, inputStride, count, output); + + dxContext->Unmap(mBuffer, 0); + + return gl::Error(GL_NO_ERROR); +} + +gl::Error VertexBuffer11::getSpaceRequired(const gl::VertexAttribute &attrib, GLsizei count, + GLsizei instances, unsigned int *outSpaceRequired) const +{ + unsigned int elementCount = 0; + if (attrib.enabled) + { + if (instances == 0 || attrib.divisor == 0) + { + elementCount = count; + } + else + { + // Round up to divisor, if possible + elementCount = rx::UnsignedCeilDivide(static_cast<unsigned int>(instances), attrib.divisor); + } + + gl::VertexFormat vertexFormat(attrib); + const d3d11::VertexFormat &vertexFormatInfo = d3d11::GetVertexFormatInfo(vertexFormat); + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(vertexFormatInfo.nativeFormat); + unsigned int elementSize = dxgiFormatInfo.pixelBytes; + if (elementSize <= std::numeric_limits<unsigned int>::max() / elementCount) + { + if (outSpaceRequired) + { + *outSpaceRequired = elementSize * elementCount; + } + return gl::Error(GL_NO_ERROR); + } + else + { + return gl::Error(GL_OUT_OF_MEMORY, "New vertex buffer size would result in an overflow."); + } + } + else + { + const unsigned int elementSize = 4; + if (outSpaceRequired) + { + *outSpaceRequired = elementSize * 4; + } + return gl::Error(GL_NO_ERROR); + } +} + +unsigned int VertexBuffer11::getBufferSize() const +{ + return mBufferSize; +} + +gl::Error VertexBuffer11::setBufferSize(unsigned int size) +{ + if (size > mBufferSize) + { + return initialize(size, mDynamicUsage); + } + else + { + return gl::Error(GL_NO_ERROR); + } +} + +gl::Error VertexBuffer11::discard() +{ + if (!mBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal vertex buffer is not initialized."); + } + + ID3D11DeviceContext *dxContext = mRenderer->getDeviceContext(); + + D3D11_MAPPED_SUBRESOURCE mappedResource; + HRESULT result = dxContext->Map(mBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to map internal buffer for discarding, HRESULT: 0x%08x", result); + } + + dxContext->Unmap(mBuffer, 0); + + return gl::Error(GL_NO_ERROR); +} + +ID3D11Buffer *VertexBuffer11::getBuffer() const +{ + return mBuffer; +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/VertexBuffer11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/VertexBuffer11.h new file mode 100644 index 0000000000..0e10da1df8 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/VertexBuffer11.h @@ -0,0 +1,52 @@ +// +// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// VertexBuffer11.h: Defines the D3D11 VertexBuffer implementation. + +#ifndef LIBGLESV2_RENDERER_VERTEXBUFFER11_H_ +#define LIBGLESV2_RENDERER_VERTEXBUFFER11_H_ + +#include "libGLESv2/renderer/d3d/VertexBuffer.h" + +namespace rx +{ +class Renderer11; + +class VertexBuffer11 : public VertexBuffer +{ + public: + explicit VertexBuffer11(rx::Renderer11 *const renderer); + virtual ~VertexBuffer11(); + + virtual gl::Error initialize(unsigned int size, bool dynamicUsage); + + static VertexBuffer11 *makeVertexBuffer11(VertexBuffer *vetexBuffer); + + virtual gl::Error storeVertexAttributes(const gl::VertexAttribute &attrib, const gl::VertexAttribCurrentValueData ¤tValue, + GLint start, GLsizei count, GLsizei instances, unsigned int offset); + + virtual gl::Error getSpaceRequired(const gl::VertexAttribute &attrib, GLsizei count, GLsizei instances, + unsigned int *outSpaceRequired) const; + + virtual unsigned int getBufferSize() const; + virtual gl::Error setBufferSize(unsigned int size); + virtual gl::Error discard(); + + ID3D11Buffer *getBuffer() const; + + private: + DISALLOW_COPY_AND_ASSIGN(VertexBuffer11); + + rx::Renderer11 *const mRenderer; + + ID3D11Buffer *mBuffer; + unsigned int mBufferSize; + bool mDynamicUsage; +}; + +} + +#endif // LIBGLESV2_RENDERER_VERTEXBUFFER11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/formatutils11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/formatutils11.cpp new file mode 100644 index 0000000000..c07828757d --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/formatutils11.cpp @@ -0,0 +1,1075 @@ +// +// Copyright (c) 2013-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// formatutils11.cpp: Queries for GL image formats and their translations to D3D11 +// formats. + +#include "libGLESv2/renderer/d3d/d3d11/formatutils11.h" +#include "libGLESv2/renderer/generatemip.h" +#include "libGLESv2/renderer/loadimage.h" +#include "libGLESv2/renderer/copyimage.h" +#include "libGLESv2/renderer/Renderer.h" +#include "libGLESv2/renderer/copyvertex.h" + +namespace rx +{ + +namespace d3d11 +{ + +typedef std::map<DXGI_FORMAT, GLenum> DXGIToESFormatMap; + +inline void AddDXGIToESEntry(DXGIToESFormatMap *map, DXGI_FORMAT key, GLenum value) +{ + map->insert(std::make_pair(key, value)); +} + +static DXGIToESFormatMap BuildDXGIToESFormatMap() +{ + DXGIToESFormatMap map; + + AddDXGIToESEntry(&map, DXGI_FORMAT_UNKNOWN, GL_NONE); + + AddDXGIToESEntry(&map, DXGI_FORMAT_A8_UNORM, GL_ALPHA8_EXT); + AddDXGIToESEntry(&map, DXGI_FORMAT_R8_UNORM, GL_R8); + AddDXGIToESEntry(&map, DXGI_FORMAT_R8G8_UNORM, GL_RG8); + AddDXGIToESEntry(&map, DXGI_FORMAT_R8G8B8A8_UNORM, GL_RGBA8); + AddDXGIToESEntry(&map, DXGI_FORMAT_R8G8B8A8_UNORM_SRGB, GL_SRGB8_ALPHA8); + AddDXGIToESEntry(&map, DXGI_FORMAT_B8G8R8A8_UNORM, GL_BGRA8_EXT); + + AddDXGIToESEntry(&map, DXGI_FORMAT_R8_SNORM, GL_R8_SNORM); + AddDXGIToESEntry(&map, DXGI_FORMAT_R8G8_SNORM, GL_RG8_SNORM); + AddDXGIToESEntry(&map, DXGI_FORMAT_R8G8B8A8_SNORM, GL_RGBA8_SNORM); + + AddDXGIToESEntry(&map, DXGI_FORMAT_R8_UINT, GL_R8UI); + AddDXGIToESEntry(&map, DXGI_FORMAT_R16_UINT, GL_R16UI); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32_UINT, GL_R32UI); + AddDXGIToESEntry(&map, DXGI_FORMAT_R8G8_UINT, GL_RG8UI); + AddDXGIToESEntry(&map, DXGI_FORMAT_R16G16_UINT, GL_RG16UI); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32G32_UINT, GL_RG32UI); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32G32B32_UINT, GL_RGB32UI); + AddDXGIToESEntry(&map, DXGI_FORMAT_R8G8B8A8_UINT, GL_RGBA8UI); + AddDXGIToESEntry(&map, DXGI_FORMAT_R16G16B16A16_UINT, GL_RGBA16UI); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32G32B32A32_UINT, GL_RGBA32UI); + + AddDXGIToESEntry(&map, DXGI_FORMAT_R8_SINT, GL_R8I); + AddDXGIToESEntry(&map, DXGI_FORMAT_R16_SINT, GL_R16I); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32_SINT, GL_R32I); + AddDXGIToESEntry(&map, DXGI_FORMAT_R8G8_SINT, GL_RG8I); + AddDXGIToESEntry(&map, DXGI_FORMAT_R16G16_SINT, GL_RG16I); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32G32_SINT, GL_RG32I); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32G32B32_SINT, GL_RGB32I); + AddDXGIToESEntry(&map, DXGI_FORMAT_R8G8B8A8_SINT, GL_RGBA8I); + AddDXGIToESEntry(&map, DXGI_FORMAT_R16G16B16A16_SINT, GL_RGBA16I); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32G32B32A32_SINT, GL_RGBA32I); + + AddDXGIToESEntry(&map, DXGI_FORMAT_R10G10B10A2_UNORM, GL_RGB10_A2); + AddDXGIToESEntry(&map, DXGI_FORMAT_R10G10B10A2_UINT, GL_RGB10_A2UI); + + AddDXGIToESEntry(&map, DXGI_FORMAT_R16_FLOAT, GL_R16F); + AddDXGIToESEntry(&map, DXGI_FORMAT_R16G16_FLOAT, GL_RG16F); + AddDXGIToESEntry(&map, DXGI_FORMAT_R16G16B16A16_FLOAT, GL_RGBA16F); + + AddDXGIToESEntry(&map, DXGI_FORMAT_R32_FLOAT, GL_R32F); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32G32_FLOAT, GL_RG32F); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32G32B32_FLOAT, GL_RGB32F); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32G32B32A32_FLOAT, GL_RGBA32F); + + AddDXGIToESEntry(&map, DXGI_FORMAT_R9G9B9E5_SHAREDEXP, GL_RGB9_E5); + AddDXGIToESEntry(&map, DXGI_FORMAT_R11G11B10_FLOAT, GL_R11F_G11F_B10F); + + AddDXGIToESEntry(&map, DXGI_FORMAT_R16_TYPELESS, GL_DEPTH_COMPONENT16); + AddDXGIToESEntry(&map, DXGI_FORMAT_R16_UNORM, GL_DEPTH_COMPONENT16); + AddDXGIToESEntry(&map, DXGI_FORMAT_D16_UNORM, GL_DEPTH_COMPONENT16); + AddDXGIToESEntry(&map, DXGI_FORMAT_R24G8_TYPELESS, GL_DEPTH24_STENCIL8_OES); + AddDXGIToESEntry(&map, DXGI_FORMAT_R24_UNORM_X8_TYPELESS, GL_DEPTH24_STENCIL8_OES); + AddDXGIToESEntry(&map, DXGI_FORMAT_D24_UNORM_S8_UINT, GL_DEPTH24_STENCIL8_OES); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32G8X24_TYPELESS, GL_DEPTH32F_STENCIL8); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32_FLOAT_X8X24_TYPELESS, GL_DEPTH32F_STENCIL8); + AddDXGIToESEntry(&map, DXGI_FORMAT_D32_FLOAT_S8X24_UINT, GL_DEPTH32F_STENCIL8); + AddDXGIToESEntry(&map, DXGI_FORMAT_R32_TYPELESS, GL_DEPTH_COMPONENT32F); + AddDXGIToESEntry(&map, DXGI_FORMAT_D32_FLOAT, GL_DEPTH_COMPONENT32F); + + AddDXGIToESEntry(&map, DXGI_FORMAT_BC1_UNORM, GL_COMPRESSED_RGBA_S3TC_DXT1_EXT); + AddDXGIToESEntry(&map, DXGI_FORMAT_BC2_UNORM, GL_COMPRESSED_RGBA_S3TC_DXT3_ANGLE); + AddDXGIToESEntry(&map, DXGI_FORMAT_BC3_UNORM, GL_COMPRESSED_RGBA_S3TC_DXT5_ANGLE); + + return map; +} + +struct D3D11FastCopyFormat +{ + GLenum destFormat; + GLenum destType; + ColorCopyFunction copyFunction; + + D3D11FastCopyFormat(GLenum destFormat, GLenum destType, ColorCopyFunction copyFunction) + : destFormat(destFormat), destType(destType), copyFunction(copyFunction) + { } + + bool operator<(const D3D11FastCopyFormat& other) const + { + return memcmp(this, &other, sizeof(D3D11FastCopyFormat)) < 0; + } +}; + +typedef std::multimap<DXGI_FORMAT, D3D11FastCopyFormat> D3D11FastCopyMap; + +static D3D11FastCopyMap BuildFastCopyMap() +{ + D3D11FastCopyMap map; + + map.insert(std::make_pair(DXGI_FORMAT_B8G8R8A8_UNORM, D3D11FastCopyFormat(GL_RGBA, GL_UNSIGNED_BYTE, CopyBGRA8ToRGBA8))); + + return map; +} + +struct DXGIDepthStencilInfo +{ + unsigned int depthBits; + unsigned int depthOffset; + unsigned int stencilBits; + unsigned int stencilOffset; +}; + +typedef std::map<DXGI_FORMAT, DXGIDepthStencilInfo> DepthStencilInfoMap; +typedef std::pair<DXGI_FORMAT, DXGIDepthStencilInfo> DepthStencilInfoPair; + +static inline void InsertDXGIDepthStencilInfo(DepthStencilInfoMap *map, DXGI_FORMAT format, unsigned int depthBits, + unsigned int depthOffset, unsigned int stencilBits, unsigned int stencilOffset) +{ + DXGIDepthStencilInfo info; + info.depthBits = depthBits; + info.depthOffset = depthOffset; + info.stencilBits = stencilBits; + info.stencilOffset = stencilOffset; + + map->insert(std::make_pair(format, info)); +} + +static DepthStencilInfoMap BuildDepthStencilInfoMap() +{ + DepthStencilInfoMap map; + + InsertDXGIDepthStencilInfo(&map, DXGI_FORMAT_R16_TYPELESS, 16, 0, 0, 0); + InsertDXGIDepthStencilInfo(&map, DXGI_FORMAT_R16_UNORM, 16, 0, 0, 0); + InsertDXGIDepthStencilInfo(&map, DXGI_FORMAT_D16_UNORM, 16, 0, 0, 0); + + InsertDXGIDepthStencilInfo(&map, DXGI_FORMAT_R24G8_TYPELESS, 24, 0, 8, 24); + InsertDXGIDepthStencilInfo(&map, DXGI_FORMAT_R24_UNORM_X8_TYPELESS, 24, 0, 8, 24); + InsertDXGIDepthStencilInfo(&map, DXGI_FORMAT_D24_UNORM_S8_UINT, 24, 0, 8, 24); + + InsertDXGIDepthStencilInfo(&map, DXGI_FORMAT_R32_TYPELESS, 32, 0, 0, 0); + InsertDXGIDepthStencilInfo(&map, DXGI_FORMAT_R32_FLOAT, 32, 0, 0, 0); + InsertDXGIDepthStencilInfo(&map, DXGI_FORMAT_D32_FLOAT, 32, 0, 0, 0); + + InsertDXGIDepthStencilInfo(&map, DXGI_FORMAT_R32G8X24_TYPELESS, 32, 0, 8, 32); + InsertDXGIDepthStencilInfo(&map, DXGI_FORMAT_R32_FLOAT_X8X24_TYPELESS, 32, 0, 8, 32); + InsertDXGIDepthStencilInfo(&map, DXGI_FORMAT_D32_FLOAT_S8X24_UINT, 32, 0, 8, 32); + + return map; +} + +typedef std::map<DXGI_FORMAT, DXGIFormat> DXGIFormatInfoMap; + +DXGIFormat::DXGIFormat() + : pixelBytes(0), + blockWidth(0), + blockHeight(0), + depthBits(0), + depthOffset(0), + stencilBits(0), + stencilOffset(0), + internalFormat(GL_NONE), + componentType(GL_NONE), + mipGenerationFunction(NULL), + colorReadFunction(NULL), + fastCopyFunctions() +{ +} + +ColorCopyFunction DXGIFormat::getFastCopyFunction(GLenum format, GLenum type) const +{ + FastCopyFunctionMap::const_iterator iter = fastCopyFunctions.find(std::make_pair(format, type)); + return (iter != fastCopyFunctions.end()) ? iter->second : NULL; +} + +void AddDXGIFormat(DXGIFormatInfoMap *map, DXGI_FORMAT dxgiFormat, GLuint pixelBits, GLuint blockWidth, GLuint blockHeight, + GLenum componentType, MipGenerationFunction mipFunc, ColorReadFunction readFunc) +{ + DXGIFormat info; + info.pixelBytes = pixelBits / 8; + info.blockWidth = blockWidth; + info.blockHeight = blockHeight; + + static const DepthStencilInfoMap dsInfoMap = BuildDepthStencilInfoMap(); + DepthStencilInfoMap::const_iterator dsInfoIter = dsInfoMap.find(dxgiFormat); + if (dsInfoIter != dsInfoMap.end()) + { + info.depthBits = dsInfoIter->second.depthBits; + info.depthOffset = dsInfoIter->second.depthOffset; + info.stencilBits = dsInfoIter->second.stencilBits; + info.stencilOffset = dsInfoIter->second.stencilOffset; + } + else + { + info.depthBits = 0; + info.depthOffset = 0; + info.stencilBits = 0; + info.stencilOffset = 0; + } + + static const DXGIToESFormatMap dxgiToESMap = BuildDXGIToESFormatMap(); + DXGIToESFormatMap::const_iterator dxgiToESIter = dxgiToESMap.find(dxgiFormat); + info.internalFormat = (dxgiToESIter != dxgiToESMap.end()) ? dxgiToESIter->second : GL_NONE; + + info.componentType = componentType; + + info.mipGenerationFunction = mipFunc; + info.colorReadFunction = readFunc; + + static const D3D11FastCopyMap fastCopyMap = BuildFastCopyMap(); + std::pair<D3D11FastCopyMap::const_iterator, D3D11FastCopyMap::const_iterator> fastCopyIter = fastCopyMap.equal_range(dxgiFormat); + for (D3D11FastCopyMap::const_iterator i = fastCopyIter.first; i != fastCopyIter.second; i++) + { + info.fastCopyFunctions.insert(std::make_pair(std::make_pair(i->second.destFormat, i->second.destType), i->second.copyFunction)); + } + + map->insert(std::make_pair(dxgiFormat, info)); +} + +// A map to determine the pixel size and mipmap generation function of a given DXGI format +static DXGIFormatInfoMap BuildDXGIFormatInfoMap() +{ + DXGIFormatInfoMap map; + + // | DXGI format |S |W |H |Component Type | Mip generation function | Color read function + AddDXGIFormat(&map, DXGI_FORMAT_UNKNOWN, 0, 0, 0, GL_NONE, NULL, NULL); + + AddDXGIFormat(&map, DXGI_FORMAT_A8_UNORM, 8, 1, 1, GL_UNSIGNED_NORMALIZED, GenerateMip<A8>, ReadColor<A8, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_R8_UNORM, 8, 1, 1, GL_UNSIGNED_NORMALIZED, GenerateMip<R8>, ReadColor<R8, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_R8G8_UNORM, 16, 1, 1, GL_UNSIGNED_NORMALIZED, GenerateMip<R8G8>, ReadColor<R8G8, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_R8G8B8A8_UNORM, 32, 1, 1, GL_UNSIGNED_NORMALIZED, GenerateMip<R8G8B8A8>, ReadColor<R8G8B8A8, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_R8G8B8A8_UNORM_SRGB, 32, 1, 1, GL_UNSIGNED_NORMALIZED, GenerateMip<R8G8B8A8>, ReadColor<R8G8B8A8, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_B8G8R8A8_UNORM, 32, 1, 1, GL_UNSIGNED_NORMALIZED, GenerateMip<B8G8R8A8>, ReadColor<B8G8R8A8, GLfloat>); + + AddDXGIFormat(&map, DXGI_FORMAT_R8_SNORM, 8, 1, 1, GL_SIGNED_NORMALIZED, GenerateMip<R8S>, ReadColor<R8S, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_R8G8_SNORM, 16, 1, 1, GL_SIGNED_NORMALIZED, GenerateMip<R8G8S>, ReadColor<R8G8S, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_R8G8B8A8_SNORM, 32, 1, 1, GL_SIGNED_NORMALIZED, GenerateMip<R8G8B8A8S>, ReadColor<R8G8B8A8S, GLfloat>); + + AddDXGIFormat(&map, DXGI_FORMAT_R8_UINT, 8, 1, 1, GL_UNSIGNED_INT, GenerateMip<R8>, ReadColor<R8, GLuint>); + AddDXGIFormat(&map, DXGI_FORMAT_R16_UINT, 16, 1, 1, GL_UNSIGNED_INT, GenerateMip<R16>, ReadColor<R16, GLuint>); + AddDXGIFormat(&map, DXGI_FORMAT_R32_UINT, 32, 1, 1, GL_UNSIGNED_INT, GenerateMip<R32>, ReadColor<R32, GLuint>); + AddDXGIFormat(&map, DXGI_FORMAT_R8G8_UINT, 16, 1, 1, GL_UNSIGNED_INT, GenerateMip<R8G8>, ReadColor<R8G8, GLuint>); + AddDXGIFormat(&map, DXGI_FORMAT_R16G16_UINT, 32, 1, 1, GL_UNSIGNED_INT, GenerateMip<R16G16>, ReadColor<R16G16, GLuint>); + AddDXGIFormat(&map, DXGI_FORMAT_R32G32_UINT, 64, 1, 1, GL_UNSIGNED_INT, GenerateMip<R32G32>, ReadColor<R32G32, GLuint>); + AddDXGIFormat(&map, DXGI_FORMAT_R32G32B32_UINT, 96, 1, 1, GL_UNSIGNED_INT, GenerateMip<R32G32B32>, ReadColor<R32G32B32, GLuint>); + AddDXGIFormat(&map, DXGI_FORMAT_R8G8B8A8_UINT, 32, 1, 1, GL_UNSIGNED_INT, GenerateMip<R8G8B8A8>, ReadColor<R8G8B8A8, GLuint>); + AddDXGIFormat(&map, DXGI_FORMAT_R16G16B16A16_UINT, 64, 1, 1, GL_UNSIGNED_INT, GenerateMip<R16G16B16A16>, ReadColor<R16G16B16A16, GLuint>); + AddDXGIFormat(&map, DXGI_FORMAT_R32G32B32A32_UINT, 128, 1, 1, GL_UNSIGNED_INT, GenerateMip<R32G32B32A32>, ReadColor<R32G32B32A32, GLuint>); + + AddDXGIFormat(&map, DXGI_FORMAT_R8_SINT, 8, 1, 1, GL_INT, GenerateMip<R8S>, ReadColor<R8S, GLint>); + AddDXGIFormat(&map, DXGI_FORMAT_R16_SINT, 16, 1, 1, GL_INT, GenerateMip<R16S>, ReadColor<R16S, GLint>); + AddDXGIFormat(&map, DXGI_FORMAT_R32_SINT, 32, 1, 1, GL_INT, GenerateMip<R32S>, ReadColor<R32S, GLint>); + AddDXGIFormat(&map, DXGI_FORMAT_R8G8_SINT, 16, 1, 1, GL_INT, GenerateMip<R8G8S>, ReadColor<R8G8S, GLint>); + AddDXGIFormat(&map, DXGI_FORMAT_R16G16_SINT, 32, 1, 1, GL_INT, GenerateMip<R16G16S>, ReadColor<R16G16S, GLint>); + AddDXGIFormat(&map, DXGI_FORMAT_R32G32_SINT, 64, 1, 1, GL_INT, GenerateMip<R32G32S>, ReadColor<R32G32S, GLint>); + AddDXGIFormat(&map, DXGI_FORMAT_R32G32B32_SINT, 96, 1, 1, GL_INT, GenerateMip<R32G32B32S>, ReadColor<R32G32B32S, GLint>); + AddDXGIFormat(&map, DXGI_FORMAT_R8G8B8A8_SINT, 32, 1, 1, GL_INT, GenerateMip<R8G8B8A8S>, ReadColor<R8G8B8A8S, GLint>); + AddDXGIFormat(&map, DXGI_FORMAT_R16G16B16A16_SINT, 64, 1, 1, GL_INT, GenerateMip<R16G16B16A16S>, ReadColor<R16G16B16A16S, GLint>); + AddDXGIFormat(&map, DXGI_FORMAT_R32G32B32A32_SINT, 128, 1, 1, GL_INT, GenerateMip<R32G32B32A32S>, ReadColor<R32G32B32A32S, GLint>); + + AddDXGIFormat(&map, DXGI_FORMAT_R10G10B10A2_UNORM, 32, 1, 1, GL_UNSIGNED_NORMALIZED, GenerateMip<R10G10B10A2>, ReadColor<R10G10B10A2, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_R10G10B10A2_UINT, 32, 1, 1, GL_UNSIGNED_INT, GenerateMip<R10G10B10A2>, ReadColor<R10G10B10A2, GLuint>); + + AddDXGIFormat(&map, DXGI_FORMAT_R16_FLOAT, 16, 1, 1, GL_FLOAT, GenerateMip<R16F>, ReadColor<R16F, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_R16G16_FLOAT, 32, 1, 1, GL_FLOAT, GenerateMip<R16G16F>, ReadColor<R16G16F, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_R16G16B16A16_FLOAT, 64, 1, 1, GL_FLOAT, GenerateMip<R16G16B16A16F>, ReadColor<R16G16B16A16F, GLfloat>); + + AddDXGIFormat(&map, DXGI_FORMAT_R32_FLOAT, 32, 1, 1, GL_FLOAT, GenerateMip<R32F>, ReadColor<R32F, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_R32G32_FLOAT, 64, 1, 1, GL_FLOAT, GenerateMip<R32G32F>, ReadColor<R32G32F, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_R32G32B32_FLOAT, 96, 1, 1, GL_FLOAT, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_R32G32B32A32_FLOAT, 128, 1, 1, GL_FLOAT, GenerateMip<R32G32B32A32F>, ReadColor<R32G32B32A32F, GLfloat>); + + AddDXGIFormat(&map, DXGI_FORMAT_R9G9B9E5_SHAREDEXP, 32, 1, 1, GL_FLOAT, GenerateMip<R9G9B9E5>, ReadColor<R9G9B9E5, GLfloat>); + AddDXGIFormat(&map, DXGI_FORMAT_R11G11B10_FLOAT, 32, 1, 1, GL_FLOAT, GenerateMip<R11G11B10F>, ReadColor<R11G11B10F, GLfloat>); + + AddDXGIFormat(&map, DXGI_FORMAT_R16_TYPELESS, 16, 1, 1, GL_NONE, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_R16_UNORM, 16, 1, 1, GL_UNSIGNED_NORMALIZED, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_D16_UNORM, 16, 1, 1, GL_UNSIGNED_NORMALIZED, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_R24G8_TYPELESS, 32, 1, 1, GL_NONE, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_R24_UNORM_X8_TYPELESS, 32, 1, 1, GL_NONE, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_D24_UNORM_S8_UINT, 32, 1, 1, GL_UNSIGNED_INT, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_R32G8X24_TYPELESS, 64, 1, 1, GL_NONE, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_R32_FLOAT_X8X24_TYPELESS, 64, 1, 1, GL_NONE, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_D32_FLOAT_S8X24_UINT, 64, 1, 1, GL_UNSIGNED_INT, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_R32_TYPELESS, 32, 1, 1, GL_NONE, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_D32_FLOAT, 32, 1, 1, GL_FLOAT, NULL, NULL); + + AddDXGIFormat(&map, DXGI_FORMAT_BC1_UNORM, 64, 4, 4, GL_UNSIGNED_NORMALIZED, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_BC2_UNORM, 128, 4, 4, GL_UNSIGNED_NORMALIZED, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_BC3_UNORM, 128, 4, 4, GL_UNSIGNED_NORMALIZED, NULL, NULL); + + // Useful formats for vertex buffers + AddDXGIFormat(&map, DXGI_FORMAT_R16_UNORM, 16, 1, 1, GL_UNSIGNED_NORMALIZED, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_R16_SNORM, 16, 1, 1, GL_SIGNED_NORMALIZED, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_R16G16_UNORM, 32, 1, 1, GL_UNSIGNED_NORMALIZED, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_R16G16_SNORM, 32, 1, 1, GL_SIGNED_NORMALIZED, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_R16G16B16A16_UNORM, 64, 1, 1, GL_UNSIGNED_NORMALIZED, NULL, NULL); + AddDXGIFormat(&map, DXGI_FORMAT_R16G16B16A16_SNORM, 64, 1, 1, GL_SIGNED_NORMALIZED, NULL, NULL); + + return map; +} + +const DXGIFormat &GetDXGIFormatInfo(DXGI_FORMAT format) +{ + static const DXGIFormatInfoMap infoMap = BuildDXGIFormatInfoMap(); + DXGIFormatInfoMap::const_iterator iter = infoMap.find(format); + if (iter != infoMap.end()) + { + return iter->second; + } + else + { + static DXGIFormat defaultInfo; + return defaultInfo; + } +} + +struct SwizzleSizeType +{ + size_t maxComponentSize; + GLenum componentType; + + SwizzleSizeType() + : maxComponentSize(0), componentType(GL_NONE) + { } + + SwizzleSizeType(size_t maxComponentSize, GLenum componentType) + : maxComponentSize(maxComponentSize), componentType(componentType) + { } + + bool operator<(const SwizzleSizeType& other) const + { + return (maxComponentSize != other.maxComponentSize) ? (maxComponentSize < other.maxComponentSize) + : (componentType < other.componentType); + } +}; + +struct SwizzleFormatInfo +{ + DXGI_FORMAT mTexFormat; + DXGI_FORMAT mSRVFormat; + DXGI_FORMAT mRTVFormat; + + SwizzleFormatInfo() + : mTexFormat(DXGI_FORMAT_UNKNOWN), mSRVFormat(DXGI_FORMAT_UNKNOWN), mRTVFormat(DXGI_FORMAT_UNKNOWN) + { } + + SwizzleFormatInfo(DXGI_FORMAT texFormat, DXGI_FORMAT srvFormat, DXGI_FORMAT rtvFormat) + : mTexFormat(texFormat), mSRVFormat(srvFormat), mRTVFormat(rtvFormat) + { } +}; + +typedef std::map<SwizzleSizeType, SwizzleFormatInfo> SwizzleInfoMap; +typedef std::pair<SwizzleSizeType, SwizzleFormatInfo> SwizzleInfoPair; + +static SwizzleInfoMap BuildSwizzleInfoMap() +{ + SwizzleInfoMap map; + + map.insert(SwizzleInfoPair(SwizzleSizeType( 8, GL_UNSIGNED_NORMALIZED), SwizzleFormatInfo(DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM ))); + map.insert(SwizzleInfoPair(SwizzleSizeType(16, GL_UNSIGNED_NORMALIZED), SwizzleFormatInfo(DXGI_FORMAT_R16G16B16A16_UNORM, DXGI_FORMAT_R16G16B16A16_UNORM, DXGI_FORMAT_R16G16B16A16_UNORM))); + map.insert(SwizzleInfoPair(SwizzleSizeType(24, GL_UNSIGNED_NORMALIZED), SwizzleFormatInfo(DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT))); + map.insert(SwizzleInfoPair(SwizzleSizeType(32, GL_UNSIGNED_NORMALIZED), SwizzleFormatInfo(DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT))); + + map.insert(SwizzleInfoPair(SwizzleSizeType( 8, GL_SIGNED_NORMALIZED ), SwizzleFormatInfo(DXGI_FORMAT_R8G8B8A8_SNORM, DXGI_FORMAT_R8G8B8A8_SNORM, DXGI_FORMAT_R8G8B8A8_SNORM ))); + + map.insert(SwizzleInfoPair(SwizzleSizeType(16, GL_FLOAT ), SwizzleFormatInfo(DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_R16G16B16A16_FLOAT))); + map.insert(SwizzleInfoPair(SwizzleSizeType(32, GL_FLOAT ), SwizzleFormatInfo(DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT))); + + map.insert(SwizzleInfoPair(SwizzleSizeType( 8, GL_UNSIGNED_INT ), SwizzleFormatInfo(DXGI_FORMAT_R8G8B8A8_UINT, DXGI_FORMAT_R8G8B8A8_UINT, DXGI_FORMAT_R8G8B8A8_UINT ))); + map.insert(SwizzleInfoPair(SwizzleSizeType(16, GL_UNSIGNED_INT ), SwizzleFormatInfo(DXGI_FORMAT_R16G16B16A16_UINT, DXGI_FORMAT_R16G16B16A16_UINT, DXGI_FORMAT_R16G16B16A16_UINT ))); + map.insert(SwizzleInfoPair(SwizzleSizeType(32, GL_UNSIGNED_INT ), SwizzleFormatInfo(DXGI_FORMAT_R32G32B32A32_UINT, DXGI_FORMAT_R32G32B32A32_UINT, DXGI_FORMAT_R32G32B32A32_UINT ))); + + map.insert(SwizzleInfoPair(SwizzleSizeType( 8, GL_INT ), SwizzleFormatInfo(DXGI_FORMAT_R8G8B8A8_SINT, DXGI_FORMAT_R8G8B8A8_SINT, DXGI_FORMAT_R8G8B8A8_SINT ))); + map.insert(SwizzleInfoPair(SwizzleSizeType(16, GL_INT ), SwizzleFormatInfo(DXGI_FORMAT_R16G16B16A16_SINT, DXGI_FORMAT_R16G16B16A16_SINT, DXGI_FORMAT_R16G16B16A16_SINT ))); + map.insert(SwizzleInfoPair(SwizzleSizeType(32, GL_INT ), SwizzleFormatInfo(DXGI_FORMAT_R32G32B32A32_SINT, DXGI_FORMAT_R32G32B32A32_SINT, DXGI_FORMAT_R32G32B32A32_SINT ))); + + return map; +} + +typedef std::pair<GLint, InitializeTextureDataFunction> InternalFormatInitializerPair; +typedef std::map<GLint, InitializeTextureDataFunction> InternalFormatInitializerMap; + +static InternalFormatInitializerMap BuildInternalFormatInitializerMap() +{ + InternalFormatInitializerMap map; + + map.insert(InternalFormatInitializerPair(GL_RGB8, Initialize4ComponentData<GLubyte, 0x00, 0x00, 0x00, 0xFF> )); + map.insert(InternalFormatInitializerPair(GL_RGB565, Initialize4ComponentData<GLubyte, 0x00, 0x00, 0x00, 0xFF> )); + map.insert(InternalFormatInitializerPair(GL_SRGB8, Initialize4ComponentData<GLubyte, 0x00, 0x00, 0x00, 0xFF> )); + map.insert(InternalFormatInitializerPair(GL_RGB16F, Initialize4ComponentData<GLhalf, 0x0000, 0x0000, 0x0000, gl::Float16One>)); + map.insert(InternalFormatInitializerPair(GL_RGB32F, Initialize4ComponentData<GLfloat, 0x00000000, 0x00000000, 0x00000000, gl::Float32One>)); + map.insert(InternalFormatInitializerPair(GL_RGB8UI, Initialize4ComponentData<GLubyte, 0x00, 0x00, 0x00, 0x01> )); + map.insert(InternalFormatInitializerPair(GL_RGB8I, Initialize4ComponentData<GLbyte, 0x00, 0x00, 0x00, 0x01> )); + map.insert(InternalFormatInitializerPair(GL_RGB16UI, Initialize4ComponentData<GLushort, 0x0000, 0x0000, 0x0000, 0x0001> )); + map.insert(InternalFormatInitializerPair(GL_RGB16I, Initialize4ComponentData<GLshort, 0x0000, 0x0000, 0x0000, 0x0001> )); + map.insert(InternalFormatInitializerPair(GL_RGB32UI, Initialize4ComponentData<GLuint, 0x00000000, 0x00000000, 0x00000000, 0x00000001> )); + map.insert(InternalFormatInitializerPair(GL_RGB32I, Initialize4ComponentData<GLint, 0x00000000, 0x00000000, 0x00000000, 0x00000001> )); + + return map; +} + +// ES3 image loading functions vary based on the internal format and data type given, +// this map type determines the loading function from the internal format and type supplied +// to glTex*Image*D and the destination DXGI_FORMAT. Source formats and types are taken from +// Tables 3.2 and 3.3 of the ES 3 spec. +typedef std::pair<GLenum, LoadImageFunction> TypeLoadFunctionPair; +typedef std::map<GLenum, std::vector<TypeLoadFunctionPair> > D3D11LoadFunctionMap; + +static void UnimplementedLoadFunction(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + UNIMPLEMENTED(); +} + +static void UnreachableLoadFunction(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + UNREACHABLE(); +} + +// A helper function to insert data into the D3D11LoadFunctionMap with fewer characters. +static inline void InsertLoadFunction(D3D11LoadFunctionMap *map, GLenum internalFormat, GLenum type, + LoadImageFunction loadFunc) +{ + (*map)[internalFormat].push_back(TypeLoadFunctionPair(type, loadFunc)); +} + +D3D11LoadFunctionMap BuildD3D11LoadFunctionMap() +{ + D3D11LoadFunctionMap map; + + // | Internal format | Type | Load function | + InsertLoadFunction(&map, GL_RGBA8, GL_UNSIGNED_BYTE, LoadToNative<GLubyte, 4> ); + InsertLoadFunction(&map, GL_RGB5_A1, GL_UNSIGNED_BYTE, LoadToNative<GLubyte, 4> ); + InsertLoadFunction(&map, GL_RGBA4, GL_UNSIGNED_BYTE, LoadToNative<GLubyte, 4> ); + InsertLoadFunction(&map, GL_SRGB8_ALPHA8, GL_UNSIGNED_BYTE, LoadToNative<GLubyte, 4> ); + InsertLoadFunction(&map, GL_RGBA8_SNORM, GL_BYTE, LoadToNative<GLbyte, 4> ); + InsertLoadFunction(&map, GL_RGBA4, GL_UNSIGNED_SHORT_4_4_4_4, LoadRGBA4ToRGBA8 ); + InsertLoadFunction(&map, GL_RGB10_A2, GL_UNSIGNED_INT_2_10_10_10_REV, LoadToNative<GLuint, 1> ); + InsertLoadFunction(&map, GL_RGB5_A1, GL_UNSIGNED_SHORT_5_5_5_1, LoadRGB5A1ToRGBA8 ); + InsertLoadFunction(&map, GL_RGB5_A1, GL_UNSIGNED_INT_2_10_10_10_REV, LoadRGB10A2ToRGBA8 ); + InsertLoadFunction(&map, GL_RGBA16F, GL_HALF_FLOAT, LoadToNative<GLhalf, 4> ); + InsertLoadFunction(&map, GL_RGBA16F, GL_HALF_FLOAT_OES, LoadToNative<GLhalf, 4> ); + InsertLoadFunction(&map, GL_RGBA32F, GL_FLOAT, LoadToNative<GLfloat, 4> ); + InsertLoadFunction(&map, GL_RGBA16F, GL_FLOAT, Load32FTo16F<4> ); + InsertLoadFunction(&map, GL_RGBA8UI, GL_UNSIGNED_BYTE, LoadToNative<GLubyte, 4> ); + InsertLoadFunction(&map, GL_RGBA8I, GL_BYTE, LoadToNative<GLbyte, 4> ); + InsertLoadFunction(&map, GL_RGBA16UI, GL_UNSIGNED_SHORT, LoadToNative<GLushort, 4> ); + InsertLoadFunction(&map, GL_RGBA16I, GL_SHORT, LoadToNative<GLshort, 4> ); + InsertLoadFunction(&map, GL_RGBA32UI, GL_UNSIGNED_INT, LoadToNative<GLuint, 4> ); + InsertLoadFunction(&map, GL_RGBA32I, GL_INT, LoadToNative<GLint, 4> ); + InsertLoadFunction(&map, GL_RGB10_A2UI, GL_UNSIGNED_INT_2_10_10_10_REV, LoadToNative<GLuint, 1> ); + InsertLoadFunction(&map, GL_RGB8, GL_UNSIGNED_BYTE, LoadToNative3To4<GLubyte, 0xFF> ); + InsertLoadFunction(&map, GL_RGB565, GL_UNSIGNED_BYTE, LoadToNative3To4<GLubyte, 0xFF> ); + InsertLoadFunction(&map, GL_SRGB8, GL_UNSIGNED_BYTE, LoadToNative3To4<GLubyte, 0xFF> ); + InsertLoadFunction(&map, GL_RGB8_SNORM, GL_BYTE, LoadToNative3To4<GLbyte, 0x7F> ); + InsertLoadFunction(&map, GL_RGB565, GL_UNSIGNED_SHORT_5_6_5, LoadR5G6B5ToRGBA8 ); + InsertLoadFunction(&map, GL_R11F_G11F_B10F, GL_UNSIGNED_INT_10F_11F_11F_REV, LoadToNative<GLuint, 1> ); + InsertLoadFunction(&map, GL_RGB9_E5, GL_UNSIGNED_INT_5_9_9_9_REV, LoadToNative<GLuint, 1> ); + InsertLoadFunction(&map, GL_RGB16F, GL_HALF_FLOAT, LoadToNative3To4<GLhalf, gl::Float16One>); + InsertLoadFunction(&map, GL_RGB16F, GL_HALF_FLOAT_OES, LoadToNative3To4<GLhalf, gl::Float16One>); + InsertLoadFunction(&map, GL_R11F_G11F_B10F, GL_HALF_FLOAT, LoadRGB16FToRG11B10F ); + InsertLoadFunction(&map, GL_R11F_G11F_B10F, GL_HALF_FLOAT_OES, LoadRGB16FToRG11B10F ); + InsertLoadFunction(&map, GL_RGB9_E5, GL_HALF_FLOAT, LoadRGB16FToRGB9E5 ); + InsertLoadFunction(&map, GL_RGB9_E5, GL_HALF_FLOAT_OES, LoadRGB16FToRGB9E5 ); + InsertLoadFunction(&map, GL_RGB32F, GL_FLOAT, LoadToNative3To4<GLfloat, gl::Float32One>); + InsertLoadFunction(&map, GL_RGB16F, GL_FLOAT, LoadRGB32FToRGBA16F ); + InsertLoadFunction(&map, GL_R11F_G11F_B10F, GL_FLOAT, LoadRGB32FToRG11B10F ); + InsertLoadFunction(&map, GL_RGB9_E5, GL_FLOAT, LoadRGB32FToRGB9E5 ); + InsertLoadFunction(&map, GL_RGB8UI, GL_UNSIGNED_BYTE, LoadToNative3To4<GLubyte, 0x01> ); + InsertLoadFunction(&map, GL_RGB8I, GL_BYTE, LoadToNative3To4<GLbyte, 0x01> ); + InsertLoadFunction(&map, GL_RGB16UI, GL_UNSIGNED_SHORT, LoadToNative3To4<GLushort, 0x0001> ); + InsertLoadFunction(&map, GL_RGB16I, GL_SHORT, LoadToNative3To4<GLshort, 0x0001> ); + InsertLoadFunction(&map, GL_RGB32UI, GL_UNSIGNED_INT, LoadToNative3To4<GLuint, 0x00000001> ); + InsertLoadFunction(&map, GL_RGB32I, GL_INT, LoadToNative3To4<GLint, 0x00000001> ); + InsertLoadFunction(&map, GL_RG8, GL_UNSIGNED_BYTE, LoadToNative<GLubyte, 2> ); + InsertLoadFunction(&map, GL_RG8_SNORM, GL_BYTE, LoadToNative<GLbyte, 2> ); + InsertLoadFunction(&map, GL_RG16F, GL_HALF_FLOAT, LoadToNative<GLhalf, 2> ); + InsertLoadFunction(&map, GL_RG16F, GL_HALF_FLOAT_OES, LoadToNative<GLhalf, 2> ); + InsertLoadFunction(&map, GL_RG32F, GL_FLOAT, LoadToNative<GLfloat, 2> ); + InsertLoadFunction(&map, GL_RG16F, GL_FLOAT, Load32FTo16F<2> ); + InsertLoadFunction(&map, GL_RG8UI, GL_UNSIGNED_BYTE, LoadToNative<GLubyte, 2> ); + InsertLoadFunction(&map, GL_RG8I, GL_BYTE, LoadToNative<GLbyte, 2> ); + InsertLoadFunction(&map, GL_RG16UI, GL_UNSIGNED_SHORT, LoadToNative<GLushort, 2> ); + InsertLoadFunction(&map, GL_RG16I, GL_SHORT, LoadToNative<GLshort, 2> ); + InsertLoadFunction(&map, GL_RG32UI, GL_UNSIGNED_INT, LoadToNative<GLuint, 2> ); + InsertLoadFunction(&map, GL_RG32I, GL_INT, LoadToNative<GLint, 2> ); + InsertLoadFunction(&map, GL_R8, GL_UNSIGNED_BYTE, LoadToNative<GLubyte, 1> ); + InsertLoadFunction(&map, GL_R8_SNORM, GL_BYTE, LoadToNative<GLbyte, 1> ); + InsertLoadFunction(&map, GL_R16F, GL_HALF_FLOAT, LoadToNative<GLhalf, 1> ); + InsertLoadFunction(&map, GL_R16F, GL_HALF_FLOAT_OES, LoadToNative<GLhalf, 1> ); + InsertLoadFunction(&map, GL_R32F, GL_FLOAT, LoadToNative<GLfloat, 1> ); + InsertLoadFunction(&map, GL_R16F, GL_FLOAT, Load32FTo16F<1> ); + InsertLoadFunction(&map, GL_R8UI, GL_UNSIGNED_BYTE, LoadToNative<GLubyte, 1> ); + InsertLoadFunction(&map, GL_R8I, GL_BYTE, LoadToNative<GLbyte, 1> ); + InsertLoadFunction(&map, GL_R16UI, GL_UNSIGNED_SHORT, LoadToNative<GLushort, 1> ); + InsertLoadFunction(&map, GL_R16I, GL_SHORT, LoadToNative<GLshort, 1> ); + InsertLoadFunction(&map, GL_R32UI, GL_UNSIGNED_INT, LoadToNative<GLuint, 1> ); + InsertLoadFunction(&map, GL_R32I, GL_INT, LoadToNative<GLint, 1> ); + InsertLoadFunction(&map, GL_DEPTH_COMPONENT16, GL_UNSIGNED_SHORT, LoadToNative<GLushort, 1> ); + InsertLoadFunction(&map, GL_DEPTH_COMPONENT24, GL_UNSIGNED_INT, LoadR32ToR24G8 ); + InsertLoadFunction(&map, GL_DEPTH_COMPONENT16, GL_UNSIGNED_INT, LoadR32ToR16 ); + InsertLoadFunction(&map, GL_DEPTH_COMPONENT32F, GL_FLOAT, LoadToNative<GLfloat, 1> ); + InsertLoadFunction(&map, GL_DEPTH24_STENCIL8, GL_UNSIGNED_INT_24_8, LoadR32ToR24G8 ); + InsertLoadFunction(&map, GL_DEPTH32F_STENCIL8, GL_FLOAT_32_UNSIGNED_INT_24_8_REV, LoadToNative<GLuint, 2> ); + + // Unsized formats + // Load functions are unreachable because they are converted to sized internal formats based on + // the format and type before loading takes place. + InsertLoadFunction(&map, GL_RGBA, GL_UNSIGNED_BYTE, UnreachableLoadFunction ); + InsertLoadFunction(&map, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4, UnreachableLoadFunction ); + InsertLoadFunction(&map, GL_RGBA, GL_UNSIGNED_SHORT_5_5_5_1, UnreachableLoadFunction ); + InsertLoadFunction(&map, GL_RGB, GL_UNSIGNED_BYTE, UnreachableLoadFunction ); + InsertLoadFunction(&map, GL_RGB, GL_UNSIGNED_SHORT_5_6_5, UnreachableLoadFunction ); + InsertLoadFunction(&map, GL_LUMINANCE_ALPHA, GL_UNSIGNED_BYTE, UnreachableLoadFunction ); + InsertLoadFunction(&map, GL_LUMINANCE, GL_UNSIGNED_BYTE, UnreachableLoadFunction ); + InsertLoadFunction(&map, GL_ALPHA, GL_UNSIGNED_BYTE, UnreachableLoadFunction ); + + // From GL_OES_texture_float + InsertLoadFunction(&map, GL_LUMINANCE_ALPHA, GL_FLOAT, LoadLA32FToRGBA32F ); + InsertLoadFunction(&map, GL_LUMINANCE, GL_FLOAT, LoadL32FToRGBA32F ); + InsertLoadFunction(&map, GL_ALPHA, GL_FLOAT, LoadA32FToRGBA32F ); + + // From GL_OES_texture_half_float + InsertLoadFunction(&map, GL_LUMINANCE_ALPHA, GL_HALF_FLOAT, LoadLA16FToRGBA16F ); + InsertLoadFunction(&map, GL_LUMINANCE_ALPHA, GL_HALF_FLOAT_OES, LoadLA16FToRGBA16F ); + InsertLoadFunction(&map, GL_LUMINANCE, GL_HALF_FLOAT, LoadL16FToRGBA16F ); + InsertLoadFunction(&map, GL_LUMINANCE, GL_HALF_FLOAT_OES, LoadL16FToRGBA16F ); + InsertLoadFunction(&map, GL_ALPHA, GL_HALF_FLOAT, LoadA16FToRGBA16F ); + InsertLoadFunction(&map, GL_ALPHA, GL_HALF_FLOAT_OES, LoadA16FToRGBA16F ); + + // From GL_EXT_texture_storage + InsertLoadFunction(&map, GL_ALPHA8_EXT, GL_UNSIGNED_BYTE, LoadA8ToRGBA8 ); + InsertLoadFunction(&map, GL_LUMINANCE8_EXT, GL_UNSIGNED_BYTE, LoadL8ToRGBA8 ); + InsertLoadFunction(&map, GL_LUMINANCE8_ALPHA8_EXT, GL_UNSIGNED_BYTE, LoadLA8ToRGBA8 ); + InsertLoadFunction(&map, GL_ALPHA32F_EXT, GL_FLOAT, LoadA32FToRGBA32F ); + InsertLoadFunction(&map, GL_LUMINANCE32F_EXT, GL_FLOAT, LoadL32FToRGBA32F ); + InsertLoadFunction(&map, GL_LUMINANCE_ALPHA32F_EXT, GL_FLOAT, LoadLA32FToRGBA32F ); + InsertLoadFunction(&map, GL_ALPHA16F_EXT, GL_HALF_FLOAT, LoadA16FToRGBA16F ); + InsertLoadFunction(&map, GL_ALPHA16F_EXT, GL_HALF_FLOAT_OES, LoadA16FToRGBA16F ); + InsertLoadFunction(&map, GL_LUMINANCE16F_EXT, GL_HALF_FLOAT, LoadL16FToRGBA16F ); + InsertLoadFunction(&map, GL_LUMINANCE16F_EXT, GL_HALF_FLOAT_OES, LoadL16FToRGBA16F ); + InsertLoadFunction(&map, GL_LUMINANCE_ALPHA16F_EXT, GL_HALF_FLOAT, LoadLA16FToRGBA16F ); + InsertLoadFunction(&map, GL_LUMINANCE_ALPHA16F_EXT, GL_HALF_FLOAT_OES, LoadLA16FToRGBA16F ); + + // From GL_ANGLE_depth_texture + InsertLoadFunction(&map, GL_DEPTH_COMPONENT32_OES, GL_UNSIGNED_INT, LoadR32ToR24G8 ); + + // From GL_EXT_texture_format_BGRA8888 + InsertLoadFunction(&map, GL_BGRA8_EXT, GL_UNSIGNED_BYTE, LoadToNative<GLubyte, 4> ); + InsertLoadFunction(&map, GL_BGRA4_ANGLEX, GL_UNSIGNED_SHORT_4_4_4_4_REV_EXT, LoadRGBA4ToRGBA8 ); + InsertLoadFunction(&map, GL_BGRA4_ANGLEX, GL_UNSIGNED_BYTE, LoadToNative<GLubyte, 4> ); + InsertLoadFunction(&map, GL_BGR5_A1_ANGLEX, GL_UNSIGNED_SHORT_1_5_5_5_REV_EXT, LoadRGB5A1ToRGBA8 ); + InsertLoadFunction(&map, GL_BGR5_A1_ANGLEX, GL_UNSIGNED_BYTE, LoadToNative<GLubyte, 4> ); + + // Compressed formats + // From ES 3.0.1 spec, table 3.16 + // | Internal format | Type | Load function | + InsertLoadFunction(&map, GL_COMPRESSED_R11_EAC, GL_UNSIGNED_BYTE, UnimplementedLoadFunction ); + InsertLoadFunction(&map, GL_COMPRESSED_R11_EAC, GL_UNSIGNED_BYTE, UnimplementedLoadFunction ); + InsertLoadFunction(&map, GL_COMPRESSED_SIGNED_R11_EAC, GL_UNSIGNED_BYTE, UnimplementedLoadFunction ); + InsertLoadFunction(&map, GL_COMPRESSED_RG11_EAC, GL_UNSIGNED_BYTE, UnimplementedLoadFunction ); + InsertLoadFunction(&map, GL_COMPRESSED_SIGNED_RG11_EAC, GL_UNSIGNED_BYTE, UnimplementedLoadFunction ); + InsertLoadFunction(&map, GL_COMPRESSED_RGB8_ETC2, GL_UNSIGNED_BYTE, UnimplementedLoadFunction ); + InsertLoadFunction(&map, GL_COMPRESSED_SRGB8_ETC2, GL_UNSIGNED_BYTE, UnimplementedLoadFunction ); + InsertLoadFunction(&map, GL_COMPRESSED_RGB8_PUNCHTHROUGH_ALPHA1_ETC2, GL_UNSIGNED_BYTE, UnimplementedLoadFunction ); + InsertLoadFunction(&map, GL_COMPRESSED_SRGB8_PUNCHTHROUGH_ALPHA1_ETC2, GL_UNSIGNED_BYTE, UnimplementedLoadFunction ); + InsertLoadFunction(&map, GL_COMPRESSED_RGBA8_ETC2_EAC, GL_UNSIGNED_BYTE, UnimplementedLoadFunction ); + InsertLoadFunction(&map, GL_COMPRESSED_SRGB8_ALPHA8_ETC2_EAC, GL_UNSIGNED_BYTE, UnimplementedLoadFunction ); + + // From GL_EXT_texture_compression_dxt1 + InsertLoadFunction(&map, GL_COMPRESSED_RGB_S3TC_DXT1_EXT, GL_UNSIGNED_BYTE, LoadCompressedToNative<4, 4, 8>); + InsertLoadFunction(&map, GL_COMPRESSED_RGBA_S3TC_DXT1_EXT, GL_UNSIGNED_BYTE, LoadCompressedToNative<4, 4, 8>); + + // From GL_ANGLE_texture_compression_dxt3 + InsertLoadFunction(&map, GL_COMPRESSED_RGBA_S3TC_DXT3_ANGLE, GL_UNSIGNED_BYTE, LoadCompressedToNative<4, 4, 16>); + + // From GL_ANGLE_texture_compression_dxt5 + InsertLoadFunction(&map, GL_COMPRESSED_RGBA_S3TC_DXT5_ANGLE, GL_UNSIGNED_BYTE, LoadCompressedToNative<4, 4, 16>); + + return map; +} + +// For sized GL internal formats, there is only one corresponding D3D11 format. This map type allows +// querying for the DXGI texture formats to use for textures, SRVs, RTVs and DSVs given a GL internal +// format. +typedef std::map<GLenum, TextureFormat> D3D11ES3FormatMap; + +TextureFormat::TextureFormat() + : texFormat(DXGI_FORMAT_UNKNOWN), + srvFormat(DXGI_FORMAT_UNKNOWN), + rtvFormat(DXGI_FORMAT_UNKNOWN), + dsvFormat(DXGI_FORMAT_UNKNOWN), + renderFormat(DXGI_FORMAT_UNKNOWN), + swizzleTexFormat(DXGI_FORMAT_UNKNOWN), + swizzleSRVFormat(DXGI_FORMAT_UNKNOWN), + swizzleRTVFormat(DXGI_FORMAT_UNKNOWN), + dataInitializerFunction(NULL), + loadFunctions() +{ +} + +static inline void InsertD3D11FormatInfo(D3D11ES3FormatMap *map, GLenum internalFormat, DXGI_FORMAT texFormat, + DXGI_FORMAT srvFormat, DXGI_FORMAT rtvFormat, DXGI_FORMAT dsvFormat) +{ + TextureFormat info; + info.texFormat = texFormat; + info.srvFormat = srvFormat; + info.rtvFormat = rtvFormat; + info.dsvFormat = dsvFormat; + + // Given a GL internal format, the renderFormat is the DSV format if it is depth- or stencil-renderable, + // the RTV format if it is color-renderable, and the (nonrenderable) texture format otherwise. + if (dsvFormat != DXGI_FORMAT_UNKNOWN) + { + info.renderFormat = dsvFormat; + } + else if (rtvFormat != DXGI_FORMAT_UNKNOWN) + { + info.renderFormat = rtvFormat; + } + else if (texFormat != DXGI_FORMAT_UNKNOWN) + { + info.renderFormat = texFormat; + } + else + { + info.renderFormat = DXGI_FORMAT_UNKNOWN; + } + + // Compute the swizzle formats + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(internalFormat); + if (internalFormat != GL_NONE && formatInfo.pixelBytes > 0) + { + if (formatInfo.componentCount != 4 || texFormat == DXGI_FORMAT_UNKNOWN || + srvFormat == DXGI_FORMAT_UNKNOWN || rtvFormat == DXGI_FORMAT_UNKNOWN) + { + // Get the maximum sized component + unsigned int maxBits = 1; + if (formatInfo.compressed) + { + unsigned int compressedBitsPerBlock = formatInfo.pixelBytes * 8; + unsigned int blockSize = formatInfo.compressedBlockWidth * formatInfo.compressedBlockHeight; + maxBits = std::max(compressedBitsPerBlock / blockSize, maxBits); + } + else + { + maxBits = std::max(maxBits, formatInfo.alphaBits); + maxBits = std::max(maxBits, formatInfo.redBits); + maxBits = std::max(maxBits, formatInfo.greenBits); + maxBits = std::max(maxBits, formatInfo.blueBits); + maxBits = std::max(maxBits, formatInfo.luminanceBits); + maxBits = std::max(maxBits, formatInfo.depthBits); + } + + maxBits = roundUp(maxBits, 8U); + + static const SwizzleInfoMap swizzleMap = BuildSwizzleInfoMap(); + SwizzleInfoMap::const_iterator swizzleIter = swizzleMap.find(SwizzleSizeType(maxBits, formatInfo.componentType)); + ASSERT(swizzleIter != swizzleMap.end()); + + const SwizzleFormatInfo &swizzleInfo = swizzleIter->second; + info.swizzleTexFormat = swizzleInfo.mTexFormat; + info.swizzleSRVFormat = swizzleInfo.mSRVFormat; + info.swizzleRTVFormat = swizzleInfo.mRTVFormat; + } + else + { + // The original texture format is suitable for swizzle operations + info.swizzleTexFormat = texFormat; + info.swizzleSRVFormat = srvFormat; + info.swizzleRTVFormat = rtvFormat; + } + } + else + { + // Not possible to swizzle with this texture format since it is either unsized or GL_NONE + info.swizzleTexFormat = DXGI_FORMAT_UNKNOWN; + info.swizzleSRVFormat = DXGI_FORMAT_UNKNOWN; + info.swizzleRTVFormat = DXGI_FORMAT_UNKNOWN; + } + + // Check if there is an initialization function for this texture format + static const InternalFormatInitializerMap initializerMap = BuildInternalFormatInitializerMap(); + InternalFormatInitializerMap::const_iterator initializerIter = initializerMap.find(internalFormat); + info.dataInitializerFunction = (initializerIter != initializerMap.end()) ? initializerIter->second : NULL; + + // Gather all the load functions for this internal format + static const D3D11LoadFunctionMap loadFunctions = BuildD3D11LoadFunctionMap(); + D3D11LoadFunctionMap::const_iterator loadFunctionIter = loadFunctions.find(internalFormat); + if (loadFunctionIter != loadFunctions.end()) + { + const std::vector<TypeLoadFunctionPair> &loadFunctionVector = loadFunctionIter->second; + for (size_t i = 0; i < loadFunctionVector.size(); i++) + { + GLenum type = loadFunctionVector[i].first; + LoadImageFunction function = loadFunctionVector[i].second; + info.loadFunctions.insert(std::make_pair(type, function)); + } + } + + map->insert(std::make_pair(internalFormat, info)); +} + +static D3D11ES3FormatMap BuildD3D11FormatMap() +{ + D3D11ES3FormatMap map; + + // | GL internal format | D3D11 texture format | D3D11 SRV format | D3D11 RTV format | D3D11 DSV format | + InsertD3D11FormatInfo(&map, GL_NONE, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_R8, DXGI_FORMAT_R8_UNORM, DXGI_FORMAT_R8_UNORM, DXGI_FORMAT_R8_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_R8_SNORM, DXGI_FORMAT_R8_SNORM, DXGI_FORMAT_R8_SNORM, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RG8, DXGI_FORMAT_R8G8_UNORM, DXGI_FORMAT_R8G8_UNORM, DXGI_FORMAT_R8G8_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RG8_SNORM, DXGI_FORMAT_R8G8_SNORM, DXGI_FORMAT_R8G8_SNORM, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB8, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB8_SNORM, DXGI_FORMAT_R8G8B8A8_SNORM, DXGI_FORMAT_R8G8B8A8_SNORM, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB565, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGBA4, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB5_A1, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGBA8, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGBA8_SNORM, DXGI_FORMAT_R8G8B8A8_SNORM, DXGI_FORMAT_R8G8B8A8_SNORM, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB10_A2, DXGI_FORMAT_R10G10B10A2_UNORM, DXGI_FORMAT_R10G10B10A2_UNORM, DXGI_FORMAT_R10G10B10A2_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB10_A2UI, DXGI_FORMAT_R10G10B10A2_UINT, DXGI_FORMAT_R10G10B10A2_UINT, DXGI_FORMAT_R10G10B10A2_UINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_SRGB8, DXGI_FORMAT_R8G8B8A8_UNORM_SRGB, DXGI_FORMAT_R8G8B8A8_UNORM_SRGB, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_SRGB8_ALPHA8, DXGI_FORMAT_R8G8B8A8_UNORM_SRGB, DXGI_FORMAT_R8G8B8A8_UNORM_SRGB, DXGI_FORMAT_R8G8B8A8_UNORM_SRGB, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_R16F, DXGI_FORMAT_R16_FLOAT, DXGI_FORMAT_R16_FLOAT, DXGI_FORMAT_R16_FLOAT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RG16F, DXGI_FORMAT_R16G16_FLOAT, DXGI_FORMAT_R16G16_FLOAT, DXGI_FORMAT_R16G16_FLOAT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB16F, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGBA16F, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_R32F, DXGI_FORMAT_R32_FLOAT, DXGI_FORMAT_R32_FLOAT, DXGI_FORMAT_R32_FLOAT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RG32F, DXGI_FORMAT_R32G32_FLOAT, DXGI_FORMAT_R32G32_FLOAT, DXGI_FORMAT_R32G32_FLOAT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB32F, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGBA32F, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_R11F_G11F_B10F, DXGI_FORMAT_R11G11B10_FLOAT, DXGI_FORMAT_R11G11B10_FLOAT, DXGI_FORMAT_R11G11B10_FLOAT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB9_E5, DXGI_FORMAT_R9G9B9E5_SHAREDEXP, DXGI_FORMAT_R9G9B9E5_SHAREDEXP, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_R8I, DXGI_FORMAT_R8_SINT, DXGI_FORMAT_R8_SINT, DXGI_FORMAT_R8_SINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_R8UI, DXGI_FORMAT_R8_UINT, DXGI_FORMAT_R8_UINT, DXGI_FORMAT_R8_UINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_R16I, DXGI_FORMAT_R16_SINT, DXGI_FORMAT_R16_SINT, DXGI_FORMAT_R16_SINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_R16UI, DXGI_FORMAT_R16_UINT, DXGI_FORMAT_R16_UINT, DXGI_FORMAT_R16_UINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_R32I, DXGI_FORMAT_R32_SINT, DXGI_FORMAT_R32_SINT, DXGI_FORMAT_R32_SINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_R32UI, DXGI_FORMAT_R32_UINT, DXGI_FORMAT_R32_UINT, DXGI_FORMAT_R32_UINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RG8I, DXGI_FORMAT_R8G8_SINT, DXGI_FORMAT_R8G8_SINT, DXGI_FORMAT_R8G8_SINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RG8UI, DXGI_FORMAT_R8G8_UINT, DXGI_FORMAT_R8G8_UINT, DXGI_FORMAT_R8G8_UINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RG16I, DXGI_FORMAT_R16G16_SINT, DXGI_FORMAT_R16G16_SINT, DXGI_FORMAT_R16G16_SINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RG16UI, DXGI_FORMAT_R16G16_UINT, DXGI_FORMAT_R16G16_UINT, DXGI_FORMAT_R16G16_UINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RG32I, DXGI_FORMAT_R32G32_SINT, DXGI_FORMAT_R32G32_SINT, DXGI_FORMAT_R32G32_SINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RG32UI, DXGI_FORMAT_R32G32_UINT, DXGI_FORMAT_R32G32_UINT, DXGI_FORMAT_R32G32_UINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB8I, DXGI_FORMAT_R8G8B8A8_SINT, DXGI_FORMAT_R8G8B8A8_SINT, DXGI_FORMAT_R8G8B8A8_SINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB8UI, DXGI_FORMAT_R8G8B8A8_UINT, DXGI_FORMAT_R8G8B8A8_UINT, DXGI_FORMAT_R8G8B8A8_UINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB16I, DXGI_FORMAT_R16G16B16A16_SINT, DXGI_FORMAT_R16G16B16A16_SINT, DXGI_FORMAT_R16G16B16A16_SINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB16UI, DXGI_FORMAT_R16G16B16A16_UINT, DXGI_FORMAT_R16G16B16A16_UINT, DXGI_FORMAT_R16G16B16A16_UINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB32I, DXGI_FORMAT_R32G32B32A32_SINT, DXGI_FORMAT_R32G32B32A32_SINT, DXGI_FORMAT_R32G32B32A32_SINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB32UI, DXGI_FORMAT_R32G32B32A32_UINT, DXGI_FORMAT_R32G32B32A32_UINT, DXGI_FORMAT_R32G32B32A32_UINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGBA8I, DXGI_FORMAT_R8G8B8A8_SINT, DXGI_FORMAT_R8G8B8A8_SINT, DXGI_FORMAT_R8G8B8A8_SINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGBA8UI, DXGI_FORMAT_R8G8B8A8_UINT, DXGI_FORMAT_R8G8B8A8_UINT, DXGI_FORMAT_R8G8B8A8_UINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGBA16I, DXGI_FORMAT_R16G16B16A16_SINT, DXGI_FORMAT_R16G16B16A16_SINT, DXGI_FORMAT_R16G16B16A16_SINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGBA16UI, DXGI_FORMAT_R16G16B16A16_UINT, DXGI_FORMAT_R16G16B16A16_UINT, DXGI_FORMAT_R16G16B16A16_UINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGBA32I, DXGI_FORMAT_R32G32B32A32_SINT, DXGI_FORMAT_R32G32B32A32_SINT, DXGI_FORMAT_R32G32B32A32_SINT, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGBA32UI, DXGI_FORMAT_R32G32B32A32_UINT, DXGI_FORMAT_R32G32B32A32_UINT, DXGI_FORMAT_R32G32B32A32_UINT, DXGI_FORMAT_UNKNOWN); + + // Unsized formats, TODO: Are types of float and half float allowed for the unsized types? Would it change the DXGI format? + InsertD3D11FormatInfo(&map, GL_ALPHA, DXGI_FORMAT_A8_UNORM, DXGI_FORMAT_A8_UNORM, DXGI_FORMAT_A8_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_LUMINANCE, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_LUMINANCE_ALPHA, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGB, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_RGBA, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_BGRA_EXT, DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_UNKNOWN); + + // From GL_EXT_texture_storage + // | GL internal format | D3D11 texture format | D3D11 SRV format | D3D11 RTV format | D3D11 DSV format | + InsertD3D11FormatInfo(&map, GL_ALPHA8_EXT, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_UNKNOWN ); + InsertD3D11FormatInfo(&map, GL_LUMINANCE8_EXT, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_UNKNOWN ); + InsertD3D11FormatInfo(&map, GL_ALPHA32F_EXT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_UNKNOWN ); + InsertD3D11FormatInfo(&map, GL_LUMINANCE32F_EXT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_UNKNOWN ); + InsertD3D11FormatInfo(&map, GL_ALPHA16F_EXT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_UNKNOWN ); + InsertD3D11FormatInfo(&map, GL_LUMINANCE16F_EXT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_UNKNOWN ); + InsertD3D11FormatInfo(&map, GL_LUMINANCE8_ALPHA8_EXT, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_R8G8B8A8_UNORM, DXGI_FORMAT_UNKNOWN ); + InsertD3D11FormatInfo(&map, GL_LUMINANCE_ALPHA32F_EXT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_R32G32B32A32_FLOAT, DXGI_FORMAT_UNKNOWN ); + InsertD3D11FormatInfo(&map, GL_LUMINANCE_ALPHA16F_EXT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_R16G16B16A16_FLOAT, DXGI_FORMAT_UNKNOWN ); + InsertD3D11FormatInfo(&map, GL_BGRA8_EXT, DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_UNKNOWN ); + InsertD3D11FormatInfo(&map, GL_BGRA4_ANGLEX, DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_UNKNOWN ); + InsertD3D11FormatInfo(&map, GL_BGR5_A1_ANGLEX, DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_UNKNOWN ); + + // Depth stencil formats + InsertD3D11FormatInfo(&map, GL_DEPTH_COMPONENT16, DXGI_FORMAT_R16_TYPELESS, DXGI_FORMAT_R16_UNORM, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_D16_UNORM ); + InsertD3D11FormatInfo(&map, GL_DEPTH_COMPONENT24, DXGI_FORMAT_R24G8_TYPELESS, DXGI_FORMAT_R24_UNORM_X8_TYPELESS, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_D24_UNORM_S8_UINT ); + InsertD3D11FormatInfo(&map, GL_DEPTH_COMPONENT32F, DXGI_FORMAT_R32_TYPELESS, DXGI_FORMAT_R32_FLOAT, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_D32_FLOAT ); + InsertD3D11FormatInfo(&map, GL_DEPTH24_STENCIL8, DXGI_FORMAT_R24G8_TYPELESS, DXGI_FORMAT_R24_UNORM_X8_TYPELESS, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_D24_UNORM_S8_UINT ); + InsertD3D11FormatInfo(&map, GL_DEPTH32F_STENCIL8, DXGI_FORMAT_R32G8X24_TYPELESS, DXGI_FORMAT_R32_FLOAT_X8X24_TYPELESS, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_D32_FLOAT_S8X24_UINT); + InsertD3D11FormatInfo(&map, GL_STENCIL_INDEX8, DXGI_FORMAT_R24G8_TYPELESS, DXGI_FORMAT_X24_TYPELESS_G8_UINT, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_D24_UNORM_S8_UINT ); + + // From GL_ANGLE_depth_texture + // Since D3D11 doesn't have a D32_UNORM format, use D24S8 which has comparable precision and matches the ES3 format. + InsertD3D11FormatInfo(&map, GL_DEPTH_COMPONENT32_OES, DXGI_FORMAT_R24G8_TYPELESS, DXGI_FORMAT_R24_UNORM_X8_TYPELESS, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_D24_UNORM_S8_UINT); + + // Compressed formats, From ES 3.0.1 spec, table 3.16 + // | GL internal format | D3D11 texture format | D3D11 SRV format | D3D11 RTV format | D3D11 DSV format | + InsertD3D11FormatInfo(&map, GL_COMPRESSED_R11_EAC, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_COMPRESSED_SIGNED_R11_EAC, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_COMPRESSED_RG11_EAC, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_COMPRESSED_SIGNED_RG11_EAC, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_COMPRESSED_RGB8_ETC2, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_COMPRESSED_SRGB8_ETC2, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_COMPRESSED_RGB8_PUNCHTHROUGH_ALPHA1_ETC2, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_COMPRESSED_SRGB8_PUNCHTHROUGH_ALPHA1_ETC2, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_COMPRESSED_RGBA8_ETC2_EAC, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_COMPRESSED_SRGB8_ALPHA8_ETC2_EAC, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + + // From GL_EXT_texture_compression_dxt1 + InsertD3D11FormatInfo(&map, GL_COMPRESSED_RGB_S3TC_DXT1_EXT, DXGI_FORMAT_BC1_UNORM, DXGI_FORMAT_BC1_UNORM, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + InsertD3D11FormatInfo(&map, GL_COMPRESSED_RGBA_S3TC_DXT1_EXT, DXGI_FORMAT_BC1_UNORM, DXGI_FORMAT_BC1_UNORM, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + + // From GL_ANGLE_texture_compression_dxt3 + InsertD3D11FormatInfo(&map, GL_COMPRESSED_RGBA_S3TC_DXT3_ANGLE, DXGI_FORMAT_BC2_UNORM, DXGI_FORMAT_BC2_UNORM, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + + // From GL_ANGLE_texture_compression_dxt5 + InsertD3D11FormatInfo(&map, GL_COMPRESSED_RGBA_S3TC_DXT5_ANGLE, DXGI_FORMAT_BC3_UNORM, DXGI_FORMAT_BC3_UNORM, DXGI_FORMAT_UNKNOWN, DXGI_FORMAT_UNKNOWN); + + return map; +} + +const TextureFormat &GetTextureFormatInfo(GLenum internalFormat) +{ + static const D3D11ES3FormatMap formatMap = BuildD3D11FormatMap(); + D3D11ES3FormatMap::const_iterator iter = formatMap.find(internalFormat); + if (iter != formatMap.end()) + { + return iter->second; + } + else + { + static const TextureFormat defaultInfo; + return defaultInfo; + } +} + +typedef std::map<gl::VertexFormat, VertexFormat> D3D11VertexFormatInfoMap; +typedef std::pair<gl::VertexFormat, VertexFormat> D3D11VertexFormatPair; + +VertexFormat::VertexFormat() + : conversionType(VERTEX_CONVERT_NONE), + nativeFormat(DXGI_FORMAT_UNKNOWN), + copyFunction(NULL) +{ +} + +static void AddVertexFormatInfo(D3D11VertexFormatInfoMap *map, GLenum inputType, GLboolean normalized, GLuint componentCount, + VertexConversionType conversionType, DXGI_FORMAT nativeFormat, VertexCopyFunction copyFunction) +{ + gl::VertexFormat inputFormat(inputType, normalized, componentCount, false); + + VertexFormat info; + info.conversionType = conversionType; + info.nativeFormat = nativeFormat; + info.copyFunction = copyFunction; + + map->insert(D3D11VertexFormatPair(inputFormat, info)); +} + +static void AddIntegerVertexFormatInfo(D3D11VertexFormatInfoMap *map, GLenum inputType, GLuint componentCount, + VertexConversionType conversionType, DXGI_FORMAT nativeFormat, VertexCopyFunction copyFunction) +{ + gl::VertexFormat inputFormat(inputType, GL_FALSE, componentCount, true); + + VertexFormat info; + info.conversionType = conversionType; + info.nativeFormat = nativeFormat; + info.copyFunction = copyFunction; + + map->insert(D3D11VertexFormatPair(inputFormat, info)); +} + +static D3D11VertexFormatInfoMap BuildD3D11VertexFormatInfoMap() +{ + D3D11VertexFormatInfoMap map; + + // TODO: column legend + + // + // Float formats + // + + // GL_BYTE -- un-normalized + AddVertexFormatInfo(&map, GL_BYTE, GL_FALSE, 1, VERTEX_CONVERT_GPU, DXGI_FORMAT_R8_SINT, &CopyNativeVertexData<GLbyte, 1, 0>); + AddVertexFormatInfo(&map, GL_BYTE, GL_FALSE, 2, VERTEX_CONVERT_GPU, DXGI_FORMAT_R8G8_SINT, &CopyNativeVertexData<GLbyte, 2, 0>); + AddVertexFormatInfo(&map, GL_BYTE, GL_FALSE, 3, VERTEX_CONVERT_BOTH, DXGI_FORMAT_R8G8B8A8_SINT, &CopyNativeVertexData<GLbyte, 3, 1>); + AddVertexFormatInfo(&map, GL_BYTE, GL_FALSE, 4, VERTEX_CONVERT_GPU, DXGI_FORMAT_R8G8B8A8_SINT, &CopyNativeVertexData<GLbyte, 4, 0>); + + // GL_BYTE -- normalized + AddVertexFormatInfo(&map, GL_BYTE, GL_TRUE, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R8_SNORM, &CopyNativeVertexData<GLbyte, 1, 0>); + AddVertexFormatInfo(&map, GL_BYTE, GL_TRUE, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R8G8_SNORM, &CopyNativeVertexData<GLbyte, 2, 0>); + AddVertexFormatInfo(&map, GL_BYTE, GL_TRUE, 3, VERTEX_CONVERT_CPU, DXGI_FORMAT_R8G8B8A8_SNORM, &CopyNativeVertexData<GLbyte, 3, INT8_MAX>); + AddVertexFormatInfo(&map, GL_BYTE, GL_TRUE, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R8G8B8A8_SNORM, &CopyNativeVertexData<GLbyte, 4, 0>); + + // GL_UNSIGNED_BYTE -- un-normalized + AddVertexFormatInfo(&map, GL_UNSIGNED_BYTE, GL_FALSE, 1, VERTEX_CONVERT_GPU, DXGI_FORMAT_R8_UINT, &CopyNativeVertexData<GLubyte, 1, 0>); + AddVertexFormatInfo(&map, GL_UNSIGNED_BYTE, GL_FALSE, 2, VERTEX_CONVERT_GPU, DXGI_FORMAT_R8G8_UINT, &CopyNativeVertexData<GLubyte, 2, 0>); + AddVertexFormatInfo(&map, GL_UNSIGNED_BYTE, GL_FALSE, 3, VERTEX_CONVERT_BOTH, DXGI_FORMAT_R8G8B8A8_UINT, &CopyNativeVertexData<GLubyte, 3, 1>); + AddVertexFormatInfo(&map, GL_UNSIGNED_BYTE, GL_FALSE, 4, VERTEX_CONVERT_GPU, DXGI_FORMAT_R8G8B8A8_UINT, &CopyNativeVertexData<GLubyte, 4, 0>); + + // GL_UNSIGNED_BYTE -- normalized + AddVertexFormatInfo(&map, GL_UNSIGNED_BYTE, GL_TRUE, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R8_UNORM, &CopyNativeVertexData<GLubyte, 1, 0>); + AddVertexFormatInfo(&map, GL_UNSIGNED_BYTE, GL_TRUE, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R8G8_UNORM, &CopyNativeVertexData<GLubyte, 2, 0>); + AddVertexFormatInfo(&map, GL_UNSIGNED_BYTE, GL_TRUE, 3, VERTEX_CONVERT_CPU, DXGI_FORMAT_R8G8B8A8_UNORM, &CopyNativeVertexData<GLubyte, 3, UINT8_MAX>); + AddVertexFormatInfo(&map, GL_UNSIGNED_BYTE, GL_TRUE, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R8G8B8A8_UNORM, &CopyNativeVertexData<GLubyte, 4, 0>); + + // GL_SHORT -- un-normalized + AddVertexFormatInfo(&map, GL_SHORT, GL_FALSE, 1, VERTEX_CONVERT_GPU, DXGI_FORMAT_R16_SINT, &CopyNativeVertexData<GLshort, 1, 0>); + AddVertexFormatInfo(&map, GL_SHORT, GL_FALSE, 2, VERTEX_CONVERT_GPU, DXGI_FORMAT_R16G16_SINT, &CopyNativeVertexData<GLshort, 2, 0>); + AddVertexFormatInfo(&map, GL_SHORT, GL_FALSE, 3, VERTEX_CONVERT_BOTH, DXGI_FORMAT_R16G16B16A16_SINT, &CopyNativeVertexData<GLshort, 4, 1>); + AddVertexFormatInfo(&map, GL_SHORT, GL_FALSE, 4, VERTEX_CONVERT_GPU, DXGI_FORMAT_R16G16B16A16_SINT, &CopyNativeVertexData<GLshort, 4, 0>); + + // GL_SHORT -- normalized + AddVertexFormatInfo(&map, GL_SHORT, GL_TRUE, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16_SNORM, &CopyNativeVertexData<GLshort, 1, 0>); + AddVertexFormatInfo(&map, GL_SHORT, GL_TRUE, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16G16_SNORM, &CopyNativeVertexData<GLshort, 2, 0>); + AddVertexFormatInfo(&map, GL_SHORT, GL_TRUE, 3, VERTEX_CONVERT_CPU, DXGI_FORMAT_R16G16B16A16_SNORM, &CopyNativeVertexData<GLshort, 3, INT16_MAX>); + AddVertexFormatInfo(&map, GL_SHORT, GL_TRUE, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16G16B16A16_SNORM, &CopyNativeVertexData<GLshort, 4, 0>); + + // GL_UNSIGNED_SHORT -- un-normalized + AddVertexFormatInfo(&map, GL_UNSIGNED_SHORT, GL_FALSE, 1, VERTEX_CONVERT_GPU, DXGI_FORMAT_R16_UINT, &CopyNativeVertexData<GLushort, 1, 0>); + AddVertexFormatInfo(&map, GL_UNSIGNED_SHORT, GL_FALSE, 2, VERTEX_CONVERT_GPU, DXGI_FORMAT_R16G16_UINT, &CopyNativeVertexData<GLushort, 2, 0>); + AddVertexFormatInfo(&map, GL_UNSIGNED_SHORT, GL_FALSE, 3, VERTEX_CONVERT_BOTH, DXGI_FORMAT_R16G16B16A16_UINT, &CopyNativeVertexData<GLushort, 3, 1>); + AddVertexFormatInfo(&map, GL_UNSIGNED_SHORT, GL_FALSE, 4, VERTEX_CONVERT_GPU, DXGI_FORMAT_R16G16B16A16_UINT, &CopyNativeVertexData<GLushort, 4, 0>); + + // GL_UNSIGNED_SHORT -- normalized + AddVertexFormatInfo(&map, GL_UNSIGNED_SHORT, GL_TRUE, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16_UNORM, &CopyNativeVertexData<GLushort, 1, 0>); + AddVertexFormatInfo(&map, GL_UNSIGNED_SHORT, GL_TRUE, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16G16_UNORM, &CopyNativeVertexData<GLushort, 2, 0>); + AddVertexFormatInfo(&map, GL_UNSIGNED_SHORT, GL_TRUE, 3, VERTEX_CONVERT_CPU, DXGI_FORMAT_R16G16B16A16_UNORM, &CopyNativeVertexData<GLushort, 3, UINT16_MAX>); + AddVertexFormatInfo(&map, GL_UNSIGNED_SHORT, GL_TRUE, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16G16B16A16_UNORM, &CopyNativeVertexData<GLushort, 4, 0>); + + // GL_INT -- un-normalized + AddVertexFormatInfo(&map, GL_INT, GL_FALSE, 1, VERTEX_CONVERT_GPU, DXGI_FORMAT_R32_SINT, &CopyNativeVertexData<GLint, 1, 0>); + AddVertexFormatInfo(&map, GL_INT, GL_FALSE, 2, VERTEX_CONVERT_GPU, DXGI_FORMAT_R32G32_SINT, &CopyNativeVertexData<GLint, 2, 0>); + AddVertexFormatInfo(&map, GL_INT, GL_FALSE, 3, VERTEX_CONVERT_GPU, DXGI_FORMAT_R32G32B32_SINT, &CopyNativeVertexData<GLint, 3, 0>); + AddVertexFormatInfo(&map, GL_INT, GL_FALSE, 4, VERTEX_CONVERT_GPU, DXGI_FORMAT_R32G32B32A32_SINT, &CopyNativeVertexData<GLint, 4, 0>); + + // GL_INT -- normalized + AddVertexFormatInfo(&map, GL_INT, GL_TRUE, 1, VERTEX_CONVERT_CPU, DXGI_FORMAT_R32_FLOAT, &CopyTo32FVertexData<GLint, 1, true>); + AddVertexFormatInfo(&map, GL_INT, GL_TRUE, 2, VERTEX_CONVERT_CPU, DXGI_FORMAT_R32G32_FLOAT, &CopyTo32FVertexData<GLint, 2, true>); + AddVertexFormatInfo(&map, GL_INT, GL_TRUE, 3, VERTEX_CONVERT_CPU, DXGI_FORMAT_R32G32B32_FLOAT, &CopyTo32FVertexData<GLint, 3, true>); + AddVertexFormatInfo(&map, GL_INT, GL_TRUE, 4, VERTEX_CONVERT_CPU, DXGI_FORMAT_R32G32B32A32_FLOAT, &CopyTo32FVertexData<GLint, 4, true>); + + // GL_UNSIGNED_INT -- un-normalized + AddVertexFormatInfo(&map, GL_UNSIGNED_INT, GL_FALSE, 1, VERTEX_CONVERT_GPU, DXGI_FORMAT_R32_UINT, &CopyNativeVertexData<GLuint, 1, 0>); + AddVertexFormatInfo(&map, GL_UNSIGNED_INT, GL_FALSE, 2, VERTEX_CONVERT_GPU, DXGI_FORMAT_R32G32_UINT, &CopyNativeVertexData<GLuint, 2, 0>); + AddVertexFormatInfo(&map, GL_UNSIGNED_INT, GL_FALSE, 3, VERTEX_CONVERT_GPU, DXGI_FORMAT_R32G32B32_UINT, &CopyNativeVertexData<GLuint, 3, 0>); + AddVertexFormatInfo(&map, GL_UNSIGNED_INT, GL_FALSE, 4, VERTEX_CONVERT_GPU, DXGI_FORMAT_R32G32B32A32_UINT, &CopyNativeVertexData<GLuint, 4, 0>); + + // GL_UNSIGNED_INT -- normalized + AddVertexFormatInfo(&map, GL_UNSIGNED_INT, GL_TRUE, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32_FLOAT, &CopyTo32FVertexData<GLuint, 1, true>); + AddVertexFormatInfo(&map, GL_UNSIGNED_INT, GL_TRUE, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32G32_FLOAT, &CopyTo32FVertexData<GLuint, 2, true>); + AddVertexFormatInfo(&map, GL_UNSIGNED_INT, GL_TRUE, 3, VERTEX_CONVERT_CPU, DXGI_FORMAT_R32G32B32_FLOAT, &CopyTo32FVertexData<GLuint, 3, true>); + AddVertexFormatInfo(&map, GL_UNSIGNED_INT, GL_TRUE, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32G32B32A32_FLOAT, &CopyTo32FVertexData<GLuint, 4, true>); + + // GL_FIXED + AddVertexFormatInfo(&map, GL_FIXED, GL_FALSE, 1, VERTEX_CONVERT_CPU, DXGI_FORMAT_R32_FLOAT, &Copy32FixedTo32FVertexData<1>); + AddVertexFormatInfo(&map, GL_FIXED, GL_FALSE, 2, VERTEX_CONVERT_CPU, DXGI_FORMAT_R32G32_FLOAT, &Copy32FixedTo32FVertexData<2>); + AddVertexFormatInfo(&map, GL_FIXED, GL_FALSE, 3, VERTEX_CONVERT_CPU, DXGI_FORMAT_R32G32B32_FLOAT, &Copy32FixedTo32FVertexData<3>); + AddVertexFormatInfo(&map, GL_FIXED, GL_FALSE, 4, VERTEX_CONVERT_CPU, DXGI_FORMAT_R32G32B32A32_FLOAT, &Copy32FixedTo32FVertexData<4>); + + // GL_HALF_FLOAT + AddVertexFormatInfo(&map, GL_HALF_FLOAT, GL_FALSE, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16_FLOAT, &CopyNativeVertexData<GLhalf, 1, 0>); + AddVertexFormatInfo(&map, GL_HALF_FLOAT, GL_FALSE, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16G16_FLOAT, &CopyNativeVertexData<GLhalf, 2, 0>); + AddVertexFormatInfo(&map, GL_HALF_FLOAT, GL_FALSE, 3, VERTEX_CONVERT_CPU, DXGI_FORMAT_R16G16B16A16_FLOAT, &CopyNativeVertexData<GLhalf, 3, gl::Float16One>); + AddVertexFormatInfo(&map, GL_HALF_FLOAT, GL_FALSE, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16G16B16A16_FLOAT, &CopyNativeVertexData<GLhalf, 4, 0>); + + // GL_FLOAT + AddVertexFormatInfo(&map, GL_FLOAT, GL_FALSE, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32_FLOAT, &CopyNativeVertexData<GLfloat, 1, 0>); + AddVertexFormatInfo(&map, GL_FLOAT, GL_FALSE, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32G32_FLOAT, &CopyNativeVertexData<GLfloat, 2, 0>); + AddVertexFormatInfo(&map, GL_FLOAT, GL_FALSE, 3, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32G32B32_FLOAT, &CopyNativeVertexData<GLfloat, 3, 0>); + AddVertexFormatInfo(&map, GL_FLOAT, GL_FALSE, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32G32B32A32_FLOAT, &CopyNativeVertexData<GLfloat, 4, 0>); + + // GL_INT_2_10_10_10_REV + AddVertexFormatInfo(&map, GL_INT_2_10_10_10_REV, GL_FALSE, 4, VERTEX_CONVERT_CPU, DXGI_FORMAT_R32G32B32A32_FLOAT, &CopyXYZ10W2ToXYZW32FVertexData<true, false, true>); + AddVertexFormatInfo(&map, GL_INT_2_10_10_10_REV, GL_TRUE, 4, VERTEX_CONVERT_CPU, DXGI_FORMAT_R32G32B32A32_FLOAT, &CopyXYZ10W2ToXYZW32FVertexData<true, true, true>); + + // GL_UNSIGNED_INT_2_10_10_10_REV + AddVertexFormatInfo(&map, GL_UNSIGNED_INT_2_10_10_10_REV, GL_FALSE, 4, VERTEX_CONVERT_CPU, DXGI_FORMAT_R32G32B32A32_FLOAT, &CopyXYZ10W2ToXYZW32FVertexData<false, false, true>); + AddVertexFormatInfo(&map, GL_UNSIGNED_INT_2_10_10_10_REV, GL_TRUE, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R10G10B10A2_UNORM, &CopyNativeVertexData<GLuint, 1, 0>); + + // + // Integer Formats + // + + // GL_BYTE + AddIntegerVertexFormatInfo(&map, GL_BYTE, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R8_SINT, &CopyNativeVertexData<GLbyte, 1, 0>); + AddIntegerVertexFormatInfo(&map, GL_BYTE, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R8G8_SINT, &CopyNativeVertexData<GLbyte, 2, 0>); + AddIntegerVertexFormatInfo(&map, GL_BYTE, 3, VERTEX_CONVERT_CPU, DXGI_FORMAT_R8G8B8A8_SINT, &CopyNativeVertexData<GLbyte, 3, 1>); + AddIntegerVertexFormatInfo(&map, GL_BYTE, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R8G8B8A8_SINT, &CopyNativeVertexData<GLbyte, 4, 0>); + + // GL_UNSIGNED_BYTE + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_BYTE, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R8_UINT, &CopyNativeVertexData<GLubyte, 1, 0>); + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_BYTE, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R8G8_UINT, &CopyNativeVertexData<GLubyte, 2, 0>); + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_BYTE, 3, VERTEX_CONVERT_CPU, DXGI_FORMAT_R8G8B8A8_UINT, &CopyNativeVertexData<GLubyte, 3, 1>); + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_BYTE, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R8G8B8A8_UINT, &CopyNativeVertexData<GLubyte, 4, 0>); + + // GL_SHORT + AddIntegerVertexFormatInfo(&map, GL_SHORT, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16_SINT, &CopyNativeVertexData<GLshort, 1, 0>); + AddIntegerVertexFormatInfo(&map, GL_SHORT, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16G16_SINT, &CopyNativeVertexData<GLshort, 2, 0>); + AddIntegerVertexFormatInfo(&map, GL_SHORT, 3, VERTEX_CONVERT_CPU, DXGI_FORMAT_R16G16B16A16_SINT, &CopyNativeVertexData<GLshort, 3, 1>); + AddIntegerVertexFormatInfo(&map, GL_SHORT, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16G16B16A16_SINT, &CopyNativeVertexData<GLshort, 4, 0>); + + // GL_UNSIGNED_SHORT + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_SHORT, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16_UINT, &CopyNativeVertexData<GLushort, 1, 0>); + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_SHORT, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16G16_UINT, &CopyNativeVertexData<GLushort, 2, 0>); + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_SHORT, 3, VERTEX_CONVERT_CPU, DXGI_FORMAT_R16G16B16A16_UINT, &CopyNativeVertexData<GLushort, 3, 1>); + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_SHORT, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R16G16B16A16_UINT, &CopyNativeVertexData<GLushort, 4, 0>); + + // GL_INT + AddIntegerVertexFormatInfo(&map, GL_INT, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32_SINT, &CopyNativeVertexData<GLint, 1, 0>); + AddIntegerVertexFormatInfo(&map, GL_INT, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32G32_SINT, &CopyNativeVertexData<GLint, 2, 0>); + AddIntegerVertexFormatInfo(&map, GL_INT, 3, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32G32B32_SINT, &CopyNativeVertexData<GLint, 3, 0>); + AddIntegerVertexFormatInfo(&map, GL_INT, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32G32B32A32_SINT, &CopyNativeVertexData<GLint, 4, 0>); + + // GL_UNSIGNED_INT + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_INT, 1, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32_SINT, &CopyNativeVertexData<GLuint, 1, 0>); + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_INT, 2, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32G32_SINT, &CopyNativeVertexData<GLuint, 2, 0>); + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_INT, 3, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32G32B32_SINT, &CopyNativeVertexData<GLuint, 3, 0>); + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_INT, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R32G32B32A32_SINT, &CopyNativeVertexData<GLuint, 4, 0>); + + // GL_INT_2_10_10_10_REV + AddIntegerVertexFormatInfo(&map, GL_INT_2_10_10_10_REV, 4, VERTEX_CONVERT_CPU, DXGI_FORMAT_R16G16B16A16_SINT, &CopyXYZ10W2ToXYZW32FVertexData<true, true, false>); + + // GL_UNSIGNED_INT_2_10_10_10_REV + AddIntegerVertexFormatInfo(&map, GL_UNSIGNED_INT_2_10_10_10_REV, 4, VERTEX_CONVERT_NONE, DXGI_FORMAT_R10G10B10A2_UINT, &CopyNativeVertexData<GLuint, 1, 0>); + + return map; +} + +const VertexFormat &GetVertexFormatInfo(const gl::VertexFormat &vertexFormat) +{ + static const D3D11VertexFormatInfoMap vertexFormatMap = BuildD3D11VertexFormatInfoMap(); + + D3D11VertexFormatInfoMap::const_iterator iter = vertexFormatMap.find(vertexFormat); + if (iter != vertexFormatMap.end()) + { + return iter->second; + } + else + { + static const VertexFormat defaultInfo; + return defaultInfo; + } +} + +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/formatutils11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/formatutils11.h new file mode 100644 index 0000000000..ea11aaa74c --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/formatutils11.h @@ -0,0 +1,84 @@ +// +// Copyright (c) 2013-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// formatutils11.h: Queries for GL image formats and their translations to D3D11 +// formats. + +#ifndef LIBGLESV2_RENDERER_FORMATUTILS11_H_ +#define LIBGLESV2_RENDERER_FORMATUTILS11_H_ + +#include "libGLESv2/formatutils.h" + +#include <map> + +namespace rx +{ + +namespace d3d11 +{ + +typedef std::map<std::pair<GLenum, GLenum>, ColorCopyFunction> FastCopyFunctionMap; + +struct DXGIFormat +{ + DXGIFormat(); + + GLuint pixelBytes; + GLuint blockWidth; + GLuint blockHeight; + + GLuint depthBits; + GLuint depthOffset; + GLuint stencilBits; + GLuint stencilOffset; + + GLenum internalFormat; + GLenum componentType; + + MipGenerationFunction mipGenerationFunction; + ColorReadFunction colorReadFunction; + + FastCopyFunctionMap fastCopyFunctions; + ColorCopyFunction getFastCopyFunction(GLenum format, GLenum type) const; +}; +const DXGIFormat &GetDXGIFormatInfo(DXGI_FORMAT format); + +struct TextureFormat +{ + TextureFormat(); + + DXGI_FORMAT texFormat; + DXGI_FORMAT srvFormat; + DXGI_FORMAT rtvFormat; + DXGI_FORMAT dsvFormat; + DXGI_FORMAT renderFormat; + + DXGI_FORMAT swizzleTexFormat; + DXGI_FORMAT swizzleSRVFormat; + DXGI_FORMAT swizzleRTVFormat; + + InitializeTextureDataFunction dataInitializerFunction; + + typedef std::map<GLenum, LoadImageFunction> LoadFunctionMap; + LoadFunctionMap loadFunctions; +}; +const TextureFormat &GetTextureFormatInfo(GLenum internalFormat); + +struct VertexFormat +{ + VertexFormat(); + + VertexConversionType conversionType; + DXGI_FORMAT nativeFormat; + VertexCopyFunction copyFunction; +}; +const VertexFormat &GetVertexFormatInfo(const gl::VertexFormat &vertexFormat); + +} + +} + +#endif // LIBGLESV2_RENDERER_FORMATUTILS11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/renderer11_utils.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/renderer11_utils.cpp new file mode 100644 index 0000000000..2af97e73f0 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/renderer11_utils.cpp @@ -0,0 +1,1216 @@ +// +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// renderer11_utils.cpp: Conversion functions and other utility routines +// specific to the D3D11 renderer. + +#include "libGLESv2/renderer/d3d/d3d11/renderer11_utils.h" +#include "libGLESv2/renderer/d3d/d3d11/formatutils11.h" +#include "libGLESv2/renderer/d3d/d3d11/RenderTarget11.h" +#include "libGLESv2/ProgramBinary.h" +#include "libGLESv2/Framebuffer.h" + +#include "common/debug.h" + +#include <algorithm> + +#ifndef D3D_FL9_1_DEFAULT_MAX_ANISOTROPY +# define D3D_FL9_1_DEFAULT_MAX_ANISOTROPY 2 +#endif +#ifndef D3D_FL9_1_SIMULTANEOUS_RENDER_TARGET_COUNT +# define D3D_FL9_1_SIMULTANEOUS_RENDER_TARGET_COUNT 1 +#endif +#ifndef D3D_FL9_3_SIMULTANEOUS_RENDER_TARGET_COUNT +# define D3D_FL9_3_SIMULTANEOUS_RENDER_TARGET_COUNT 4 +#endif +#ifndef D3D_FL9_1_IA_PRIMITIVE_MAX_COUNT +# define D3D_FL9_1_IA_PRIMITIVE_MAX_COUNT 65535 +#endif +#ifndef D3D_FL9_2_IA_PRIMITIVE_MAX_COUNT +# define D3D_FL9_2_IA_PRIMITIVE_MAX_COUNT 1048575 +#endif +#ifndef D3D_FL9_1_REQ_TEXTURECUBE_DIMENSION +# define D3D_FL9_1_REQ_TEXTURECUBE_DIMENSION 512 +#endif +#ifndef D3D_FL9_3_REQ_TEXTURECUBE_DIMENSION +# define D3D_FL9_3_REQ_TEXTURECUBE_DIMENSION 4096 +#endif +#ifndef D3D_FL9_1_REQ_TEXTURE2D_U_OR_V_DIMENSION +# define D3D_FL9_1_REQ_TEXTURE2D_U_OR_V_DIMENSION 2048 +#endif +#ifndef D3D_FL9_1_REQ_TEXTURE3D_U_V_OR_W_DIMENSION +# define D3D_FL9_1_REQ_TEXTURE3D_U_V_OR_W_DIMENSION 256 +#endif +#ifndef D3D_FL9_3_REQ_TEXTURE2D_U_OR_V_DIMENSION +# define D3D_FL9_3_REQ_TEXTURE2D_U_OR_V_DIMENSION 4096 +#endif +#ifndef D3D11_REQ_TEXTURECUBE_DIMENSION +# define D3D11_REQ_TEXTURECUBE_DIMENSION 16384 +#endif +#ifndef D3D11_REQ_TEXTURE2D_ARRAY_AXIS_DIMENSION +# define D3D11_REQ_TEXTURE2D_ARRAY_AXIS_DIMENSION 2048 +#endif +#ifndef D3D11_REQ_TEXTURE3D_U_V_OR_W_DIMENSION +# define D3D11_REQ_TEXTURE3D_U_V_OR_W_DIMENSION 2048 +#endif +#ifndef D3D11_REQ_DRAWINDEXED_INDEX_COUNT_2_TO_EXP +# define D3D11_REQ_DRAWINDEXED_INDEX_COUNT_2_TO_EXP 32 +#endif +#ifndef D3D11_REQ_DRAW_VERTEX_COUNT_2_TO_EXP +# define D3D11_REQ_DRAW_VERTEX_COUNT_2_TO_EXP 32 +#endif +#ifndef D3D10_1_STANDARD_VERTEX_ELEMENT_COUNT +# define D3D10_1_STANDARD_VERTEX_ELEMENT_COUNT 32 +#endif +#ifndef D3D11_STANDARD_VERTEX_ELEMENT_COUNT +# define D3D11_STANDARD_VERTEX_ELEMENT_COUNT 32 +#endif +#ifndef D3D10_1_SO_BUFFER_SLOT_COUNT +# define D3D10_1_SO_BUFFER_SLOT_COUNT 4 +#endif +#ifndef D3D11_SO_BUFFER_SLOT_COUNT +# define D3D11_SO_BUFFER_SLOT_COUNT 4 +#endif +#ifndef D3D11_COMMONSHADER_CONSTANT_BUFFER_API_SLOT_COUNT +# define D3D11_COMMONSHADER_CONSTANT_BUFFER_API_SLOT_COUNT 14 +#endif +#ifndef D3D11_COMMONSHADER_SAMPLER_SLOT_COUNT +# define D3D11_COMMONSHADER_SAMPLER_SLOT_COUNT 16 +#endif +#ifndef D3D11_COMMONSHADER_TEXEL_OFFSET_MAX_NEGATIVE +# define D3D11_COMMONSHADER_TEXEL_OFFSET_MAX_NEGATIVE -8 +#endif +#ifndef D3D11_COMMONSHADER_TEXEL_OFFSET_MAX_POSITIVE +# define D3D11_COMMONSHADER_TEXEL_OFFSET_MAX_POSITIVE 7 +#endif +#ifndef D3D11_REQ_CONSTANT_BUFFER_ELEMENT_COUNT +# define D3D11_REQ_CONSTANT_BUFFER_ELEMENT_COUNT 4096 +#endif +#ifndef D3D11_PS_INPUT_REGISTER_COUNT +# define D3D11_PS_INPUT_REGISTER_COUNT 32 +#endif +#ifndef D3D10_1_VS_OUTPUT_REGISTER_COUNT +# define D3D10_1_VS_OUTPUT_REGISTER_COUNT 32 +#endif +#ifndef D3D11_VS_OUTPUT_REGISTER_COUNT +# define D3D11_VS_OUTPUT_REGISTER_COUNT 32 +#endif + +namespace rx +{ + +namespace gl_d3d11 +{ + +D3D11_BLEND ConvertBlendFunc(GLenum glBlend, bool isAlpha) +{ + D3D11_BLEND d3dBlend = D3D11_BLEND_ZERO; + + switch (glBlend) + { + case GL_ZERO: d3dBlend = D3D11_BLEND_ZERO; break; + case GL_ONE: d3dBlend = D3D11_BLEND_ONE; break; + case GL_SRC_COLOR: d3dBlend = (isAlpha ? D3D11_BLEND_SRC_ALPHA : D3D11_BLEND_SRC_COLOR); break; + case GL_ONE_MINUS_SRC_COLOR: d3dBlend = (isAlpha ? D3D11_BLEND_INV_SRC_ALPHA : D3D11_BLEND_INV_SRC_COLOR); break; + case GL_DST_COLOR: d3dBlend = (isAlpha ? D3D11_BLEND_DEST_ALPHA : D3D11_BLEND_DEST_COLOR); break; + case GL_ONE_MINUS_DST_COLOR: d3dBlend = (isAlpha ? D3D11_BLEND_INV_DEST_ALPHA : D3D11_BLEND_INV_DEST_COLOR); break; + case GL_SRC_ALPHA: d3dBlend = D3D11_BLEND_SRC_ALPHA; break; + case GL_ONE_MINUS_SRC_ALPHA: d3dBlend = D3D11_BLEND_INV_SRC_ALPHA; break; + case GL_DST_ALPHA: d3dBlend = D3D11_BLEND_DEST_ALPHA; break; + case GL_ONE_MINUS_DST_ALPHA: d3dBlend = D3D11_BLEND_INV_DEST_ALPHA; break; + case GL_CONSTANT_COLOR: d3dBlend = D3D11_BLEND_BLEND_FACTOR; break; + case GL_ONE_MINUS_CONSTANT_COLOR: d3dBlend = D3D11_BLEND_INV_BLEND_FACTOR; break; + case GL_CONSTANT_ALPHA: d3dBlend = D3D11_BLEND_BLEND_FACTOR; break; + case GL_ONE_MINUS_CONSTANT_ALPHA: d3dBlend = D3D11_BLEND_INV_BLEND_FACTOR; break; + case GL_SRC_ALPHA_SATURATE: d3dBlend = D3D11_BLEND_SRC_ALPHA_SAT; break; + default: UNREACHABLE(); + } + + return d3dBlend; +} + +D3D11_BLEND_OP ConvertBlendOp(GLenum glBlendOp) +{ + D3D11_BLEND_OP d3dBlendOp = D3D11_BLEND_OP_ADD; + + switch (glBlendOp) + { + case GL_FUNC_ADD: d3dBlendOp = D3D11_BLEND_OP_ADD; break; + case GL_FUNC_SUBTRACT: d3dBlendOp = D3D11_BLEND_OP_SUBTRACT; break; + case GL_FUNC_REVERSE_SUBTRACT: d3dBlendOp = D3D11_BLEND_OP_REV_SUBTRACT; break; + case GL_MIN: d3dBlendOp = D3D11_BLEND_OP_MIN; break; + case GL_MAX: d3dBlendOp = D3D11_BLEND_OP_MAX; break; + default: UNREACHABLE(); + } + + return d3dBlendOp; +} + +UINT8 ConvertColorMask(bool red, bool green, bool blue, bool alpha) +{ + UINT8 mask = 0; + if (red) + { + mask |= D3D11_COLOR_WRITE_ENABLE_RED; + } + if (green) + { + mask |= D3D11_COLOR_WRITE_ENABLE_GREEN; + } + if (blue) + { + mask |= D3D11_COLOR_WRITE_ENABLE_BLUE; + } + if (alpha) + { + mask |= D3D11_COLOR_WRITE_ENABLE_ALPHA; + } + return mask; +} + +D3D11_CULL_MODE ConvertCullMode(bool cullEnabled, GLenum cullMode) +{ + D3D11_CULL_MODE cull = D3D11_CULL_NONE; + + if (cullEnabled) + { + switch (cullMode) + { + case GL_FRONT: cull = D3D11_CULL_FRONT; break; + case GL_BACK: cull = D3D11_CULL_BACK; break; + case GL_FRONT_AND_BACK: cull = D3D11_CULL_NONE; break; + default: UNREACHABLE(); + } + } + else + { + cull = D3D11_CULL_NONE; + } + + return cull; +} + +D3D11_COMPARISON_FUNC ConvertComparison(GLenum comparison) +{ + D3D11_COMPARISON_FUNC d3dComp = D3D11_COMPARISON_NEVER; + switch (comparison) + { + case GL_NEVER: d3dComp = D3D11_COMPARISON_NEVER; break; + case GL_ALWAYS: d3dComp = D3D11_COMPARISON_ALWAYS; break; + case GL_LESS: d3dComp = D3D11_COMPARISON_LESS; break; + case GL_LEQUAL: d3dComp = D3D11_COMPARISON_LESS_EQUAL; break; + case GL_EQUAL: d3dComp = D3D11_COMPARISON_EQUAL; break; + case GL_GREATER: d3dComp = D3D11_COMPARISON_GREATER; break; + case GL_GEQUAL: d3dComp = D3D11_COMPARISON_GREATER_EQUAL; break; + case GL_NOTEQUAL: d3dComp = D3D11_COMPARISON_NOT_EQUAL; break; + default: UNREACHABLE(); + } + + return d3dComp; +} + +D3D11_DEPTH_WRITE_MASK ConvertDepthMask(bool depthWriteEnabled) +{ + return depthWriteEnabled ? D3D11_DEPTH_WRITE_MASK_ALL : D3D11_DEPTH_WRITE_MASK_ZERO; +} + +UINT8 ConvertStencilMask(GLuint stencilmask) +{ + return static_cast<UINT8>(stencilmask); +} + +D3D11_STENCIL_OP ConvertStencilOp(GLenum stencilOp) +{ + D3D11_STENCIL_OP d3dStencilOp = D3D11_STENCIL_OP_KEEP; + + switch (stencilOp) + { + case GL_ZERO: d3dStencilOp = D3D11_STENCIL_OP_ZERO; break; + case GL_KEEP: d3dStencilOp = D3D11_STENCIL_OP_KEEP; break; + case GL_REPLACE: d3dStencilOp = D3D11_STENCIL_OP_REPLACE; break; + case GL_INCR: d3dStencilOp = D3D11_STENCIL_OP_INCR_SAT; break; + case GL_DECR: d3dStencilOp = D3D11_STENCIL_OP_DECR_SAT; break; + case GL_INVERT: d3dStencilOp = D3D11_STENCIL_OP_INVERT; break; + case GL_INCR_WRAP: d3dStencilOp = D3D11_STENCIL_OP_INCR; break; + case GL_DECR_WRAP: d3dStencilOp = D3D11_STENCIL_OP_DECR; break; + default: UNREACHABLE(); + } + + return d3dStencilOp; +} + +D3D11_FILTER ConvertFilter(GLenum minFilter, GLenum magFilter, float maxAnisotropy, GLenum comparisonMode) +{ + bool comparison = comparisonMode != GL_NONE; + + if (maxAnisotropy > 1.0f) + { + return D3D11_ENCODE_ANISOTROPIC_FILTER(static_cast<D3D11_COMPARISON_FUNC>(comparison)); + } + else + { + D3D11_FILTER_TYPE dxMin = D3D11_FILTER_TYPE_POINT; + D3D11_FILTER_TYPE dxMip = D3D11_FILTER_TYPE_POINT; + switch (minFilter) + { + case GL_NEAREST: dxMin = D3D11_FILTER_TYPE_POINT; dxMip = D3D11_FILTER_TYPE_POINT; break; + case GL_LINEAR: dxMin = D3D11_FILTER_TYPE_LINEAR; dxMip = D3D11_FILTER_TYPE_POINT; break; + case GL_NEAREST_MIPMAP_NEAREST: dxMin = D3D11_FILTER_TYPE_POINT; dxMip = D3D11_FILTER_TYPE_POINT; break; + case GL_LINEAR_MIPMAP_NEAREST: dxMin = D3D11_FILTER_TYPE_LINEAR; dxMip = D3D11_FILTER_TYPE_POINT; break; + case GL_NEAREST_MIPMAP_LINEAR: dxMin = D3D11_FILTER_TYPE_POINT; dxMip = D3D11_FILTER_TYPE_LINEAR; break; + case GL_LINEAR_MIPMAP_LINEAR: dxMin = D3D11_FILTER_TYPE_LINEAR; dxMip = D3D11_FILTER_TYPE_LINEAR; break; + default: UNREACHABLE(); + } + + D3D11_FILTER_TYPE dxMag = D3D11_FILTER_TYPE_POINT; + switch (magFilter) + { + case GL_NEAREST: dxMag = D3D11_FILTER_TYPE_POINT; break; + case GL_LINEAR: dxMag = D3D11_FILTER_TYPE_LINEAR; break; + default: UNREACHABLE(); + } + + return D3D11_ENCODE_BASIC_FILTER(dxMin, dxMag, dxMip, static_cast<D3D11_COMPARISON_FUNC>(comparison)); + } +} + +D3D11_TEXTURE_ADDRESS_MODE ConvertTextureWrap(GLenum wrap) +{ + switch (wrap) + { + case GL_REPEAT: return D3D11_TEXTURE_ADDRESS_WRAP; + case GL_CLAMP_TO_EDGE: return D3D11_TEXTURE_ADDRESS_CLAMP; + case GL_MIRRORED_REPEAT: return D3D11_TEXTURE_ADDRESS_MIRROR; + default: UNREACHABLE(); + } + + return D3D11_TEXTURE_ADDRESS_WRAP; +} + +D3D11_QUERY ConvertQueryType(GLenum queryType) +{ + switch (queryType) + { + case GL_ANY_SAMPLES_PASSED_EXT: + case GL_ANY_SAMPLES_PASSED_CONSERVATIVE_EXT: return D3D11_QUERY_OCCLUSION; + case GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN: return D3D11_QUERY_SO_STATISTICS; + default: UNREACHABLE(); return D3D11_QUERY_EVENT; + } +} + +} + + +namespace d3d11_gl +{ + +static gl::TextureCaps GenerateTextureFormatCaps(GLenum internalFormat, ID3D11Device *device) +{ + gl::TextureCaps textureCaps; + + const d3d11::TextureFormat &formatInfo = d3d11::GetTextureFormatInfo(internalFormat); + + UINT formatSupport; + if (SUCCEEDED(device->CheckFormatSupport(formatInfo.texFormat, &formatSupport))) + { + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(internalFormat); + if (formatInfo.depthBits > 0 || formatInfo.stencilBits > 0) + { + textureCaps.texturable = ((formatSupport & D3D11_FORMAT_SUPPORT_TEXTURE2D) != 0); + } + else + { + textureCaps.texturable = ((formatSupport & D3D11_FORMAT_SUPPORT_TEXTURE2D) != 0) && + ((formatSupport & D3D11_FORMAT_SUPPORT_TEXTURECUBE) != 0) && + ((formatSupport & D3D11_FORMAT_SUPPORT_TEXTURE3D) != 0); + } + } + + if (SUCCEEDED(device->CheckFormatSupport(formatInfo.renderFormat, &formatSupport)) && + ((formatSupport & D3D11_FORMAT_SUPPORT_MULTISAMPLE_RENDERTARGET) != 0)) + { + for (size_t sampleCount = 1; sampleCount <= D3D11_MAX_MULTISAMPLE_SAMPLE_COUNT; sampleCount++) + { + UINT qualityCount = 0; + if (SUCCEEDED(device->CheckMultisampleQualityLevels(formatInfo.renderFormat, sampleCount, &qualityCount)) && + qualityCount > 0) + { + textureCaps.sampleCounts.insert(sampleCount); + } + } + } + + textureCaps.filterable = SUCCEEDED(device->CheckFormatSupport(formatInfo.srvFormat, &formatSupport)) && + ((formatSupport & D3D11_FORMAT_SUPPORT_SHADER_SAMPLE)) != 0; + textureCaps.renderable = (SUCCEEDED(device->CheckFormatSupport(formatInfo.rtvFormat, &formatSupport)) && + ((formatSupport & D3D11_FORMAT_SUPPORT_RENDER_TARGET)) != 0) || + (SUCCEEDED(device->CheckFormatSupport(formatInfo.dsvFormat, &formatSupport)) && + ((formatSupport & D3D11_FORMAT_SUPPORT_DEPTH_STENCIL) != 0)); + + return textureCaps; +} + +static bool GetNPOTTextureSupport(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return true; + + // From http://msdn.microsoft.com/en-us/library/windows/desktop/ff476876.aspx + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return true; // Provided that mipmaps & wrap modes are not used + + default: UNREACHABLE(); return false; + } +} + +static float GetMaximumAnisotropy(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_MAX_MAXANISOTROPY; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_MAX_MAXANISOTROPY; + + // From http://msdn.microsoft.com/en-us/library/windows/desktop/ff476876.aspx + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: return 16; + + case D3D_FEATURE_LEVEL_9_1: return D3D_FL9_1_DEFAULT_MAX_ANISOTROPY; + + default: UNREACHABLE(); return 0; + } +} + +static bool GetOcclusionQuerySupport(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return true; + + // From http://msdn.microsoft.com/en-us/library/windows/desktop/ff476150.aspx ID3D11Device::CreateQuery + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: return true; + case D3D_FEATURE_LEVEL_9_1: return false; + + default: UNREACHABLE(); return false; + } +} + +static bool GetEventQuerySupport(D3D_FEATURE_LEVEL featureLevel) +{ + // From http://msdn.microsoft.com/en-us/library/windows/desktop/ff476150.aspx ID3D11Device::CreateQuery + + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return true; + + default: UNREACHABLE(); return false; + } +} + +static bool GetInstancingSupport(D3D_FEATURE_LEVEL featureLevel) +{ + // From http://msdn.microsoft.com/en-us/library/windows/desktop/ff476150.aspx ID3D11Device::CreateInputLayout + + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: + case D3D_FEATURE_LEVEL_9_3: return true; + + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return false; + + default: UNREACHABLE(); return false; + } +} + +static bool GetDerivativeInstructionSupport(D3D_FEATURE_LEVEL featureLevel) +{ + // http://msdn.microsoft.com/en-us/library/windows/desktop/bb509588.aspx states that shader model + // ps_2_x is required for the ddx (and other derivative functions). + + // http://msdn.microsoft.com/en-us/library/windows/desktop/ff476876.aspx states that feature level + // 9.3 supports shader model ps_2_x. + + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: + case D3D_FEATURE_LEVEL_9_3: return true; + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return false; + + default: UNREACHABLE(); return false; + } +} + +static size_t GetMaximumSimultaneousRenderTargets(D3D_FEATURE_LEVEL featureLevel) +{ + // From http://msdn.microsoft.com/en-us/library/windows/desktop/ff476150.aspx ID3D11Device::CreateInputLayout + + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_SIMULTANEOUS_RENDER_TARGET_COUNT; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_SIMULTANEOUS_RENDER_TARGET_COUNT; + + case D3D_FEATURE_LEVEL_9_3: return D3D_FL9_3_SIMULTANEOUS_RENDER_TARGET_COUNT; + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return D3D_FL9_1_SIMULTANEOUS_RENDER_TARGET_COUNT; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximum2DTextureSize(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_REQ_TEXTURE2D_U_OR_V_DIMENSION; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_REQ_TEXTURE2D_U_OR_V_DIMENSION; + + case D3D_FEATURE_LEVEL_9_3: return D3D_FL9_3_REQ_TEXTURE2D_U_OR_V_DIMENSION; + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return D3D_FL9_1_REQ_TEXTURE2D_U_OR_V_DIMENSION; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumCubeMapTextureSize(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_REQ_TEXTURECUBE_DIMENSION; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_REQ_TEXTURECUBE_DIMENSION; + + case D3D_FEATURE_LEVEL_9_3: return D3D_FL9_3_REQ_TEXTURECUBE_DIMENSION; + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return D3D_FL9_1_REQ_TEXTURECUBE_DIMENSION; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximum2DTextureArraySize(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_REQ_TEXTURE2D_ARRAY_AXIS_DIMENSION; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_REQ_TEXTURE2D_ARRAY_AXIS_DIMENSION; + + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 0; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximum3DTextureSize(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_REQ_TEXTURE3D_U_V_OR_W_DIMENSION; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_REQ_TEXTURE3D_U_V_OR_W_DIMENSION; + + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return D3D_FL9_1_REQ_TEXTURE3D_U_V_OR_W_DIMENSION; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumViewportSize(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_VIEWPORT_BOUNDS_MAX; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_VIEWPORT_BOUNDS_MAX; + + // No constants for D3D9 viewport size limits, use the maximum texture sizes + case D3D_FEATURE_LEVEL_9_3: return D3D_FL9_3_REQ_TEXTURE2D_U_OR_V_DIMENSION; + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return D3D_FL9_1_REQ_TEXTURE2D_U_OR_V_DIMENSION; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumDrawIndexedIndexCount(D3D_FEATURE_LEVEL featureLevel) +{ + // D3D11 allows up to 2^32 elements, but we report max signed int for convenience since that's what's + // returned from glGetInteger + META_ASSERT(D3D11_REQ_DRAWINDEXED_INDEX_COUNT_2_TO_EXP == 32); + META_ASSERT(D3D10_REQ_DRAWINDEXED_INDEX_COUNT_2_TO_EXP == 32); + + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return std::numeric_limits<GLint>::max(); + + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: return D3D_FL9_2_IA_PRIMITIVE_MAX_COUNT; + case D3D_FEATURE_LEVEL_9_1: return D3D_FL9_1_IA_PRIMITIVE_MAX_COUNT; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumDrawVertexCount(D3D_FEATURE_LEVEL featureLevel) +{ + // D3D11 allows up to 2^32 elements, but we report max signed int for convenience since that's what's + // returned from glGetInteger + META_ASSERT(D3D11_REQ_DRAW_VERTEX_COUNT_2_TO_EXP == 32); + META_ASSERT(D3D10_REQ_DRAW_VERTEX_COUNT_2_TO_EXP == 32); + + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return std::numeric_limits<GLint>::max(); + + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: return D3D_FL9_2_IA_PRIMITIVE_MAX_COUNT; + case D3D_FEATURE_LEVEL_9_1: return D3D_FL9_1_IA_PRIMITIVE_MAX_COUNT; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumVertexInputSlots(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_STANDARD_VERTEX_ELEMENT_COUNT; + + case D3D_FEATURE_LEVEL_10_1: return D3D10_1_STANDARD_VERTEX_ELEMENT_COUNT; + case D3D_FEATURE_LEVEL_10_0: return D3D10_STANDARD_VERTEX_ELEMENT_COUNT; + + // From http://http://msdn.microsoft.com/en-us/library/windows/desktop/ff476876.aspx "Max Input Slots" + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 16; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumVertexUniformVectors(D3D_FEATURE_LEVEL featureLevel) +{ + // TODO(geofflang): Remove hard-coded limit once the gl-uniform-arrays test can pass + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return 1024; // D3D11_REQ_CONSTANT_BUFFER_ELEMENT_COUNT; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return 1024; // D3D10_REQ_CONSTANT_BUFFER_ELEMENT_COUNT; + + // From http://msdn.microsoft.com/en-us/library/windows/desktop/ff476149.aspx ID3D11DeviceContext::VSSetConstantBuffers + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 255; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetReservedVertexUniformBuffers() +{ + // Reserve one buffer for the application uniforms, and one for driver uniforms + return 2; +} + +static size_t GetMaximumVertexUniformBlocks(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_COMMONSHADER_CONSTANT_BUFFER_API_SLOT_COUNT - GetReservedVertexUniformBuffers(); + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_COMMONSHADER_CONSTANT_BUFFER_API_SLOT_COUNT - GetReservedVertexUniformBuffers(); + + // Uniform blocks not supported in D3D9 feature levels + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 0; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetReservedVertexOutputVectors() +{ + // We potentially reserve varyings for gl_Position, dx_Position, gl_FragCoord and gl_PointSize + return 4; +} + +static size_t GetMaximumVertexOutputVectors(D3D_FEATURE_LEVEL featureLevel) +{ + META_ASSERT(gl::IMPLEMENTATION_MAX_VARYING_VECTORS == D3D11_VS_OUTPUT_REGISTER_COUNT); + + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_VS_OUTPUT_REGISTER_COUNT - GetReservedVertexOutputVectors(); + + case D3D_FEATURE_LEVEL_10_1: return D3D10_1_VS_OUTPUT_REGISTER_COUNT - GetReservedVertexOutputVectors(); + case D3D_FEATURE_LEVEL_10_0: return D3D10_VS_OUTPUT_REGISTER_COUNT - GetReservedVertexOutputVectors(); + + // Use D3D9 SM3 and SM2 limits + case D3D_FEATURE_LEVEL_9_3: return 10 - GetReservedVertexOutputVectors(); + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 8 - GetReservedVertexOutputVectors(); + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumVertexTextureUnits(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_COMMONSHADER_SAMPLER_SLOT_COUNT; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_COMMONSHADER_SAMPLER_SLOT_COUNT; + + // Vertex textures not supported in D3D9 feature levels according to + // http://msdn.microsoft.com/en-us/library/windows/desktop/ff476149.aspx + // ID3D11DeviceContext::VSSetSamplers and ID3D11DeviceContext::VSSetShaderResources + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 0; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumPixelUniformVectors(D3D_FEATURE_LEVEL featureLevel) +{ + // TODO(geofflang): Remove hard-coded limit once the gl-uniform-arrays test can pass + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return 1024; // D3D11_REQ_CONSTANT_BUFFER_ELEMENT_COUNT; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return 1024; // D3D10_REQ_CONSTANT_BUFFER_ELEMENT_COUNT; + + // From http://msdn.microsoft.com/en-us/library/windows/desktop/ff476149.aspx ID3D11DeviceContext::PSSetConstantBuffers + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 32; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetReservedPixelUniformBuffers() +{ + // Reserve one buffer for the application uniforms, and one for driver uniforms + return 2; +} + +static size_t GetMaximumPixelUniformBlocks(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_COMMONSHADER_CONSTANT_BUFFER_API_SLOT_COUNT - GetReservedPixelUniformBuffers(); + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_COMMONSHADER_CONSTANT_BUFFER_API_SLOT_COUNT - GetReservedPixelUniformBuffers(); + + // Uniform blocks not supported in D3D9 feature levels + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 0; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumPixelInputVectors(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_PS_INPUT_REGISTER_COUNT - GetReservedVertexOutputVectors(); + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_PS_INPUT_REGISTER_COUNT - GetReservedVertexOutputVectors(); + + // Use D3D9 SM3 and SM2 limits + case D3D_FEATURE_LEVEL_9_3: return 10 - GetReservedVertexOutputVectors(); + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 8 - GetReservedVertexOutputVectors(); + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumPixelTextureUnits(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_COMMONSHADER_SAMPLER_SLOT_COUNT; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_COMMONSHADER_SAMPLER_SLOT_COUNT; + + // http://msdn.microsoft.com/en-us/library/windows/desktop/ff476149.aspx ID3D11DeviceContext::PSSetShaderResources + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 16; + + default: UNREACHABLE(); return 0; + } +} + +static int GetMinimumTexelOffset(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_COMMONSHADER_TEXEL_OFFSET_MAX_NEGATIVE; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_COMMONSHADER_TEXEL_OFFSET_MAX_NEGATIVE; + + // Sampling functions with offsets are not available below shader model 4.0. + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 0; + + default: UNREACHABLE(); return 0; + } +} + +static int GetMaximumTexelOffset(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_COMMONSHADER_TEXEL_OFFSET_MAX_POSITIVE; + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D11_COMMONSHADER_TEXEL_OFFSET_MAX_POSITIVE; + + // Sampling functions with offsets are not available below shader model 4.0. + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 0; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumConstantBufferSize(D3D_FEATURE_LEVEL featureLevel) +{ + // Returns a size_t despite the limit being a GLuint64 because size_t is the maximum size of + // any buffer that could be allocated. + + const size_t bytesPerComponent = 4 * sizeof(float); + + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_REQ_CONSTANT_BUFFER_ELEMENT_COUNT * bytesPerComponent; + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return D3D10_REQ_CONSTANT_BUFFER_ELEMENT_COUNT * bytesPerComponent; + + // Limits from http://msdn.microsoft.com/en-us/library/windows/desktop/ff476501.aspx remarks section + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 4096 * bytesPerComponent; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumStreamOutputBuffers(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return D3D11_SO_BUFFER_SLOT_COUNT; + + case D3D_FEATURE_LEVEL_10_1: return D3D10_1_SO_BUFFER_SLOT_COUNT; + case D3D_FEATURE_LEVEL_10_0: return D3D10_SO_BUFFER_SLOT_COUNT; + + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 0; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumStreamOutputInterleavedComponenets(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: + + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return GetMaximumVertexOutputVectors(featureLevel) * 4; + + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 0; + + default: UNREACHABLE(); return 0; + } +} + +static size_t GetMaximumStreamOutputSeparateCompeonents(D3D_FEATURE_LEVEL featureLevel) +{ + switch (featureLevel) + { +#if _MSC_VER >= 1700 + case D3D_FEATURE_LEVEL_11_1: +#endif + case D3D_FEATURE_LEVEL_11_0: return GetMaximumStreamOutputInterleavedComponenets(featureLevel) / + GetMaximumStreamOutputBuffers(featureLevel); + + + // D3D 10 and 10.1 only allow one output per output slot if an output slot other than zero is used. + case D3D_FEATURE_LEVEL_10_1: + case D3D_FEATURE_LEVEL_10_0: return 4; + + case D3D_FEATURE_LEVEL_9_3: + case D3D_FEATURE_LEVEL_9_2: + case D3D_FEATURE_LEVEL_9_1: return 0; + + default: UNREACHABLE(); return 0; + } +} + +void GenerateCaps(ID3D11Device *device, gl::Caps *caps, gl::TextureCapsMap *textureCapsMap, gl::Extensions *extensions) +{ + GLuint maxSamples = 0; + const gl::FormatSet &allFormats = gl::GetAllSizedInternalFormats(); + for (gl::FormatSet::const_iterator internalFormat = allFormats.begin(); internalFormat != allFormats.end(); ++internalFormat) + { + gl::TextureCaps textureCaps = GenerateTextureFormatCaps(*internalFormat, device); + textureCapsMap->insert(*internalFormat, textureCaps); + + maxSamples = std::max(maxSamples, textureCaps.getMaxSamples()); + + if (gl::GetInternalFormatInfo(*internalFormat).compressed) + { + caps->compressedTextureFormats.push_back(*internalFormat); + } + } + + D3D_FEATURE_LEVEL featureLevel = device->GetFeatureLevel(); + + // GL core feature limits + caps->maxElementIndex = static_cast<GLint64>(std::numeric_limits<unsigned int>::max()); + caps->max3DTextureSize = GetMaximum3DTextureSize(featureLevel); + caps->max2DTextureSize = GetMaximum2DTextureSize(featureLevel); + caps->maxCubeMapTextureSize = GetMaximumCubeMapTextureSize(featureLevel); + caps->maxArrayTextureLayers = GetMaximum2DTextureArraySize(featureLevel); + + // Unimplemented, set to minimum required + caps->maxLODBias = 2.0f; + + // No specific limits on render target size, maximum 2D texture size is equivalent + caps->maxRenderbufferSize = caps->max2DTextureSize; + + // Maximum draw buffers and color attachments are the same, max color attachments could eventually be + // increased to 16 + caps->maxDrawBuffers = GetMaximumSimultaneousRenderTargets(featureLevel); + caps->maxColorAttachments = GetMaximumSimultaneousRenderTargets(featureLevel); + + // D3D11 has the same limit for viewport width and height + caps->maxViewportWidth = GetMaximumViewportSize(featureLevel); + caps->maxViewportHeight = caps->maxViewportWidth; + + // Choose a reasonable maximum, enforced in the shader. + caps->minAliasedPointSize = 1.0f; + caps->maxAliasedPointSize = 1024.0f; + + // Wide lines not supported + caps->minAliasedLineWidth = 1.0f; + caps->maxAliasedLineWidth = 1.0f; + + // Primitive count limits + caps->maxElementsIndices = GetMaximumDrawIndexedIndexCount(featureLevel); + caps->maxElementsVertices = GetMaximumDrawVertexCount(featureLevel); + + // Program and shader binary formats (no supported shader binary formats) + caps->programBinaryFormats.push_back(GL_PROGRAM_BINARY_ANGLE); + + // We do not wait for server fence objects internally, so report a max timeout of zero. + caps->maxServerWaitTimeout = 0; + + // Vertex shader limits + caps->maxVertexAttributes = GetMaximumVertexInputSlots(featureLevel); + caps->maxVertexUniformComponents = GetMaximumVertexUniformVectors(featureLevel) * 4; + caps->maxVertexUniformVectors = GetMaximumVertexUniformVectors(featureLevel); + caps->maxVertexUniformBlocks = GetMaximumVertexUniformBlocks(featureLevel); + caps->maxVertexOutputComponents = GetMaximumVertexOutputVectors(featureLevel) * 4; + caps->maxVertexTextureImageUnits = GetMaximumVertexTextureUnits(featureLevel); + + // Fragment shader limits + caps->maxFragmentUniformComponents = GetMaximumPixelUniformVectors(featureLevel) * 4; + caps->maxFragmentUniformVectors = GetMaximumPixelUniformVectors(featureLevel); + caps->maxFragmentUniformBlocks = GetMaximumPixelUniformBlocks(featureLevel); + caps->maxFragmentInputComponents = GetMaximumPixelInputVectors(featureLevel) * 4; + caps->maxTextureImageUnits = GetMaximumPixelTextureUnits(featureLevel); + caps->minProgramTexelOffset = GetMinimumTexelOffset(featureLevel); + caps->maxProgramTexelOffset = GetMaximumTexelOffset(featureLevel); + + // Aggregate shader limits + caps->maxUniformBufferBindings = caps->maxVertexUniformBlocks + caps->maxFragmentUniformBlocks; + caps->maxUniformBlockSize = GetMaximumConstantBufferSize(featureLevel); + + // Setting a large alignment forces uniform buffers to bind with zero offset + caps->uniformBufferOffsetAlignment = static_cast<GLuint>(std::numeric_limits<GLint>::max()); + + caps->maxCombinedUniformBlocks = caps->maxVertexUniformBlocks + caps->maxFragmentUniformBlocks; + caps->maxCombinedVertexUniformComponents = (static_cast<GLint64>(caps->maxVertexUniformBlocks) * static_cast<GLint64>(caps->maxUniformBlockSize / 4)) + + static_cast<GLint64>(caps->maxVertexUniformComponents); + caps->maxCombinedFragmentUniformComponents = (static_cast<GLint64>(caps->maxFragmentUniformBlocks) * static_cast<GLint64>(caps->maxUniformBlockSize / 4)) + + static_cast<GLint64>(caps->maxFragmentUniformComponents); + caps->maxVaryingComponents = GetMaximumVertexOutputVectors(featureLevel) * 4; + caps->maxVaryingVectors = GetMaximumVertexOutputVectors(featureLevel); + caps->maxCombinedTextureImageUnits = caps->maxVertexTextureImageUnits + caps->maxTextureImageUnits; + + // Transform feedback limits + caps->maxTransformFeedbackInterleavedComponents = GetMaximumStreamOutputInterleavedComponenets(featureLevel); + caps->maxTransformFeedbackSeparateAttributes = GetMaximumStreamOutputBuffers(featureLevel); + caps->maxTransformFeedbackSeparateComponents = GetMaximumStreamOutputSeparateCompeonents(featureLevel); + + // GL extension support + extensions->setTextureExtensionSupport(*textureCapsMap); + extensions->elementIndexUint = true; + extensions->packedDepthStencil = true; + extensions->getProgramBinary = true; + extensions->rgb8rgba8 = true; + extensions->readFormatBGRA = true; + extensions->pixelBufferObject = true; + extensions->mapBuffer = true; + extensions->mapBufferRange = true; + extensions->textureNPOT = GetNPOTTextureSupport(featureLevel); + extensions->drawBuffers = GetMaximumSimultaneousRenderTargets(featureLevel) > 1; + extensions->textureStorage = true; + extensions->textureFilterAnisotropic = true; + extensions->maxTextureAnisotropy = GetMaximumAnisotropy(featureLevel); + extensions->occlusionQueryBoolean = GetOcclusionQuerySupport(featureLevel); + extensions->fence = GetEventQuerySupport(featureLevel); + extensions->timerQuery = false; // Unimplemented + extensions->robustness = true; + extensions->blendMinMax = true; + extensions->framebufferBlit = true; + extensions->framebufferMultisample = true; + extensions->maxSamples = maxSamples; + extensions->instancedArrays = GetInstancingSupport(featureLevel); + extensions->packReverseRowOrder = true; + extensions->standardDerivatives = GetDerivativeInstructionSupport(featureLevel); + extensions->shaderTextureLOD = true; + extensions->fragDepth = true; + extensions->textureUsage = true; // This could be false since it has no effect in D3D11 + extensions->translatedShaderSource = true; +} + +} + +namespace d3d11 +{ + +void MakeValidSize(bool isImage, DXGI_FORMAT format, GLsizei *requestWidth, GLsizei *requestHeight, int *levelOffset) +{ + const DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(format); + + int upsampleCount = 0; + // Don't expand the size of full textures that are at least (blockWidth x blockHeight) already. + if (isImage || *requestWidth < static_cast<GLsizei>(dxgiFormatInfo.blockWidth) || + *requestHeight < static_cast<GLsizei>(dxgiFormatInfo.blockHeight)) + { + while (*requestWidth % dxgiFormatInfo.blockWidth != 0 || *requestHeight % dxgiFormatInfo.blockHeight != 0) + { + *requestWidth <<= 1; + *requestHeight <<= 1; + upsampleCount++; + } + } + *levelOffset = upsampleCount; +} + +void GenerateInitialTextureData(GLint internalFormat, GLuint width, GLuint height, GLuint depth, + GLuint mipLevels, std::vector<D3D11_SUBRESOURCE_DATA> *outSubresourceData, + std::vector< std::vector<BYTE> > *outData) +{ + const d3d11::TextureFormat &d3dFormatInfo = d3d11::GetTextureFormatInfo(internalFormat); + ASSERT(d3dFormatInfo.dataInitializerFunction != NULL); + + const d3d11::DXGIFormat &dxgiFormatInfo = d3d11::GetDXGIFormatInfo(d3dFormatInfo.texFormat); + + outSubresourceData->resize(mipLevels); + outData->resize(mipLevels); + + for (unsigned int i = 0; i < mipLevels; i++) + { + unsigned int mipWidth = std::max(width >> i, 1U); + unsigned int mipHeight = std::max(height >> i, 1U); + unsigned int mipDepth = std::max(depth >> i, 1U); + + unsigned int rowWidth = dxgiFormatInfo.pixelBytes * mipWidth; + unsigned int imageSize = rowWidth * height; + + outData->at(i).resize(rowWidth * mipHeight * mipDepth); + d3dFormatInfo.dataInitializerFunction(mipWidth, mipHeight, mipDepth, outData->at(i).data(), rowWidth, imageSize); + + outSubresourceData->at(i).pSysMem = outData->at(i).data(); + outSubresourceData->at(i).SysMemPitch = rowWidth; + outSubresourceData->at(i).SysMemSlicePitch = imageSize; + } +} + +void SetPositionTexCoordVertex(PositionTexCoordVertex* vertex, float x, float y, float u, float v) +{ + vertex->x = x; + vertex->y = y; + vertex->u = u; + vertex->v = v; +} + +void SetPositionLayerTexCoord3DVertex(PositionLayerTexCoord3DVertex* vertex, float x, float y, + unsigned int layer, float u, float v, float s) +{ + vertex->x = x; + vertex->y = y; + vertex->l = layer; + vertex->u = u; + vertex->v = v; + vertex->s = s; +} + +HRESULT SetDebugName(ID3D11DeviceChild *resource, const char *name) +{ +#if !defined(__MINGW32__) && defined(_DEBUG) + return resource->SetPrivateData(WKPDID_D3DDebugObjectName, strlen(name), name); +#else + return S_OK; +#endif +} + +RenderTarget11 *GetAttachmentRenderTarget(gl::FramebufferAttachment *attachment) +{ + RenderTarget *renderTarget = rx::GetAttachmentRenderTarget(attachment); + return RenderTarget11::makeRenderTarget11(renderTarget); +} + +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/renderer11_utils.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/renderer11_utils.h new file mode 100644 index 0000000000..4c05eb9256 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/renderer11_utils.h @@ -0,0 +1,185 @@ +// +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// renderer11_utils.h: Conversion functions and other utility routines +// specific to the D3D11 renderer. + +#ifndef LIBGLESV2_RENDERER_RENDERER11_UTILS_H +#define LIBGLESV2_RENDERER_RENDERER11_UTILS_H + +#include "libGLESv2/angletypes.h" +#include "libGLESv2/Caps.h" + +#include <vector> + +namespace gl +{ +class FramebufferAttachment; +} + +namespace rx +{ +class RenderTarget11; + +namespace gl_d3d11 +{ + +D3D11_BLEND ConvertBlendFunc(GLenum glBlend, bool isAlpha); +D3D11_BLEND_OP ConvertBlendOp(GLenum glBlendOp); +UINT8 ConvertColorMask(bool maskRed, bool maskGreen, bool maskBlue, bool maskAlpha); + +D3D11_CULL_MODE ConvertCullMode(bool cullEnabled, GLenum cullMode); + +D3D11_COMPARISON_FUNC ConvertComparison(GLenum comparison); +D3D11_DEPTH_WRITE_MASK ConvertDepthMask(bool depthWriteEnabled); +UINT8 ConvertStencilMask(GLuint stencilmask); +D3D11_STENCIL_OP ConvertStencilOp(GLenum stencilOp); + +D3D11_FILTER ConvertFilter(GLenum minFilter, GLenum magFilter, float maxAnisotropy, GLenum comparisonMode); +D3D11_TEXTURE_ADDRESS_MODE ConvertTextureWrap(GLenum wrap); + +D3D11_QUERY ConvertQueryType(GLenum queryType); + +} + +namespace d3d11_gl +{ + +void GenerateCaps(ID3D11Device *device, gl::Caps *caps, gl::TextureCapsMap *textureCapsMap, gl::Extensions *extensions); + +} + +namespace d3d11 +{ + +void MakeValidSize(bool isImage, DXGI_FORMAT format, GLsizei *requestWidth, GLsizei *requestHeight, int *levelOffset); + +void GenerateInitialTextureData(GLint internalFormat, GLuint width, GLuint height, GLuint depth, + GLuint mipLevels, std::vector<D3D11_SUBRESOURCE_DATA> *outSubresourceData, + std::vector< std::vector<BYTE> > *outData); + +struct PositionTexCoordVertex +{ + float x, y; + float u, v; +}; +void SetPositionTexCoordVertex(PositionTexCoordVertex* vertex, float x, float y, float u, float v); + +struct PositionLayerTexCoord3DVertex +{ + float x, y; + unsigned int l; + float u, v, s; +}; +void SetPositionLayerTexCoord3DVertex(PositionLayerTexCoord3DVertex* vertex, float x, float y, + unsigned int layer, float u, float v, float s); + +template <typename T> +struct PositionDepthColorVertex +{ + float x, y, z; + T r, g, b, a; +}; + +template <typename T> +void SetPositionDepthColorVertex(PositionDepthColorVertex<T>* vertex, float x, float y, float z, + const gl::Color<T> &color) +{ + vertex->x = x; + vertex->y = y; + vertex->z = z; + vertex->r = color.red; + vertex->g = color.green; + vertex->b = color.blue; + vertex->a = color.alpha; +} + +HRESULT SetDebugName(ID3D11DeviceChild *resource, const char *name); + +template <typename outType> +outType* DynamicCastComObject(IUnknown* object) +{ + outType *outObject = NULL; + HRESULT result = object->QueryInterface(__uuidof(outType), reinterpret_cast<void**>(&outObject)); + if (SUCCEEDED(result)) + { + return outObject; + } + else + { + SafeRelease(outObject); + return NULL; + } +} + +inline bool isDeviceLostError(HRESULT errorCode) +{ + switch (errorCode) + { + case DXGI_ERROR_DEVICE_HUNG: + case DXGI_ERROR_DEVICE_REMOVED: + case DXGI_ERROR_DEVICE_RESET: + case DXGI_ERROR_DRIVER_INTERNAL_ERROR: + case DXGI_ERROR_NOT_CURRENTLY_AVAILABLE: + return true; + default: + return false; + } +} + +template <unsigned int N> +inline ID3D11VertexShader *CompileVS(ID3D11Device *device, const BYTE (&byteCode)[N], const char *name) +{ + ID3D11VertexShader *vs = NULL; + HRESULT result = device->CreateVertexShader(byteCode, N, NULL, &vs); + UNUSED_ASSERTION_VARIABLE(result); + ASSERT(SUCCEEDED(result)); + SetDebugName(vs, name); + return vs; +} + +template <unsigned int N> +inline ID3D11GeometryShader *CompileGS(ID3D11Device *device, const BYTE (&byteCode)[N], const char *name) +{ + ID3D11GeometryShader *gs = NULL; + HRESULT result = device->CreateGeometryShader(byteCode, N, NULL, &gs); + UNUSED_ASSERTION_VARIABLE(result); + ASSERT(SUCCEEDED(result)); + SetDebugName(gs, name); + return gs; +} + +template <unsigned int N> +inline ID3D11PixelShader *CompilePS(ID3D11Device *device, const BYTE (&byteCode)[N], const char *name) +{ + ID3D11PixelShader *ps = NULL; + HRESULT result = device->CreatePixelShader(byteCode, N, NULL, &ps); + UNUSED_ASSERTION_VARIABLE(result); + ASSERT(SUCCEEDED(result)); + SetDebugName(ps, name); + return ps; +} + +// Copy data to small D3D11 buffers, such as for small constant buffers, which use one struct to +// represent an entire buffer. +template <class T> +inline void SetBufferData(ID3D11DeviceContext *context, ID3D11Buffer *constantBuffer, const T &value) +{ + D3D11_MAPPED_SUBRESOURCE mappedResource; + context->Map(constantBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + + memcpy(mappedResource.pData, &value, sizeof(T)); + + context->Unmap(constantBuffer, 0); +} + +RenderTarget11 *GetAttachmentRenderTarget(gl::FramebufferAttachment *attachment); + +} + +} + +#endif // LIBGLESV2_RENDERER_RENDERER11_UTILS_H diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/BufferToTexture11.hlsl b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/BufferToTexture11.hlsl new file mode 100644 index 0000000000..20e6623a30 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/BufferToTexture11.hlsl @@ -0,0 +1,76 @@ +Buffer<float4> Buffer4F : register(t0); +Buffer<int4> Buffer4I : register(t0); +Buffer<uint4> Buffer4UI : register(t0); + +struct VS_OUTPUT +{ + float4 position : SV_Position; + uint index : TEXCOORD0; + uint slice : LAYER; +}; + +struct GS_OUTPUT +{ + float4 position : SV_Position; + uint index : TEXCOORD0; + uint slice : SV_RenderTargetArrayIndex; +}; + +cbuffer BufferCopyParams : register(b0) +{ + uint FirstPixelOffset; + uint PixelsPerRow; + uint RowStride; + uint RowsPerSlice; + float2 PositionOffset; + float2 PositionScale; + int2 TexLocationOffset; + int2 TexLocationScale; +} + +void ComputePositionAndIndex(uint vertexID, out VS_OUTPUT outVertex) +{ + uint PixelsPerSlice = PixelsPerRow * RowsPerSlice; + uint SliceStride = RowStride * RowsPerSlice; + + uint slice = vertexID / PixelsPerSlice; + uint sliceOffset = slice * PixelsPerSlice; + uint row = (vertexID - sliceOffset) / PixelsPerRow; + uint col = vertexID - sliceOffset - (row * PixelsPerRow); + + float2 coords = float2(float(col), float(row)); + + outVertex.position = float4(PositionOffset + PositionScale * coords, 0.0f, 1.0f); + outVertex.index = FirstPixelOffset + slice * SliceStride + row * RowStride + col; + outVertex.slice = slice; +} + +void VS_BufferToTexture(in uint vertexID : SV_VertexID, out VS_OUTPUT outVertex) +{ + ComputePositionAndIndex(vertexID, outVertex); +} + +[maxvertexcount(1)] +void GS_BufferToTexture(point VS_OUTPUT inVertex[1], inout PointStream<GS_OUTPUT> outStream) +{ + GS_OUTPUT outVertex; + outVertex.position = inVertex[0].position; + outVertex.index = inVertex[0].index; + outVertex.slice = inVertex[0].slice; + outStream.Append(outVertex); +} + +float4 PS_BufferToTexture_4F(in float4 inPosition : SV_Position, in uint inIndex : TEXCOORD0) : SV_Target +{ + return Buffer4F.Load(inIndex); +} + +int4 PS_BufferToTexture_4I(in float4 inPosition : SV_Position, in uint inIndex : TEXCOORD0) : SV_Target +{ + return Buffer4I.Load(inIndex); +} + +uint4 PS_BufferToTexture_4UI(in float4 inPosition : SV_Position, in uint inIndex : TEXCOORD0) : SV_Target +{ + return Buffer4UI.Load(inIndex); +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/Clear11.hlsl b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/Clear11.hlsl new file mode 100644 index 0000000000..b4cf38076e --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/Clear11.hlsl @@ -0,0 +1,106 @@ +// Assume we are in SM4+, which has 8 color outputs + +void VS_ClearFloat( in float3 inPosition : POSITION, in float4 inColor : COLOR, + out float4 outPosition : SV_POSITION, out float4 outColor : COLOR) +{ + outPosition = float4(inPosition, 1.0f); + outColor = inColor; +} + +struct PS_OutputFloat +{ + float4 color0 : SV_TARGET0; + float4 color1 : SV_TARGET1; + float4 color2 : SV_TARGET2; + float4 color3 : SV_TARGET3; +#if SM4 + float4 color4 : SV_TARGET4; + float4 color5 : SV_TARGET5; + float4 color6 : SV_TARGET6; + float4 color7 : SV_TARGET7; +#endif +}; + +PS_OutputFloat PS_ClearFloat(in float4 inPosition : SV_POSITION, in float4 inColor : COLOR) +{ + PS_OutputFloat outColor; + outColor.color0 = inColor; + outColor.color1 = inColor; + outColor.color2 = inColor; + outColor.color3 = inColor; +#if SM4 + outColor.color4 = inColor; + outColor.color5 = inColor; + outColor.color6 = inColor; + outColor.color7 = inColor; +#endif + return outColor; +} + + +void VS_ClearUint( in float3 inPosition : POSITION, in uint4 inColor : COLOR, + out float4 outPosition : SV_POSITION, out uint4 outColor : COLOR) +{ + outPosition = float4(inPosition, 1.0f); + outColor = inColor; +} + +struct PS_OutputUint +{ + uint4 color0 : SV_TARGET0; + uint4 color1 : SV_TARGET1; + uint4 color2 : SV_TARGET2; + uint4 color3 : SV_TARGET3; + uint4 color4 : SV_TARGET4; + uint4 color5 : SV_TARGET5; + uint4 color6 : SV_TARGET6; + uint4 color7 : SV_TARGET7; +}; + +PS_OutputUint PS_ClearUint(in float4 inPosition : SV_POSITION, in uint4 inColor : COLOR) +{ + PS_OutputUint outColor; + outColor.color0 = inColor; + outColor.color1 = inColor; + outColor.color2 = inColor; + outColor.color3 = inColor; + outColor.color4 = inColor; + outColor.color5 = inColor; + outColor.color6 = inColor; + outColor.color7 = inColor; + return outColor; +} + + +void VS_ClearSint( in float3 inPosition : POSITION, in int4 inColor : COLOR, + out float4 outPosition : SV_POSITION, out int4 outColor : COLOR) +{ + outPosition = float4(inPosition, 1.0f); + outColor = inColor; +} + +struct PS_OutputSint +{ + int4 color0 : SV_TARGET0; + int4 color1 : SV_TARGET1; + int4 color2 : SV_TARGET2; + int4 color3 : SV_TARGET3; + int4 color4 : SV_TARGET4; + int4 color5 : SV_TARGET5; + int4 color6 : SV_TARGET6; + int4 color7 : SV_TARGET7; +}; + +PS_OutputSint PS_ClearSint(in float4 inPosition : SV_POSITION, in int4 inColor : COLOR) +{ + PS_OutputSint outColor; + outColor.color0 = inColor; + outColor.color1 = inColor; + outColor.color2 = inColor; + outColor.color3 = inColor; + outColor.color4 = inColor; + outColor.color5 = inColor; + outColor.color6 = inColor; + outColor.color7 = inColor; + return outColor; +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/Passthrough2D11.hlsl b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/Passthrough2D11.hlsl new file mode 100644 index 0000000000..8671c39fb7 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/Passthrough2D11.hlsl @@ -0,0 +1,111 @@ +Texture2D<float4> TextureF : register(t0); +Texture2D<uint4> TextureUI : register(t0); +Texture2D<int4> TextureI : register(t0); + +SamplerState Sampler : register(s0); + +void VS_Passthrough2D( in float2 inPosition : POSITION, in float2 inTexCoord : TEXCOORD0, + out float4 outPosition : SV_POSITION, out float2 outTexCoord : TEXCOORD0) +{ + outPosition = float4(inPosition, 0.0f, 1.0f); + outTexCoord = inTexCoord; +} + +float PS_PassthroughDepth2D(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_DEPTH +{ + return TextureF.Sample(Sampler, inTexCoord).r; +} + +float4 PS_PassthroughRGBA2D(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + return TextureF.Sample(Sampler, inTexCoord).rgba; +} + +uint4 PS_PassthroughRGBA2DUI(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint2 size; + TextureUI.GetDimensions(size.x, size.y); + + return TextureUI.Load(int3(size * inTexCoord, 0)).rgba; +} + +int4 PS_PassthroughRGBA2DI(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint2 size; + TextureI.GetDimensions(size.x, size.y); + + return TextureI.Load(int3(size * inTexCoord, 0)).rgba; +} + +float4 PS_PassthroughRGB2D(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + return float4(TextureF.Sample(Sampler, inTexCoord).rgb, 1.0f); +} + +uint4 PS_PassthroughRGB2DUI(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint2 size; + TextureUI.GetDimensions(size.x, size.y); + + return uint4(TextureUI.Load(int3(size * inTexCoord, 0)).rgb, 0); +} + +int4 PS_PassthroughRGB2DI(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint2 size; + TextureI.GetDimensions(size.x, size.y); + + return int4(TextureI.Load(int3(size * inTexCoord, 0)).rgb, 0); +} + +float4 PS_PassthroughRG2D(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + return float4(TextureF.Sample(Sampler, inTexCoord).rg, 0.0f, 1.0f); +} + +uint4 PS_PassthroughRG2DUI(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint2 size; + TextureUI.GetDimensions(size.x, size.y); + + return uint4(TextureUI.Load(int3(size * inTexCoord, 0)).rg, 0, 0); +} + +int4 PS_PassthroughRG2DI(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint2 size; + TextureI.GetDimensions(size.x, size.y); + + return int4(TextureI.Load(int3(size * inTexCoord, 0)).rg, 0, 0); +} + +float4 PS_PassthroughR2D(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + return float4(TextureF.Sample(Sampler, inTexCoord).r, 0.0f, 0.0f, 1.0f); +} + +uint4 PS_PassthroughR2DUI(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint2 size; + TextureUI.GetDimensions(size.x, size.y); + + return uint4(TextureUI.Load(int3(size * inTexCoord, 0)).r, 0, 0, 0); +} + +int4 PS_PassthroughR2DI(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint2 size; + TextureI.GetDimensions(size.x, size.y); + + return int4(TextureI.Load(int3(size * inTexCoord, 0)).r, 0, 0, 0); +} + +float4 PS_PassthroughLum2D(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + return float4(TextureF.Sample(Sampler, inTexCoord).rrr, 1.0f); +} + +float4 PS_PassthroughLumAlpha2D(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + return TextureF.Sample(Sampler, inTexCoord).rrra; +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/Passthrough3D11.hlsl b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/Passthrough3D11.hlsl new file mode 100644 index 0000000000..c23c9032ec --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/Passthrough3D11.hlsl @@ -0,0 +1,146 @@ +Texture3D<float4> TextureF : register(t0); +Texture3D<uint4> TextureUI : register(t0); +Texture3D<int4> TextureI : register(t0); + +SamplerState Sampler : register(s0); + +struct VS_INPUT +{ + float2 Position : POSITION; + uint Layer : LAYER; + float3 TexCoord : TEXCOORD; +}; + +struct VS_OUTPUT +{ + float4 Position : SV_POSITION; + uint Layer : LAYER; + float3 TexCoord : TEXCOORD; +}; + +struct GS_OUTPUT +{ + float4 Position : SV_POSITION; + uint Layer : SV_RENDERTARGETARRAYINDEX; + float3 TexCoord : TEXCOORD; +}; + +VS_OUTPUT VS_Passthrough3D(VS_INPUT input) +{ + VS_OUTPUT output; + + output.Position = float4(input.Position, 0.0f, 1.0f); + output.Layer = input.Layer; + output.TexCoord = input.TexCoord; + + return output; +} + +[maxvertexcount(3)] +void GS_Passthrough3D(triangle VS_OUTPUT input[3], inout TriangleStream<GS_OUTPUT> outputStream) +{ + GS_OUTPUT output; + + for (int i = 0; i < 3; i++) + { + output.Position = input[i].Position; + output.Layer = input[i].Layer; + output.TexCoord = input[i].TexCoord; + + outputStream.Append(output); + } +} + +float4 PS_PassthroughRGBA3D(GS_OUTPUT input) : SV_TARGET0 +{ + return TextureF.Sample(Sampler, input.TexCoord).rgba; +} + +uint4 PS_PassthroughRGBA3DUI(GS_OUTPUT input) : SV_TARGET0 +{ + uint3 size; + TextureUI.GetDimensions(size.x, size.y, size.z); + + return TextureUI.Load(int4(size * input.TexCoord, 0)).rgba; +} + +int4 PS_PassthroughRGBA3DI(GS_OUTPUT input) : SV_TARGET0 +{ + uint3 size; + TextureI.GetDimensions(size.x, size.y, size.z); + + return TextureI.Load(int4(size * input.TexCoord, 0)).rgba; +} + +float4 PS_PassthroughRGB3D(GS_OUTPUT input) : SV_TARGET0 +{ + return float4(TextureF.Sample(Sampler, input.TexCoord).rgb, 1.0f); +} + +uint4 PS_PassthroughRGB3DUI(GS_OUTPUT input) : SV_TARGET0 +{ + uint3 size; + TextureUI.GetDimensions(size.x, size.y, size.z); + + return uint4(TextureUI.Load(int4(size * input.TexCoord, 0)).rgb, 0); +} + +int4 PS_PassthroughRGB3DI(GS_OUTPUT input) : SV_TARGET0 +{ + uint3 size; + TextureI.GetDimensions(size.x, size.y, size.z); + + return int4(TextureI.Load(int4(size * input.TexCoord, 0)).rgb, 0); +} + +float4 PS_PassthroughRG3D(GS_OUTPUT input) : SV_TARGET0 +{ + return float4(TextureF.Sample(Sampler, input.TexCoord).rg, 0.0f, 1.0f); +} + +uint4 PS_PassthroughRG3DUI(GS_OUTPUT input) : SV_TARGET0 +{ + uint3 size; + TextureUI.GetDimensions(size.x, size.y, size.z); + + return uint4(TextureUI.Load(int4(size * input.TexCoord, 0)).rg, 0, 0); +} + +int4 PS_PassthroughRG3DI(GS_OUTPUT input) : SV_TARGET0 +{ + uint3 size; + TextureI.GetDimensions(size.x, size.y, size.z); + + return int4(TextureI.Load(int4(size * input.TexCoord, 0)).rg, 0, 0); +} + +float4 PS_PassthroughR3D(GS_OUTPUT input) : SV_TARGET0 +{ + return float4(TextureF.Sample(Sampler, input.TexCoord).r, 0.0f, 0.0f, 1.0f); +} + +uint4 PS_PassthroughR3DUI(GS_OUTPUT input) : SV_TARGET0 +{ + uint3 size; + TextureUI.GetDimensions(size.x, size.y, size.z); + + return uint4(TextureUI.Load(int4(size * input.TexCoord, 0)).r, 0, 0, 0); +} + +int4 PS_PassthroughR3DI(GS_OUTPUT input) : SV_TARGET0 +{ + uint3 size; + TextureI.GetDimensions(size.x, size.y, size.z); + + return int4(TextureI.Load(int4(size * input.TexCoord, 0)).r, 0, 0, 0); +} + +float4 PS_PassthroughLum3D(GS_OUTPUT input) : SV_TARGET0 +{ + return float4(TextureF.Sample(Sampler, input.TexCoord).rrr, 1.0f); +} + +float4 PS_PassthroughLumAlpha3D(GS_OUTPUT input) : SV_TARGET0 +{ + return TextureF.Sample(Sampler, input.TexCoord).rrra; +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/Swizzle11.hlsl b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/Swizzle11.hlsl new file mode 100644 index 0000000000..505e222137 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d11/shaders/Swizzle11.hlsl @@ -0,0 +1,99 @@ +Texture2D<float4> TextureF2D : register(t0); +Texture2D<uint4> TextureUI2D : register(t0); +Texture2D<int4> TextureI2D : register(t0); + +Texture3D<float4> TextureF3D : register(t0); +Texture3D<uint4> TextureUI3D : register(t0); +Texture3D<int4> TextureI3D : register(t0); + +Texture2DArray<float4> TextureF2DArray : register(t0); +Texture2DArray<uint4> TextureUI2DArray : register(t0); +Texture2DArray<int4> TextureI2DArray : register(t0); + +SamplerState Sampler : register(s0); + +cbuffer SwizzleProperties : register(b0) +{ + uint4 SwizzleIndices : packoffset(c0); +} + +float4 SwizzleLookup(in float4 sample) +{ + float lookup[6] = { sample[0], sample[1], sample[2], sample[3], 0.0f, 1.0f }; + return float4(lookup[SwizzleIndices[0]], lookup[SwizzleIndices[1]], lookup[SwizzleIndices[2]], lookup[SwizzleIndices[3]]); +} + +int4 SwizzleLookup(in int4 sample) +{ + int lookup[6] = { sample[0], sample[1], sample[2], sample[3], 0.0f, 1.0f }; + return int4(lookup[SwizzleIndices[0]], lookup[SwizzleIndices[1]], lookup[SwizzleIndices[2]], lookup[SwizzleIndices[3]]); +} + +uint4 SwizzleLookup(in uint4 sample) +{ + uint lookup[6] = { sample[0], sample[1], sample[2], sample[3], 0.0f, 1.0f }; + return uint4(lookup[SwizzleIndices[0]], lookup[SwizzleIndices[1]], lookup[SwizzleIndices[2]], lookup[SwizzleIndices[3]]); +} + +float4 PS_SwizzleF2D(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + return SwizzleLookup(TextureF2D.Sample(Sampler, inTexCoord)); +} + +int4 PS_SwizzleI2D(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint2 size; + TextureI2D.GetDimensions(size.x, size.y); + + return SwizzleLookup(TextureI2D.Load(int3(size * inTexCoord, 0))); +} + +uint4 PS_SwizzleUI2D(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint2 size; + TextureUI2D.GetDimensions(size.x, size.y); + + return SwizzleLookup(TextureUI2D.Load(int3(size * inTexCoord, 0))); +} + +float4 PS_SwizzleF3D(in float4 inPosition : SV_POSITION, in uint inLayer : SV_RENDERTARGETARRAYINDEX, in float3 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + return SwizzleLookup(TextureF3D.Sample(Sampler, inTexCoord)); +} + +int4 PS_SwizzleI3D(in float4 inPosition : SV_POSITION, in uint inLayer : SV_RENDERTARGETARRAYINDEX, in float3 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint3 size; + TextureI3D.GetDimensions(size.x, size.y, size.z); + + return SwizzleLookup(TextureI3D.Load(int4(size * inTexCoord, 0))); +} + +uint4 PS_SwizzleUI3D(in float4 inPosition : SV_POSITION, in uint inLayer : SV_RENDERTARGETARRAYINDEX, in float3 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint3 size; + TextureUI3D.GetDimensions(size.x, size.y, size.z); + + return SwizzleLookup(TextureUI3D.Load(int4(size * inTexCoord, 0))); +} + +float4 PS_SwizzleF2DArray(in float4 inPosition : SV_POSITION, in uint inLayer : SV_RENDERTARGETARRAYINDEX, in float3 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + return SwizzleLookup(TextureF2DArray.Sample(Sampler, float3(inTexCoord.xy, inLayer))); +} + +int4 PS_SwizzleI2DArray(in float4 inPosition : SV_POSITION, in uint inLayer : SV_RENDERTARGETARRAYINDEX, in float3 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint3 size; + TextureI2DArray.GetDimensions(size.x, size.y, size.z); + + return SwizzleLookup(TextureI2DArray.Load(int4(size.xy * inTexCoord.xy, inLayer, 0))); +} + +uint4 PS_SwizzleUI2DArray(in float4 inPosition : SV_POSITION, in uint inLayer : SV_RENDERTARGETARRAYINDEX, in float3 inTexCoord : TEXCOORD0) : SV_TARGET0 +{ + uint3 size; + TextureUI2DArray.GetDimensions(size.x, size.y, size.z); + + return SwizzleLookup(TextureUI2DArray.Load(int4(size.xy * inTexCoord.xy, inLayer, 0))); +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Blit.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Blit9.cpp index d73df6418d..f061a32c52 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Blit.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Blit9.cpp @@ -1,74 +1,71 @@ -#include "precompiled.h" // // Copyright (c) 2002-2010 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // -// Blit.cpp: Surface copy utility class. +// Blit9.cpp: Surface copy utility class. -#include "libGLESv2/renderer/d3d9/Blit.h" - -#include "libGLESv2/main.h" -#include "libGLESv2/renderer/d3d9/renderer9_utils.h" -#include "libGLESv2/renderer/d3d9/TextureStorage9.h" -#include "libGLESv2/renderer/d3d9/RenderTarget9.h" -#include "libGLESv2/renderer/d3d9/Renderer9.h" +#include "libGLESv2/renderer/d3d/d3d9/Blit9.h" +#include "libGLESv2/renderer/d3d/d3d9/renderer9_utils.h" +#include "libGLESv2/renderer/d3d/d3d9/formatutils9.h" +#include "libGLESv2/renderer/d3d/d3d9/TextureStorage9.h" +#include "libGLESv2/renderer/d3d/d3d9/RenderTarget9.h" +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" #include "libGLESv2/Framebuffer.h" -#include "libGLESv2/Renderbuffer.h" +#include "libGLESv2/FramebufferAttachment.h" +#include "libGLESv2/main.h" namespace { -#include "libGLESv2/renderer/d3d9/shaders/compiled/standardvs.h" -#include "libGLESv2/renderer/d3d9/shaders/compiled/flipyvs.h" -#include "libGLESv2/renderer/d3d9/shaders/compiled/passthroughps.h" -#include "libGLESv2/renderer/d3d9/shaders/compiled/luminanceps.h" -#include "libGLESv2/renderer/d3d9/shaders/compiled/componentmaskps.h" +// Precompiled shaders +#include "libGLESv2/renderer/d3d/d3d9/shaders/compiled/standardvs.h" +#include "libGLESv2/renderer/d3d/d3d9/shaders/compiled/flipyvs.h" +#include "libGLESv2/renderer/d3d/d3d9/shaders/compiled/passthroughps.h" +#include "libGLESv2/renderer/d3d/d3d9/shaders/compiled/luminanceps.h" +#include "libGLESv2/renderer/d3d/d3d9/shaders/compiled/componentmaskps.h" const BYTE* const g_shaderCode[] = { - g_vs20_standardvs, - g_vs20_flipyvs, - g_ps20_passthroughps, - g_ps20_luminanceps, - g_ps20_componentmaskps + g_vs20_VS_standard, + g_vs20_VS_flipy, + g_ps20_PS_passthrough, + g_ps20_PS_luminance, + g_ps20_PS_componentmask }; const size_t g_shaderSize[] = { - sizeof(g_vs20_standardvs), - sizeof(g_vs20_flipyvs), - sizeof(g_ps20_passthroughps), - sizeof(g_ps20_luminanceps), - sizeof(g_ps20_componentmaskps) + sizeof(g_vs20_VS_standard), + sizeof(g_vs20_VS_flipy), + sizeof(g_ps20_PS_passthrough), + sizeof(g_ps20_PS_luminance), + sizeof(g_ps20_PS_componentmask) }; } namespace rx { -Blit::Blit(rx::Renderer9 *renderer) +Blit9::Blit9(rx::Renderer9 *renderer) : mRenderer(renderer), mQuadVertexBuffer(NULL), mQuadVertexDeclaration(NULL), mSavedStateBlock(NULL), mSavedRenderTarget(NULL), mSavedDepthStencil(NULL) { initGeometry(); memset(mCompiledShaders, 0, sizeof(mCompiledShaders)); } -Blit::~Blit() +Blit9::~Blit9() { - if (mSavedStateBlock) mSavedStateBlock->Release(); - if (mQuadVertexBuffer) mQuadVertexBuffer->Release(); - if (mQuadVertexDeclaration) mQuadVertexDeclaration->Release(); + SafeRelease(mSavedStateBlock); + SafeRelease(mQuadVertexBuffer); + SafeRelease(mQuadVertexDeclaration); for (int i = 0; i < SHADER_COUNT; i++) { - if (mCompiledShaders[i]) - { - mCompiledShaders[i]->Release(); - } + SafeRelease(mCompiledShaders[i]); } } -void Blit::initGeometry() +void Blit9::initGeometry() { static const float quad[] = { @@ -116,7 +113,7 @@ void Blit::initGeometry() } template <class D3DShaderType> -bool Blit::setShader(ShaderId source, const char *profile, +bool Blit9::setShader(ShaderId source, const char *profile, D3DShaderType *(rx::Renderer9::*createShader)(const DWORD *, size_t length), HRESULT (WINAPI IDirect3DDevice9::*setShader)(D3DShaderType*)) { @@ -154,17 +151,17 @@ bool Blit::setShader(ShaderId source, const char *profile, return true; } -bool Blit::setVertexShader(ShaderId shader) +bool Blit9::setVertexShader(ShaderId shader) { return setShader<IDirect3DVertexShader9>(shader, "vs_2_0", &rx::Renderer9::createVertexShader, &IDirect3DDevice9::SetVertexShader); } -bool Blit::setPixelShader(ShaderId shader) +bool Blit9::setPixelShader(ShaderId shader) { return setShader<IDirect3DPixelShader9>(shader, "ps_2_0", &rx::Renderer9::createPixelShader, &IDirect3DDevice9::SetPixelShader); } -RECT Blit::getSurfaceRect(IDirect3DSurface9 *surface) const +RECT Blit9::getSurfaceRect(IDirect3DSurface9 *surface) const { D3DSURFACE_DESC desc; surface->GetDesc(&desc); @@ -178,7 +175,7 @@ RECT Blit::getSurfaceRect(IDirect3DSurface9 *surface) const return rect; } -bool Blit::boxFilter(IDirect3DSurface9 *source, IDirect3DSurface9 *dest) +bool Blit9::boxFilter(IDirect3DSurface9 *source, IDirect3DSurface9 *dest) { IDirect3DTexture9 *texture = copySurfaceToTexture(source, getSurfaceRect(source)); if (!texture) @@ -204,24 +201,24 @@ bool Blit::boxFilter(IDirect3DSurface9 *source, IDirect3DSurface9 *dest) render(); - texture->Release(); + SafeRelease(texture); restoreState(); return true; } -bool Blit::copy(gl::Framebuffer *framebuffer, const RECT &sourceRect, GLenum destFormat, GLint xoffset, GLint yoffset, TextureStorageInterface2D *storage, GLint level) +bool Blit9::copy2D(gl::Framebuffer *framebuffer, const RECT &sourceRect, GLenum destFormat, GLint xoffset, GLint yoffset, TextureStorage *storage, GLint level) { RenderTarget9 *renderTarget = NULL; IDirect3DSurface9 *source = NULL; - gl::Renderbuffer *colorbuffer = framebuffer->getColorbuffer(0); + gl::FramebufferAttachment *colorbuffer = framebuffer->getColorbuffer(0); if (colorbuffer) { - renderTarget = RenderTarget9::makeRenderTarget9(colorbuffer->getRenderTarget()); + renderTarget = d3d9::GetAttachmentRenderTarget(colorbuffer); } - + if (renderTarget) { source = renderTarget->getSurface(); @@ -233,31 +230,31 @@ bool Blit::copy(gl::Framebuffer *framebuffer, const RECT &sourceRect, GLenum des return gl::error(GL_OUT_OF_MEMORY, false); } - TextureStorage9_2D *storage9 = TextureStorage9_2D::makeTextureStorage9_2D(storage->getStorageInstance()); + TextureStorage9_2D *storage9 = TextureStorage9_2D::makeTextureStorage9_2D(storage); IDirect3DSurface9 *destSurface = storage9->getSurfaceLevel(level, true); bool result = false; - + if (destSurface) { result = copy(source, sourceRect, destFormat, xoffset, yoffset, destSurface); - destSurface->Release(); + SafeRelease(destSurface); } - source->Release(); + SafeRelease(source); return result; } -bool Blit::copy(gl::Framebuffer *framebuffer, const RECT &sourceRect, GLenum destFormat, GLint xoffset, GLint yoffset, TextureStorageInterfaceCube *storage, GLenum target, GLint level) +bool Blit9::copyCube(gl::Framebuffer *framebuffer, const RECT &sourceRect, GLenum destFormat, GLint xoffset, GLint yoffset, TextureStorage *storage, GLenum target, GLint level) { RenderTarget9 *renderTarget = NULL; IDirect3DSurface9 *source = NULL; - gl::Renderbuffer *colorbuffer = framebuffer->getColorbuffer(0); + gl::FramebufferAttachment *colorbuffer = framebuffer->getColorbuffer(0); if (colorbuffer) { - renderTarget = RenderTarget9::makeRenderTarget9(colorbuffer->getRenderTarget()); + renderTarget = d3d9::GetAttachmentRenderTarget(colorbuffer); } - + if (renderTarget) { source = renderTarget->getSurface(); @@ -269,21 +266,21 @@ bool Blit::copy(gl::Framebuffer *framebuffer, const RECT &sourceRect, GLenum des return gl::error(GL_OUT_OF_MEMORY, false); } - TextureStorage9_Cube *storage9 = TextureStorage9_Cube::makeTextureStorage9_Cube(storage->getStorageInstance()); + TextureStorage9_Cube *storage9 = TextureStorage9_Cube::makeTextureStorage9_Cube(storage); IDirect3DSurface9 *destSurface = storage9->getCubeMapSurface(target, level, true); bool result = false; if (destSurface) { result = copy(source, sourceRect, destFormat, xoffset, yoffset, destSurface); - destSurface->Release(); + SafeRelease(destSurface); } - source->Release(); + SafeRelease(source); return result; } -bool Blit::copy(IDirect3DSurface9 *source, const RECT &sourceRect, GLenum destFormat, GLint xoffset, GLint yoffset, IDirect3DSurface9 *dest) +bool Blit9::copy(IDirect3DSurface9 *source, const RECT &sourceRect, GLenum destFormat, GLint xoffset, GLint yoffset, IDirect3DSurface9 *dest) { if (!dest) { @@ -316,7 +313,7 @@ bool Blit::copy(IDirect3DSurface9 *source, const RECT &sourceRect, GLenum destFo return true; } -bool Blit::formatConvert(IDirect3DSurface9 *source, const RECT &sourceRect, GLenum destFormat, GLint xoffset, GLint yoffset, IDirect3DSurface9 *dest) +bool Blit9::formatConvert(IDirect3DSurface9 *source, const RECT &sourceRect, GLenum destFormat, GLint xoffset, GLint yoffset, IDirect3DSurface9 *dest) { IDirect3DTexture9 *texture = copySurfaceToTexture(source, sourceRect); if (!texture) @@ -339,14 +336,14 @@ bool Blit::formatConvert(IDirect3DSurface9 *source, const RECT &sourceRect, GLen render(); } - texture->Release(); + SafeRelease(texture); restoreState(); return true; } -bool Blit::setFormatConvertShaders(GLenum destFormat) +bool Blit9::setFormatConvertShaders(GLenum destFormat) { bool okay = setVertexShader(SHADER_VS_STANDARD); @@ -356,6 +353,8 @@ bool Blit::setFormatConvertShaders(GLenum destFormat) case GL_RGBA: case GL_BGRA_EXT: case GL_RGB: + case GL_RG_EXT: + case GL_RED_EXT: case GL_ALPHA: okay = okay && setPixelShader(SHADER_PS_COMPONENTMASK); break; @@ -375,41 +374,99 @@ bool Blit::setFormatConvertShaders(GLenum destFormat) // The meaning of this constant depends on the shader that was selected. // See the shader assembly code above for details. - float psConst0[4] = { 0, 0, 0, 0 }; + // Allocate one array for both registers and split it into two float4's. + float psConst[8] = { 0 }; + float *multConst = &psConst[0]; + float *addConst = &psConst[4]; switch (destFormat) { default: UNREACHABLE(); case GL_RGBA: case GL_BGRA_EXT: - psConst0[X] = 1; - psConst0[Z] = 1; + multConst[X] = 1; + multConst[Y] = 1; + multConst[Z] = 1; + multConst[W] = 1; + addConst[X] = 0; + addConst[Y] = 0; + addConst[Z] = 0; + addConst[W] = 0; break; case GL_RGB: - psConst0[X] = 1; - psConst0[W] = 1; + multConst[X] = 1; + multConst[Y] = 1; + multConst[Z] = 1; + multConst[W] = 0; + addConst[X] = 0; + addConst[Y] = 0; + addConst[Z] = 0; + addConst[W] = 1; + break; + + case GL_RG_EXT: + multConst[X] = 1; + multConst[Y] = 1; + multConst[Z] = 0; + multConst[W] = 0; + addConst[X] = 0; + addConst[Y] = 0; + addConst[Z] = 0; + addConst[W] = 1; + break; + + case GL_RED_EXT: + multConst[X] = 1; + multConst[Y] = 0; + multConst[Z] = 0; + multConst[W] = 0; + addConst[X] = 0; + addConst[Y] = 0; + addConst[Z] = 0; + addConst[W] = 1; break; case GL_ALPHA: - psConst0[Z] = 1; + multConst[X] = 0; + multConst[Y] = 0; + multConst[Z] = 0; + multConst[W] = 1; + addConst[X] = 0; + addConst[Y] = 0; + addConst[Z] = 0; + addConst[W] = 0; break; case GL_LUMINANCE: - psConst0[Y] = 1; + multConst[X] = 1; + multConst[Y] = 0; + multConst[Z] = 0; + multConst[W] = 0; + addConst[X] = 0; + addConst[Y] = 0; + addConst[Z] = 0; + addConst[W] = 1; break; case GL_LUMINANCE_ALPHA: - psConst0[X] = 1; + multConst[X] = 1; + multConst[Y] = 0; + multConst[Z] = 0; + multConst[W] = 1; + addConst[X] = 0; + addConst[Y] = 0; + addConst[Z] = 0; + addConst[W] = 0; break; } - mRenderer->getDevice()->SetPixelShaderConstantF(0, psConst0, 1); + mRenderer->getDevice()->SetPixelShaderConstantF(0, psConst, 2); return true; } -IDirect3DTexture9 *Blit::copySurfaceToTexture(IDirect3DSurface9 *surface, const RECT &sourceRect) +IDirect3DTexture9 *Blit9::copySurfaceToTexture(IDirect3DSurface9 *surface, const RECT &sourceRect) { if (!surface) { @@ -437,26 +494,26 @@ IDirect3DTexture9 *Blit::copySurfaceToTexture(IDirect3DSurface9 *surface, const if (FAILED(result)) { ASSERT(result == D3DERR_OUTOFVIDEOMEMORY || result == E_OUTOFMEMORY); - texture->Release(); + SafeRelease(texture); return gl::error(GL_OUT_OF_MEMORY, (IDirect3DTexture9*)NULL); } mRenderer->endScene(); result = device->StretchRect(surface, &sourceRect, textureSurface, NULL, D3DTEXF_NONE); - textureSurface->Release(); + SafeRelease(textureSurface); if (FAILED(result)) { ASSERT(result == D3DERR_OUTOFVIDEOMEMORY || result == E_OUTOFMEMORY); - texture->Release(); + SafeRelease(texture); return gl::error(GL_OUT_OF_MEMORY, (IDirect3DTexture9*)NULL); } return texture; } -void Blit::setViewport(const RECT &sourceRect, GLint xoffset, GLint yoffset) +void Blit9::setViewport(const RECT &sourceRect, GLint xoffset, GLint yoffset) { IDirect3DDevice9 *device = mRenderer->getDevice(); @@ -473,7 +530,7 @@ void Blit::setViewport(const RECT &sourceRect, GLint xoffset, GLint yoffset) device->SetVertexShaderConstantF(0, halfPixelAdjust, 1); } -void Blit::setCommonBlitState() +void Blit9::setCommonBlitState() { IDirect3DDevice9 *device = mRenderer->getDevice(); @@ -503,7 +560,7 @@ void Blit::setCommonBlitState() } } -void Blit::render() +void Blit9::render() { IDirect3DDevice9 *device = mRenderer->getDevice(); @@ -514,7 +571,7 @@ void Blit::render() hr = device->DrawPrimitive(D3DPT_TRIANGLESTRIP, 0, 2); } -void Blit::saveState() +void Blit9::saveState() { IDirect3DDevice9 *device = mRenderer->getDevice(); @@ -530,12 +587,12 @@ void Blit::saveState() setCommonBlitState(); - static const float dummyConst[4] = { 0, 0, 0, 0 }; + static const float dummyConst[8] = { 0 }; device->SetVertexShader(NULL); - device->SetVertexShaderConstantF(0, dummyConst, 1); + device->SetVertexShaderConstantF(0, dummyConst, 2); device->SetPixelShader(NULL); - device->SetPixelShaderConstantF(0, dummyConst, 1); + device->SetPixelShaderConstantF(0, dummyConst, 2); D3DVIEWPORT9 dummyVp; dummyVp.X = 0; @@ -566,23 +623,15 @@ void Blit::saveState() } } -void Blit::restoreState() +void Blit9::restoreState() { IDirect3DDevice9 *device = mRenderer->getDevice(); device->SetDepthStencilSurface(mSavedDepthStencil); - if (mSavedDepthStencil != NULL) - { - mSavedDepthStencil->Release(); - mSavedDepthStencil = NULL; - } + SafeRelease(mSavedDepthStencil); device->SetRenderTarget(0, mSavedRenderTarget); - if (mSavedRenderTarget != NULL) - { - mSavedRenderTarget->Release(); - mSavedRenderTarget = NULL; - } + SafeRelease(mSavedRenderTarget); ASSERT(mSavedStateBlock != NULL); diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Blit.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Blit9.h index 3718028e66..46a3ee1cf3 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Blit.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Blit9.h @@ -4,13 +4,15 @@ // found in the LICENSE file. // -// Blit.cpp: Surface copy utility class. +// Blit9.cpp: Surface copy utility class. -#ifndef LIBGLESV2_BLIT_H_ -#define LIBGLESV2_BLIT_H_ +#ifndef LIBGLESV2_BLIT9_H_ +#define LIBGLESV2_BLIT9_H_ #include "common/angleutils.h" +#include <GLES2/gl2.h> + namespace gl { class Framebuffer; @@ -19,19 +21,18 @@ class Framebuffer; namespace rx { class Renderer9; -class TextureStorageInterface2D; -class TextureStorageInterfaceCube; +class TextureStorage; -class Blit +class Blit9 { public: - explicit Blit(Renderer9 *renderer); - ~Blit(); + explicit Blit9(Renderer9 *renderer); + ~Blit9(); // Copy from source surface to dest surface. // sourceRect, xoffset, yoffset are in D3D coordinates (0,0 in upper-left) - bool copy(gl::Framebuffer *framebuffer, const RECT &sourceRect, GLenum destFormat, GLint xoffset, GLint yoffset, TextureStorageInterface2D *storage, GLint level); - bool copy(gl::Framebuffer *framebuffer, const RECT &sourceRect, GLenum destFormat, GLint xoffset, GLint yoffset, TextureStorageInterfaceCube *storage, GLenum target, GLint level); + bool copy2D(gl::Framebuffer *framebuffer, const RECT &sourceRect, GLenum destFormat, GLint xoffset, GLint yoffset, TextureStorage *storage, GLint level); + bool copyCube(gl::Framebuffer *framebuffer, const RECT &sourceRect, GLenum destFormat, GLint xoffset, GLint yoffset, TextureStorage *storage, GLenum target, GLint level); // Copy from source surface to dest surface. // sourceRect, xoffset, yoffset are in D3D coordinates (0,0 in upper-left) @@ -87,8 +88,8 @@ class Blit IDirect3DSurface9 *mSavedRenderTarget; IDirect3DSurface9 *mSavedDepthStencil; - DISALLOW_COPY_AND_ASSIGN(Blit); + DISALLOW_COPY_AND_ASSIGN(Blit9); }; } -#endif // LIBGLESV2_BLIT_H_ +#endif // LIBGLESV2_BLIT9_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Buffer9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Buffer9.cpp new file mode 100644 index 0000000000..c02db515a2 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Buffer9.cpp @@ -0,0 +1,121 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Buffer9.cpp Defines the Buffer9 class. + +#include "libGLESv2/renderer/d3d/d3d9/Buffer9.h" +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" +#include "libGLESv2/main.h" + +namespace rx +{ + +Buffer9::Buffer9(rx::Renderer9 *renderer) + : BufferD3D(), + mRenderer(renderer), + mSize(0) +{} + +Buffer9::~Buffer9() +{ + mSize = 0; +} + +Buffer9 *Buffer9::makeBuffer9(BufferImpl *buffer) +{ + ASSERT(HAS_DYNAMIC_TYPE(Buffer9*, buffer)); + return static_cast<Buffer9*>(buffer); +} + +gl::Error Buffer9::setData(const void* data, size_t size, GLenum usage) +{ + if (size > mMemory.size()) + { + if (!mMemory.resize(size)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to resize internal buffer."); + } + } + + mSize = size; + if (data) + { + memcpy(mMemory.data(), data, size); + } + + invalidateStaticData(); + + if (usage == GL_STATIC_DRAW) + { + initializeStaticData(); + } + + return gl::Error(GL_NO_ERROR); +} + +void *Buffer9::getData() +{ + return mMemory.data(); +} + +gl::Error Buffer9::setSubData(const void* data, size_t size, size_t offset) +{ + if (offset + size > mMemory.size()) + { + if (!mMemory.resize(offset + size)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to resize internal buffer."); + } + } + + mSize = std::max(mSize, offset + size); + if (data) + { + memcpy(mMemory.data() + offset, data, size); + } + + invalidateStaticData(); + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Buffer9::copySubData(BufferImpl* source, GLintptr sourceOffset, GLintptr destOffset, GLsizeiptr size) +{ + // Note: this method is currently unreachable + Buffer9* sourceBuffer = makeBuffer9(source); + ASSERT(sourceBuffer); + + memcpy(mMemory.data() + destOffset, sourceBuffer->mMemory.data() + sourceOffset, size); + + invalidateStaticData(); + + return gl::Error(GL_NO_ERROR); +} + +// We do not support buffer mapping in D3D9 +gl::Error Buffer9::map(size_t offset, size_t length, GLbitfield access, GLvoid **mapPtr) +{ + UNREACHABLE(); + return gl::Error(GL_INVALID_OPERATION); +} + +gl::Error Buffer9::unmap() +{ + UNREACHABLE(); + return gl::Error(GL_INVALID_OPERATION); +} + +void Buffer9::markTransformFeedbackUsage() +{ + UNREACHABLE(); +} + +Renderer* Buffer9::getRenderer() +{ + return mRenderer; +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Buffer9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Buffer9.h new file mode 100644 index 0000000000..e78182f905 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Buffer9.h @@ -0,0 +1,52 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Buffer9.h: Defines the rx::Buffer9 class which implements rx::BufferImpl via rx::BufferD3D. + +#ifndef LIBGLESV2_RENDERER_BUFFER9_H_ +#define LIBGLESV2_RENDERER_BUFFER9_H_ + +#include "libGLESv2/renderer/d3d/BufferD3D.h" +#include "libGLESv2/renderer/d3d/MemoryBuffer.h" +#include "libGLESv2/angletypes.h" + +namespace rx +{ +class Renderer9; + +class Buffer9 : public BufferD3D +{ + public: + Buffer9(rx::Renderer9 *renderer); + virtual ~Buffer9(); + + static Buffer9 *makeBuffer9(BufferImpl *buffer); + + // BufferD3D implementation + virtual size_t getSize() const { return mSize; } + virtual bool supportsDirectBinding() const { return false; } + virtual Renderer* getRenderer(); + + // BufferImpl implementation + virtual gl::Error setData(const void* data, size_t size, GLenum usage); + virtual void *getData(); + virtual gl::Error setSubData(const void* data, size_t size, size_t offset); + virtual gl::Error copySubData(BufferImpl* source, GLintptr sourceOffset, GLintptr destOffset, GLsizeiptr size); + virtual gl::Error map(size_t offset, size_t length, GLbitfield access, GLvoid **mapPtr); + virtual gl::Error unmap(); + virtual void markTransformFeedbackUsage(); + + private: + DISALLOW_COPY_AND_ASSIGN(Buffer9); + + rx::Renderer9 *mRenderer; + MemoryBuffer mMemory; + size_t mSize; +}; + +} + +#endif // LIBGLESV2_RENDERER_BUFFER9_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Fence9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Fence9.cpp new file mode 100644 index 0000000000..e352a5f50a --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Fence9.cpp @@ -0,0 +1,72 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Fence9.cpp: Defines the rx::Fence9 class. + +#include "libGLESv2/renderer/d3d/d3d9/Fence9.h" +#include "libGLESv2/renderer/d3d/d3d9/renderer9_utils.h" +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" +#include "libGLESv2/main.h" + +namespace rx +{ + +Fence9::Fence9(rx::Renderer9 *renderer) +{ + mRenderer = renderer; + mQuery = NULL; +} + +Fence9::~Fence9() +{ + SafeRelease(mQuery); +} + +bool Fence9::isSet() const +{ + return mQuery != NULL; +} + +void Fence9::set() +{ + if (!mQuery) + { + mQuery = mRenderer->allocateEventQuery(); + if (!mQuery) + { + return gl::error(GL_OUT_OF_MEMORY); + } + } + + HRESULT result = mQuery->Issue(D3DISSUE_END); + UNUSED_ASSERTION_VARIABLE(result); + ASSERT(SUCCEEDED(result)); +} + +bool Fence9::test(bool flushCommandBuffer) +{ + ASSERT(mQuery); + + DWORD getDataFlags = (flushCommandBuffer ? D3DGETDATA_FLUSH : 0); + HRESULT result = mQuery->GetData(NULL, 0, getDataFlags); + + if (d3d9::isDeviceLostError(result)) + { + mRenderer->notifyDeviceLost(); + return gl::error(GL_OUT_OF_MEMORY, true); + } + + ASSERT(result == S_OK || result == S_FALSE); + + return (result == S_OK); +} + +bool Fence9::hasError() const +{ + return mRenderer->isDeviceLost(); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Fence9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Fence9.h index 9f17641e51..e923a2178c 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Fence9.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Fence9.h @@ -21,11 +21,10 @@ class Fence9 : public FenceImpl explicit Fence9(rx::Renderer9 *renderer); virtual ~Fence9(); - GLboolean isFence(); - void setFence(GLenum condition); - GLboolean testFence(); - void finishFence(); - void getFenceiv(GLenum pname, GLint *params); + bool isSet() const; + void set(); + bool test(bool flushCommandBuffer); + bool hasError() const; private: DISALLOW_COPY_AND_ASSIGN(Fence9); diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Image9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Image9.cpp index cd12d8cc9e..18383fba78 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Image9.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Image9.cpp @@ -1,4 +1,3 @@ -#include "precompiled.h" // // Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be @@ -8,17 +7,17 @@ // Image9.cpp: Implements the rx::Image9 class, which acts as the interface to // the actual underlying surfaces of a Texture. -#include "libGLESv2/renderer/d3d9/Image9.h" - +#include "libGLESv2/renderer/d3d/d3d9/Image9.h" +#include "libGLESv2/renderer/d3d/d3d9/renderer9_utils.h" +#include "libGLESv2/renderer/d3d/d3d9/formatutils9.h" +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" +#include "libGLESv2/renderer/d3d/d3d9/RenderTarget9.h" +#include "libGLESv2/renderer/d3d/d3d9/TextureStorage9.h" #include "libGLESv2/main.h" #include "libGLESv2/Framebuffer.h" +#include "libGLESv2/FramebufferAttachment.h" #include "libGLESv2/Renderbuffer.h" -#include "libGLESv2/renderer/d3d9/Renderer9.h" -#include "libGLESv2/renderer/d3d9/RenderTarget9.h" -#include "libGLESv2/renderer/d3d9/TextureStorage9.h" -#include "libGLESv2/renderer/d3d9/renderer9_utils.h" -#include "libGLESv2/renderer/generatemip.h" namespace rx { @@ -34,10 +33,7 @@ Image9::Image9() Image9::~Image9() { - if (mSurface) - { - mSurface->Release(); - } + SafeRelease(mSurface); } void Image9::generateMip(IDirect3DSurface9 *destSurface, IDirect3DSurface9 *sourceSurface) @@ -54,6 +50,9 @@ void Image9::generateMip(IDirect3DSurface9 *destSurface, IDirect3DSurface9 *sour ASSERT(sourceDesc.Width == 1 || sourceDesc.Width / 2 == destDesc.Width); ASSERT(sourceDesc.Height == 1 || sourceDesc.Height / 2 == destDesc.Height); + const d3d9::D3DFormat &d3dFormatInfo = d3d9::GetD3DFormatInfo(sourceDesc.Format); + ASSERT(d3dFormatInfo.mipGenerationFunction != NULL); + D3DLOCKED_RECT sourceLocked = {0}; result = sourceSurface->LockRect(&sourceLocked, NULL, D3DLOCK_READONLY); ASSERT(SUCCEEDED(result)); @@ -62,37 +61,17 @@ void Image9::generateMip(IDirect3DSurface9 *destSurface, IDirect3DSurface9 *sour result = destSurface->LockRect(&destLocked, NULL, 0); ASSERT(SUCCEEDED(result)); - const unsigned char *sourceData = reinterpret_cast<const unsigned char*>(sourceLocked.pBits); - unsigned char *destData = reinterpret_cast<unsigned char*>(destLocked.pBits); + const uint8_t *sourceData = reinterpret_cast<const uint8_t*>(sourceLocked.pBits); + uint8_t *destData = reinterpret_cast<uint8_t*>(destLocked.pBits); if (sourceData && destData) { - switch (sourceDesc.Format) - { - case D3DFMT_L8: - GenerateMip<L8>(sourceDesc.Width, sourceDesc.Height, sourceData, sourceLocked.Pitch, destData, destLocked.Pitch); - break; - case D3DFMT_A8L8: - GenerateMip<A8L8>(sourceDesc.Width, sourceDesc.Height, sourceData, sourceLocked.Pitch, destData, destLocked.Pitch); - break; - case D3DFMT_A8R8G8B8: - case D3DFMT_X8R8G8B8: - GenerateMip<A8R8G8B8>(sourceDesc.Width, sourceDesc.Height, sourceData, sourceLocked.Pitch, destData, destLocked.Pitch); - break; - case D3DFMT_A16B16G16R16F: - GenerateMip<A16B16G16R16F>(sourceDesc.Width, sourceDesc.Height, sourceData, sourceLocked.Pitch, destData, destLocked.Pitch); - break; - case D3DFMT_A32B32G32R32F: - GenerateMip<A32B32G32R32F>(sourceDesc.Width, sourceDesc.Height, sourceData, sourceLocked.Pitch, destData, destLocked.Pitch); - break; - default: - UNREACHABLE(); - break; - } - - destSurface->UnlockRect(); - sourceSurface->UnlockRect(); + d3dFormatInfo.mipGenerationFunction(sourceDesc.Width, sourceDesc.Height, 1, sourceData, sourceLocked.Pitch, 0, + destData, destLocked.Pitch, 0); } + + destSurface->UnlockRect(); + sourceSurface->UnlockRect(); } Image9 *Image9::makeImage9(Image *img) @@ -117,20 +96,23 @@ void Image9::copyLockableSurfaces(IDirect3DSurface9 *dest, IDirect3DSurface9 *so { D3DLOCKED_RECT sourceLock = {0}; D3DLOCKED_RECT destLock = {0}; - + source->LockRect(&sourceLock, NULL, 0); dest->LockRect(&destLock, NULL, 0); - + if (sourceLock.pBits && destLock.pBits) { D3DSURFACE_DESC desc; source->GetDesc(&desc); - int rows = d3d9::IsCompressedFormat(desc.Format) ? desc.Height / 4 : desc.Height; - int bytes = d3d9::ComputeRowSize(desc.Format, desc.Width); - ASSERT(bytes <= sourceLock.Pitch && bytes <= destLock.Pitch); + const d3d9::D3DFormat &d3dFormatInfo = d3d9::GetD3DFormatInfo(desc.Format); + unsigned int rows = desc.Height / d3dFormatInfo.blockHeight; + + unsigned int bytes = d3d9::ComputeBlockSize(desc.Format, desc.Width, d3dFormatInfo.blockHeight); + ASSERT(bytes <= static_cast<unsigned int>(sourceLock.Pitch) && + bytes <= static_cast<unsigned int>(destLock.Pitch)); - for(int i = 0; i < rows; i++) + for(unsigned int i = 0; i < rows; i++) { memcpy((char*)destLock.pBits + destLock.Pitch * i, (char*)sourceLock.pBits + sourceLock.Pitch * i, bytes); } @@ -141,10 +123,17 @@ void Image9::copyLockableSurfaces(IDirect3DSurface9 *dest, IDirect3DSurface9 *so else UNREACHABLE(); } -bool Image9::redefine(rx::Renderer *renderer, GLint internalformat, GLsizei width, GLsizei height, bool forceRelease) +bool Image9::redefine(rx::Renderer *renderer, GLenum target, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, bool forceRelease) { + // 3D textures are not supported by the D3D9 backend. + ASSERT(depth <= 1); + + // Only 2D and cube texture are supported by the D3D9 backend. + ASSERT(target == GL_TEXTURE_2D || target == GL_TEXTURE_CUBE_MAP); + if (mWidth != width || mHeight != height || + mDepth != depth || mInternalFormat != internalformat || forceRelease) { @@ -152,16 +141,18 @@ bool Image9::redefine(rx::Renderer *renderer, GLint internalformat, GLsizei widt mWidth = width; mHeight = height; + mDepth = depth; mInternalFormat = internalformat; + // compute the d3d format that will be used - mD3DFormat = mRenderer->ConvertTextureInternalFormat(internalformat); - mActualFormat = d3d9_gl::GetEquivalentFormat(mD3DFormat); + const d3d9::TextureFormat &d3d9FormatInfo = d3d9::GetTextureFormatInfo(internalformat); + const d3d9::D3DFormat &d3dFormatInfo = d3d9::GetD3DFormatInfo(d3d9FormatInfo.texFormat); + mD3DFormat = d3d9FormatInfo.texFormat; + mActualFormat = d3dFormatInfo.internalFormat; + mRenderable = (d3d9FormatInfo.renderFormat != D3DFMT_UNKNOWN); - if (mSurface) - { - mSurface->Release(); - mSurface = NULL; - } + SafeRelease(mSurface); + mDirty = (d3d9FormatInfo.dataInitializerFunction != NULL); return true; } @@ -180,14 +171,13 @@ void Image9::createSurface() IDirect3DSurface9 *newSurface = NULL; const D3DPOOL poolToUse = D3DPOOL_SYSTEMMEM; const D3DFORMAT d3dFormat = getD3DFormat(); - ASSERT(d3dFormat != D3DFMT_INTZ); // We should never get here for depth textures if (mWidth != 0 && mHeight != 0) { int levelToFetch = 0; GLsizei requestWidth = mWidth; GLsizei requestHeight = mHeight; - gl::MakeValidSize(true, gl::IsCompressed(mInternalFormat), &requestWidth, &requestHeight, &levelToFetch); + d3d9::MakeValidSize(true, d3dFormat, &requestWidth, &requestHeight, &levelToFetch); IDirect3DDevice9 *device = mRenderer->getDevice(); @@ -202,7 +192,27 @@ void Image9::createSurface() } newTexture->GetSurfaceLevel(levelToFetch, &newSurface); - newTexture->Release(); + SafeRelease(newTexture); + + const d3d9::TextureFormat &d3dFormatInfo = d3d9::GetTextureFormatInfo(mInternalFormat); + if (d3dFormatInfo.dataInitializerFunction != NULL) + { + RECT entireRect; + entireRect.left = 0; + entireRect.right = mWidth; + entireRect.top = 0; + entireRect.bottom = mHeight; + + D3DLOCKED_RECT lockedRect; + result = newSurface->LockRect(&lockedRect, &entireRect, 0); + ASSERT(SUCCEEDED(result)); + + d3dFormatInfo.dataInitializerFunction(mWidth, mHeight, 1, reinterpret_cast<uint8_t*>(lockedRect.pBits), + lockedRect.Pitch, 0); + + result = newSurface->UnlockRect(); + ASSERT(SUCCEEDED(result)); + } } mSurface = newSurface; @@ -232,15 +242,11 @@ void Image9::unlock() if (mSurface) { HRESULT result = mSurface->UnlockRect(); + UNUSED_ASSERTION_VARIABLE(result); ASSERT(SUCCEEDED(result)); } } -bool Image9::isRenderableFormat() const -{ - return TextureStorage9::IsTextureFormatRenderable(getD3DFormat()); -} - D3DFORMAT Image9::getD3DFormat() const { // this should only happen if the image hasn't been redefined first @@ -250,6 +256,13 @@ D3DFORMAT Image9::getD3DFormat() const return mD3DFormat; } +bool Image9::isDirty() const +{ + // Make sure to that this image is marked as dirty even if the staging texture hasn't been created yet + // if initialization is required before use. + return (mSurface || d3d9::GetTextureFormatInfo(mInternalFormat).dataInitializerFunction != NULL) && mDirty; +} + IDirect3DSurface9 *Image9::getSurface() { createSurface(); @@ -257,15 +270,15 @@ IDirect3DSurface9 *Image9::getSurface() return mSurface; } -void Image9::setManagedSurface(TextureStorageInterface2D *storage, int level) +void Image9::setManagedSurface2D(TextureStorage *storage, int level) { - TextureStorage9_2D *storage9 = TextureStorage9_2D::makeTextureStorage9_2D(storage->getStorageInstance()); + TextureStorage9_2D *storage9 = TextureStorage9_2D::makeTextureStorage9_2D(storage); setManagedSurface(storage9->getSurfaceLevel(level, false)); } -void Image9::setManagedSurface(TextureStorageInterfaceCube *storage, int face, int level) +void Image9::setManagedSurfaceCube(TextureStorage *storage, int face, int level) { - TextureStorage9_Cube *storage9 = TextureStorage9_Cube::makeTextureStorage9_Cube(storage->getStorageInstance()); + TextureStorage9_Cube *storage9 = TextureStorage9_Cube::makeTextureStorage9_Cube(storage); setManagedSurface(storage9->getCubeMapSurface(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face, level, false)); } @@ -280,7 +293,7 @@ void Image9::setManagedSurface(IDirect3DSurface9 *surface) if (mSurface) { copyLockableSurfaces(surface, mSurface); - mSurface->Release(); + SafeRelease(mSurface); } mSurface = surface; @@ -288,22 +301,38 @@ void Image9::setManagedSurface(IDirect3DSurface9 *surface) } } -bool Image9::updateSurface(TextureStorageInterface2D *storage, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) +bool Image9::copyToStorage2D(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) { ASSERT(getSurface() != NULL); - TextureStorage9_2D *storage9 = TextureStorage9_2D::makeTextureStorage9_2D(storage->getStorageInstance()); - return updateSurface(storage9->getSurfaceLevel(level, true), xoffset, yoffset, width, height); + TextureStorage9_2D *storage9 = TextureStorage9_2D::makeTextureStorage9_2D(storage); + return copyToSurface(storage9->getSurfaceLevel(level, true), xoffset, yoffset, width, height); } -bool Image9::updateSurface(TextureStorageInterfaceCube *storage, int face, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) +bool Image9::copyToStorageCube(TextureStorage *storage, int face, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) { ASSERT(getSurface() != NULL); - TextureStorage9_Cube *storage9 = TextureStorage9_Cube::makeTextureStorage9_Cube(storage->getStorageInstance()); - return updateSurface(storage9->getCubeMapSurface(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face, level, true), xoffset, yoffset, width, height); + TextureStorage9_Cube *storage9 = TextureStorage9_Cube::makeTextureStorage9_Cube(storage); + return copyToSurface(storage9->getCubeMapSurface(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face, level, true), xoffset, yoffset, width, height); +} + +bool Image9::copyToStorage3D(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth) +{ + // 3D textures are not supported by the D3D9 backend. + UNREACHABLE(); + return false; } -bool Image9::updateSurface(IDirect3DSurface9 *destSurface, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) +bool Image9::copyToStorage2DArray(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height) { + // 2D array textures are not supported by the D3D9 backend. + UNREACHABLE(); + return false; +} + +bool Image9::copyToSurface(IDirect3DSurface9 *destSurface, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) +{ + ASSERT(width > 0 && height > 0); + if (!destSurface) return false; @@ -334,26 +363,36 @@ bool Image9::updateSurface(IDirect3DSurface9 *destSurface, GLint xoffset, GLint copyLockableSurfaces(surf, sourceSurface); result = device->UpdateSurface(surf, &rect, destSurface, &point); ASSERT(SUCCEEDED(result)); - surf->Release(); + SafeRelease(surf); } } else { - // UpdateSurface: source must be SYSTEMMEM, dest must be DEFAULT pools + // UpdateSurface: source must be SYSTEMMEM, dest must be DEFAULT pools HRESULT result = device->UpdateSurface(sourceSurface, &rect, destSurface, &point); + UNUSED_ASSERTION_VARIABLE(result); ASSERT(SUCCEEDED(result)); } } - destSurface->Release(); + SafeRelease(destSurface); return true; } // Store the pixel rectangle designated by xoffset,yoffset,width,height with pixels stored as format/type at input // into the target pixel rectangle. -void Image9::loadData(GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, - GLint unpackAlignment, const void *input) +void Image9::loadData(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, + GLint unpackAlignment, GLenum type, const void *input) { + // 3D textures are not supported by the D3D9 backend. + ASSERT(zoffset == 0 && depth == 1); + + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(mInternalFormat); + GLsizei inputRowPitch = formatInfo.computeRowPitch(type, width, unpackAlignment); + + const d3d9::TextureFormat &d3dFormatInfo = d3d9::GetTextureFormatInfo(mInternalFormat); + ASSERT(d3dFormatInfo.loadFunction != NULL); + RECT lockRect = { xoffset, yoffset, @@ -367,100 +406,32 @@ void Image9::loadData(GLint xoffset, GLint yoffset, GLsizei width, GLsizei heigh return; } - - GLsizei inputPitch = gl::ComputePitch(width, mInternalFormat, unpackAlignment); - - switch (mInternalFormat) - { - case GL_ALPHA8_EXT: -#if defined(__SSE2__) - if (gl::supportsSSE2()) - { - loadAlphaDataToBGRASSE2(width, height, inputPitch, input, locked.Pitch, locked.pBits); - } - else -#endif - { - loadAlphaDataToBGRA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - } - break; - case GL_LUMINANCE8_EXT: - loadLuminanceDataToNativeOrBGRA(width, height, inputPitch, input, locked.Pitch, locked.pBits, getD3DFormat() == D3DFMT_L8); - break; - case GL_ALPHA32F_EXT: - loadAlphaFloatDataToRGBA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_LUMINANCE32F_EXT: - loadLuminanceFloatDataToRGBA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_ALPHA16F_EXT: - loadAlphaHalfFloatDataToRGBA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_LUMINANCE16F_EXT: - loadLuminanceHalfFloatDataToRGBA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_LUMINANCE8_ALPHA8_EXT: - loadLuminanceAlphaDataToNativeOrBGRA(width, height, inputPitch, input, locked.Pitch, locked.pBits, getD3DFormat() == D3DFMT_A8L8); - break; - case GL_LUMINANCE_ALPHA32F_EXT: - loadLuminanceAlphaFloatDataToRGBA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_LUMINANCE_ALPHA16F_EXT: - loadLuminanceAlphaHalfFloatDataToRGBA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_RGB8_OES: - loadRGBUByteDataToBGRX(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_RGB565: - loadRGB565DataToBGRA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_RGBA8_OES: -#if defined(__SSE2__) - if (gl::supportsSSE2()) - { - loadRGBAUByteDataToBGRASSE2(width, height, inputPitch, input, locked.Pitch, locked.pBits); - } - else -#endif - { - loadRGBAUByteDataToBGRA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - } - break; - case GL_RGBA4: - loadRGBA4444DataToBGRA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_RGB5_A1: - loadRGBA5551DataToBGRA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_BGRA8_EXT: - loadBGRADataToBGRA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - // float textures are converted to RGBA, not BGRA, as they're stored that way in D3D - case GL_RGB32F_EXT: - loadRGBFloatDataToRGBA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_RGB16F_EXT: - loadRGBHalfFloatDataToRGBA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_RGBA32F_EXT: - loadRGBAFloatDataToRGBA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - case GL_RGBA16F_EXT: - loadRGBAHalfFloatDataToRGBA(width, height, inputPitch, input, locked.Pitch, locked.pBits); - break; - default: UNREACHABLE(); - } + d3dFormatInfo.loadFunction(width, height, depth, + reinterpret_cast<const uint8_t*>(input), inputRowPitch, 0, + reinterpret_cast<uint8_t*>(locked.pBits), locked.Pitch, 0); unlock(); } -void Image9::loadCompressedData(GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, +void Image9::loadCompressedData(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, const void *input) { - ASSERT(xoffset % 4 == 0); - ASSERT(yoffset % 4 == 0); + // 3D textures are not supported by the D3D9 backend. + ASSERT(zoffset == 0 && depth == 1); + + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(mInternalFormat); + GLsizei inputRowPitch = formatInfo.computeRowPitch(GL_UNSIGNED_BYTE, width, 1); + GLsizei inputDepthPitch = formatInfo.computeDepthPitch(GL_UNSIGNED_BYTE, width, height, 1); + + const d3d9::TextureFormat &d3d9FormatInfo = d3d9::GetTextureFormatInfo(mInternalFormat); + + ASSERT(xoffset % d3d9::GetD3DFormatInfo(d3d9FormatInfo.texFormat).blockWidth == 0); + ASSERT(yoffset % d3d9::GetD3DFormatInfo(d3d9FormatInfo.texFormat).blockHeight == 0); - RECT lockRect = { + ASSERT(d3d9FormatInfo.loadFunction != NULL); + + RECT lockRect = + { xoffset, yoffset, xoffset + width, yoffset + height }; @@ -472,29 +443,28 @@ void Image9::loadCompressedData(GLint xoffset, GLint yoffset, GLsizei width, GLs return; } - GLsizei inputSize = gl::ComputeCompressedSize(width, height, mInternalFormat); - GLsizei inputPitch = gl::ComputeCompressedPitch(width, mInternalFormat); - int rows = inputSize / inputPitch; - for (int i = 0; i < rows; ++i) - { - memcpy((void*)((BYTE*)locked.pBits + i * locked.Pitch), (void*)((BYTE*)input + i * inputPitch), inputPitch); - } + d3d9FormatInfo.loadFunction(width, height, depth, + reinterpret_cast<const uint8_t*>(input), inputRowPitch, inputDepthPitch, + reinterpret_cast<uint8_t*>(locked.pBits), locked.Pitch, 0); unlock(); } // This implements glCopyTex[Sub]Image2D for non-renderable internal texture formats and incomplete textures -void Image9::copy(GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) +void Image9::copy(GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) { + // ES3.0 only behaviour to copy into a 3d texture + ASSERT(zoffset == 0); + RenderTarget9 *renderTarget = NULL; IDirect3DSurface9 *surface = NULL; - gl::Renderbuffer *colorbuffer = source->getColorbuffer(0); + gl::FramebufferAttachment *colorbuffer = source->getColorbuffer(0); if (colorbuffer) { - renderTarget = RenderTarget9::makeRenderTarget9(colorbuffer->getRenderTarget()); + renderTarget = d3d9::GetAttachmentRenderTarget(colorbuffer); } - + if (renderTarget) { surface = renderTarget->getSurface(); @@ -511,13 +481,13 @@ void Image9::copy(GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, IDirect3DSurface9 *renderTargetData = NULL; D3DSURFACE_DESC description; surface->GetDesc(&description); - + HRESULT result = device->CreateOffscreenPlainSurface(description.Width, description.Height, description.Format, D3DPOOL_SYSTEMMEM, &renderTargetData, NULL); if (FAILED(result)) { ERR("Could not create matching destination surface."); - surface->Release(); + SafeRelease(surface); return gl::error(GL_OUT_OF_MEMORY); } @@ -526,8 +496,8 @@ void Image9::copy(GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, if (FAILED(result)) { ERR("GetRenderTargetData unexpectedly failed."); - renderTargetData->Release(); - surface->Release(); + SafeRelease(renderTargetData); + SafeRelease(surface); return gl::error(GL_OUT_OF_MEMORY); } @@ -540,20 +510,20 @@ void Image9::copy(GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, if (FAILED(result)) { ERR("Failed to lock the source surface (rectangle might be invalid)."); - renderTargetData->Release(); - surface->Release(); + SafeRelease(renderTargetData); + SafeRelease(surface); return gl::error(GL_OUT_OF_MEMORY); } D3DLOCKED_RECT destLock = {0}; result = lock(&destLock, &destRect); - + if (FAILED(result)) { ERR("Failed to lock the destination surface (rectangle might be invalid)."); renderTargetData->UnlockRect(); - renderTargetData->Release(); - surface->Release(); + SafeRelease(renderTargetData); + SafeRelease(surface); return gl::error(GL_OUT_OF_MEMORY); } @@ -727,8 +697,8 @@ void Image9::copy(GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, unlock(); renderTargetData->UnlockRect(); - renderTargetData->Release(); - surface->Release(); + SafeRelease(renderTargetData); + SafeRelease(surface); mDirty = true; } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Image9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Image9.h new file mode 100644 index 0000000000..08d8ee3545 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Image9.h @@ -0,0 +1,78 @@ +// +// Copyright (c) 2002-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Image9.h: Defines the rx::Image9 class, which acts as the interface to +// the actual underlying surfaces of a Texture. + +#ifndef LIBGLESV2_RENDERER_IMAGE9_H_ +#define LIBGLESV2_RENDERER_IMAGE9_H_ + +#include "libGLESv2/renderer/d3d/ImageD3D.h" +#include "common/debug.h" + +namespace gl +{ +class Framebuffer; +} + +namespace rx +{ +class Renderer; +class Renderer9; + +class Image9 : public ImageD3D +{ + public: + Image9(); + ~Image9(); + + static Image9 *makeImage9(Image *img); + + static void generateMipmap(Image9 *dest, Image9 *source); + static void generateMip(IDirect3DSurface9 *destSurface, IDirect3DSurface9 *sourceSurface); + static void copyLockableSurfaces(IDirect3DSurface9 *dest, IDirect3DSurface9 *source); + + virtual bool redefine(Renderer *renderer, GLenum target, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, bool forceRelease); + + D3DFORMAT getD3DFormat() const; + + virtual bool isDirty() const; + IDirect3DSurface9 *getSurface(); + + virtual void setManagedSurface2D(TextureStorage *storage, int level); + virtual void setManagedSurfaceCube(TextureStorage *storage, int face, int level); + virtual bool copyToStorage2D(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); + virtual bool copyToStorageCube(TextureStorage *storage, int face, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); + virtual bool copyToStorage3D(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth); + virtual bool copyToStorage2DArray(TextureStorage *storage, int level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height); + + virtual void loadData(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, + GLint unpackAlignment, GLenum type, const void *input); + virtual void loadCompressedData(GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, + const void *input); + + virtual void copy(GLint xoffset, GLint yoffset, GLint zoffset,GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source); + + private: + DISALLOW_COPY_AND_ASSIGN(Image9); + + void createSurface(); + void setManagedSurface(IDirect3DSurface9 *surface); + bool copyToSurface(IDirect3DSurface9 *dest, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); + + HRESULT lock(D3DLOCKED_RECT *lockedRect, const RECT *rect); + void unlock(); + + Renderer9 *mRenderer; + + D3DPOOL mD3DPool; // can only be D3DPOOL_SYSTEMMEM or D3DPOOL_MANAGED since it needs to be lockable. + D3DFORMAT mD3DFormat; + + IDirect3DSurface9 *mSurface; +}; +} + +#endif // LIBGLESV2_RENDERER_IMAGE9_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/IndexBuffer9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/IndexBuffer9.cpp new file mode 100644 index 0000000000..1c51b9e985 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/IndexBuffer9.cpp @@ -0,0 +1,173 @@ +// +// Copyright (c) 2002-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Indexffer9.cpp: Defines the D3D9 IndexBuffer implementation. + +#include "libGLESv2/renderer/d3d/d3d9/IndexBuffer9.h" +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" + +namespace rx +{ + +IndexBuffer9::IndexBuffer9(Renderer9 *const renderer) : mRenderer(renderer) +{ + mIndexBuffer = NULL; + mBufferSize = 0; + mIndexType = 0; + mDynamic = false; +} + +IndexBuffer9::~IndexBuffer9() +{ + SafeRelease(mIndexBuffer); +} + +gl::Error IndexBuffer9::initialize(unsigned int bufferSize, GLenum indexType, bool dynamic) +{ + SafeRelease(mIndexBuffer); + + updateSerial(); + + if (bufferSize > 0) + { + D3DFORMAT format = D3DFMT_UNKNOWN; + if (indexType == GL_UNSIGNED_SHORT || indexType == GL_UNSIGNED_BYTE) + { + format = D3DFMT_INDEX16; + } + else if (indexType == GL_UNSIGNED_INT) + { + ASSERT(mRenderer->getRendererExtensions().elementIndexUint); + format = D3DFMT_INDEX32; + } + else UNREACHABLE(); + + DWORD usageFlags = D3DUSAGE_WRITEONLY; + if (dynamic) + { + usageFlags |= D3DUSAGE_DYNAMIC; + } + + HRESULT result = mRenderer->createIndexBuffer(bufferSize, usageFlags, format, &mIndexBuffer); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to allocate internal index buffer of size, %lu.", bufferSize); + } + } + + mBufferSize = bufferSize; + mIndexType = indexType; + mDynamic = dynamic; + + return gl::Error(GL_NO_ERROR); +} + +IndexBuffer9 *IndexBuffer9::makeIndexBuffer9(IndexBuffer *indexBuffer) +{ + ASSERT(HAS_DYNAMIC_TYPE(IndexBuffer9*, indexBuffer)); + return static_cast<IndexBuffer9*>(indexBuffer); +} + +gl::Error IndexBuffer9::mapBuffer(unsigned int offset, unsigned int size, void** outMappedMemory) +{ + if (!mIndexBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal index buffer is not initialized."); + } + + DWORD lockFlags = mDynamic ? D3DLOCK_NOOVERWRITE : 0; + + void *mapPtr = NULL; + HRESULT result = mIndexBuffer->Lock(offset, size, &mapPtr, lockFlags); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to lock internal index buffer, HRESULT: 0x%08x.", result); + } + + *outMappedMemory = mapPtr; + return gl::Error(GL_NO_ERROR); +} + +gl::Error IndexBuffer9::unmapBuffer() +{ + if (!mIndexBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal index buffer is not initialized."); + } + + HRESULT result = mIndexBuffer->Unlock(); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to unlock internal index buffer, HRESULT: 0x%08x.", result); + } + + return gl::Error(GL_NO_ERROR); +} + +GLenum IndexBuffer9::getIndexType() const +{ + return mIndexType; +} + +unsigned int IndexBuffer9::getBufferSize() const +{ + return mBufferSize; +} + +gl::Error IndexBuffer9::setSize(unsigned int bufferSize, GLenum indexType) +{ + if (bufferSize > mBufferSize || indexType != mIndexType) + { + return initialize(bufferSize, indexType, mDynamic); + } + else + { + return gl::Error(GL_NO_ERROR); + } +} + +gl::Error IndexBuffer9::discard() +{ + if (!mIndexBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal index buffer is not initialized."); + } + + void *dummy; + HRESULT result; + + result = mIndexBuffer->Lock(0, 1, &dummy, D3DLOCK_DISCARD); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to lock internal index buffer, HRESULT: 0x%08x.", result); + } + + result = mIndexBuffer->Unlock(); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to unlock internal index buffer, HRESULT: 0x%08x.", result); + } + + return gl::Error(GL_NO_ERROR); +} + +D3DFORMAT IndexBuffer9::getIndexFormat() const +{ + switch (mIndexType) + { + case GL_UNSIGNED_BYTE: return D3DFMT_INDEX16; + case GL_UNSIGNED_SHORT: return D3DFMT_INDEX16; + case GL_UNSIGNED_INT: return D3DFMT_INDEX32; + default: UNREACHABLE(); return D3DFMT_UNKNOWN; + } +} + +IDirect3DIndexBuffer9 * IndexBuffer9::getBuffer() const +{ + return mIndexBuffer; +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/IndexBuffer9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/IndexBuffer9.h index 6801867532..d0970d6ac5 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/IndexBuffer9.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/IndexBuffer9.h @@ -9,7 +9,7 @@ #ifndef LIBGLESV2_RENDERER_INDEXBUFFER9_H_ #define LIBGLESV2_RENDERER_INDEXBUFFER9_H_ -#include "libGLESv2/renderer/IndexBuffer.h" +#include "libGLESv2/renderer/d3d/IndexBuffer.h" namespace rx { @@ -21,18 +21,18 @@ class IndexBuffer9 : public IndexBuffer explicit IndexBuffer9(Renderer9 *const renderer); virtual ~IndexBuffer9(); - virtual bool initialize(unsigned int bufferSize, GLenum indexType, bool dynamic); + virtual gl::Error initialize(unsigned int bufferSize, GLenum indexType, bool dynamic); static IndexBuffer9 *makeIndexBuffer9(IndexBuffer *indexBuffer); - virtual bool mapBuffer(unsigned int offset, unsigned int size, void** outMappedMemory); - virtual bool unmapBuffer(); + virtual gl::Error mapBuffer(unsigned int offset, unsigned int size, void** outMappedMemory); + virtual gl::Error unmapBuffer(); virtual GLenum getIndexType() const; virtual unsigned int getBufferSize() const; - virtual bool setSize(unsigned int bufferSize, GLenum indexType); + virtual gl::Error setSize(unsigned int bufferSize, GLenum indexType); - virtual bool discard(); + virtual gl::Error discard(); D3DFORMAT getIndexFormat() const; IDirect3DIndexBuffer9 *getBuffer() const; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Query9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Query9.cpp new file mode 100644 index 0000000000..815fc01a9b --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Query9.cpp @@ -0,0 +1,144 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Query9.cpp: Defines the rx::Query9 class which implements rx::QueryImpl. + +#include "libGLESv2/renderer/d3d/d3d9/Query9.h" +#include "libGLESv2/renderer/d3d/d3d9/renderer9_utils.h" +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" +#include "libGLESv2/main.h" + +#include <GLES2/gl2ext.h> + +namespace rx +{ +Query9::Query9(rx::Renderer9 *renderer, GLenum type) + : QueryImpl(type), + mResult(GL_FALSE), + mQueryFinished(false), + mRenderer(renderer), + mQuery(NULL) +{ +} + +Query9::~Query9() +{ + SafeRelease(mQuery); +} + +gl::Error Query9::begin() +{ + if (mQuery == NULL) + { + HRESULT result = mRenderer->getDevice()->CreateQuery(D3DQUERYTYPE_OCCLUSION, &mQuery); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal query creation failed, result: 0x%X.", result); + } + } + + HRESULT result = mQuery->Issue(D3DISSUE_BEGIN); + ASSERT(SUCCEEDED(result)); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to begin internal query, result: 0x%X.", result); + } + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Query9::end() +{ + ASSERT(mQuery); + + HRESULT result = mQuery->Issue(D3DISSUE_END); + ASSERT(SUCCEEDED(result)); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to end internal query, result: 0x%X.", result); + } + + mQueryFinished = false; + mResult = GL_FALSE; + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Query9::getResult(GLuint *params) +{ + while (!mQueryFinished) + { + gl::Error error = testQuery(); + if (error.isError()) + { + return error; + } + + if (!mQueryFinished) + { + Sleep(0); + } + } + + ASSERT(mQueryFinished); + *params = mResult; + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Query9::isResultAvailable(GLuint *available) +{ + gl::Error error = testQuery(); + if (error.isError()) + { + return error; + } + + *available = (mQueryFinished ? GL_TRUE : GL_FALSE); + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Query9::testQuery() +{ + if (!mQueryFinished) + { + ASSERT(mQuery); + + DWORD numPixels = 0; + + HRESULT hres = mQuery->GetData(&numPixels, sizeof(DWORD), D3DGETDATA_FLUSH); + if (hres == S_OK) + { + mQueryFinished = true; + + switch (getType()) + { + case GL_ANY_SAMPLES_PASSED_EXT: + case GL_ANY_SAMPLES_PASSED_CONSERVATIVE_EXT: + mResult = (numPixels > 0) ? GL_TRUE : GL_FALSE; + break; + + default: + UNREACHABLE(); + break; + } + } + else if (d3d9::isDeviceLostError(hres)) + { + mRenderer->notifyDeviceLost(); + return gl::Error(GL_OUT_OF_MEMORY, "Failed to test get query result, device is lost."); + } + else if (mRenderer->testDeviceLost(true)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to test get query result, device is lost."); + } + } + + return gl::Error(GL_NO_ERROR); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Query9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Query9.h index 47eef89336..513e0ba6fd 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Query9.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Query9.h @@ -21,15 +21,18 @@ class Query9 : public QueryImpl Query9(rx::Renderer9 *renderer, GLenum type); virtual ~Query9(); - void begin(); - void end(); - GLuint getResult(); - GLboolean isResultAvailable(); + virtual gl::Error begin(); + virtual gl::Error end(); + virtual gl::Error getResult(GLuint *params); + virtual gl::Error isResultAvailable(GLuint *available); private: DISALLOW_COPY_AND_ASSIGN(Query9); - GLboolean testQuery(); + gl::Error testQuery(); + + GLuint mResult; + bool mQueryFinished; rx::Renderer9 *mRenderer; IDirect3DQuery9 *mQuery; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/RenderTarget9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/RenderTarget9.cpp new file mode 100644 index 0000000000..13321ac8cd --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/RenderTarget9.cpp @@ -0,0 +1,133 @@ +// +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// RenderTarget9.cpp: Implements a D3D9-specific wrapper for IDirect3DSurface9 +// pointers retained by renderbuffers. + +#include "libGLESv2/renderer/d3d/d3d9/RenderTarget9.h" +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" +#include "libGLESv2/renderer/d3d/d3d9/renderer9_utils.h" +#include "libGLESv2/renderer/d3d/d3d9/formatutils9.h" +#include "libGLESv2/main.h" + +namespace rx +{ + +// TODO: AddRef the incoming surface to take ownership instead of expecting that its ref is being given. +RenderTarget9::RenderTarget9(Renderer *renderer, IDirect3DSurface9 *surface) +{ + mRenderer = Renderer9::makeRenderer9(renderer); + mRenderTarget = surface; + + if (mRenderTarget) + { + D3DSURFACE_DESC description; + mRenderTarget->GetDesc(&description); + + mWidth = description.Width; + mHeight = description.Height; + mDepth = 1; + + const d3d9::D3DFormat &d3dFormatInfo = d3d9::GetD3DFormatInfo(description.Format); + mInternalFormat = d3dFormatInfo.internalFormat; + mActualFormat = d3dFormatInfo.internalFormat; + mSamples = d3d9_gl::GetSamplesCount(description.MultiSampleType); + } +} + +RenderTarget9::RenderTarget9(Renderer *renderer, GLsizei width, GLsizei height, GLenum internalFormat, GLsizei samples) +{ + mRenderer = Renderer9::makeRenderer9(renderer); + mRenderTarget = NULL; + + const d3d9::TextureFormat &d3d9FormatInfo = d3d9::GetTextureFormatInfo(internalFormat); + const d3d9::D3DFormat &d3dFormatInfo = d3d9::GetD3DFormatInfo(d3d9FormatInfo.renderFormat); + + const gl::TextureCaps &textureCaps = mRenderer->getRendererTextureCaps().get(internalFormat); + GLuint supportedSamples = textureCaps.getNearestSamples(samples); + + HRESULT result = D3DERR_INVALIDCALL; + + if (width > 0 && height > 0) + { + IDirect3DDevice9 *device = mRenderer->getDevice(); + + bool requiresInitialization = false; + + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(internalFormat); + if (formatInfo.depthBits > 0 || formatInfo.stencilBits > 0) + { + result = device->CreateDepthStencilSurface(width, height, d3d9FormatInfo.renderFormat, + gl_d3d9::GetMultisampleType(supportedSamples), + 0, FALSE, &mRenderTarget, NULL); + } + else + { + requiresInitialization = (d3d9FormatInfo.dataInitializerFunction != NULL); + result = device->CreateRenderTarget(width, height, d3d9FormatInfo.renderFormat, + gl_d3d9::GetMultisampleType(supportedSamples), + 0, FALSE, &mRenderTarget, NULL); + } + + if (result == D3DERR_OUTOFVIDEOMEMORY || result == E_OUTOFMEMORY) + { + gl::error(GL_OUT_OF_MEMORY); + + return; + } + + ASSERT(SUCCEEDED(result)); + + if (requiresInitialization) + { + // This format requires that the data be initialized before the render target can be used + // Unfortunately this requires a Get call on the d3d device but it is far better than having + // to mark the render target as lockable and copy data to the gpu. + IDirect3DSurface9 *prevRenderTarget = NULL; + device->GetRenderTarget(0, &prevRenderTarget); + device->SetRenderTarget(0, mRenderTarget); + device->Clear(0, NULL, D3DCLEAR_TARGET, D3DCOLOR_RGBA(0, 0, 0, 255), 0.0f, 0); + device->SetRenderTarget(0, prevRenderTarget); + } + } + + mWidth = width; + mHeight = height; + mDepth = 1; + mInternalFormat = internalFormat; + mSamples = supportedSamples; + mActualFormat = d3dFormatInfo.internalFormat; +} + +RenderTarget9::~RenderTarget9() +{ + SafeRelease(mRenderTarget); +} + +RenderTarget9 *RenderTarget9::makeRenderTarget9(RenderTarget *target) +{ + ASSERT(HAS_DYNAMIC_TYPE(rx::RenderTarget9*, target)); + return static_cast<rx::RenderTarget9*>(target); +} + +void RenderTarget9::invalidate(GLint x, GLint y, GLsizei width, GLsizei height) +{ + // Currently a no-op +} + +IDirect3DSurface9 *RenderTarget9::getSurface() +{ + // Caller is responsible for releasing the returned surface reference. + // TODO: remove the AddRef to match RenderTarget11 + if (mRenderTarget) + { + mRenderTarget->AddRef(); + } + + return mRenderTarget; +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/RenderTarget9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/RenderTarget9.h index faf8ad1c6d..68d7adb49e 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/RenderTarget9.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/RenderTarget9.h @@ -21,10 +21,13 @@ class RenderTarget9 : public RenderTarget { public: RenderTarget9(Renderer *renderer, IDirect3DSurface9 *surface); - RenderTarget9(Renderer *renderer, GLsizei width, GLsizei height, GLenum format, GLsizei samples); + RenderTarget9(Renderer *renderer, GLsizei width, GLsizei height, GLenum internalFormat, GLsizei samples); virtual ~RenderTarget9(); static RenderTarget9 *makeRenderTarget9(RenderTarget *renderTarget); + + virtual void invalidate(GLint x, GLint y, GLsizei width, GLsizei height); + IDirect3DSurface9 *getSurface(); private: diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Renderer9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Renderer9.cpp index 97a10d64bf..d63f9b8582 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Renderer9.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Renderer9.cpp @@ -1,36 +1,45 @@ -#include "precompiled.h" // -// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // // Renderer9.cpp: Implements a back-end specific class for the D3D9 renderer. +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" +#include "libGLESv2/renderer/d3d/d3d9/renderer9_utils.h" +#include "libGLESv2/renderer/d3d/d3d9/formatutils9.h" +#include "libGLESv2/renderer/d3d/d3d9/ShaderExecutable9.h" +#include "libGLESv2/renderer/d3d/d3d9/SwapChain9.h" +#include "libGLESv2/renderer/d3d/d3d9/TextureStorage9.h" +#include "libGLESv2/renderer/d3d/d3d9/Image9.h" +#include "libGLESv2/renderer/d3d/d3d9/Blit9.h" +#include "libGLESv2/renderer/d3d/d3d9/RenderTarget9.h" +#include "libGLESv2/renderer/d3d/d3d9/VertexBuffer9.h" +#include "libGLESv2/renderer/d3d/d3d9/IndexBuffer9.h" +#include "libGLESv2/renderer/d3d/d3d9/Buffer9.h" +#include "libGLESv2/renderer/d3d/d3d9/Query9.h" +#include "libGLESv2/renderer/d3d/d3d9/Fence9.h" +#include "libGLESv2/renderer/d3d/d3d9/VertexArray9.h" +#include "libGLESv2/renderer/d3d/IndexDataManager.h" +#include "libGLESv2/renderer/d3d/ProgramD3D.h" +#include "libGLESv2/renderer/d3d/ShaderD3D.h" +#include "libGLESv2/renderer/d3d/TextureD3D.h" +#include "libGLESv2/renderer/d3d/TransformFeedbackD3D.h" #include "libGLESv2/main.h" #include "libGLESv2/Buffer.h" #include "libGLESv2/Texture.h" #include "libGLESv2/Framebuffer.h" +#include "libGLESv2/FramebufferAttachment.h" #include "libGLESv2/Renderbuffer.h" #include "libGLESv2/ProgramBinary.h" -#include "libGLESv2/renderer/IndexDataManager.h" -#include "libGLESv2/renderer/d3d9/Renderer9.h" -#include "libGLESv2/renderer/d3d9/renderer9_utils.h" -#include "libGLESv2/renderer/d3d9/ShaderExecutable9.h" -#include "libGLESv2/renderer/d3d9/SwapChain9.h" -#include "libGLESv2/renderer/d3d9/TextureStorage9.h" -#include "libGLESv2/renderer/d3d9/Image9.h" -#include "libGLESv2/renderer/d3d9/Blit.h" -#include "libGLESv2/renderer/d3d9/RenderTarget9.h" -#include "libGLESv2/renderer/d3d9/VertexBuffer9.h" -#include "libGLESv2/renderer/d3d9/IndexBuffer9.h" -#include "libGLESv2/renderer/d3d9/BufferStorage9.h" -#include "libGLESv2/renderer/d3d9/Query9.h" -#include "libGLESv2/renderer/d3d9/Fence9.h" +#include "libGLESv2/angletypes.h" #include "libEGL/Display.h" -#include "third_party/trace_event/trace_event.h" +#include "common/utilities.h" + +#include <sstream> // Can also be enabled by defining FORCE_REF_RAST in the project's predefined macros #define REF_RAST 0 @@ -43,6 +52,13 @@ #define ANGLE_ENABLE_D3D9EX 1 #endif // !defined(ANGLE_ENABLE_D3D9EX) +#if !defined(ANGLE_COMPILE_OPTIMIZATION_LEVEL) +#define ANGLE_COMPILE_OPTIMIZATION_LEVEL D3DCOMPILE_OPTIMIZATION_LEVEL3 +#endif + +const D3DFORMAT D3DFMT_INTZ = ((D3DFORMAT)(MAKEFOURCC('I','N','T','Z'))); +const D3DFORMAT D3DFMT_NULL = ((D3DFORMAT)(MAKEFOURCC('N','U','L','L'))); + namespace rx { static const D3DFORMAT RenderTargetFormats[] = @@ -80,7 +96,9 @@ enum MAX_TEXTURE_IMAGE_UNITS_VTF_SM3 = 4 }; -Renderer9::Renderer9(egl::Display *display, HDC hDc, bool softwareDevice) : Renderer(display), mDc(hDc), mSoftwareDevice(softwareDevice) +Renderer9::Renderer9(egl::Display *display, EGLNativeDisplayType hDc, EGLint requestedDisplay) + : Renderer(display), + mDc(hDc) { mD3d9Module = NULL; @@ -101,13 +119,12 @@ Renderer9::Renderer9(egl::Display *display, HDC hDc, bool softwareDevice) : Rend mDeviceLost = false; - mMaxSupportedSamples = 0; - mMaskedClearSavedState = NULL; mVertexDataManager = NULL; mIndexDataManager = NULL; mLineLoopIB = NULL; + mCountingIB = NULL; mMaxNullColorbufferLRU = 0; for (int i = 0; i < NUM_NULL_COLORBUFFER_CACHE_ENTRIES; i++) @@ -117,6 +134,10 @@ Renderer9::Renderer9(egl::Display *display, HDC hDc, bool softwareDevice) : Rend mNullColorbufferCache[i].height = 0; mNullColorbufferCache[i].buffer = NULL; } + + mAppliedVertexShader = NULL; + mAppliedPixelShader = NULL; + mAppliedProgramSerial = 0; } Renderer9::~Renderer9() @@ -130,11 +151,12 @@ Renderer9::~Renderer9() } } - deinitialize(); + release(); } -void Renderer9::deinitialize() +void Renderer9::release() { + releaseShaderCompiler(); releaseDeviceResources(); SafeRelease(mDevice); @@ -142,22 +164,15 @@ void Renderer9::deinitialize() SafeRelease(mD3d9); SafeRelease(mD3d9Ex); + mCompiler.release(); + if (mDeviceWindow) { DestroyWindow(mDeviceWindow); mDeviceWindow = NULL; } - if (mD3d9Module) - { - mD3d9Module = NULL; - } - - while (!mMultiSampleSupport.empty()) - { - delete [] mMultiSampleSupport.begin()->second; - mMultiSampleSupport.erase(mMultiSampleSupport.begin()); - } + mD3d9Module = NULL; } Renderer9 *Renderer9::makeRenderer9(Renderer *renderer) @@ -168,21 +183,12 @@ Renderer9 *Renderer9::makeRenderer9(Renderer *renderer) EGLint Renderer9::initialize() { - if (!initializeCompiler()) + if (!mCompiler.initialize()) { return EGL_NOT_INITIALIZED; } - if (mSoftwareDevice) - { - TRACE_EVENT0("gpu", "GetModuleHandle_swiftshader"); - mD3d9Module = GetModuleHandle(TEXT("swiftshader_d3d9.dll")); - } - else - { - TRACE_EVENT0("gpu", "GetModuleHandle_d3d9"); - mD3d9Module = GetModuleHandle(TEXT("d3d9.dll")); - } + mD3d9Module = GetModuleHandle(TEXT("d3d9.dll")); if (mD3d9Module == NULL) { @@ -198,14 +204,12 @@ EGLint Renderer9::initialize() // desktop. Direct3D9Ex is available in Windows Vista and later if suitable drivers are available. if (ANGLE_ENABLE_D3D9EX && Direct3DCreate9ExPtr && SUCCEEDED(Direct3DCreate9ExPtr(D3D_SDK_VERSION, &mD3d9Ex))) { - TRACE_EVENT0("gpu", "D3d9Ex_QueryInterface"); ASSERT(mD3d9Ex); mD3d9Ex->QueryInterface(IID_IDirect3D9, reinterpret_cast<void**>(&mD3d9)); ASSERT(mD3d9); } else { - TRACE_EVENT0("gpu", "Direct3DCreate9"); mD3d9 = Direct3DCreate9(D3D_SDK_VERSION); } @@ -224,7 +228,6 @@ EGLint Renderer9::initialize() // Give up on getting device caps after about one second. { - TRACE_EVENT0("gpu", "GetDeviceCaps"); for (int i = 0; i < 10; ++i) { result = mD3d9->GetDeviceCaps(mAdapter, mDeviceType, &mDeviceCaps); @@ -251,7 +254,7 @@ EGLint Renderer9::initialize() } // When DirectX9 is running with an older DirectX8 driver, a StretchRect from a regular texture to a render target texture is not supported. - // This is required by Texture2D::convertToRenderTarget. + // This is required by Texture2D::ensureRenderTarget. if ((mDeviceCaps.DevCaps2 & D3DDEVCAPS2_CAN_STRETCHRECT_FROM_TEXTURES) == 0) { ERR("Renderer does not support stretctrect from textures!\n"); @@ -259,19 +262,9 @@ EGLint Renderer9::initialize() } { - TRACE_EVENT0("gpu", "GetAdapterIdentifier"); mD3d9->GetAdapterIdentifier(mAdapter, 0, &mAdapterIdentifier); } - // ATI cards on XP have problems with non-power-of-two textures. - mSupportsNonPower2Textures = !(mDeviceCaps.TextureCaps & D3DPTEXTURECAPS_POW2) && - !(mDeviceCaps.TextureCaps & D3DPTEXTURECAPS_CUBEMAP_POW2) && - !(mDeviceCaps.TextureCaps & D3DPTEXTURECAPS_NONPOW2CONDITIONAL) && - !(getComparableOSVersion() < versionWindowsVista && mAdapterIdentifier.VendorId == VENDOR_ID_AMD); - - // Must support a minimum of 2:1 anisotropy for max anisotropy to be considered supported, per the spec - mSupportsTextureFilterAnisotropy = ((mDeviceCaps.RasterCaps & D3DPRASTERCAPS_ANISOTROPY) && (mDeviceCaps.MaxAnisotropy >= 2)); - mMinSwapInterval = 4; mMaxSwapInterval = 0; @@ -301,61 +294,21 @@ EGLint Renderer9::initialize() mMaxSwapInterval = std::max(mMaxSwapInterval, 4); } - int max = 0; - { - TRACE_EVENT0("gpu", "getMultiSampleSupport"); - for (unsigned int i = 0; i < ArraySize(RenderTargetFormats); ++i) - { - bool *multisampleArray = new bool[D3DMULTISAMPLE_16_SAMPLES + 1]; - getMultiSampleSupport(RenderTargetFormats[i], multisampleArray); - mMultiSampleSupport[RenderTargetFormats[i]] = multisampleArray; - - for (int j = D3DMULTISAMPLE_16_SAMPLES; j >= 0; --j) - { - if (multisampleArray[j] && j != D3DMULTISAMPLE_NONMASKABLE && j > max) - { - max = j; - } - } - } - } - - { - TRACE_EVENT0("gpu", "getMultiSampleSupport2"); - for (unsigned int i = 0; i < ArraySize(DepthStencilFormats); ++i) - { - if (DepthStencilFormats[i] == D3DFMT_UNKNOWN) - continue; - - bool *multisampleArray = new bool[D3DMULTISAMPLE_16_SAMPLES + 1]; - getMultiSampleSupport(DepthStencilFormats[i], multisampleArray); - mMultiSampleSupport[DepthStencilFormats[i]] = multisampleArray; - - for (int j = D3DMULTISAMPLE_16_SAMPLES; j >= 0; --j) - { - if (multisampleArray[j] && j != D3DMULTISAMPLE_NONMASKABLE && j > max) - { - max = j; - } - } - } - } - - mMaxSupportedSamples = max; - static const TCHAR windowName[] = TEXT("AngleHiddenWindow"); static const TCHAR className[] = TEXT("STATIC"); { - TRACE_EVENT0("gpu", "CreateWindowEx"); mDeviceWindow = CreateWindowEx(WS_EX_NOACTIVATE, className, windowName, WS_DISABLED | WS_POPUP, 0, 0, 1, 1, HWND_MESSAGE, NULL, GetModuleHandle(NULL), NULL); } D3DPRESENT_PARAMETERS presentParameters = getDefaultPresentParameters(); DWORD behaviorFlags = D3DCREATE_FPU_PRESERVE | D3DCREATE_NOWINDOWCHANGES; + static wchar_t *qt_d3dcreate_multihreaded_var = _wgetenv(L"QT_D3DCREATE_MULTITHREADED"); + if (qt_d3dcreate_multihreaded_var && wcsstr(qt_d3dcreate_multihreaded_var, L"1")) + behaviorFlags |= D3DCREATE_MULTITHREADED; + { - TRACE_EVENT0("gpu", "D3d9_CreateDevice"); result = mD3d9->CreateDevice(mAdapter, mDeviceType, mDeviceWindow, behaviorFlags | D3DCREATE_HARDWARE_VERTEXPROCESSING | D3DCREATE_PUREDEVICE, &presentParameters, &mDevice); } if (result == D3DERR_OUTOFVIDEOMEMORY || result == E_OUTOFMEMORY || result == D3DERR_DEVICELOST) @@ -365,7 +318,6 @@ EGLint Renderer9::initialize() if (FAILED(result)) { - TRACE_EVENT0("gpu", "D3d9_CreateDevice2"); result = mD3d9->CreateDevice(mAdapter, mDeviceType, mDeviceWindow, behaviorFlags | D3DCREATE_SOFTWARE_VERTEXPROCESSING, &presentParameters, &mDevice); if (FAILED(result)) @@ -377,47 +329,15 @@ EGLint Renderer9::initialize() if (mD3d9Ex) { - TRACE_EVENT0("gpu", "mDevice_QueryInterface"); - result = mDevice->QueryInterface(IID_IDirect3DDevice9Ex, (void**) &mDeviceEx); + result = mDevice->QueryInterface(IID_IDirect3DDevice9Ex, (void**)&mDeviceEx); ASSERT(SUCCEEDED(result)); } { - TRACE_EVENT0("gpu", "ShaderCache initialize"); mVertexShaderCache.initialize(mDevice); mPixelShaderCache.initialize(mDevice); } - // Check occlusion query support - IDirect3DQuery9 *occlusionQuery = NULL; - { - TRACE_EVENT0("gpu", "device_CreateQuery"); - if (SUCCEEDED(mDevice->CreateQuery(D3DQUERYTYPE_OCCLUSION, &occlusionQuery)) && occlusionQuery) - { - occlusionQuery->Release(); - mOcclusionQuerySupport = true; - } - else - { - mOcclusionQuerySupport = false; - } - } - - // Check event query support - IDirect3DQuery9 *eventQuery = NULL; - { - TRACE_EVENT0("gpu", "device_CreateQuery2"); - if (SUCCEEDED(mDevice->CreateQuery(D3DQUERYTYPE_EVENT, &eventQuery)) && eventQuery) - { - eventQuery->Release(); - mEventQuerySupport = true; - } - else - { - mEventQuerySupport = false; - } - } - D3DDISPLAYMODE currentDisplayMode; mD3d9->GetAdapterDisplayMode(mAdapter, ¤tDisplayMode); @@ -428,70 +348,6 @@ EGLint Renderer9::initialize() SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, D3DUSAGE_QUERY_VERTEXTEXTURE, D3DRTYPE_TEXTURE, D3DFMT_R16F)); - // Check depth texture support - // we use INTZ for depth textures in Direct3D9 - // we also want NULL texture support to ensure the we can make depth-only FBOs - // see http://aras-p.info/texts/D3D9GPUHacks.html - mDepthTextureSupport = SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, - D3DUSAGE_DEPTHSTENCIL, D3DRTYPE_TEXTURE, D3DFMT_INTZ)) && - SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, - D3DUSAGE_RENDERTARGET, D3DRTYPE_SURFACE, D3DFMT_NULL)); - - // Check 32 bit floating point texture support - mFloat32FilterSupport = SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, D3DUSAGE_QUERY_FILTER, - D3DRTYPE_TEXTURE, D3DFMT_A32B32G32R32F)) && - SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, D3DUSAGE_QUERY_FILTER, - D3DRTYPE_CUBETEXTURE, D3DFMT_A32B32G32R32F)); - - mFloat32RenderSupport = SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, D3DUSAGE_RENDERTARGET, - D3DRTYPE_TEXTURE, D3DFMT_A32B32G32R32F)) && - SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, D3DUSAGE_RENDERTARGET, - D3DRTYPE_CUBETEXTURE, D3DFMT_A32B32G32R32F)); - - if (!mFloat32FilterSupport && !mFloat32RenderSupport) - { - mFloat32TextureSupport = SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, 0, - D3DRTYPE_TEXTURE, D3DFMT_A32B32G32R32F)) && - SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, 0, - D3DRTYPE_CUBETEXTURE, D3DFMT_A32B32G32R32F)); - } - else - { - mFloat32TextureSupport = true; - } - - // Check 16 bit floating point texture support - mFloat16FilterSupport = SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, D3DUSAGE_QUERY_FILTER, - D3DRTYPE_TEXTURE, D3DFMT_A16B16G16R16F)) && - SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, D3DUSAGE_QUERY_FILTER, - D3DRTYPE_CUBETEXTURE, D3DFMT_A16B16G16R16F)); - - mFloat16RenderSupport = SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, D3DUSAGE_RENDERTARGET, - D3DRTYPE_TEXTURE, D3DFMT_A16B16G16R16F)) && - SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, D3DUSAGE_RENDERTARGET, - D3DRTYPE_CUBETEXTURE, D3DFMT_A16B16G16R16F)); - - if (!mFloat16FilterSupport && !mFloat16RenderSupport) - { - mFloat16TextureSupport = SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, 0, - D3DRTYPE_TEXTURE, D3DFMT_A16B16G16R16F)) && - SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, 0, - D3DRTYPE_CUBETEXTURE, D3DFMT_A16B16G16R16F)); - } - else - { - mFloat16TextureSupport = true; - } - - // Check DXT texture support - mDXT1TextureSupport = SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, 0, D3DRTYPE_TEXTURE, D3DFMT_DXT1)); - mDXT3TextureSupport = SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, 0, D3DRTYPE_TEXTURE, D3DFMT_DXT3)); - mDXT5TextureSupport = SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, 0, D3DRTYPE_TEXTURE, D3DFMT_DXT5)); - - // Check luminance[alpha] texture support - mLuminanceTextureSupport = SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, 0, D3DRTYPE_TEXTURE, D3DFMT_L8)); - mLuminanceAlphaTextureSupport = SUCCEEDED(mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, 0, D3DRTYPE_TEXTURE, D3DFMT_A8L8)); - initializeDevice(); return EGL_SUCCESS; @@ -515,12 +371,23 @@ void Renderer9::initializeDevice() mDevice->SetRenderState(D3DRS_POINTSIZE_MAX, 0x3F800000); // 1.0f } + const gl::Caps &rendererCaps = getRendererCaps(); + + mForceSetVertexSamplerStates.resize(rendererCaps.maxVertexTextureImageUnits); + mCurVertexSamplerStates.resize(rendererCaps.maxVertexTextureImageUnits); + + mForceSetPixelSamplerStates.resize(rendererCaps.maxTextureImageUnits); + mCurPixelSamplerStates.resize(rendererCaps.maxTextureImageUnits); + + mCurVertexTextureSerials.resize(rendererCaps.maxVertexTextureImageUnits); + mCurPixelTextureSerials.resize(rendererCaps.maxTextureImageUnits); + markAllStateDirty(); mSceneStarted = false; ASSERT(!mBlit && !mVertexDataManager && !mIndexDataManager); - mBlit = new Blit(this); + mBlit = new Blit9(this); mVertexDataManager = new rx::VertexDataManager(this); mIndexDataManager = new rx::IndexDataManager(this); } @@ -559,39 +426,24 @@ int Renderer9::generateConfigs(ConfigDesc **configDescList) for (unsigned int formatIndex = 0; formatIndex < numRenderFormats; formatIndex++) { - D3DFORMAT renderTargetFormat = RenderTargetFormats[formatIndex]; - - HRESULT result = mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, D3DUSAGE_RENDERTARGET, D3DRTYPE_SURFACE, renderTargetFormat); - - if (SUCCEEDED(result)) + const d3d9::D3DFormat &renderTargetFormatInfo = d3d9::GetD3DFormatInfo(RenderTargetFormats[formatIndex]); + const gl::TextureCaps &renderTargetFormatCaps = getRendererTextureCaps().get(renderTargetFormatInfo.internalFormat); + if (renderTargetFormatCaps.renderable) { for (unsigned int depthStencilIndex = 0; depthStencilIndex < numDepthFormats; depthStencilIndex++) { - D3DFORMAT depthStencilFormat = DepthStencilFormats[depthStencilIndex]; - HRESULT result = D3D_OK; - - if(depthStencilFormat != D3DFMT_UNKNOWN) + const d3d9::D3DFormat &depthStencilFormatInfo = d3d9::GetD3DFormatInfo(DepthStencilFormats[depthStencilIndex]); + const gl::TextureCaps &depthStencilFormatCaps = getRendererTextureCaps().get(depthStencilFormatInfo.internalFormat); + if (depthStencilFormatCaps.renderable || DepthStencilFormats[depthStencilIndex] == D3DFMT_UNKNOWN) { - result = mD3d9->CheckDeviceFormat(mAdapter, mDeviceType, currentDisplayMode.Format, D3DUSAGE_DEPTHSTENCIL, D3DRTYPE_SURFACE, depthStencilFormat); - } - - if (SUCCEEDED(result)) - { - if(depthStencilFormat != D3DFMT_UNKNOWN) - { - result = mD3d9->CheckDepthStencilMatch(mAdapter, mDeviceType, currentDisplayMode.Format, renderTargetFormat, depthStencilFormat); - } - - if (SUCCEEDED(result)) - { - ConfigDesc newConfig; - newConfig.renderTargetFormat = d3d9_gl::ConvertBackBufferFormat(renderTargetFormat); - newConfig.depthStencilFormat = d3d9_gl::ConvertDepthStencilFormat(depthStencilFormat); - newConfig.multiSample = 0; // FIXME: enumerate multi-sampling - newConfig.fastConfig = (currentDisplayMode.Format == renderTargetFormat); - - (*configDescList)[numConfigs++] = newConfig; - } + ConfigDesc newConfig; + newConfig.renderTargetFormat = renderTargetFormatInfo.internalFormat; + newConfig.depthStencilFormat = depthStencilFormatInfo.internalFormat; + newConfig.multiSample = 0; // FIXME: enumerate multi-sampling + newConfig.fastConfig = (currentDisplayMode.Format == RenderTargetFormats[formatIndex]); + newConfig.es3Capable = false; + + (*configDescList)[numConfigs++] = newConfig; } } } @@ -681,6 +533,7 @@ IDirect3DQuery9* Renderer9::allocateEventQuery() if (mEventQueryPool.empty()) { HRESULT result = mDevice->CreateQuery(D3DQUERYTYPE_EVENT, &query); + UNUSED_ASSERTION_VARIABLE(result); ASSERT(SUCCEEDED(result)); } else @@ -696,7 +549,7 @@ void Renderer9::freeEventQuery(IDirect3DQuery9* query) { if (mEventQueryPool.size() > 1000) { - query->Release(); + SafeRelease(query); } else { @@ -736,9 +589,14 @@ IndexBuffer *Renderer9::createIndexBuffer() return new IndexBuffer9(this); } -BufferStorage *Renderer9::createBufferStorage() +BufferImpl *Renderer9::createBuffer() +{ + return new Buffer9(this); +} + +VertexArrayImpl *Renderer9::createVertexArray() { - return new BufferStorage9(); + return new VertexArray9(this); } QueryImpl *Renderer9::createQuery(GLenum type) @@ -751,10 +609,36 @@ FenceImpl *Renderer9::createFence() return new Fence9(this); } -void Renderer9::setSamplerState(gl::SamplerType type, int index, const gl::SamplerState &samplerState) +TransformFeedbackImpl* Renderer9::createTransformFeedback() +{ + return new TransformFeedbackD3D(); +} + +bool Renderer9::supportsFastCopyBufferToTexture(GLenum internalFormat) const +{ + // Pixel buffer objects are not supported in D3D9, since D3D9 is ES2-only and PBOs are ES3. + return false; +} + +bool Renderer9::fastCopyBufferToTexture(const gl::PixelUnpackState &unpack, unsigned int offset, RenderTarget *destRenderTarget, + GLenum destinationFormat, GLenum sourcePixelsType, const gl::Box &destArea) +{ + // Pixel buffer objects are not supported in D3D9, since D3D9 is ES2-only and PBOs are ES3. + UNREACHABLE(); + return false; +} + +gl::Error Renderer9::generateSwizzle(gl::Texture *texture) { - bool *forceSetSamplers = (type == gl::SAMPLER_PIXEL) ? mForceSetPixelSamplerStates : mForceSetVertexSamplerStates; - gl::SamplerState *appliedSamplers = (type == gl::SAMPLER_PIXEL) ? mCurPixelSamplerStates: mCurVertexSamplerStates; + // Swizzled textures are not available in ES2 or D3D9 + UNREACHABLE(); + return gl::Error(GL_INVALID_OPERATION); +} + +gl::Error Renderer9::setSamplerState(gl::SamplerType type, int index, const gl::SamplerState &samplerState) +{ + std::vector<bool> &forceSetSamplers = (type == gl::SAMPLER_PIXEL) ? mForceSetPixelSamplerStates : mForceSetVertexSamplerStates; + std::vector<gl::SamplerState> &appliedSamplers = (type == gl::SAMPLER_PIXEL) ? mCurPixelSamplerStates: mCurVertexSamplerStates; if (forceSetSamplers[index] || memcmp(&samplerState, &appliedSamplers[index], sizeof(gl::SamplerState)) != 0) { @@ -769,8 +653,8 @@ void Renderer9::setSamplerState(gl::SamplerType type, int index, const gl::Sampl gl_d3d9::ConvertMinFilter(samplerState.minFilter, &d3dMinFilter, &d3dMipFilter, samplerState.maxAnisotropy); mDevice->SetSamplerState(d3dSampler, D3DSAMP_MINFILTER, d3dMinFilter); mDevice->SetSamplerState(d3dSampler, D3DSAMP_MIPFILTER, d3dMipFilter); - mDevice->SetSamplerState(d3dSampler, D3DSAMP_MAXMIPLEVEL, samplerState.lodOffset); - if (mSupportsTextureFilterAnisotropy) + mDevice->SetSamplerState(d3dSampler, D3DSAMP_MAXMIPLEVEL, samplerState.baseLevel); + if (getRendererExtensions().textureFilterAnisotropic) { mDevice->SetSamplerState(d3dSampler, D3DSAMP_MAXANISOTROPY, (DWORD)samplerState.maxAnisotropy); } @@ -778,9 +662,11 @@ void Renderer9::setSamplerState(gl::SamplerType type, int index, const gl::Sampl forceSetSamplers[index] = false; appliedSamplers[index] = samplerState; + + return gl::Error(GL_NO_ERROR); } -void Renderer9::setTexture(gl::SamplerType type, int index, gl::Texture *texture) +gl::Error Renderer9::setTexture(gl::SamplerType type, int index, gl::Texture *texture) { int d3dSamplerOffset = (type == gl::SAMPLER_PIXEL) ? 0 : D3DVERTEXTEXTURESAMPLER0; int d3dSampler = index + d3dSamplerOffset; @@ -788,14 +674,16 @@ void Renderer9::setTexture(gl::SamplerType type, int index, gl::Texture *texture unsigned int serial = 0; bool forceSetTexture = false; - unsigned int *appliedSerials = (type == gl::SAMPLER_PIXEL) ? mCurPixelTextureSerials : mCurVertexTextureSerials; + std::vector<unsigned int> &appliedSerials = (type == gl::SAMPLER_PIXEL) ? mCurPixelTextureSerials : mCurVertexTextureSerials; if (texture) { - TextureStorageInterface *texStorage = texture->getNativeTexture(); + TextureD3D* textureImpl = TextureD3D::makeTextureD3D(texture->getImplementation()); + + TextureStorage *texStorage = textureImpl->getNativeTexture(); if (texStorage) { - TextureStorage9 *storage9 = TextureStorage9::makeTextureStorage9(texStorage->getStorageInstance()); + TextureStorage9 *storage9 = TextureStorage9::makeTextureStorage9(texStorage); d3dTexture = storage9->getBaseTexture(); } // If we get NULL back from getBaseTexture here, something went wrong @@ -803,7 +691,8 @@ void Renderer9::setTexture(gl::SamplerType type, int index, gl::Texture *texture ASSERT(d3dTexture != NULL); serial = texture->getTextureSerial(); - forceSetTexture = texture->hasDirtyImages(); + forceSetTexture = textureImpl->hasDirtyImages(); + textureImpl->resetDirty(); } if (forceSetTexture || appliedSerials[index] != serial) @@ -812,9 +701,17 @@ void Renderer9::setTexture(gl::SamplerType type, int index, gl::Texture *texture } appliedSerials[index] = serial; + + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer9::setUniformBuffers(const gl::Buffer* /*vertexUniformBuffers*/[], const gl::Buffer* /*fragmentUniformBuffers*/[]) +{ + // No effect in ES2/D3D9 + return gl::Error(GL_NO_ERROR); } -void Renderer9::setRasterizerState(const gl::RasterizerState &rasterState) +gl::Error Renderer9::setRasterizerState(const gl::RasterizerState &rasterState) { bool rasterStateChanged = mForceSetRasterState || memcmp(&rasterState, &mCurRasterState, sizeof(gl::RasterizerState)) != 0; @@ -850,12 +747,15 @@ void Renderer9::setRasterizerState(const gl::RasterizerState &rasterState) } mForceSetRasterState = false; + + return gl::Error(GL_NO_ERROR); } -void Renderer9::setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState &blendState, const gl::Color &blendColor, unsigned int sampleMask) +gl::Error Renderer9::setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState &blendState, const gl::ColorF &blendColor, + unsigned int sampleMask) { bool blendStateChanged = mForceSetBlendState || memcmp(&blendState, &mCurBlendState, sizeof(gl::BlendState)) != 0; - bool blendColorChanged = mForceSetBlendState || memcmp(&blendColor, &mCurBlendColor, sizeof(gl::Color)) != 0; + bool blendColorChanged = mForceSetBlendState || memcmp(&blendColor, &mCurBlendColor, sizeof(gl::ColorF)) != 0; bool sampleMaskChanged = mForceSetBlendState || sampleMask != mCurSampleMask; if (blendStateChanged || blendColorChanged) @@ -906,6 +806,9 @@ void Renderer9::setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState FIXME("Sample alpha to coverage is unimplemented."); } + gl::FramebufferAttachment *attachment = framebuffer->getFirstColorbuffer(); + GLenum internalFormat = attachment ? attachment->getInternalFormat() : GL_NONE; + // Set the color mask bool zeroColorMaskAllowed = getAdapterVendor() != VENDOR_ID_AMD; // Apparently some ATI cards have a bug where a draw with a zero color @@ -914,8 +817,11 @@ void Renderer9::setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState // drawing is done. // http://code.google.com/p/angleproject/issues/detail?id=169 - DWORD colorMask = gl_d3d9::ConvertColorMask(blendState.colorMaskRed, blendState.colorMaskGreen, - blendState.colorMaskBlue, blendState.colorMaskAlpha); + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(internalFormat); + DWORD colorMask = gl_d3d9::ConvertColorMask(formatInfo.redBits > 0 && blendState.colorMaskRed, + formatInfo.greenBits > 0 && blendState.colorMaskGreen, + formatInfo.blueBits > 0 && blendState.colorMaskBlue, + formatInfo.alphaBits > 0 && blendState.colorMaskAlpha); if (colorMask == 0 && !zeroColorMaskAllowed) { // Enable green channel, but set blending so nothing will be drawn. @@ -947,10 +853,12 @@ void Renderer9::setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState } mForceSetBlendState = false; + + return gl::Error(GL_NO_ERROR); } -void Renderer9::setDepthStencilState(const gl::DepthStencilState &depthStencilState, int stencilRef, - int stencilBackRef, bool frontFaceCCW) +gl::Error Renderer9::setDepthStencilState(const gl::DepthStencilState &depthStencilState, int stencilRef, + int stencilBackRef, bool frontFaceCCW) { bool depthStencilStateChanged = mForceSetDepthStencilState || memcmp(&depthStencilState, &mCurDepthStencilState, sizeof(gl::DepthStencilState)) != 0; @@ -984,13 +892,10 @@ void Renderer9::setDepthStencilState(const gl::DepthStencilState &depthStencilSt const D3DRENDERSTATETYPE D3DRS_CCW_STENCILREF = D3DRS_STENCILREF; const D3DRENDERSTATETYPE D3DRS_CCW_STENCILMASK = D3DRS_STENCILMASK; const D3DRENDERSTATETYPE D3DRS_CCW_STENCILWRITEMASK = D3DRS_STENCILWRITEMASK; - if (depthStencilState.stencilWritemask != depthStencilState.stencilBackWritemask || - stencilRef != stencilBackRef || - depthStencilState.stencilMask != depthStencilState.stencilBackMask) - { - ERR("Separate front/back stencil writemasks, reference values, or stencil mask values are invalid under WebGL."); - return gl::error(GL_INVALID_OPERATION); - } + + ASSERT(depthStencilState.stencilWritemask == depthStencilState.stencilBackWritemask); + ASSERT(stencilRef == stencilBackRef); + ASSERT(depthStencilState.stencilMask == depthStencilState.stencilBackMask); // get the maximum size of the stencil ref unsigned int maxStencil = (1 << mCurStencilSize) - 1; @@ -1042,6 +947,8 @@ void Renderer9::setDepthStencilState(const gl::DepthStencilState &depthStencilSt } mForceSetDepthStencilState = false; + + return gl::Error(GL_NO_ERROR); } void Renderer9::setScissorRectangle(const gl::Rectangle &scissor, bool enabled) @@ -1071,7 +978,7 @@ void Renderer9::setScissorRectangle(const gl::Rectangle &scissor, bool enabled) mForceSetScissor = false; } -bool Renderer9::setViewport(const gl::Rectangle &viewport, float zNear, float zFar, GLenum drawMode, GLenum frontFace, +void Renderer9::setViewport(const gl::Rectangle &viewport, float zNear, float zFar, GLenum drawMode, GLenum frontFace, bool ignoreViewport) { gl::Rectangle actualViewport = viewport; @@ -1095,11 +1002,6 @@ bool Renderer9::setViewport(const gl::Rectangle &viewport, float zNear, float zF dxViewport.MinZ = actualZNear; dxViewport.MaxZ = actualZFar; - if (dxViewport.Width <= 0 || dxViewport.Height <= 0) - { - return false; // Nothing to render - } - float depthFront = !gl::IsTriangleMode(drawMode) ? 0.0f : (frontFace == GL_CCW ? 1.0f : -1.0f); bool viewportChanged = mForceSetViewport || memcmp(&actualViewport, &mCurViewport, sizeof(gl::Rectangle)) != 0 || @@ -1152,7 +1054,6 @@ bool Renderer9::setViewport(const gl::Rectangle &viewport, float zNear, float zF } mForceSetViewport = false; - return true; } bool Renderer9::applyPrimitiveType(GLenum mode, GLsizei count) @@ -1188,14 +1089,15 @@ bool Renderer9::applyPrimitiveType(GLenum mode, GLsizei count) mPrimitiveCount = count - 2; break; default: - return gl::error(GL_INVALID_ENUM, false); + UNREACHABLE(); + return false; } return mPrimitiveCount > 0; } -gl::Renderbuffer *Renderer9::getNullColorbuffer(gl::Renderbuffer *depthbuffer) +gl::FramebufferAttachment *Renderer9::getNullColorbuffer(gl::FramebufferAttachment *depthbuffer) { if (!depthbuffer) { @@ -1218,7 +1120,8 @@ gl::Renderbuffer *Renderer9::getNullColorbuffer(gl::Renderbuffer *depthbuffer) } } - gl::Renderbuffer *nullbuffer = new gl::Renderbuffer(this, 0, new gl::Colorbuffer(this, width, height, GL_NONE, 0)); + gl::Renderbuffer *nullRenderbuffer = new gl::Renderbuffer(0, new gl::Colorbuffer(this, width, height, GL_NONE, 0)); + gl::RenderbufferAttachment *nullbuffer = new gl::RenderbufferAttachment(GL_NONE, nullRenderbuffer); // add nullbuffer to the cache NullColorbufferCacheEntry *oldest = &mNullColorbufferCache[0]; @@ -1239,75 +1142,56 @@ gl::Renderbuffer *Renderer9::getNullColorbuffer(gl::Renderbuffer *depthbuffer) return nullbuffer; } -bool Renderer9::applyRenderTarget(gl::Framebuffer *framebuffer) +gl::Error Renderer9::applyRenderTarget(gl::Framebuffer *framebuffer) { // if there is no color attachment we must synthesize a NULL colorattachment // to keep the D3D runtime happy. This should only be possible if depth texturing. - gl::Renderbuffer *renderbufferObject = NULL; - if (framebuffer->getColorbufferType(0) != GL_NONE) - { - renderbufferObject = framebuffer->getColorbuffer(0); - } - else + gl::FramebufferAttachment *attachment = framebuffer->getColorbuffer(0); + if (!attachment) { - renderbufferObject = getNullColorbuffer(framebuffer->getDepthbuffer()); + attachment = getNullColorbuffer(framebuffer->getDepthbuffer()); } - if (!renderbufferObject) + if (!attachment) { - ERR("unable to locate renderbuffer for FBO."); - return false; + return gl::Error(GL_OUT_OF_MEMORY, "Unable to locate renderbuffer for FBO."); } bool renderTargetChanged = false; - unsigned int renderTargetSerial = renderbufferObject->getSerial(); + unsigned int renderTargetSerial = GetAttachmentSerial(attachment); if (renderTargetSerial != mAppliedRenderTargetSerial) { // Apply the render target on the device IDirect3DSurface9 *renderTargetSurface = NULL; - RenderTarget *renderTarget = renderbufferObject->getRenderTarget(); + RenderTarget9 *renderTarget = d3d9::GetAttachmentRenderTarget(attachment); if (renderTarget) { - renderTargetSurface = RenderTarget9::makeRenderTarget9(renderTarget)->getSurface(); + renderTargetSurface = renderTarget->getSurface(); } if (!renderTargetSurface) { - ERR("render target pointer unexpectedly null."); - return false; // Context must be lost + return gl::Error(GL_OUT_OF_MEMORY, "Internal render target pointer unexpectedly null."); } mDevice->SetRenderTarget(0, renderTargetSurface); - renderTargetSurface->Release(); + SafeRelease(renderTargetSurface); mAppliedRenderTargetSerial = renderTargetSerial; renderTargetChanged = true; } - gl::Renderbuffer *depthStencil = NULL; + gl::FramebufferAttachment *depthStencil = framebuffer->getDepthbuffer(); unsigned int depthbufferSerial = 0; unsigned int stencilbufferSerial = 0; - if (framebuffer->getDepthbufferType() != GL_NONE) + if (depthStencil) { - depthStencil = framebuffer->getDepthbuffer(); - if (!depthStencil) - { - ERR("Depth stencil pointer unexpectedly null."); - return false; - } - - depthbufferSerial = depthStencil->getSerial(); + depthbufferSerial = GetAttachmentSerial(depthStencil); } - else if (framebuffer->getStencilbufferType() != GL_NONE) + else if (framebuffer->getStencilbuffer()) { depthStencil = framebuffer->getStencilbuffer(); - if (!depthStencil) - { - ERR("Depth stencil pointer unexpectedly null."); - return false; - } - - stencilbufferSerial = depthStencil->getSerial(); + stencilbufferSerial = GetAttachmentSerial(depthStencil); } if (depthbufferSerial != mAppliedDepthbufferSerial || @@ -1321,21 +1205,20 @@ bool Renderer9::applyRenderTarget(gl::Framebuffer *framebuffer) if (depthStencil) { IDirect3DSurface9 *depthStencilSurface = NULL; - RenderTarget *depthStencilRenderTarget = depthStencil->getDepthStencil(); + rx::RenderTarget9 *depthStencilRenderTarget = d3d9::GetAttachmentRenderTarget(depthStencil); if (depthStencilRenderTarget) { - depthStencilSurface = RenderTarget9::makeRenderTarget9(depthStencilRenderTarget)->getSurface(); + depthStencilSurface = depthStencilRenderTarget->getSurface(); } if (!depthStencilSurface) { - ERR("depth stencil pointer unexpectedly null."); - return false; // Context must be lost + return gl::Error(GL_OUT_OF_MEMORY, "Internal depth stencil pointer unexpectedly null."); } mDevice->SetDepthStencilSurface(depthStencilSurface); - depthStencilSurface->Release(); + SafeRelease(depthStencilSurface); depthSize = depthStencil->getDepthSize(); stencilSize = depthStencil->getStencilSize(); @@ -1366,160 +1249,172 @@ bool Renderer9::applyRenderTarget(gl::Framebuffer *framebuffer) { mForceSetScissor = true; mForceSetViewport = true; + mForceSetBlendState = true; - mRenderTargetDesc.width = renderbufferObject->getWidth(); - mRenderTargetDesc.height = renderbufferObject->getHeight(); - mRenderTargetDesc.format = renderbufferObject->getActualFormat(); + mRenderTargetDesc.width = attachment->getWidth(); + mRenderTargetDesc.height = attachment->getHeight(); + mRenderTargetDesc.format = attachment->getActualFormat(); mRenderTargetDescInitialized = true; } - return true; + return gl::Error(GL_NO_ERROR); } -GLenum Renderer9::applyVertexBuffer(gl::ProgramBinary *programBinary, gl::VertexAttribute vertexAttributes[], GLint first, GLsizei count, GLsizei instances) +gl::Error Renderer9::applyVertexBuffer(gl::ProgramBinary *programBinary, const gl::VertexAttribute vertexAttributes[], const gl::VertexAttribCurrentValueData currentValues[], + GLint first, GLsizei count, GLsizei instances) { TranslatedAttribute attributes[gl::MAX_VERTEX_ATTRIBS]; - GLenum err = mVertexDataManager->prepareVertexData(vertexAttributes, programBinary, first, count, attributes, instances); - if (err != GL_NO_ERROR) + gl::Error error = mVertexDataManager->prepareVertexData(vertexAttributes, currentValues, programBinary, first, count, attributes, instances); + if (error.isError()) { - return err; + return error; } - + return mVertexDeclarationCache.applyDeclaration(mDevice, attributes, programBinary, instances, &mRepeatDraw); } // Applies the indices and element array bindings to the Direct3D 9 device -GLenum Renderer9::applyIndexBuffer(const GLvoid *indices, gl::Buffer *elementArrayBuffer, GLsizei count, GLenum mode, GLenum type, TranslatedIndexData *indexInfo) +gl::Error Renderer9::applyIndexBuffer(const GLvoid *indices, gl::Buffer *elementArrayBuffer, GLsizei count, GLenum mode, GLenum type, TranslatedIndexData *indexInfo) { - GLenum err = mIndexDataManager->prepareIndexData(type, count, elementArrayBuffer, indices, indexInfo); - - if (err == GL_NO_ERROR) + gl::Error error = mIndexDataManager->prepareIndexData(type, count, elementArrayBuffer, indices, indexInfo); + if (error.isError()) { - // Directly binding the storage buffer is not supported for d3d9 - ASSERT(indexInfo->storage == NULL); + return error; + } - if (indexInfo->serial != mAppliedIBSerial) - { - IndexBuffer9* indexBuffer = IndexBuffer9::makeIndexBuffer9(indexInfo->indexBuffer); + // Directly binding the storage buffer is not supported for d3d9 + ASSERT(indexInfo->storage == NULL); - mDevice->SetIndices(indexBuffer->getBuffer()); - mAppliedIBSerial = indexInfo->serial; - } + if (indexInfo->serial != mAppliedIBSerial) + { + IndexBuffer9* indexBuffer = IndexBuffer9::makeIndexBuffer9(indexInfo->indexBuffer); + + mDevice->SetIndices(indexBuffer->getBuffer()); + mAppliedIBSerial = indexInfo->serial; } - return err; + return gl::Error(GL_NO_ERROR); +} + +void Renderer9::applyTransformFeedbackBuffers(gl::Buffer *transformFeedbackBuffers[], GLintptr offsets[]) +{ + UNREACHABLE(); } -void Renderer9::drawArrays(GLenum mode, GLsizei count, GLsizei instances) +gl::Error Renderer9::drawArrays(GLenum mode, GLsizei count, GLsizei instances, bool transformFeedbackActive) { + ASSERT(!transformFeedbackActive); + startScene(); - + if (mode == GL_LINE_LOOP) { - drawLineLoop(count, GL_NONE, NULL, 0, NULL); + return drawLineLoop(count, GL_NONE, NULL, 0, NULL); } else if (instances > 0) { - StaticIndexBufferInterface *countingIB = mIndexDataManager->getCountingIndices(count); - if (countingIB) + StaticIndexBufferInterface *countingIB = NULL; + gl::Error error = getCountingIB(count, &countingIB); + if (error.isError()) { - if (mAppliedIBSerial != countingIB->getSerial()) - { - IndexBuffer9 *indexBuffer = IndexBuffer9::makeIndexBuffer9(countingIB->getIndexBuffer()); + return error; + } - mDevice->SetIndices(indexBuffer->getBuffer()); - mAppliedIBSerial = countingIB->getSerial(); - } + if (mAppliedIBSerial != countingIB->getSerial()) + { + IndexBuffer9 *indexBuffer = IndexBuffer9::makeIndexBuffer9(countingIB->getIndexBuffer()); - for (int i = 0; i < mRepeatDraw; i++) - { - mDevice->DrawIndexedPrimitive(mPrimitiveType, 0, 0, count, 0, mPrimitiveCount); - } + mDevice->SetIndices(indexBuffer->getBuffer()); + mAppliedIBSerial = countingIB->getSerial(); } - else + + for (int i = 0; i < mRepeatDraw; i++) { - ERR("Could not create a counting index buffer for glDrawArraysInstanced."); - return gl::error(GL_OUT_OF_MEMORY); + mDevice->DrawIndexedPrimitive(mPrimitiveType, 0, 0, count, 0, mPrimitiveCount); } + + return gl::Error(GL_NO_ERROR); } else // Regular case { mDevice->DrawPrimitive(mPrimitiveType, 0, mPrimitiveCount); + return gl::Error(GL_NO_ERROR); } } -void Renderer9::drawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, gl::Buffer *elementArrayBuffer, const TranslatedIndexData &indexInfo, GLsizei /*instances*/) +gl::Error Renderer9::drawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, + gl::Buffer *elementArrayBuffer, const TranslatedIndexData &indexInfo, GLsizei /*instances*/) { startScene(); + int minIndex = static_cast<int>(indexInfo.indexRange.start); + if (mode == GL_POINTS) { - drawIndexedPoints(count, type, indices, indexInfo.minIndex, elementArrayBuffer); + return drawIndexedPoints(count, type, indices, minIndex, elementArrayBuffer); } else if (mode == GL_LINE_LOOP) { - drawLineLoop(count, type, indices, indexInfo.minIndex, elementArrayBuffer); + return drawLineLoop(count, type, indices, minIndex, elementArrayBuffer); } else { for (int i = 0; i < mRepeatDraw; i++) { - GLsizei vertexCount = indexInfo.maxIndex - indexInfo.minIndex + 1; - mDevice->DrawIndexedPrimitive(mPrimitiveType, -(INT)indexInfo.minIndex, indexInfo.minIndex, vertexCount, indexInfo.startIndex, mPrimitiveCount); + GLsizei vertexCount = static_cast<int>(indexInfo.indexRange.length()) + 1; + mDevice->DrawIndexedPrimitive(mPrimitiveType, -minIndex, minIndex, vertexCount, indexInfo.startIndex, mPrimitiveCount); } + return gl::Error(GL_NO_ERROR); } } -void Renderer9::drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer) +gl::Error Renderer9::drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer) { // Get the raw indices for an indexed draw if (type != GL_NONE && elementArrayBuffer) { gl::Buffer *indexBuffer = elementArrayBuffer; - BufferStorage *storage = indexBuffer->getStorage(); + BufferImpl *storage = indexBuffer->getImplementation(); intptr_t offset = reinterpret_cast<intptr_t>(indices); indices = static_cast<const GLubyte*>(storage->getData()) + offset; } unsigned int startIndex = 0; - if (get32BitIndexSupport()) + if (getRendererExtensions().elementIndexUint) { if (!mLineLoopIB) { mLineLoopIB = new StreamingIndexBufferInterface(this); - if (!mLineLoopIB->reserveBufferSpace(INITIAL_INDEX_BUFFER_SIZE, GL_UNSIGNED_INT)) + gl::Error error = mLineLoopIB->reserveBufferSpace(INITIAL_INDEX_BUFFER_SIZE, GL_UNSIGNED_INT); + if (error.isError()) { - delete mLineLoopIB; - mLineLoopIB = NULL; - - ERR("Could not create a 32-bit looping index buffer for GL_LINE_LOOP."); - return gl::error(GL_OUT_OF_MEMORY); + SafeDelete(mLineLoopIB); + return error; } } + // Checked by Renderer9::applyPrimitiveType + ASSERT(count >= 0); + if (static_cast<unsigned int>(count) + 1 > (std::numeric_limits<unsigned int>::max() / sizeof(unsigned int))) { - ERR("Could not create a 32-bit looping index buffer for GL_LINE_LOOP, too many indices required."); - return gl::error(GL_OUT_OF_MEMORY); + return gl::Error(GL_OUT_OF_MEMORY, "Failed to create a 32-bit looping index buffer for GL_LINE_LOOP, too many indices required."); } - // Checked by Renderer9::applyPrimitiveType - ASSERT(count >= 0); - - const unsigned int spaceNeeded = (static_cast<unsigned int>(count) + 1) * sizeof(unsigned int); - if (!mLineLoopIB->reserveBufferSpace(spaceNeeded, GL_UNSIGNED_INT)) + const unsigned int spaceNeeded = (static_cast<unsigned int>(count)+1) * sizeof(unsigned int); + gl::Error error = mLineLoopIB->reserveBufferSpace(spaceNeeded, GL_UNSIGNED_INT); + if (error.isError()) { - ERR("Could not reserve enough space in looping index buffer for GL_LINE_LOOP."); - return gl::error(GL_OUT_OF_MEMORY); + return error; } void* mappedMemory = NULL; unsigned int offset = 0; - if (!mLineLoopIB->mapBuffer(spaceNeeded, &mappedMemory, &offset)) + error = mLineLoopIB->mapBuffer(spaceNeeded, &mappedMemory, &offset); + if (error.isError()) { - ERR("Could not map index buffer for GL_LINE_LOOP."); - return gl::error(GL_OUT_OF_MEMORY); + return error; } startIndex = static_cast<unsigned int>(offset) / 4; @@ -1558,10 +1453,10 @@ void Renderer9::drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, default: UNREACHABLE(); } - if (!mLineLoopIB->unmapBuffer()) + error = mLineLoopIB->unmapBuffer(); + if (error.isError()) { - ERR("Could not unmap index buffer for GL_LINE_LOOP."); - return gl::error(GL_OUT_OF_MEMORY); + return error; } } else @@ -1569,13 +1464,11 @@ void Renderer9::drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, if (!mLineLoopIB) { mLineLoopIB = new StreamingIndexBufferInterface(this); - if (!mLineLoopIB->reserveBufferSpace(INITIAL_INDEX_BUFFER_SIZE, GL_UNSIGNED_SHORT)) + gl::Error error = mLineLoopIB->reserveBufferSpace(INITIAL_INDEX_BUFFER_SIZE, GL_UNSIGNED_SHORT); + if (error.isError()) { - delete mLineLoopIB; - mLineLoopIB = NULL; - - ERR("Could not create a 16-bit looping index buffer for GL_LINE_LOOP."); - return gl::error(GL_OUT_OF_MEMORY); + SafeDelete(mLineLoopIB); + return error; } } @@ -1584,23 +1477,22 @@ void Renderer9::drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, if (static_cast<unsigned int>(count) + 1 > (std::numeric_limits<unsigned short>::max() / sizeof(unsigned short))) { - ERR("Could not create a 16-bit looping index buffer for GL_LINE_LOOP, too many indices required."); - return gl::error(GL_OUT_OF_MEMORY); + return gl::Error(GL_OUT_OF_MEMORY, "Failed to create a 16-bit looping index buffer for GL_LINE_LOOP, too many indices required."); } const unsigned int spaceNeeded = (static_cast<unsigned int>(count) + 1) * sizeof(unsigned short); - if (!mLineLoopIB->reserveBufferSpace(spaceNeeded, GL_UNSIGNED_SHORT)) + gl::Error error = mLineLoopIB->reserveBufferSpace(spaceNeeded, GL_UNSIGNED_SHORT); + if (error.isError()) { - ERR("Could not reserve enough space in looping index buffer for GL_LINE_LOOP."); - return gl::error(GL_OUT_OF_MEMORY); + return error; } void* mappedMemory = NULL; unsigned int offset; - if (mLineLoopIB->mapBuffer(spaceNeeded, &mappedMemory, &offset)) + error = mLineLoopIB->mapBuffer(spaceNeeded, &mappedMemory, &offset); + if (error.isError()) { - ERR("Could not map index buffer for GL_LINE_LOOP."); - return gl::error(GL_OUT_OF_MEMORY); + return error; } startIndex = static_cast<unsigned int>(offset) / 2; @@ -1639,10 +1531,10 @@ void Renderer9::drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, default: UNREACHABLE(); } - if (!mLineLoopIB->unmapBuffer()) + error = mLineLoopIB->unmapBuffer(); + if (error.isError()) { - ERR("Could not unmap index buffer for GL_LINE_LOOP."); - return gl::error(GL_OUT_OF_MEMORY); + return error; } } @@ -1655,67 +1547,162 @@ void Renderer9::drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, } mDevice->DrawIndexedPrimitive(D3DPT_LINESTRIP, -minIndex, minIndex, count, startIndex, count); + + return gl::Error(GL_NO_ERROR); } template <typename T> -static void drawPoints(IDirect3DDevice9* device, GLsizei count, const GLvoid *indices, int minIndex) +static gl::Error drawPoints(IDirect3DDevice9* device, GLsizei count, const GLvoid *indices, int minIndex) { for (int i = 0; i < count; i++) { unsigned int indexValue = static_cast<unsigned int>(static_cast<const T*>(indices)[i]) - minIndex; device->DrawPrimitive(D3DPT_POINTLIST, indexValue, 1); } + + return gl::Error(GL_NO_ERROR); } -void Renderer9::drawIndexedPoints(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer) +gl::Error Renderer9::drawIndexedPoints(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer) { // Drawing index point lists is unsupported in d3d9, fall back to a regular DrawPrimitive call // for each individual point. This call is not expected to happen often. if (elementArrayBuffer) { - BufferStorage *storage = elementArrayBuffer->getStorage(); + BufferImpl *storage = elementArrayBuffer->getImplementation(); intptr_t offset = reinterpret_cast<intptr_t>(indices); indices = static_cast<const GLubyte*>(storage->getData()) + offset; } switch (type) { - case GL_UNSIGNED_BYTE: drawPoints<GLubyte>(mDevice, count, indices, minIndex); break; - case GL_UNSIGNED_SHORT: drawPoints<GLushort>(mDevice, count, indices, minIndex); break; - case GL_UNSIGNED_INT: drawPoints<GLuint>(mDevice, count, indices, minIndex); break; - default: UNREACHABLE(); + case GL_UNSIGNED_BYTE: return drawPoints<GLubyte>(mDevice, count, indices, minIndex); + case GL_UNSIGNED_SHORT: return drawPoints<GLushort>(mDevice, count, indices, minIndex); + case GL_UNSIGNED_INT: return drawPoints<GLuint>(mDevice, count, indices, minIndex); + default: UNREACHABLE(); return gl::Error(GL_INVALID_OPERATION); } } -void Renderer9::applyShaders(gl::ProgramBinary *programBinary) +gl::Error Renderer9::getCountingIB(size_t count, StaticIndexBufferInterface **outIB) { - unsigned int programBinarySerial = programBinary->getSerial(); - if (programBinarySerial != mAppliedProgramBinarySerial) + // Update the counting index buffer if it is not large enough or has not been created yet. + if (count <= 65536) // 16-bit indices { - ShaderExecutable *vertexExe = programBinary->getVertexExecutable(); - ShaderExecutable *pixelExe = programBinary->getPixelExecutable(); + const unsigned int spaceNeeded = count * sizeof(unsigned short); - IDirect3DVertexShader9 *vertexShader = NULL; - if (vertexExe) vertexShader = ShaderExecutable9::makeShaderExecutable9(vertexExe)->getVertexShader(); + if (!mCountingIB || mCountingIB->getBufferSize() < spaceNeeded) + { + SafeDelete(mCountingIB); + mCountingIB = new StaticIndexBufferInterface(this); + mCountingIB->reserveBufferSpace(spaceNeeded, GL_UNSIGNED_SHORT); - IDirect3DPixelShader9 *pixelShader = NULL; - if (pixelExe) pixelShader = ShaderExecutable9::makeShaderExecutable9(pixelExe)->getPixelShader(); + void *mappedMemory = NULL; + gl::Error error = mCountingIB->mapBuffer(spaceNeeded, &mappedMemory, NULL); + if (error.isError()) + { + return error; + } - mDevice->SetPixelShader(pixelShader); + unsigned short *data = reinterpret_cast<unsigned short*>(mappedMemory); + for (size_t i = 0; i < count; i++) + { + data[i] = i; + } + + error = mCountingIB->unmapBuffer(); + if (error.isError()) + { + return error; + } + } + } + else if (getRendererExtensions().elementIndexUint) + { + const unsigned int spaceNeeded = count * sizeof(unsigned int); + + if (!mCountingIB || mCountingIB->getBufferSize() < spaceNeeded) + { + SafeDelete(mCountingIB); + mCountingIB = new StaticIndexBufferInterface(this); + mCountingIB->reserveBufferSpace(spaceNeeded, GL_UNSIGNED_INT); + + void *mappedMemory = NULL; + gl::Error error = mCountingIB->mapBuffer(spaceNeeded, &mappedMemory, NULL); + if (error.isError()) + { + return error; + } + + unsigned int *data = reinterpret_cast<unsigned int*>(mappedMemory); + for (size_t i = 0; i < count; i++) + { + data[i] = i; + } + + error = mCountingIB->unmapBuffer(); + if (error.isError()) + { + return error; + } + } + } + else + { + return gl::Error(GL_OUT_OF_MEMORY, "Could not create a counting index buffer for glDrawArraysInstanced."); + } + + *outIB = mCountingIB; + return gl::Error(GL_NO_ERROR); +} + +gl::Error Renderer9::applyShaders(gl::ProgramBinary *programBinary, const gl::VertexFormat inputLayout[], const gl::Framebuffer *framebuffer, + bool rasterizerDiscard, bool transformFeedbackActive) +{ + ASSERT(!transformFeedbackActive); + ASSERT(!rasterizerDiscard); + + ShaderExecutable *vertexExe = programBinary->getVertexExecutableForInputLayout(inputLayout); + ShaderExecutable *pixelExe = programBinary->getPixelExecutableForFramebuffer(framebuffer); + + IDirect3DVertexShader9 *vertexShader = (vertexExe ? ShaderExecutable9::makeShaderExecutable9(vertexExe)->getVertexShader() : NULL); + IDirect3DPixelShader9 *pixelShader = (pixelExe ? ShaderExecutable9::makeShaderExecutable9(pixelExe)->getPixelShader() : NULL); + + if (vertexShader != mAppliedVertexShader) + { mDevice->SetVertexShader(vertexShader); + mAppliedVertexShader = vertexShader; + } + + if (pixelShader != mAppliedPixelShader) + { + mDevice->SetPixelShader(pixelShader); + mAppliedPixelShader = pixelShader; + } + + // D3D9 has a quirk where creating multiple shaders with the same content + // can return the same shader pointer. Because GL programs store different data + // per-program, checking the program serial guarantees we upload fresh + // uniform data even if our shader pointers are the same. + // https://code.google.com/p/angleproject/issues/detail?id=661 + unsigned int programSerial = programBinary->getSerial(); + if (programSerial != mAppliedProgramSerial) + { programBinary->dirtyAllUniforms(); mDxUniformsDirty = true; - - mAppliedProgramBinarySerial = programBinarySerial; + mAppliedProgramSerial = programSerial; } + + return gl::Error(GL_NO_ERROR); } -void Renderer9::applyUniforms(gl::ProgramBinary *programBinary, gl::UniformArray *uniformArray) +gl::Error Renderer9::applyUniforms(const gl::ProgramBinary &programBinary) { - for (std::vector<gl::Uniform*>::const_iterator ub = uniformArray->begin(), ue = uniformArray->end(); ub != ue; ++ub) + const std::vector<gl::LinkedUniform*> &uniformArray = programBinary.getUniforms(); + + for (size_t uniformIndex = 0; uniformIndex < uniformArray.size(); uniformIndex++) { - gl::Uniform *targetUniform = *ub; + gl::LinkedUniform *targetUniform = uniformArray[uniformIndex]; if (targetUniform->dirty) { @@ -1751,8 +1738,6 @@ void Renderer9::applyUniforms(gl::ProgramBinary *programBinary, gl::UniformArray default: UNREACHABLE(); } - - targetUniform->dirty = false; } } @@ -1763,22 +1748,24 @@ void Renderer9::applyUniforms(gl::ProgramBinary *programBinary, gl::UniformArray mDevice->SetPixelShaderConstantF(0, (float*)&mPixelConstants, sizeof(dx_PixelConstants) / sizeof(float[4])); mDxUniformsDirty = false; } + + return gl::Error(GL_NO_ERROR); } -void Renderer9::applyUniformnfv(gl::Uniform *targetUniform, const GLfloat *v) +void Renderer9::applyUniformnfv(gl::LinkedUniform *targetUniform, const GLfloat *v) { - if (targetUniform->psRegisterIndex >= 0) + if (targetUniform->isReferencedByFragmentShader()) { mDevice->SetPixelShaderConstantF(targetUniform->psRegisterIndex, v, targetUniform->registerCount); } - if (targetUniform->vsRegisterIndex >= 0) + if (targetUniform->isReferencedByVertexShader()) { mDevice->SetVertexShaderConstantF(targetUniform->vsRegisterIndex, v, targetUniform->registerCount); } } -void Renderer9::applyUniformniv(gl::Uniform *targetUniform, const GLint *v) +void Renderer9::applyUniformniv(gl::LinkedUniform *targetUniform, const GLint *v) { ASSERT(targetUniform->registerCount <= MAX_VERTEX_CONSTANT_VECTORS_D3D9); GLfloat vector[MAX_VERTEX_CONSTANT_VECTORS_D3D9][4]; @@ -1794,7 +1781,7 @@ void Renderer9::applyUniformniv(gl::Uniform *targetUniform, const GLint *v) applyUniformnfv(targetUniform, (GLfloat*)vector); } -void Renderer9::applyUniformnbv(gl::Uniform *targetUniform, const GLint *v) +void Renderer9::applyUniformnbv(gl::LinkedUniform *targetUniform, const GLint *v) { ASSERT(targetUniform->registerCount <= MAX_VERTEX_CONSTANT_VECTORS_D3D9); GLfloat vector[MAX_VERTEX_CONSTANT_VECTORS_D3D9][4]; @@ -1810,29 +1797,60 @@ void Renderer9::applyUniformnbv(gl::Uniform *targetUniform, const GLint *v) applyUniformnfv(targetUniform, (GLfloat*)vector); } -void Renderer9::clear(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer) +gl::Error Renderer9::clear(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer) { - D3DCOLOR color = D3DCOLOR_ARGB(gl::unorm<8>(clearParams.colorClearValue.alpha), - gl::unorm<8>(clearParams.colorClearValue.red), - gl::unorm<8>(clearParams.colorClearValue.green), - gl::unorm<8>(clearParams.colorClearValue.blue)); + if (clearParams.colorClearType != GL_FLOAT) + { + // Clearing buffers with non-float values is not supported by Renderer9 and ES 2.0 + UNREACHABLE(); + return gl::Error(GL_INVALID_OPERATION); + } + + bool clearColor = clearParams.clearColor[0]; + for (unsigned int i = 0; i < ArraySize(clearParams.clearColor); i++) + { + if (clearParams.clearColor[i] != clearColor) + { + // Clearing individual buffers other than buffer zero is not supported by Renderer9 and ES 2.0 + UNREACHABLE(); + return gl::Error(GL_INVALID_OPERATION); + } + } + float depth = gl::clamp01(clearParams.depthClearValue); - int stencil = clearParams.stencilClearValue & 0x000000FF; + DWORD stencil = clearParams.stencilClearValue & 0x000000FF; unsigned int stencilUnmasked = 0x0; - if ((clearParams.mask & GL_STENCIL_BUFFER_BIT) && frameBuffer->hasStencil()) + if (clearParams.clearStencil && frameBuffer->hasStencil()) { - unsigned int stencilSize = gl::GetStencilSize(frameBuffer->getStencilbuffer()->getActualFormat()); + unsigned int stencilSize = gl::GetInternalFormatInfo((frameBuffer->getStencilbuffer()->getActualFormat())).stencilBits; stencilUnmasked = (0x1 << stencilSize) - 1; } - bool alphaUnmasked = (gl::GetAlphaSize(mRenderTargetDesc.format) == 0) || clearParams.colorMaskAlpha; - - const bool needMaskedStencilClear = (clearParams.mask & GL_STENCIL_BUFFER_BIT) && + const bool needMaskedStencilClear = clearParams.clearStencil && (clearParams.stencilWriteMask & stencilUnmasked) != stencilUnmasked; - const bool needMaskedColorClear = (clearParams.mask & GL_COLOR_BUFFER_BIT) && - !(clearParams.colorMaskRed && clearParams.colorMaskGreen && - clearParams.colorMaskBlue && alphaUnmasked); + + bool needMaskedColorClear = false; + D3DCOLOR color = D3DCOLOR_ARGB(255, 0, 0, 0); + if (clearColor) + { + const gl::FramebufferAttachment *attachment = frameBuffer->getFirstColorbuffer(); + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(attachment->getInternalFormat()); + const gl::InternalFormat &actualFormatInfo = gl::GetInternalFormatInfo(attachment->getActualFormat()); + + color = D3DCOLOR_ARGB(gl::unorm<8>((formatInfo.alphaBits == 0 && actualFormatInfo.alphaBits > 0) ? 1.0f : clearParams.colorFClearValue.alpha), + gl::unorm<8>((formatInfo.redBits == 0 && actualFormatInfo.redBits > 0) ? 0.0f : clearParams.colorFClearValue.red), + gl::unorm<8>((formatInfo.greenBits == 0 && actualFormatInfo.greenBits > 0) ? 0.0f : clearParams.colorFClearValue.green), + gl::unorm<8>((formatInfo.blueBits == 0 && actualFormatInfo.blueBits > 0) ? 0.0f : clearParams.colorFClearValue.blue)); + + if ((formatInfo.redBits > 0 && !clearParams.colorMaskRed) || + (formatInfo.greenBits > 0 && !clearParams.colorMaskGreen) || + (formatInfo.blueBits > 0 && !clearParams.colorMaskBlue) || + (formatInfo.alphaBits > 0 && !clearParams.colorMaskAlpha)) + { + needMaskedColorClear = true; + } + } if (needMaskedColorClear || needMaskedStencilClear) { @@ -1893,7 +1911,7 @@ void Renderer9::clear(const gl::ClearParameters &clearParams, gl::Framebuffer *f mDevice->SetRenderState(D3DRS_ALPHABLENDENABLE, FALSE); mDevice->SetRenderState(D3DRS_CLIPPLANEENABLE, 0); - if (clearParams.mask & GL_COLOR_BUFFER_BIT) + if (clearColor) { mDevice->SetRenderState(D3DRS_COLORWRITEENABLE, gl_d3d9::ConvertColorMask(clearParams.colorMaskRed, @@ -1906,7 +1924,7 @@ void Renderer9::clear(const gl::ClearParameters &clearParams, gl::Framebuffer *f mDevice->SetRenderState(D3DRS_COLORWRITEENABLE, 0); } - if (stencilUnmasked != 0x0 && (clearParams.mask & GL_STENCIL_BUFFER_BIT)) + if (stencilUnmasked != 0x0 && clearParams.clearStencil) { mDevice->SetRenderState(D3DRS_STENCILENABLE, TRUE); mDevice->SetRenderState(D3DRS_TWOSIDEDSTENCILMODE, FALSE); @@ -1962,7 +1980,7 @@ void Renderer9::clear(const gl::ClearParameters &clearParams, gl::Framebuffer *f startScene(); mDevice->DrawPrimitiveUP(D3DPT_TRIANGLESTRIP, 2, quad, sizeof(float[4])); - if (clearParams.mask & GL_DEPTH_BUFFER_BIT) + if (clearParams.clearDepth) { mDevice->SetRenderState(D3DRS_ZENABLE, TRUE); mDevice->SetRenderState(D3DRS_ZWRITEENABLE, TRUE); @@ -1974,24 +1992,26 @@ void Renderer9::clear(const gl::ClearParameters &clearParams, gl::Framebuffer *f mMaskedClearSavedState->Apply(); } } - else if (clearParams.mask) + else if (clearColor || clearParams.clearDepth || clearParams.clearStencil) { DWORD dxClearFlags = 0; - if (clearParams.mask & GL_COLOR_BUFFER_BIT) + if (clearColor) { dxClearFlags |= D3DCLEAR_TARGET; } - if (clearParams.mask & GL_DEPTH_BUFFER_BIT) + if (clearParams.clearDepth) { dxClearFlags |= D3DCLEAR_ZBUFFER; } - if (clearParams.mask & GL_STENCIL_BUFFER_BIT) + if (clearParams.clearStencil) { dxClearFlags |= D3DCLEAR_STENCIL; } mDevice->Clear(0, NULL, dxClearFlags, color, depth, stencil); } + + return gl::Error(GL_NO_ERROR); } void Renderer9::markAllStateDirty() @@ -2008,19 +2028,24 @@ void Renderer9::markAllStateDirty() mForceSetViewport = true; mForceSetBlendState = true; - for (unsigned int i = 0; i < gl::IMPLEMENTATION_MAX_VERTEX_TEXTURE_IMAGE_UNITS; i++) + ASSERT(mForceSetVertexSamplerStates.size() == mCurVertexTextureSerials.size()); + for (unsigned int i = 0; i < mForceSetVertexSamplerStates.size(); i++) { mForceSetVertexSamplerStates[i] = true; mCurVertexTextureSerials[i] = 0; } - for (unsigned int i = 0; i < gl::MAX_TEXTURE_IMAGE_UNITS; i++) + + ASSERT(mForceSetPixelSamplerStates.size() == mCurPixelTextureSerials.size()); + for (unsigned int i = 0; i < mForceSetPixelSamplerStates.size(); i++) { mForceSetPixelSamplerStates[i] = true; mCurPixelTextureSerials[i] = 0; } mAppliedIBSerial = 0; - mAppliedProgramBinarySerial = 0; + mAppliedVertexShader = NULL; + mAppliedPixelShader = NULL; + mAppliedProgramSerial = 0; mDxUniformsDirty = true; mVertexDeclarationCache.markStateDirty(); @@ -2028,11 +2053,11 @@ void Renderer9::markAllStateDirty() void Renderer9::releaseDeviceResources() { - while (!mEventQueryPool.empty()) + for (size_t i = 0; i < mEventQueryPool.size(); i++) { - mEventQueryPool.back()->Release(); - mEventQueryPool.pop_back(); + SafeRelease(mEventQueryPool[i]); } + mEventQueryPool.clear(); SafeRelease(mMaskedClearSavedState); @@ -2043,15 +2068,14 @@ void Renderer9::releaseDeviceResources() SafeDelete(mVertexDataManager); SafeDelete(mIndexDataManager); SafeDelete(mLineLoopIB); + SafeDelete(mCountingIB); for (int i = 0; i < NUM_NULL_COLORBUFFER_CACHE_ENTRIES; i++) { SafeDelete(mNullColorbufferCache[i].buffer); } - } - void Renderer9::notifyDeviceLost() { mDeviceLost = true; @@ -2221,7 +2245,7 @@ bool Renderer9::resetRemovedDevice() // The hardware adapter has been removed. Application must destroy the device, do enumeration of // adapters and create another Direct3D device. If application continues rendering without // calling Reset, the rendering calls will succeed. Applies to Direct3D 9Ex only. - deinitialize(); + release(); return (initialize() == EGL_SUCCESS); } @@ -2255,97 +2279,6 @@ GUID Renderer9::getAdapterIdentifier() const return mAdapterIdentifier.DeviceIdentifier; } -void Renderer9::getMultiSampleSupport(D3DFORMAT format, bool *multiSampleArray) -{ - for (int multiSampleIndex = 0; multiSampleIndex <= D3DMULTISAMPLE_16_SAMPLES; multiSampleIndex++) - { - HRESULT result = mD3d9->CheckDeviceMultiSampleType(mAdapter, mDeviceType, format, - TRUE, (D3DMULTISAMPLE_TYPE)multiSampleIndex, NULL); - - multiSampleArray[multiSampleIndex] = SUCCEEDED(result); - } -} - -bool Renderer9::getBGRATextureSupport() const -{ - // DirectX 9 always supports BGRA - return true; -} - -bool Renderer9::getDXT1TextureSupport() -{ - return mDXT1TextureSupport; -} - -bool Renderer9::getDXT3TextureSupport() -{ - return mDXT3TextureSupport; -} - -bool Renderer9::getDXT5TextureSupport() -{ - return mDXT5TextureSupport; -} - -bool Renderer9::getDepthTextureSupport() const -{ - return mDepthTextureSupport; -} - -bool Renderer9::getFloat32TextureSupport(bool *filtering, bool *renderable) -{ - *filtering = mFloat32FilterSupport; - *renderable = mFloat32RenderSupport; - return mFloat32TextureSupport; -} - -bool Renderer9::getFloat16TextureSupport(bool *filtering, bool *renderable) -{ - *filtering = mFloat16FilterSupport; - *renderable = mFloat16RenderSupport; - return mFloat16TextureSupport; -} - -bool Renderer9::getLuminanceTextureSupport() -{ - return mLuminanceTextureSupport; -} - -bool Renderer9::getLuminanceAlphaTextureSupport() -{ - return mLuminanceAlphaTextureSupport; -} - -bool Renderer9::getTextureFilterAnisotropySupport() const -{ - return mSupportsTextureFilterAnisotropy; -} - -float Renderer9::getTextureMaxAnisotropy() const -{ - if (mSupportsTextureFilterAnisotropy) - { - return static_cast<float>(mDeviceCaps.MaxAnisotropy); - } - return 1.0f; -} - -bool Renderer9::getEventQuerySupport() -{ - return mEventQuerySupport; -} - -unsigned int Renderer9::getMaxVertexTextureImageUnits() const -{ - META_ASSERT(MAX_TEXTURE_IMAGE_UNITS_VTF_SM3 <= gl::IMPLEMENTATION_MAX_VERTEX_TEXTURE_IMAGE_UNITS); - return mVertexTextureSupport ? MAX_TEXTURE_IMAGE_UNITS_VTF_SM3 : 0; -} - -unsigned int Renderer9::getMaxCombinedTextureImageUnits() const -{ - return gl::MAX_TEXTURE_IMAGE_UNITS + getMaxVertexTextureImageUnits(); -} - unsigned int Renderer9::getReservedVertexUniformVectors() const { return 2; // dx_ViewAdjust and dx_DepthRange. @@ -2356,36 +2289,14 @@ unsigned int Renderer9::getReservedFragmentUniformVectors() const return 3; // dx_ViewCoords, dx_DepthFront and dx_DepthRange. } -unsigned int Renderer9::getMaxVertexUniformVectors() const -{ - return MAX_VERTEX_CONSTANT_VECTORS_D3D9 - getReservedVertexUniformVectors(); -} - -unsigned int Renderer9::getMaxFragmentUniformVectors() const -{ - const int maxPixelConstantVectors = (getMajorShaderModel() >= 3) ? MAX_PIXEL_CONSTANT_VECTORS_SM3 : MAX_PIXEL_CONSTANT_VECTORS_SM2; - - return maxPixelConstantVectors - getReservedFragmentUniformVectors(); -} - -unsigned int Renderer9::getMaxVaryingVectors() const +unsigned int Renderer9::getReservedVertexUniformBuffers() const { - return (getMajorShaderModel() >= 3) ? MAX_VARYING_VECTORS_SM3 : MAX_VARYING_VECTORS_SM2; + return 0; } -bool Renderer9::getNonPower2TextureSupport() const +unsigned int Renderer9::getReservedFragmentUniformBuffers() const { - return mSupportsNonPower2Textures; -} - -bool Renderer9::getOcclusionQuerySupport() const -{ - return mOcclusionQuerySupport; -} - -bool Renderer9::getInstancingSupport() const -{ - return mDeviceCaps.PixelShaderVersion >= D3DPS_VERSION(3, 0); + return 0; } bool Renderer9::getShareHandleSupport() const @@ -2394,11 +2305,6 @@ bool Renderer9::getShareHandleSupport() const return (mD3d9Ex != NULL) && !gl::perfActive(); } -bool Renderer9::getDerivativeInstructionSupport() const -{ - return (mDeviceCaps.PS20Caps.Caps & D3DPS20CAPS_GRADIENTINSTRUCTIONS) != 0; -} - bool Renderer9::getPostSubBufferSupport() const { return true; @@ -2409,34 +2315,6 @@ int Renderer9::getMajorShaderModel() const return D3DSHADER_VERSION_MAJOR(mDeviceCaps.PixelShaderVersion); } -float Renderer9::getMaxPointSize() const -{ - // Point size clamped at 1.0f for SM2 - return getMajorShaderModel() == 3 ? mDeviceCaps.MaxPointSize : 1.0f; -} - -int Renderer9::getMaxViewportDimension() const -{ - int maxTextureDimension = std::min(std::min(getMaxTextureWidth(), getMaxTextureHeight()), - (int)gl::IMPLEMENTATION_MAX_TEXTURE_SIZE); - return maxTextureDimension; -} - -int Renderer9::getMaxTextureWidth() const -{ - return (int)mDeviceCaps.MaxTextureWidth; -} - -int Renderer9::getMaxTextureHeight() const -{ - return (int)mDeviceCaps.MaxTextureHeight; -} - -bool Renderer9::get32BitIndexSupport() const -{ - return mDeviceCaps.MaxVertexIndex >= (1 << 16); -} - DWORD Renderer9::getCapsDeclTypes() const { return mDeviceCaps.DeclTypes; @@ -2452,127 +2330,45 @@ int Renderer9::getMaxSwapInterval() const return mMaxSwapInterval; } -int Renderer9::getMaxSupportedSamples() const -{ - return mMaxSupportedSamples; -} - -int Renderer9::getNearestSupportedSamples(D3DFORMAT format, int requested) const -{ - if (requested == 0) - { - return requested; - } - - std::map<D3DFORMAT, bool *>::const_iterator itr = mMultiSampleSupport.find(format); - if (itr == mMultiSampleSupport.end()) - { - if (format == D3DFMT_UNKNOWN) - return 0; - return -1; - } - - for (int i = requested; i <= D3DMULTISAMPLE_16_SAMPLES; ++i) - { - if (itr->second[i] && i != D3DMULTISAMPLE_NONMASKABLE) - { - return i; - } - } - - return -1; -} - -unsigned int Renderer9::getMaxRenderTargets() const -{ - // we do not support MRT in d3d9 - return 1; -} - -D3DFORMAT Renderer9::ConvertTextureInternalFormat(GLint internalformat) -{ - switch (internalformat) - { - case GL_DEPTH_COMPONENT16: - case GL_DEPTH_COMPONENT32_OES: - case GL_DEPTH24_STENCIL8_OES: - return D3DFMT_INTZ; - case GL_COMPRESSED_RGB_S3TC_DXT1_EXT: - case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT: - return D3DFMT_DXT1; - case GL_COMPRESSED_RGBA_S3TC_DXT3_ANGLE: - return D3DFMT_DXT3; - case GL_COMPRESSED_RGBA_S3TC_DXT5_ANGLE: - return D3DFMT_DXT5; - case GL_RGBA32F_EXT: - case GL_RGB32F_EXT: - case GL_ALPHA32F_EXT: - case GL_LUMINANCE32F_EXT: - case GL_LUMINANCE_ALPHA32F_EXT: - return D3DFMT_A32B32G32R32F; - case GL_RGBA16F_EXT: - case GL_RGB16F_EXT: - case GL_ALPHA16F_EXT: - case GL_LUMINANCE16F_EXT: - case GL_LUMINANCE_ALPHA16F_EXT: - return D3DFMT_A16B16G16R16F; - case GL_LUMINANCE8_EXT: - if (getLuminanceTextureSupport()) - { - return D3DFMT_L8; - } - break; - case GL_LUMINANCE8_ALPHA8_EXT: - if (getLuminanceAlphaTextureSupport()) - { - return D3DFMT_A8L8; - } - break; - case GL_RGB8_OES: - case GL_RGB565: - return D3DFMT_X8R8G8B8; - } - - return D3DFMT_A8R8G8B8; -} - -bool Renderer9::copyToRenderTarget(TextureStorageInterface2D *dest, TextureStorageInterface2D *source) +bool Renderer9::copyToRenderTarget2D(TextureStorage *dest, TextureStorage *source) { bool result = false; if (source && dest) { - TextureStorage9_2D *source9 = TextureStorage9_2D::makeTextureStorage9_2D(source->getStorageInstance()); - TextureStorage9_2D *dest9 = TextureStorage9_2D::makeTextureStorage9_2D(dest->getStorageInstance()); + TextureStorage9_2D *source9 = TextureStorage9_2D::makeTextureStorage9_2D(source); + TextureStorage9_2D *dest9 = TextureStorage9_2D::makeTextureStorage9_2D(dest); - int levels = source9->levelCount(); + int levels = source9->getLevelCount(); for (int i = 0; i < levels; ++i) { IDirect3DSurface9 *srcSurf = source9->getSurfaceLevel(i, false); IDirect3DSurface9 *dstSurf = dest9->getSurfaceLevel(i, false); - + result = copyToRenderTarget(dstSurf, srcSurf, source9->isManaged()); - if (srcSurf) srcSurf->Release(); - if (dstSurf) dstSurf->Release(); + SafeRelease(srcSurf); + SafeRelease(dstSurf); if (!result) + { return false; + } } } return result; } -bool Renderer9::copyToRenderTarget(TextureStorageInterfaceCube *dest, TextureStorageInterfaceCube *source) +bool Renderer9::copyToRenderTargetCube(TextureStorage *dest, TextureStorage *source) { bool result = false; if (source && dest) { - TextureStorage9_Cube *source9 = TextureStorage9_Cube::makeTextureStorage9_Cube(source->getStorageInstance()); - TextureStorage9_Cube *dest9 = TextureStorage9_Cube::makeTextureStorage9_Cube(dest->getStorageInstance()); - int levels = source9->levelCount(); + TextureStorage9_Cube *source9 = TextureStorage9_Cube::makeTextureStorage9_Cube(source); + TextureStorage9_Cube *dest9 = TextureStorage9_Cube::makeTextureStorage9_Cube(dest); + int levels = source9->getLevelCount(); for (int f = 0; f < 6; f++) { for (int i = 0; i < levels; i++) @@ -2582,11 +2378,13 @@ bool Renderer9::copyToRenderTarget(TextureStorageInterfaceCube *dest, TextureSto result = copyToRenderTarget(dstSurf, srcSurf, source9->isManaged()); - if (srcSurf) srcSurf->Release(); - if (dstSurf) dstSurf->Release(); + SafeRelease(srcSurf); + SafeRelease(dstSurf); if (!result) + { return false; + } } } } @@ -2594,6 +2392,20 @@ bool Renderer9::copyToRenderTarget(TextureStorageInterfaceCube *dest, TextureSto return result; } +bool Renderer9::copyToRenderTarget3D(TextureStorage *dest, TextureStorage *source) +{ + // 3D textures are not available in the D3D9 backend. + UNREACHABLE(); + return false; +} + +bool Renderer9::copyToRenderTarget2DArray(TextureStorage *dest, TextureStorage *source) +{ + // 2D array textures are not supported by the D3D9 backend. + UNREACHABLE(); + return false; +} + D3DPOOL Renderer9::getBufferPool(DWORD usage) const { if (mD3d9Ex != NULL) @@ -2611,8 +2423,8 @@ D3DPOOL Renderer9::getBufferPool(DWORD usage) const return D3DPOOL_DEFAULT; } -bool Renderer9::copyImage(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, - GLint xoffset, GLint yoffset, TextureStorageInterface2D *storage, GLint level) +bool Renderer9::copyImage2D(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, TextureStorage *storage, GLint level) { RECT rect; rect.left = sourceRect.x; @@ -2620,11 +2432,11 @@ bool Renderer9::copyImage(gl::Framebuffer *framebuffer, const gl::Rectangle &sou rect.right = sourceRect.x + sourceRect.width; rect.bottom = sourceRect.y + sourceRect.height; - return mBlit->copy(framebuffer, rect, destFormat, xoffset, yoffset, storage, level); + return mBlit->copy2D(framebuffer, rect, destFormat, xoffset, yoffset, storage, level); } -bool Renderer9::copyImage(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, - GLint xoffset, GLint yoffset, TextureStorageInterfaceCube *storage, GLenum target, GLint level) +bool Renderer9::copyImageCube(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, TextureStorage *storage, GLenum target, GLint level) { RECT rect; rect.left = sourceRect.x; @@ -2632,18 +2444,36 @@ bool Renderer9::copyImage(gl::Framebuffer *framebuffer, const gl::Rectangle &sou rect.right = sourceRect.x + sourceRect.width; rect.bottom = sourceRect.y + sourceRect.height; - return mBlit->copy(framebuffer, rect, destFormat, xoffset, yoffset, storage, target, level); + return mBlit->copyCube(framebuffer, rect, destFormat, xoffset, yoffset, storage, target, level); +} + +bool Renderer9::copyImage3D(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, GLint zOffset, TextureStorage *storage, GLint level) +{ + // 3D textures are not available in the D3D9 backend. + UNREACHABLE(); + return false; +} + +bool Renderer9::copyImage2DArray(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, GLint zOffset, TextureStorage *storage, GLint level) +{ + // 2D array textures are not available in the D3D9 backend. + UNREACHABLE(); + return false; } bool Renderer9::blitRect(gl::Framebuffer *readFramebuffer, const gl::Rectangle &readRect, gl::Framebuffer *drawFramebuffer, const gl::Rectangle &drawRect, - bool blitRenderTarget, bool blitDepthStencil) + const gl::Rectangle *scissor, bool blitRenderTarget, bool blitDepth, bool blitStencil, GLenum filter) { + ASSERT(filter == GL_NEAREST); + endScene(); if (blitRenderTarget) { - gl::Renderbuffer *readBuffer = readFramebuffer->getColorbuffer(0); - gl::Renderbuffer *drawBuffer = drawFramebuffer->getColorbuffer(0); + gl::FramebufferAttachment *readBuffer = readFramebuffer->getColorbuffer(0); + gl::FramebufferAttachment *drawBuffer = drawFramebuffer->getColorbuffer(0); RenderTarget9 *readRenderTarget = NULL; RenderTarget9 *drawRenderTarget = NULL; IDirect3DSurface9* readSurface = NULL; @@ -2651,11 +2481,11 @@ bool Renderer9::blitRect(gl::Framebuffer *readFramebuffer, const gl::Rectangle & if (readBuffer) { - readRenderTarget = RenderTarget9::makeRenderTarget9(readBuffer->getRenderTarget()); + readRenderTarget = d3d9::GetAttachmentRenderTarget(readBuffer); } if (drawBuffer) { - drawRenderTarget = RenderTarget9::makeRenderTarget9(drawBuffer->getRenderTarget()); + drawRenderTarget = d3d9::GetAttachmentRenderTarget(drawBuffer); } if (readRenderTarget) @@ -2673,6 +2503,9 @@ bool Renderer9::blitRect(gl::Framebuffer *readFramebuffer, const gl::Rectangle & return gl::error(GL_OUT_OF_MEMORY, false); } + gl::Extents srcSize(readRenderTarget->getWidth(), readRenderTarget->getHeight(), 1); + gl::Extents dstSize(drawRenderTarget->getWidth(), drawRenderTarget->getHeight(), 1); + RECT srcRect; srcRect.left = readRect.x; srcRect.right = readRect.x + readRect.width; @@ -2685,10 +2518,79 @@ bool Renderer9::blitRect(gl::Framebuffer *readFramebuffer, const gl::Rectangle & dstRect.top = drawRect.y; dstRect.bottom = drawRect.y + drawRect.height; + // Clip the rectangles to the scissor rectangle + if (scissor) + { + if (dstRect.left < scissor->x) + { + srcRect.left += (scissor->x - dstRect.left); + dstRect.left = scissor->x; + } + if (dstRect.top < scissor->y) + { + srcRect.top += (scissor->y - dstRect.top); + dstRect.top = scissor->y; + } + if (dstRect.right > scissor->x + scissor->width) + { + srcRect.right -= (dstRect.right - (scissor->x + scissor->width)); + dstRect.right = scissor->x + scissor->width; + } + if (dstRect.bottom > scissor->y + scissor->height) + { + srcRect.bottom -= (dstRect.bottom - (scissor->y + scissor->height)); + dstRect.bottom = scissor->y + scissor->height; + } + } + + // Clip the rectangles to the destination size + if (dstRect.left < 0) + { + srcRect.left += -dstRect.left; + dstRect.left = 0; + } + if (dstRect.right > dstSize.width) + { + srcRect.right -= (dstRect.right - dstSize.width); + dstRect.right = dstSize.width; + } + if (dstRect.top < 0) + { + srcRect.top += -dstRect.top; + dstRect.top = 0; + } + if (dstRect.bottom > dstSize.height) + { + srcRect.bottom -= (dstRect.bottom - dstSize.height); + dstRect.bottom = dstSize.height; + } + + // Clip the rectangles to the source size + if (srcRect.left < 0) + { + dstRect.left += -srcRect.left; + srcRect.left = 0; + } + if (srcRect.right > srcSize.width) + { + dstRect.right -= (srcRect.right - srcSize.width); + srcRect.right = srcSize.width; + } + if (srcRect.top < 0) + { + dstRect.top += -srcRect.top; + srcRect.top = 0; + } + if (srcRect.bottom > srcSize.height) + { + dstRect.bottom -= (srcRect.bottom - srcSize.height); + srcRect.bottom = srcSize.height; + } + HRESULT result = mDevice->StretchRect(readSurface, &srcRect, drawSurface, &dstRect, D3DTEXF_NONE); - readSurface->Release(); - drawSurface->Release(); + SafeRelease(readSurface); + SafeRelease(drawSurface); if (FAILED(result)) { @@ -2697,10 +2599,10 @@ bool Renderer9::blitRect(gl::Framebuffer *readFramebuffer, const gl::Rectangle & } } - if (blitDepthStencil) + if (blitDepth || blitStencil) { - gl::Renderbuffer *readBuffer = readFramebuffer->getDepthOrStencilbuffer(); - gl::Renderbuffer *drawBuffer = drawFramebuffer->getDepthOrStencilbuffer(); + gl::FramebufferAttachment *readBuffer = readFramebuffer->getDepthOrStencilbuffer(); + gl::FramebufferAttachment *drawBuffer = drawFramebuffer->getDepthOrStencilbuffer(); RenderTarget9 *readDepthStencil = NULL; RenderTarget9 *drawDepthStencil = NULL; IDirect3DSurface9* readSurface = NULL; @@ -2708,11 +2610,11 @@ bool Renderer9::blitRect(gl::Framebuffer *readFramebuffer, const gl::Rectangle & if (readBuffer) { - readDepthStencil = RenderTarget9::makeRenderTarget9(readBuffer->getDepthStencil()); + readDepthStencil = d3d9::GetAttachmentRenderTarget(readBuffer); } if (drawBuffer) { - drawDepthStencil = RenderTarget9::makeRenderTarget9(drawBuffer->getDepthStencil()); + drawDepthStencil = d3d9::GetAttachmentRenderTarget(drawBuffer); } if (readDepthStencil) @@ -2732,8 +2634,8 @@ bool Renderer9::blitRect(gl::Framebuffer *readFramebuffer, const gl::Rectangle & HRESULT result = mDevice->StretchRect(readSurface, NULL, drawSurface, NULL, D3DTEXF_NONE); - readSurface->Release(); - drawSurface->Release(); + SafeRelease(readSurface); + SafeRelease(drawSurface); if (FAILED(result)) { @@ -2745,18 +2647,20 @@ bool Renderer9::blitRect(gl::Framebuffer *readFramebuffer, const gl::Rectangle & return true; } -void Renderer9::readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, - GLsizei outputPitch, bool packReverseRowOrder, GLint packAlignment, void* pixels) +gl::Error Renderer9::readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, + GLenum type, GLuint outputPitch, const gl::PixelPackState &pack, uint8_t *pixels) { + ASSERT(pack.pixelBuffer.get() == NULL); + RenderTarget9 *renderTarget = NULL; IDirect3DSurface9 *surface = NULL; - gl::Renderbuffer *colorbuffer = framebuffer->getColorbuffer(0); + gl::FramebufferAttachment *colorbuffer = framebuffer->getColorbuffer(0); if (colorbuffer) { - renderTarget = RenderTarget9::makeRenderTarget9(colorbuffer->getRenderTarget()); + renderTarget = d3d9::GetAttachmentRenderTarget(colorbuffer); } - + if (renderTarget) { surface = renderTarget->getSurface(); @@ -2765,7 +2669,7 @@ void Renderer9::readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsiz if (!surface) { // context must be lost - return; + return gl::Error(GL_NO_ERROR); } D3DSURFACE_DESC desc; @@ -2774,20 +2678,20 @@ void Renderer9::readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsiz if (desc.MultiSampleType != D3DMULTISAMPLE_NONE) { UNIMPLEMENTED(); // FIXME: Requires resolve using StretchRect into non-multisampled render target - surface->Release(); - return gl::error(GL_OUT_OF_MEMORY); + SafeRelease(surface); + return gl::Error(GL_OUT_OF_MEMORY, "ReadPixels is unimplemented for multisampled framebuffer attachments."); } HRESULT result; IDirect3DSurface9 *systemSurface = NULL; - bool directToPixels = !packReverseRowOrder && packAlignment <= 4 && getShareHandleSupport() && + bool directToPixels = !pack.reverseRowOrder && pack.alignment <= 4 && getShareHandleSupport() && x == 0 && y == 0 && UINT(width) == desc.Width && UINT(height) == desc.Height && desc.Format == D3DFMT_A8R8G8B8 && format == GL_BGRA_EXT && type == GL_UNSIGNED_BYTE; if (directToPixels) { // Use the pixels ptr as a shared handle to write directly into client's memory result = mDevice->CreateOffscreenPlainSurface(desc.Width, desc.Height, desc.Format, - D3DPOOL_SYSTEMMEM, &systemSurface, &pixels); + D3DPOOL_SYSTEMMEM, &systemSurface, reinterpret_cast<void**>(&pixels)); if (FAILED(result)) { // Try again without the shared handle @@ -2802,38 +2706,36 @@ void Renderer9::readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsiz if (FAILED(result)) { ASSERT(result == D3DERR_OUTOFVIDEOMEMORY || result == E_OUTOFMEMORY); - surface->Release(); - return gl::error(GL_OUT_OF_MEMORY); + SafeRelease(surface); + return gl::Error(GL_OUT_OF_MEMORY, "Failed to allocate internal texture for ReadPixels."); } } result = mDevice->GetRenderTargetData(surface, systemSurface); - surface->Release(); - surface = NULL; + SafeRelease(surface); if (FAILED(result)) { - systemSurface->Release(); + SafeRelease(systemSurface); // It turns out that D3D will sometimes produce more error // codes than those documented. if (d3d9::isDeviceLostError(result)) { notifyDeviceLost(); - return gl::error(GL_OUT_OF_MEMORY); } else { UNREACHABLE(); - return; } + return gl::Error(GL_OUT_OF_MEMORY, "Failed to read internal render target data."); } if (directToPixels) { - systemSurface->Release(); - return; + SafeRelease(systemSurface); + return gl::Error(GL_NO_ERROR); } RECT rect; @@ -2848,252 +2750,79 @@ void Renderer9::readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsiz if (FAILED(result)) { UNREACHABLE(); - systemSurface->Release(); + SafeRelease(systemSurface); - return; // No sensible error to generate + return gl::Error(GL_OUT_OF_MEMORY, "Failed to lock internal render target."); } - unsigned char *dest = (unsigned char*)pixels; - unsigned short *dest16 = (unsigned short*)pixels; - - unsigned char *source; + uint8_t *source; int inputPitch; - if (packReverseRowOrder) + if (pack.reverseRowOrder) { - source = ((unsigned char*)lock.pBits) + lock.Pitch * (rect.bottom - rect.top - 1); + source = reinterpret_cast<uint8_t*>(lock.pBits) + lock.Pitch * (rect.bottom - rect.top - 1); inputPitch = -lock.Pitch; } else { - source = (unsigned char*)lock.pBits; + source = reinterpret_cast<uint8_t*>(lock.pBits); inputPitch = lock.Pitch; } - unsigned int fastPixelSize = 0; - - if (desc.Format == D3DFMT_A8R8G8B8 && - format == GL_BGRA_EXT && - type == GL_UNSIGNED_BYTE) - { - fastPixelSize = 4; - } - else if ((desc.Format == D3DFMT_A4R4G4B4 && - format == GL_BGRA_EXT && - type == GL_UNSIGNED_SHORT_4_4_4_4_REV_EXT) || - (desc.Format == D3DFMT_A1R5G5B5 && - format == GL_BGRA_EXT && - type == GL_UNSIGNED_SHORT_1_5_5_5_REV_EXT)) - { - fastPixelSize = 2; - } - else if (desc.Format == D3DFMT_A16B16G16R16F && - format == GL_RGBA && - type == GL_HALF_FLOAT_OES) + const d3d9::D3DFormat &d3dFormatInfo = d3d9::GetD3DFormatInfo(desc.Format); + const gl::InternalFormat &sourceFormatInfo = gl::GetInternalFormatInfo(d3dFormatInfo.internalFormat); + if (sourceFormatInfo.format == format && sourceFormatInfo.type == type) { - fastPixelSize = 8; + // Direct copy possible + for (int y = 0; y < rect.bottom - rect.top; y++) + { + memcpy(pixels + y * outputPitch, source + y * inputPitch, (rect.right - rect.left) * sourceFormatInfo.pixelBytes); + } } - else if (desc.Format == D3DFMT_A32B32G32R32F && - format == GL_RGBA && - type == GL_FLOAT) + else { - fastPixelSize = 16; - } + const d3d9::D3DFormat &sourceD3DFormatInfo = d3d9::GetD3DFormatInfo(desc.Format); + ColorCopyFunction fastCopyFunc = sourceD3DFormatInfo.getFastCopyFunction(format, type); - for (int j = 0; j < rect.bottom - rect.top; j++) - { - if (fastPixelSize != 0) - { - // Fast path for formats which require no translation: - // D3DFMT_A8R8G8B8 to BGRA/UNSIGNED_BYTE - // D3DFMT_A4R4G4B4 to BGRA/UNSIGNED_SHORT_4_4_4_4_REV_EXT - // D3DFMT_A1R5G5B5 to BGRA/UNSIGNED_SHORT_1_5_5_5_REV_EXT - // D3DFMT_A16B16G16R16F to RGBA/HALF_FLOAT_OES - // D3DFMT_A32B32G32R32F to RGBA/FLOAT - // - // Note that buffers with no alpha go through the slow path below. - memcpy(dest + j * outputPitch, - source + j * inputPitch, - (rect.right - rect.left) * fastPixelSize); - continue; - } - else if (desc.Format == D3DFMT_A8R8G8B8 && - format == GL_RGBA && - type == GL_UNSIGNED_BYTE) - { - // Fast path for swapping red with blue - for (int i = 0; i < rect.right - rect.left; i++) - { - unsigned int argb = *(unsigned int*)(source + 4 * i + j * inputPitch); - *(unsigned int*)(dest + 4 * i + j * outputPitch) = - (argb & 0xFF00FF00) | // Keep alpha and green - (argb & 0x00FF0000) >> 16 | // Move red to blue - (argb & 0x000000FF) << 16; // Move blue to red - } - continue; - } + const gl::FormatType &destFormatTypeInfo = gl::GetFormatTypeInfo(format, type); + const gl::InternalFormat &destFormatInfo = gl::GetInternalFormatInfo(destFormatTypeInfo.internalFormat); - for (int i = 0; i < rect.right - rect.left; i++) + if (fastCopyFunc) { - float r; - float g; - float b; - float a; - - switch (desc.Format) + // Fast copy is possible through some special function + for (int y = 0; y < rect.bottom - rect.top; y++) { - case D3DFMT_R5G6B5: + for (int x = 0; x < rect.right - rect.left; x++) { - unsigned short rgb = *(unsigned short*)(source + 2 * i + j * inputPitch); + uint8_t *dest = pixels + y * outputPitch + x * destFormatInfo.pixelBytes; + const uint8_t *src = source + y * inputPitch + x * sourceFormatInfo.pixelBytes; - a = 1.0f; - b = (rgb & 0x001F) * (1.0f / 0x001F); - g = (rgb & 0x07E0) * (1.0f / 0x07E0); - r = (rgb & 0xF800) * (1.0f / 0xF800); + fastCopyFunc(src, dest); } - break; - case D3DFMT_A1R5G5B5: - { - unsigned short argb = *(unsigned short*)(source + 2 * i + j * inputPitch); - - a = (argb & 0x8000) ? 1.0f : 0.0f; - b = (argb & 0x001F) * (1.0f / 0x001F); - g = (argb & 0x03E0) * (1.0f / 0x03E0); - r = (argb & 0x7C00) * (1.0f / 0x7C00); - } - break; - case D3DFMT_A8R8G8B8: - { - unsigned int argb = *(unsigned int*)(source + 4 * i + j * inputPitch); - - a = (argb & 0xFF000000) * (1.0f / 0xFF000000); - b = (argb & 0x000000FF) * (1.0f / 0x000000FF); - g = (argb & 0x0000FF00) * (1.0f / 0x0000FF00); - r = (argb & 0x00FF0000) * (1.0f / 0x00FF0000); - } - break; - case D3DFMT_X8R8G8B8: - { - unsigned int xrgb = *(unsigned int*)(source + 4 * i + j * inputPitch); - - a = 1.0f; - b = (xrgb & 0x000000FF) * (1.0f / 0x000000FF); - g = (xrgb & 0x0000FF00) * (1.0f / 0x0000FF00); - r = (xrgb & 0x00FF0000) * (1.0f / 0x00FF0000); - } - break; - case D3DFMT_A2R10G10B10: - { - unsigned int argb = *(unsigned int*)(source + 4 * i + j * inputPitch); - - a = (argb & 0xC0000000) * (1.0f / 0xC0000000); - b = (argb & 0x000003FF) * (1.0f / 0x000003FF); - g = (argb & 0x000FFC00) * (1.0f / 0x000FFC00); - r = (argb & 0x3FF00000) * (1.0f / 0x3FF00000); - } - break; - case D3DFMT_A32B32G32R32F: - { - // float formats in D3D are stored rgba, rather than the other way round - r = *((float*)(source + 16 * i + j * inputPitch) + 0); - g = *((float*)(source + 16 * i + j * inputPitch) + 1); - b = *((float*)(source + 16 * i + j * inputPitch) + 2); - a = *((float*)(source + 16 * i + j * inputPitch) + 3); - } - break; - case D3DFMT_A16B16G16R16F: - { - // float formats in D3D are stored rgba, rather than the other way round - r = gl::float16ToFloat32(*((unsigned short*)(source + 8 * i + j * inputPitch) + 0)); - g = gl::float16ToFloat32(*((unsigned short*)(source + 8 * i + j * inputPitch) + 1)); - b = gl::float16ToFloat32(*((unsigned short*)(source + 8 * i + j * inputPitch) + 2)); - a = gl::float16ToFloat32(*((unsigned short*)(source + 8 * i + j * inputPitch) + 3)); - } - break; - default: - UNIMPLEMENTED(); // FIXME - UNREACHABLE(); - return; } - - switch (format) + } + else + { + uint8_t temp[sizeof(gl::ColorF)]; + for (int y = 0; y < rect.bottom - rect.top; y++) { - case GL_RGBA: - switch (type) + for (int x = 0; x < rect.right - rect.left; x++) { - case GL_UNSIGNED_BYTE: - dest[4 * i + j * outputPitch + 0] = (unsigned char)(255 * r + 0.5f); - dest[4 * i + j * outputPitch + 1] = (unsigned char)(255 * g + 0.5f); - dest[4 * i + j * outputPitch + 2] = (unsigned char)(255 * b + 0.5f); - dest[4 * i + j * outputPitch + 3] = (unsigned char)(255 * a + 0.5f); - break; - default: UNREACHABLE(); - } - break; - case GL_BGRA_EXT: - switch (type) - { - case GL_UNSIGNED_BYTE: - dest[4 * i + j * outputPitch + 0] = (unsigned char)(255 * b + 0.5f); - dest[4 * i + j * outputPitch + 1] = (unsigned char)(255 * g + 0.5f); - dest[4 * i + j * outputPitch + 2] = (unsigned char)(255 * r + 0.5f); - dest[4 * i + j * outputPitch + 3] = (unsigned char)(255 * a + 0.5f); - break; - case GL_UNSIGNED_SHORT_4_4_4_4_REV_EXT: - // According to the desktop GL spec in the "Transfer of Pixel Rectangles" section - // this type is packed as follows: - // 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 - // -------------------------------------------------------------------------------- - // | 4th | 3rd | 2nd | 1st component | - // -------------------------------------------------------------------------------- - // in the case of BGRA_EXT, B is the first component, G the second, and so forth. - dest16[i + j * outputPitch / sizeof(unsigned short)] = - ((unsigned short)(15 * a + 0.5f) << 12)| - ((unsigned short)(15 * r + 0.5f) << 8) | - ((unsigned short)(15 * g + 0.5f) << 4) | - ((unsigned short)(15 * b + 0.5f) << 0); - break; - case GL_UNSIGNED_SHORT_1_5_5_5_REV_EXT: - // According to the desktop GL spec in the "Transfer of Pixel Rectangles" section - // this type is packed as follows: - // 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 - // -------------------------------------------------------------------------------- - // | 4th | 3rd | 2nd | 1st component | - // -------------------------------------------------------------------------------- - // in the case of BGRA_EXT, B is the first component, G the second, and so forth. - dest16[i + j * outputPitch / sizeof(unsigned short)] = - ((unsigned short)( a + 0.5f) << 15) | - ((unsigned short)(31 * r + 0.5f) << 10) | - ((unsigned short)(31 * g + 0.5f) << 5) | - ((unsigned short)(31 * b + 0.5f) << 0); - break; - default: UNREACHABLE(); - } - break; - case GL_RGB: - switch (type) - { - case GL_UNSIGNED_SHORT_5_6_5: - dest16[i + j * outputPitch / sizeof(unsigned short)] = - ((unsigned short)(31 * b + 0.5f) << 0) | - ((unsigned short)(63 * g + 0.5f) << 5) | - ((unsigned short)(31 * r + 0.5f) << 11); - break; - case GL_UNSIGNED_BYTE: - dest[3 * i + j * outputPitch + 0] = (unsigned char)(255 * r + 0.5f); - dest[3 * i + j * outputPitch + 1] = (unsigned char)(255 * g + 0.5f); - dest[3 * i + j * outputPitch + 2] = (unsigned char)(255 * b + 0.5f); - break; - default: UNREACHABLE(); + uint8_t *dest = pixels + y * outputPitch + x * destFormatInfo.pixelBytes; + const uint8_t *src = source + y * inputPitch + x * sourceFormatInfo.pixelBytes; + + // readFunc and writeFunc will be using the same type of color, CopyTexImage + // will not allow the copy otherwise. + sourceD3DFormatInfo.colorReadFunction(src, temp); + destFormatTypeInfo.colorWriteFunction(temp, dest); } - break; - default: UNREACHABLE(); } } } systemSurface->UnlockRect(); + SafeRelease(systemSurface); - systemSurface->Release(); + return gl::Error(GL_NO_ERROR); } RenderTarget *Renderer9::createRenderTarget(SwapChain *swapChain, bool depth) @@ -3114,14 +2843,34 @@ RenderTarget *Renderer9::createRenderTarget(SwapChain *swapChain, bool depth) return renderTarget; } -RenderTarget *Renderer9::createRenderTarget(int width, int height, GLenum format, GLsizei samples, bool depth) +RenderTarget *Renderer9::createRenderTarget(int width, int height, GLenum format, GLsizei samples) { RenderTarget9 *renderTarget = new RenderTarget9(this, width, height, format, samples); return renderTarget; } -ShaderExecutable *Renderer9::loadExecutable(const void *function, size_t length, rx::ShaderType type) +ShaderImpl *Renderer9::createShader(GLenum type) +{ + return new ShaderD3D(type, this); +} + +ProgramImpl *Renderer9::createProgram() +{ + return new ProgramD3D(this); +} + +void Renderer9::releaseShaderCompiler() +{ + ShaderD3D::releaseCompiler(); +} + +ShaderExecutable *Renderer9::loadExecutable(const void *function, size_t length, rx::ShaderType type, + const std::vector<gl::LinkedVarying> &transformFeedbackVaryings, + bool separatedOutputBuffers) { + // Transform feedback is not supported in ES2 or D3D9 + ASSERT(transformFeedbackVaryings.size() == 0); + ShaderExecutable9 *executable = NULL; switch (type) @@ -3152,8 +2901,13 @@ ShaderExecutable *Renderer9::loadExecutable(const void *function, size_t length, return executable; } -ShaderExecutable *Renderer9::compileToExecutable(gl::InfoLog &infoLog, const char *shaderHLSL, rx::ShaderType type, D3DWorkaroundType workaround) +ShaderExecutable *Renderer9::compileToExecutable(gl::InfoLog &infoLog, const char *shaderHLSL, rx::ShaderType type, + const std::vector<gl::LinkedVarying> &transformFeedbackVaryings, + bool separatedOutputBuffers, D3DWorkaroundType workaround) { + // Transform feedback is not supported in ES2 or D3D9 + ASSERT(transformFeedbackVaryings.size() == 0); + const char *profile = NULL; switch (type) @@ -3169,20 +2923,67 @@ ShaderExecutable *Renderer9::compileToExecutable(gl::InfoLog &infoLog, const cha return NULL; } - // ANGLE issue 486: - // Work-around a D3D9 compiler bug that presents itself when using conditional discard, by disabling optimization - UINT optimizationFlags = (workaround == ANGLE_D3D_WORKAROUND_SM3_OPTIMIZER ? D3DCOMPILE_SKIP_OPTIMIZATION : ANGLE_COMPILE_OPTIMIZATION_LEVEL); + UINT flags = ANGLE_COMPILE_OPTIMIZATION_LEVEL; + + if (workaround == ANGLE_D3D_WORKAROUND_SKIP_OPTIMIZATION) + { + flags = D3DCOMPILE_SKIP_OPTIMIZATION; + } + else if (workaround == ANGLE_D3D_WORKAROUND_MAX_OPTIMIZATION) + { + flags = D3DCOMPILE_OPTIMIZATION_LEVEL3; + } + else ASSERT(workaround == ANGLE_D3D_WORKAROUND_NONE); + + if (gl::perfActive()) + { +#ifndef NDEBUG + flags = D3DCOMPILE_SKIP_OPTIMIZATION; +#endif + + flags |= D3DCOMPILE_DEBUG; + + std::string sourcePath = getTempPath(); + std::string sourceText = std::string("#line 2 \"") + sourcePath + std::string("\"\n\n") + std::string(shaderHLSL); + writeFile(sourcePath.c_str(), sourceText.c_str(), sourceText.size()); + } + + // Sometimes D3DCompile will fail with the default compilation flags for complicated shaders when it would otherwise pass with alternative options. + // Try the default flags first and if compilation fails, try some alternatives. + const UINT extraFlags[] = + { + flags, + flags | D3DCOMPILE_AVOID_FLOW_CONTROL, + flags | D3DCOMPILE_PREFER_FLOW_CONTROL + }; - ID3DBlob *binary = (ID3DBlob*)compileToBinary(infoLog, shaderHLSL, profile, optimizationFlags, true); + const static char *extraFlagNames[] = + { + "default", + "avoid flow control", + "prefer flow control" + }; + + int attempts = ArraySize(extraFlags); + + ID3DBlob *binary = (ID3DBlob*)mCompiler.compileToBinary(infoLog, shaderHLSL, profile, extraFlags, extraFlagNames, attempts); if (!binary) + { return NULL; + } - ShaderExecutable *executable = loadExecutable(binary->GetBufferPointer(), binary->GetBufferSize(), type); - binary->Release(); + ShaderExecutable *executable = loadExecutable(binary->GetBufferPointer(), binary->GetBufferSize(), type, + transformFeedbackVaryings, separatedOutputBuffers); + SafeRelease(binary); return executable; } +rx::UniformStorage *Renderer9::createUniformStorage(size_t storageSize) +{ + return new UniformStorage(storageSize); +} + bool Renderer9::boxFilter(IDirect3DSurface9 *source, IDirect3DSurface9 *dest) { return mBlit->boxFilter(source, dest); @@ -3223,7 +3024,7 @@ bool Renderer9::copyToRenderTarget(IDirect3DSurface9 *dest, IDirect3DSurface9 *s { Image9::copyLockableSurfaces(surf, source); result = mDevice->UpdateSurface(surf, NULL, dest, NULL); - surf->Release(); + SafeRelease(surf); } } else @@ -3260,14 +3061,42 @@ TextureStorage *Renderer9::createTextureStorage2D(SwapChain *swapChain) return new TextureStorage9_2D(this, swapChain9); } -TextureStorage *Renderer9::createTextureStorage2D(int levels, GLenum internalformat, GLenum usage, bool forceRenderable, GLsizei width, GLsizei height) +TextureStorage *Renderer9::createTextureStorage2D(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, int levels) +{ + return new TextureStorage9_2D(this, internalformat, renderTarget, width, height, levels); +} + +TextureStorage *Renderer9::createTextureStorageCube(GLenum internalformat, bool renderTarget, int size, int levels) +{ + return new TextureStorage9_Cube(this, internalformat, renderTarget, size, levels); +} + +TextureStorage *Renderer9::createTextureStorage3D(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, GLsizei depth, int levels) +{ + // 3D textures are not supported by the D3D9 backend. + UNREACHABLE(); + + return NULL; +} + +TextureStorage *Renderer9::createTextureStorage2DArray(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, GLsizei depth, int levels) { - return new TextureStorage9_2D(this, levels, internalformat, usage, forceRenderable, width, height); + // 2D array textures are not supported by the D3D9 backend. + UNREACHABLE(); + + return NULL; } -TextureStorage *Renderer9::createTextureStorageCube(int levels, GLenum internalformat, GLenum usage, bool forceRenderable, int size) +TextureImpl *Renderer9::createTexture(GLenum target) { - return new TextureStorage9_Cube(this, levels, internalformat, usage, forceRenderable, size); + switch(target) + { + case GL_TEXTURE_2D: return new TextureD3D_2D(this); + case GL_TEXTURE_CUBE_MAP: return new TextureD3D_Cube(this); + default: UNREACHABLE(); + } + + return NULL; } bool Renderer9::getLUID(LUID *adapterLuid) const @@ -3284,4 +3113,19 @@ bool Renderer9::getLUID(LUID *adapterLuid) const return false; } +rx::VertexConversionType Renderer9::getVertexConversionType(const gl::VertexFormat &vertexFormat) const +{ + return d3d9::GetVertexFormatInfo(getCapsDeclTypes(), vertexFormat).conversionType; +} + +GLenum Renderer9::getVertexComponentType(const gl::VertexFormat &vertexFormat) const +{ + return d3d9::GetVertexFormatInfo(getCapsDeclTypes(), vertexFormat).componentType; +} + +void Renderer9::generateCaps(gl::Caps *outCaps, gl::TextureCapsMap *outTextureCaps, gl::Extensions *outExtensions) const +{ + d3d9_gl::GenerateCaps(mD3d9, mDevice, mDeviceType, mAdapter, outCaps, outTextureCaps, outExtensions); +} + } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Renderer9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Renderer9.h new file mode 100644 index 0000000000..dd5f30268a --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/Renderer9.h @@ -0,0 +1,342 @@ +// +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// Renderer9.h: Defines a back-end specific class for the D3D9 renderer. + +#ifndef LIBGLESV2_RENDERER_RENDERER9_H_ +#define LIBGLESV2_RENDERER_RENDERER9_H_ + +#include "common/angleutils.h" +#include "common/mathutil.h" +#include "libGLESv2/renderer/d3d/HLSLCompiler.h" +#include "libGLESv2/renderer/d3d/d3d9/ShaderCache.h" +#include "libGLESv2/renderer/d3d/d3d9/VertexDeclarationCache.h" +#include "libGLESv2/renderer/Renderer.h" +#include "libGLESv2/renderer/RenderTarget.h" + +namespace gl +{ +class FramebufferAttachment; +} + +namespace rx +{ +class VertexDataManager; +class IndexDataManager; +class StreamingIndexBufferInterface; +class StaticIndexBufferInterface; +struct TranslatedAttribute; +class Blit9; + +class Renderer9 : public Renderer +{ + public: + Renderer9(egl::Display *display, EGLNativeDisplayType hDc, EGLint requestedDisplay); + virtual ~Renderer9(); + + static Renderer9 *makeRenderer9(Renderer *renderer); + + virtual EGLint initialize(); + virtual bool resetDevice(); + + virtual int generateConfigs(ConfigDesc **configDescList); + virtual void deleteConfigs(ConfigDesc *configDescList); + + void startScene(); + void endScene(); + + virtual void sync(bool block); + + virtual SwapChain *createSwapChain(HWND window, HANDLE shareHandle, GLenum backBufferFormat, GLenum depthBufferFormat); + + IDirect3DQuery9* allocateEventQuery(); + void freeEventQuery(IDirect3DQuery9* query); + + // resource creation + IDirect3DVertexShader9 *createVertexShader(const DWORD *function, size_t length); + IDirect3DPixelShader9 *createPixelShader(const DWORD *function, size_t length); + HRESULT createVertexBuffer(UINT Length, DWORD Usage, IDirect3DVertexBuffer9 **ppVertexBuffer); + HRESULT createIndexBuffer(UINT Length, DWORD Usage, D3DFORMAT Format, IDirect3DIndexBuffer9 **ppIndexBuffer); + virtual gl::Error generateSwizzle(gl::Texture *texture); + virtual gl::Error setSamplerState(gl::SamplerType type, int index, const gl::SamplerState &sampler); + virtual gl::Error setTexture(gl::SamplerType type, int index, gl::Texture *texture); + + virtual gl::Error setUniformBuffers(const gl::Buffer *vertexUniformBuffers[], const gl::Buffer *fragmentUniformBuffers[]); + + virtual gl::Error setRasterizerState(const gl::RasterizerState &rasterState); + virtual gl::Error setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState &blendState, const gl::ColorF &blendColor, + unsigned int sampleMask); + virtual gl::Error setDepthStencilState(const gl::DepthStencilState &depthStencilState, int stencilRef, + int stencilBackRef, bool frontFaceCCW); + + virtual void setScissorRectangle(const gl::Rectangle &scissor, bool enabled); + virtual void setViewport(const gl::Rectangle &viewport, float zNear, float zFar, GLenum drawMode, GLenum frontFace, + bool ignoreViewport); + + virtual gl::Error applyRenderTarget(gl::Framebuffer *frameBuffer); + virtual gl::Error applyShaders(gl::ProgramBinary *programBinary, const gl::VertexFormat inputLayout[], const gl::Framebuffer *framebuffer, + bool rasterizerDiscard, bool transformFeedbackActive); + virtual gl::Error applyUniforms(const gl::ProgramBinary &programBinary); + virtual bool applyPrimitiveType(GLenum primitiveType, GLsizei elementCount); + virtual gl::Error applyVertexBuffer(gl::ProgramBinary *programBinary, const gl::VertexAttribute vertexAttributes[], const gl::VertexAttribCurrentValueData currentValues[], + GLint first, GLsizei count, GLsizei instances); + virtual gl::Error applyIndexBuffer(const GLvoid *indices, gl::Buffer *elementArrayBuffer, GLsizei count, GLenum mode, GLenum type, TranslatedIndexData *indexInfo); + + virtual void applyTransformFeedbackBuffers(gl::Buffer *transformFeedbackBuffers[], GLintptr offsets[]); + + virtual gl::Error drawArrays(GLenum mode, GLsizei count, GLsizei instances, bool transformFeedbackActive); + virtual gl::Error drawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, + gl::Buffer *elementArrayBuffer, const TranslatedIndexData &indexInfo, GLsizei instances); + + virtual gl::Error clear(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer); + + virtual void markAllStateDirty(); + + // lost device + void notifyDeviceLost(); + virtual bool isDeviceLost(); + virtual bool testDeviceLost(bool notify); + virtual bool testDeviceResettable(); + + IDirect3DDevice9 *getDevice() { return mDevice; } + virtual DWORD getAdapterVendor() const; + virtual std::string getRendererDescription() const; + virtual GUID getAdapterIdentifier() const; + + virtual unsigned int getReservedVertexUniformVectors() const; + virtual unsigned int getReservedFragmentUniformVectors() const; + virtual unsigned int getReservedVertexUniformBuffers() const; + virtual unsigned int getReservedFragmentUniformBuffers() const; + virtual bool getShareHandleSupport() const; + virtual bool getPostSubBufferSupport() const; + + virtual int getMajorShaderModel() const; + DWORD getCapsDeclTypes() const; + virtual int getMinSwapInterval() const; + virtual int getMaxSwapInterval() const; + + // Pixel operations + virtual bool copyToRenderTarget2D(TextureStorage *dest, TextureStorage *source); + virtual bool copyToRenderTargetCube(TextureStorage *dest, TextureStorage *source); + virtual bool copyToRenderTarget3D(TextureStorage *dest, TextureStorage *source); + virtual bool copyToRenderTarget2DArray(TextureStorage *dest, TextureStorage *source); + + virtual bool copyImage2D(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, TextureStorage *storage, GLint level); + virtual bool copyImageCube(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, TextureStorage *storage, GLenum target, GLint level); + virtual bool copyImage3D(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, GLint zOffset, TextureStorage *storage, GLint level); + virtual bool copyImage2DArray(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, + GLint xoffset, GLint yoffset, GLint zOffset, TextureStorage *storage, GLint level); + + virtual bool blitRect(gl::Framebuffer *readTarget, const gl::Rectangle &readRect, gl::Framebuffer *drawTarget, const gl::Rectangle &drawRect, + const gl::Rectangle *scissor, bool blitRenderTarget, bool blitDepth, bool blitStencil, GLenum filter); + + virtual gl::Error readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, + GLenum type, GLuint outputPitch, const gl::PixelPackState &pack, uint8_t *pixels); + + // RenderTarget creation + virtual RenderTarget *createRenderTarget(SwapChain *swapChain, bool depth); + virtual RenderTarget *createRenderTarget(int width, int height, GLenum format, GLsizei samples); + + // Shader creation + virtual ShaderImpl *createShader(GLenum type); + virtual ProgramImpl *createProgram(); + + // Shader operations + virtual void releaseShaderCompiler(); + virtual ShaderExecutable *loadExecutable(const void *function, size_t length, rx::ShaderType type, + const std::vector<gl::LinkedVarying> &transformFeedbackVaryings, + bool separatedOutputBuffers); + virtual ShaderExecutable *compileToExecutable(gl::InfoLog &infoLog, const char *shaderHLSL, rx::ShaderType type, + const std::vector<gl::LinkedVarying> &transformFeedbackVaryings, + bool separatedOutputBuffers, D3DWorkaroundType workaround); + virtual UniformStorage *createUniformStorage(size_t storageSize); + + // Image operations + virtual Image *createImage(); + virtual void generateMipmap(Image *dest, Image *source); + virtual TextureStorage *createTextureStorage2D(SwapChain *swapChain); + virtual TextureStorage *createTextureStorage2D(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, int levels); + virtual TextureStorage *createTextureStorageCube(GLenum internalformat, bool renderTarget, int size, int levels); + virtual TextureStorage *createTextureStorage3D(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, GLsizei depth, int levels); + virtual TextureStorage *createTextureStorage2DArray(GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, GLsizei depth, int levels); + + // Texture creation + virtual TextureImpl *createTexture(GLenum target); + + // Buffer creation + virtual BufferImpl *createBuffer(); + virtual VertexBuffer *createVertexBuffer(); + virtual IndexBuffer *createIndexBuffer(); + + // Vertex Array creation + virtual VertexArrayImpl *createVertexArray(); + + // Query and Fence creation + virtual QueryImpl *createQuery(GLenum type); + virtual FenceImpl *createFence(); + + // Transform Feedback creation + virtual TransformFeedbackImpl* createTransformFeedback(); + + // Buffer-to-texture and Texture-to-buffer copies + virtual bool supportsFastCopyBufferToTexture(GLenum internalFormat) const; + virtual bool fastCopyBufferToTexture(const gl::PixelUnpackState &unpack, unsigned int offset, RenderTarget *destRenderTarget, + GLenum destinationFormat, GLenum sourcePixelsType, const gl::Box &destArea); + + // D3D9-renderer specific methods + bool boxFilter(IDirect3DSurface9 *source, IDirect3DSurface9 *dest); + + D3DPOOL getTexturePool(DWORD usage) const; + + virtual bool getLUID(LUID *adapterLuid) const; + virtual rx::VertexConversionType getVertexConversionType(const gl::VertexFormat &vertexFormat) const; + virtual GLenum getVertexComponentType(const gl::VertexFormat &vertexFormat) const; + + private: + DISALLOW_COPY_AND_ASSIGN(Renderer9); + + virtual void generateCaps(gl::Caps *outCaps, gl::TextureCapsMap *outTextureCaps, gl::Extensions *outExtensions) const; + + void release(); + + void applyUniformnfv(gl::LinkedUniform *targetUniform, const GLfloat *v); + void applyUniformniv(gl::LinkedUniform *targetUniform, const GLint *v); + void applyUniformnbv(gl::LinkedUniform *targetUniform, const GLint *v); + + gl::Error drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer); + gl::Error drawIndexedPoints(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer); + + gl::Error getCountingIB(size_t count, StaticIndexBufferInterface **outIB); + + bool copyToRenderTarget(IDirect3DSurface9 *dest, IDirect3DSurface9 *source, bool fromManaged); + gl::FramebufferAttachment *getNullColorbuffer(gl::FramebufferAttachment *depthbuffer); + + D3DPOOL getBufferPool(DWORD usage) const; + + HMODULE mD3d9Module; + HDC mDc; + + void initializeDevice(); + D3DPRESENT_PARAMETERS getDefaultPresentParameters(); + void releaseDeviceResources(); + + HRESULT getDeviceStatusCode(); + bool isRemovedDeviceResettable() const; + bool resetRemovedDevice(); + + UINT mAdapter; + D3DDEVTYPE mDeviceType; + IDirect3D9 *mD3d9; // Always valid after successful initialization. + IDirect3D9Ex *mD3d9Ex; // Might be null if D3D9Ex is not supported. + IDirect3DDevice9 *mDevice; + IDirect3DDevice9Ex *mDeviceEx; // Might be null if D3D9Ex is not supported. + + HLSLCompiler mCompiler; + + Blit9 *mBlit; + + HWND mDeviceWindow; + + bool mDeviceLost; + D3DCAPS9 mDeviceCaps; + D3DADAPTER_IDENTIFIER9 mAdapterIdentifier; + + D3DPRIMITIVETYPE mPrimitiveType; + int mPrimitiveCount; + GLsizei mRepeatDraw; + + bool mSceneStarted; + int mMinSwapInterval; + int mMaxSwapInterval; + + bool mVertexTextureSupport; + + // current render target states + unsigned int mAppliedRenderTargetSerial; + unsigned int mAppliedDepthbufferSerial; + unsigned int mAppliedStencilbufferSerial; + bool mDepthStencilInitialized; + bool mRenderTargetDescInitialized; + rx::RenderTarget::Desc mRenderTargetDesc; + unsigned int mCurStencilSize; + unsigned int mCurDepthSize; + + IDirect3DStateBlock9 *mMaskedClearSavedState; + + // previously set render states + bool mForceSetDepthStencilState; + gl::DepthStencilState mCurDepthStencilState; + int mCurStencilRef; + int mCurStencilBackRef; + bool mCurFrontFaceCCW; + + bool mForceSetRasterState; + gl::RasterizerState mCurRasterState; + + bool mForceSetScissor; + gl::Rectangle mCurScissor; + bool mScissorEnabled; + + bool mForceSetViewport; + gl::Rectangle mCurViewport; + float mCurNear; + float mCurFar; + float mCurDepthFront; + + bool mForceSetBlendState; + gl::BlendState mCurBlendState; + gl::ColorF mCurBlendColor; + GLuint mCurSampleMask; + + // Currently applied sampler states + std::vector<bool> mForceSetVertexSamplerStates; + std::vector<gl::SamplerState> mCurVertexSamplerStates; + + std::vector<bool> mForceSetPixelSamplerStates; + std::vector<gl::SamplerState> mCurPixelSamplerStates; + + // Currently applied textures + std::vector<unsigned int> mCurVertexTextureSerials; + std::vector<unsigned int> mCurPixelTextureSerials; + + unsigned int mAppliedIBSerial; + IDirect3DVertexShader9 *mAppliedVertexShader; + IDirect3DPixelShader9 *mAppliedPixelShader; + unsigned int mAppliedProgramSerial; + + rx::dx_VertexConstants mVertexConstants; + rx::dx_PixelConstants mPixelConstants; + bool mDxUniformsDirty; + + // A pool of event queries that are currently unused. + std::vector<IDirect3DQuery9*> mEventQueryPool; + VertexShaderCache mVertexShaderCache; + PixelShaderCache mPixelShaderCache; + + VertexDataManager *mVertexDataManager; + VertexDeclarationCache mVertexDeclarationCache; + + IndexDataManager *mIndexDataManager; + StreamingIndexBufferInterface *mLineLoopIB; + StaticIndexBufferInterface *mCountingIB; + + enum { NUM_NULL_COLORBUFFER_CACHE_ENTRIES = 12 }; + struct NullColorbufferCacheEntry + { + UINT lruCount; + int width; + int height; + gl::FramebufferAttachment *buffer; + } mNullColorbufferCache[NUM_NULL_COLORBUFFER_CACHE_ENTRIES]; + UINT mMaxNullColorbufferLRU; + +}; + +} +#endif // LIBGLESV2_RENDERER_RENDERER9_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/ShaderCache.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/ShaderCache.h index 4391ac271a..2ad3022839 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/ShaderCache.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/ShaderCache.h @@ -12,6 +12,10 @@ #include "common/debug.h" +#include <cstddef> +#include <unordered_map> +#include <string> + namespace rx { template <typename ShaderObject> @@ -53,7 +57,7 @@ class ShaderCache // Random eviction policy. if (mMap.size() >= kMaxMapSize) { - mMap.begin()->second->Release(); + SafeRelease(mMap.begin()->second); mMap.erase(mMap.begin()); } @@ -67,7 +71,7 @@ class ShaderCache { for (typename Map::iterator it = mMap.begin(); it != mMap.end(); ++it) { - it->second->Release(); + SafeRelease(it->second); } mMap.clear(); @@ -88,15 +92,7 @@ class ShaderCache return mDevice->CreatePixelShader(function, shader); } -#ifndef HASH_MAP -# ifdef _MSC_VER -# define HASH_MAP stdext::hash_map -# else -# define HASH_MAP std::unordered_map -# endif -#endif - - typedef HASH_MAP<std::string, ShaderObject*> Map; + typedef std::unordered_map<std::string, ShaderObject*> Map; Map mMap; IDirect3DDevice9 *mDevice; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/ShaderExecutable9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/ShaderExecutable9.cpp index 5decf9664d..bc7120461b 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/ShaderExecutable9.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/ShaderExecutable9.cpp @@ -1,4 +1,3 @@ -#include "precompiled.h" // // Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be @@ -8,7 +7,7 @@ // ShaderExecutable9.cpp: Implements a D3D9-specific class to contain shader // executable implementation details. -#include "libGLESv2/renderer/d3d9/ShaderExecutable9.h" +#include "libGLESv2/renderer/d3d/d3d9/ShaderExecutable9.h" #include "common/debug.h" @@ -31,14 +30,8 @@ ShaderExecutable9::ShaderExecutable9(const void *function, size_t length, IDirec ShaderExecutable9::~ShaderExecutable9() { - if (mVertexExecutable) - { - mVertexExecutable->Release(); - } - if (mPixelExecutable) - { - mPixelExecutable->Release(); - } + SafeRelease(mVertexExecutable); + SafeRelease(mPixelExecutable); } ShaderExecutable9 *ShaderExecutable9::makeShaderExecutable9(ShaderExecutable *executable) diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/ShaderExecutable9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/ShaderExecutable9.h index fa1e6c2844..fa1e6c2844 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/ShaderExecutable9.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/ShaderExecutable9.h diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/SwapChain9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/SwapChain9.cpp index dd8895d18d..0aeaabb1ca 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/SwapChain9.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/SwapChain9.cpp @@ -1,15 +1,15 @@ -#include "precompiled.h" // -// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // // SwapChain9.cpp: Implements a back-end specific class for the D3D9 swap chain. -#include "libGLESv2/renderer/d3d9/SwapChain9.h" -#include "libGLESv2/renderer/d3d9/renderer9_utils.h" -#include "libGLESv2/renderer/d3d9/Renderer9.h" +#include "libGLESv2/renderer/d3d/d3d9/SwapChain9.h" +#include "libGLESv2/renderer/d3d/d3d9/renderer9_utils.h" +#include "libGLESv2/renderer/d3d/d3d9/formatutils9.h" +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" namespace rx { @@ -35,43 +35,21 @@ SwapChain9::~SwapChain9() void SwapChain9::release() { - if (mSwapChain) - { - mSwapChain->Release(); - mSwapChain = NULL; - } - - if (mBackBuffer) - { - mBackBuffer->Release(); - mBackBuffer = NULL; - } - - if (mDepthStencil) - { - mDepthStencil->Release(); - mDepthStencil = NULL; - } - - if (mRenderTarget) - { - mRenderTarget->Release(); - mRenderTarget = NULL; - } - - if (mOffscreenTexture) - { - mOffscreenTexture->Release(); - mOffscreenTexture = NULL; - } + SafeRelease(mSwapChain); + SafeRelease(mBackBuffer); + SafeRelease(mDepthStencil); + SafeRelease(mRenderTarget); + SafeRelease(mOffscreenTexture); if (mWindow) + { mShareHandle = NULL; + } } static DWORD convertInterval(EGLint interval) { -#if ANGLE_FORCE_VSYNC_OFF +#ifdef ANGLE_FORCE_VSYNC_OFF return D3DPRESENT_INTERVAL_IMMEDIATE; #else switch(interval) @@ -111,29 +89,10 @@ EGLint SwapChain9::reset(int backbufferWidth, int backbufferHeight, EGLint swapI // Release specific resources to free up memory for the new render target, while the // old render target still exists for the purpose of preserving its contents. - if (mSwapChain) - { - mSwapChain->Release(); - mSwapChain = NULL; - } - - if (mBackBuffer) - { - mBackBuffer->Release(); - mBackBuffer = NULL; - } - - if (mOffscreenTexture) - { - mOffscreenTexture->Release(); - mOffscreenTexture = NULL; - } - - if (mDepthStencil) - { - mDepthStencil->Release(); - mDepthStencil = NULL; - } + SafeRelease(mSwapChain); + SafeRelease(mBackBuffer); + SafeRelease(mOffscreenTexture); + SafeRelease(mDepthStencil); HANDLE *pShareHandle = NULL; if (!mWindow && mRenderer->getShareHandleSupport()) @@ -141,9 +100,10 @@ EGLint SwapChain9::reset(int backbufferWidth, int backbufferHeight, EGLint swapI pShareHandle = &mShareHandle; } + const d3d9::TextureFormat &backBufferd3dFormatInfo = d3d9::GetTextureFormatInfo(mBackBufferFormat); result = device->CreateTexture(backbufferWidth, backbufferHeight, 1, D3DUSAGE_RENDERTARGET, - gl_d3d9::ConvertRenderbufferFormat(mBackBufferFormat), D3DPOOL_DEFAULT, - &mOffscreenTexture, pShareHandle); + backBufferd3dFormatInfo.texFormat, D3DPOOL_DEFAULT, &mOffscreenTexture, + pShareHandle); if (FAILED(result)) { ERR("Could not create offscreen texture: %08lX", result); @@ -187,15 +147,17 @@ EGLint SwapChain9::reset(int backbufferWidth, int backbufferHeight, EGLint swapI result = device->StretchRect(oldRenderTarget, &rect, mRenderTarget, &rect, D3DTEXF_NONE); ASSERT(SUCCEEDED(result)); - oldRenderTarget->Release(); + SafeRelease(oldRenderTarget); } + const d3d9::TextureFormat &depthBufferd3dFormatInfo = d3d9::GetTextureFormatInfo(mDepthBufferFormat); + if (mWindow) { D3DPRESENT_PARAMETERS presentParameters = {0}; - presentParameters.AutoDepthStencilFormat = gl_d3d9::ConvertRenderbufferFormat(mDepthBufferFormat); + presentParameters.AutoDepthStencilFormat = depthBufferd3dFormatInfo.renderFormat; presentParameters.BackBufferCount = 1; - presentParameters.BackBufferFormat = gl_d3d9::ConvertRenderbufferFormat(mBackBufferFormat); + presentParameters.BackBufferFormat = backBufferd3dFormatInfo.renderFormat; presentParameters.EnableAutoDepthStencil = FALSE; presentParameters.Flags = 0; presentParameters.hDeviceWindow = mWindow; @@ -247,7 +209,7 @@ EGLint SwapChain9::reset(int backbufferWidth, int backbufferHeight, EGLint swapI if (mDepthBufferFormat != GL_NONE) { result = device->CreateDepthStencilSurface(backbufferWidth, backbufferHeight, - gl_d3d9::ConvertRenderbufferFormat(mDepthBufferFormat), + depthBufferd3dFormatInfo.renderFormat, D3DMULTISAMPLE_NONE, 0, FALSE, &mDepthStencil, NULL); if (FAILED(result)) @@ -276,7 +238,7 @@ EGLint SwapChain9::reset(int backbufferWidth, int backbufferHeight, EGLint swapI } // parameters should be validated/clamped by caller -EGLint SwapChain9::swapRect(EGLint x, EGLint y, EGLint width, EGLint height) +EGLint SwapChain9::swapRect(EGLint x, EGLint y, EGLint width, EGLint height, EGLint) { if (!mSwapChain) { @@ -356,6 +318,14 @@ EGLint SwapChain9::swapRect(EGLint x, EGLint y, EGLint width, EGLint height) return EGL_BAD_ALLOC; } + // On Windows 8 systems, IDirect3DSwapChain9::Present sometimes returns 0x88760873 when the windows is + // in the process of entering/exiting fullscreen. This code doesn't seem to have any documentation. The + // device appears to be ok after emitting this error so simply return a failure to swap. + if (result == 0x88760873) + { + return EGL_BAD_MATCH; + } + // http://crbug.com/313210 // If our swap failed, trigger a device lost event. Resetting will work around an AMD-specific // device removed bug with lost contexts when reinstalling drivers. @@ -370,6 +340,7 @@ EGLint SwapChain9::swapRect(EGLint x, EGLint y, EGLint width, EGLint height) // Increments refcount on surface. // caller must Release() the returned surface +// TODO: remove the AddRef to match SwapChain11 IDirect3DSurface9 *SwapChain9::getRenderTarget() { if (mRenderTarget) @@ -382,6 +353,7 @@ IDirect3DSurface9 *SwapChain9::getRenderTarget() // Increments refcount on surface. // caller must Release() the returned surface +// TODO: remove the AddRef to match SwapChain11 IDirect3DSurface9 *SwapChain9::getDepthStencil() { if (mDepthStencil) @@ -394,6 +366,7 @@ IDirect3DSurface9 *SwapChain9::getDepthStencil() // Increments refcount on texture. // caller must Release() the returned texture +// TODO: remove the AddRef to match SwapChain11 IDirect3DTexture9 *SwapChain9::getOffscreenTexture() { if (mOffscreenTexture) @@ -434,10 +407,10 @@ void SwapChain9::recreate() return; } - mSwapChain->Release(); + SafeRelease(mSwapChain); mSwapChain = newSwapChain; - mBackBuffer->Release(); + SafeRelease(mBackBuffer); result = mSwapChain->GetBackBuffer(0, D3DBACKBUFFER_TYPE_MONO, &mBackBuffer); ASSERT(SUCCEEDED(result)); } diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/SwapChain9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/SwapChain9.h index 16a62bd86f..4d756f80d1 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/SwapChain9.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/SwapChain9.h @@ -25,7 +25,7 @@ class SwapChain9 : public SwapChain EGLint resize(EGLint backbufferWidth, EGLint backbufferHeight); virtual EGLint reset(EGLint backbufferWidth, EGLint backbufferHeight, EGLint swapInterval); - virtual EGLint swapRect(EGLint x, EGLint y, EGLint width, EGLint height); + virtual EGLint swapRect(EGLint x, EGLint y, EGLint width, EGLint height, EGLint); virtual void recreate(); virtual IDirect3DSurface9 *getRenderTarget(); diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/TextureStorage9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/TextureStorage9.cpp index 2486a9a5bf..f44e33db18 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/TextureStorage9.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/TextureStorage9.cpp @@ -1,6 +1,5 @@ -#include "precompiled.h" // -// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2012-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // @@ -9,18 +8,20 @@ // classes TextureStorage9_2D and TextureStorage9_Cube, which act as the interface to the // D3D9 texture. -#include "libGLESv2/main.h" -#include "libGLESv2/renderer/d3d9/Renderer9.h" -#include "libGLESv2/renderer/d3d9/TextureStorage9.h" -#include "libGLESv2/renderer/d3d9/SwapChain9.h" -#include "libGLESv2/renderer/d3d9/RenderTarget9.h" -#include "libGLESv2/renderer/d3d9/renderer9_utils.h" +#include "libGLESv2/renderer/d3d/d3d9/TextureStorage9.h" +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" +#include "libGLESv2/renderer/d3d/d3d9/SwapChain9.h" +#include "libGLESv2/renderer/d3d/d3d9/RenderTarget9.h" +#include "libGLESv2/renderer/d3d/d3d9/renderer9_utils.h" +#include "libGLESv2/renderer/d3d/d3d9/formatutils9.h" +#include "libGLESv2/renderer/d3d/TextureD3D.h" #include "libGLESv2/Texture.h" +#include "libGLESv2/main.h" namespace rx { TextureStorage9::TextureStorage9(Renderer *renderer, DWORD usage) - : mLodOffset(0), + : mTopLevel(0), mRenderer(Renderer9::makeRenderer9(renderer)), mD3DUsage(usage), mD3DPool(mRenderer->getTexturePool(usage)) @@ -37,46 +38,24 @@ TextureStorage9 *TextureStorage9::makeTextureStorage9(TextureStorage *storage) return static_cast<TextureStorage9*>(storage); } -DWORD TextureStorage9::GetTextureUsage(D3DFORMAT d3dfmt, GLenum glusage, bool forceRenderable) +DWORD TextureStorage9::GetTextureUsage(GLenum internalformat, bool renderTarget) { DWORD d3dusage = 0; - if (d3dfmt == D3DFMT_INTZ) + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(internalformat); + const d3d9::TextureFormat &d3dFormatInfo = d3d9::GetTextureFormatInfo(internalformat); + if (formatInfo.depthBits > 0 || formatInfo.stencilBits > 0) { d3dusage |= D3DUSAGE_DEPTHSTENCIL; } - else if(forceRenderable || (TextureStorage9::IsTextureFormatRenderable(d3dfmt) && (glusage == GL_FRAMEBUFFER_ATTACHMENT_ANGLE))) + else if (renderTarget && (d3dFormatInfo.renderFormat != D3DFMT_UNKNOWN)) { d3dusage |= D3DUSAGE_RENDERTARGET; } + return d3dusage; } -bool TextureStorage9::IsTextureFormatRenderable(D3DFORMAT format) -{ - if (format == D3DFMT_INTZ) - { - return true; - } - switch(format) - { - case D3DFMT_L8: - case D3DFMT_A8L8: - case D3DFMT_DXT1: - case D3DFMT_DXT3: - case D3DFMT_DXT5: - return false; - case D3DFMT_A8R8G8B8: - case D3DFMT_X8R8G8B8: - case D3DFMT_A16B16G16R16F: - case D3DFMT_A32B32G32R32F: - return true; - default: - UNREACHABLE(); - } - - return false; -} bool TextureStorage9::isRenderTarget() const { @@ -98,27 +77,29 @@ DWORD TextureStorage9::getUsage() const return mD3DUsage; } -int TextureStorage9::getLodOffset() const +int TextureStorage9::getTopLevel() const { - return mLodOffset; + return mTopLevel; } -int TextureStorage9::levelCount() +int TextureStorage9::getLevelCount() const { - return getBaseTexture() ? getBaseTexture()->GetLevelCount() - getLodOffset() : 0; + return getBaseTexture() ? (getBaseTexture()->GetLevelCount() - getTopLevel()) : 0; } -TextureStorage9_2D::TextureStorage9_2D(Renderer *renderer, SwapChain9 *swapchain) : TextureStorage9(renderer, D3DUSAGE_RENDERTARGET) +TextureStorage9_2D::TextureStorage9_2D(Renderer *renderer, SwapChain9 *swapchain) + : TextureStorage9(renderer, D3DUSAGE_RENDERTARGET) { IDirect3DTexture9 *surfaceTexture = swapchain->getOffscreenTexture(); mTexture = surfaceTexture; mRenderTarget = NULL; initializeRenderTarget(); + initializeSerials(1, 1); } -TextureStorage9_2D::TextureStorage9_2D(Renderer *renderer, int levels, GLenum internalformat, GLenum usage, bool forceRenderable, GLsizei width, GLsizei height) - : TextureStorage9(renderer, GetTextureUsage(Renderer9::makeRenderer9(renderer)->ConvertTextureInternalFormat(internalformat), usage, forceRenderable)) +TextureStorage9_2D::TextureStorage9_2D(Renderer *renderer, GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, int levels) + : TextureStorage9(renderer, GetTextureUsage(internalformat, renderTarget)) { mTexture = NULL; mRenderTarget = NULL; @@ -127,9 +108,11 @@ TextureStorage9_2D::TextureStorage9_2D(Renderer *renderer, int levels, GLenum in if (width > 0 && height > 0) { IDirect3DDevice9 *device = mRenderer->getDevice(); - gl::MakeValidSize(false, gl::IsCompressed(internalformat), &width, &height, &mLodOffset); - HRESULT result = device->CreateTexture(width, height, levels ? levels + mLodOffset : 0, getUsage(), - mRenderer->ConvertTextureInternalFormat(internalformat), getPool(), &mTexture, NULL); + const d3d9::TextureFormat &d3dFormatInfo = d3d9::GetTextureFormatInfo(internalformat); + d3d9::MakeValidSize(false, d3dFormatInfo.texFormat, &width, &height, &mTopLevel); + UINT creationLevels = (levels == 0) ? 0 : mTopLevel + levels; + + HRESULT result = device->CreateTexture(width, height, creationLevels, getUsage(), d3dFormatInfo.texFormat, getPool(), &mTexture, NULL); if (FAILED(result)) { @@ -139,16 +122,13 @@ TextureStorage9_2D::TextureStorage9_2D(Renderer *renderer, int levels, GLenum in } initializeRenderTarget(); + initializeSerials(getLevelCount(), 1); } TextureStorage9_2D::~TextureStorage9_2D() { - if (mTexture) - { - mTexture->Release(); - } - - delete mRenderTarget; + SafeRelease(mTexture); + SafeDelete(mRenderTarget); } TextureStorage9_2D *TextureStorage9_2D::makeTextureStorage9_2D(TextureStorage *storage) @@ -165,11 +145,12 @@ IDirect3DSurface9 *TextureStorage9_2D::getSurfaceLevel(int level, bool dirty) if (mTexture) { - HRESULT result = mTexture->GetSurfaceLevel(level + mLodOffset, &surface); + HRESULT result = mTexture->GetSurfaceLevel(level + mTopLevel, &surface); + UNUSED_ASSERTION_VARIABLE(result); ASSERT(SUCCEEDED(result)); // With managed textures the driver needs to be informed of updates to the lower mipmap levels - if (level + mLodOffset != 0 && isManaged() && dirty) + if (level + mTopLevel != 0 && isManaged() && dirty) { mTexture->AddDirtyRect(NULL); } @@ -178,23 +159,28 @@ IDirect3DSurface9 *TextureStorage9_2D::getSurfaceLevel(int level, bool dirty) return surface; } -RenderTarget *TextureStorage9_2D::getRenderTarget() +RenderTarget *TextureStorage9_2D::getRenderTarget(const gl::ImageIndex &/*index*/) { return mRenderTarget; } -void TextureStorage9_2D::generateMipmap(int level) +void TextureStorage9_2D::generateMipmaps() { - IDirect3DSurface9 *upper = getSurfaceLevel(level - 1, false); - IDirect3DSurface9 *lower = getSurfaceLevel(level, true); + // Base level must already be defined - if (upper != NULL && lower != NULL) + for (int level = 1; level < getLevelCount(); level++) { - mRenderer->boxFilter(upper, lower); - } + IDirect3DSurface9 *upper = getSurfaceLevel(level - 1, false); + IDirect3DSurface9 *lower = getSurfaceLevel(level, true); + + if (upper != NULL && lower != NULL) + { + mRenderer->boxFilter(upper, lower); + } - if (upper != NULL) upper->Release(); - if (lower != NULL) lower->Release(); + SafeRelease(upper); + SafeRelease(lower); + } } IDirect3DBaseTexture9 *TextureStorage9_2D::getBaseTexture() const @@ -214,8 +200,8 @@ void TextureStorage9_2D::initializeRenderTarget() } } -TextureStorage9_Cube::TextureStorage9_Cube(Renderer *renderer, int levels, GLenum internalformat, GLenum usage, bool forceRenderable, int size) - : TextureStorage9(renderer, GetTextureUsage(Renderer9::makeRenderer9(renderer)->ConvertTextureInternalFormat(internalformat), usage, forceRenderable)) +TextureStorage9_Cube::TextureStorage9_Cube(Renderer *renderer, GLenum internalformat, bool renderTarget, int size, int levels) + : TextureStorage9(renderer, GetTextureUsage(internalformat, renderTarget)) { mTexture = NULL; for (int i = 0; i < 6; ++i) @@ -229,9 +215,11 @@ TextureStorage9_Cube::TextureStorage9_Cube(Renderer *renderer, int levels, GLenu { IDirect3DDevice9 *device = mRenderer->getDevice(); int height = size; - gl::MakeValidSize(false, gl::IsCompressed(internalformat), &size, &height, &mLodOffset); - HRESULT result = device->CreateCubeTexture(size, levels ? levels + mLodOffset : 0, getUsage(), - mRenderer->ConvertTextureInternalFormat(internalformat), getPool(), &mTexture, NULL); + const d3d9::TextureFormat &d3dFormatInfo = d3d9::GetTextureFormatInfo(internalformat); + d3d9::MakeValidSize(false, d3dFormatInfo.texFormat, &size, &height, &mTopLevel); + UINT creationLevels = (levels == 0) ? 0 : mTopLevel + levels; + + HRESULT result = device->CreateCubeTexture(size, creationLevels, getUsage(), d3dFormatInfo.texFormat, getPool(), &mTexture, NULL); if (FAILED(result)) { @@ -241,18 +229,16 @@ TextureStorage9_Cube::TextureStorage9_Cube(Renderer *renderer, int levels, GLenu } initializeRenderTarget(); + initializeSerials(getLevelCount() * 6, 6); } TextureStorage9_Cube::~TextureStorage9_Cube() { - if (mTexture) - { - mTexture->Release(); - } + SafeRelease(mTexture); for (int i = 0; i < 6; ++i) { - delete mRenderTarget[i]; + SafeDelete(mRenderTarget[i]); } } @@ -271,7 +257,8 @@ IDirect3DSurface9 *TextureStorage9_Cube::getCubeMapSurface(GLenum faceTarget, in if (mTexture) { D3DCUBEMAP_FACES face = gl_d3d9::ConvertCubeFace(faceTarget); - HRESULT result = mTexture->GetCubeMapSurface(face, level + mLodOffset, &surface); + HRESULT result = mTexture->GetCubeMapSurface(face, level + mTopLevel, &surface); + UNUSED_ASSERTION_VARIABLE(result); ASSERT(SUCCEEDED(result)); // With managed textures the driver needs to be informed of updates to the lower mipmap levels @@ -284,23 +271,31 @@ IDirect3DSurface9 *TextureStorage9_Cube::getCubeMapSurface(GLenum faceTarget, in return surface; } -RenderTarget *TextureStorage9_Cube::getRenderTarget(GLenum faceTarget) +RenderTarget *TextureStorage9_Cube::getRenderTarget(const gl::ImageIndex &index) { - return mRenderTarget[gl::TextureCubeMap::faceIndex(faceTarget)]; + return mRenderTarget[index.layerIndex]; } -void TextureStorage9_Cube::generateMipmap(int face, int level) +void TextureStorage9_Cube::generateMipmaps() { - IDirect3DSurface9 *upper = getCubeMapSurface(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face, level - 1, false); - IDirect3DSurface9 *lower = getCubeMapSurface(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face, level, true); + // Base level must already be defined - if (upper != NULL && lower != NULL) + for (int faceIndex = 0; faceIndex < 6; faceIndex++) { - mRenderer->boxFilter(upper, lower); - } + for (int level = 1; level < getLevelCount(); level++) + { + IDirect3DSurface9 *upper = getCubeMapSurface(GL_TEXTURE_CUBE_MAP_POSITIVE_X + faceIndex, level - 1, false); + IDirect3DSurface9 *lower = getCubeMapSurface(GL_TEXTURE_CUBE_MAP_POSITIVE_X + faceIndex, level, true); - if (upper != NULL) upper->Release(); - if (lower != NULL) lower->Release(); + if (upper != NULL && lower != NULL) + { + mRenderer->boxFilter(upper, lower); + } + + SafeRelease(upper); + SafeRelease(lower); + } + } } IDirect3DBaseTexture9 *TextureStorage9_Cube::getBaseTexture() const diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/TextureStorage9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/TextureStorage9.h index 86f551a131..e698c7dd56 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/TextureStorage9.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/TextureStorage9.h @@ -1,5 +1,5 @@ // -// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // @@ -11,7 +11,7 @@ #ifndef LIBGLESV2_RENDERER_TEXTURESTORAGE9_H_ #define LIBGLESV2_RENDERER_TEXTURESTORAGE9_H_ -#include "libGLESv2/renderer/TextureStorage.h" +#include "libGLESv2/renderer/d3d/TextureStorage.h" #include "common/debug.h" namespace rx @@ -20,37 +20,34 @@ class Renderer9; class SwapChain9; class RenderTarget; class RenderTarget9; -class Blit; class TextureStorage9 : public TextureStorage { public: - TextureStorage9(Renderer *renderer, DWORD usage); virtual ~TextureStorage9(); static TextureStorage9 *makeTextureStorage9(TextureStorage *storage); - static DWORD GetTextureUsage(D3DFORMAT d3dfmt, GLenum glusage, bool forceRenderable); - static bool IsTextureFormatRenderable(D3DFORMAT format); + static DWORD GetTextureUsage(GLenum internalformat, bool renderTarget); D3DPOOL getPool() const; DWORD getUsage() const; virtual IDirect3DBaseTexture9 *getBaseTexture() const = 0; - virtual RenderTarget *getRenderTarget() { return NULL; } - virtual RenderTarget *getRenderTarget(GLenum faceTarget) { return NULL; } - virtual void generateMipmap(int level) {}; - virtual void generateMipmap(int face, int level) {}; + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index) = 0; + virtual void generateMipmaps() = 0; - virtual int getLodOffset() const; + virtual int getTopLevel() const; virtual bool isRenderTarget() const; virtual bool isManaged() const; - virtual int levelCount(); + virtual int getLevelCount() const; protected: - int mLodOffset; + int mTopLevel; Renderer9 *mRenderer; + TextureStorage9(Renderer *renderer, DWORD usage); + private: DISALLOW_COPY_AND_ASSIGN(TextureStorage9); @@ -62,15 +59,15 @@ class TextureStorage9_2D : public TextureStorage9 { public: TextureStorage9_2D(Renderer *renderer, SwapChain9 *swapchain); - TextureStorage9_2D(Renderer *renderer, int levels, GLenum internalformat, GLenum usage, bool forceRenderable, GLsizei width, GLsizei height); + TextureStorage9_2D(Renderer *renderer, GLenum internalformat, bool renderTarget, GLsizei width, GLsizei height, int levels); virtual ~TextureStorage9_2D(); static TextureStorage9_2D *makeTextureStorage9_2D(TextureStorage *storage); IDirect3DSurface9 *getSurfaceLevel(int level, bool dirty); - virtual RenderTarget *getRenderTarget(); + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index); virtual IDirect3DBaseTexture9 *getBaseTexture() const; - virtual void generateMipmap(int level); + virtual void generateMipmaps(); private: DISALLOW_COPY_AND_ASSIGN(TextureStorage9_2D); @@ -84,15 +81,15 @@ class TextureStorage9_2D : public TextureStorage9 class TextureStorage9_Cube : public TextureStorage9 { public: - TextureStorage9_Cube(Renderer *renderer, int levels, GLenum internalformat, GLenum usage, bool forceRenderable, int size); + TextureStorage9_Cube(Renderer *renderer, GLenum internalformat, bool renderTarget, int size, int levels); virtual ~TextureStorage9_Cube(); static TextureStorage9_Cube *makeTextureStorage9_Cube(TextureStorage *storage); IDirect3DSurface9 *getCubeMapSurface(GLenum faceTarget, int level, bool dirty); - virtual RenderTarget *getRenderTarget(GLenum faceTarget); + virtual RenderTarget *getRenderTarget(const gl::ImageIndex &index); virtual IDirect3DBaseTexture9 *getBaseTexture() const; - virtual void generateMipmap(int face, int level); + virtual void generateMipmaps(); private: DISALLOW_COPY_AND_ASSIGN(TextureStorage9_Cube); diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexArray9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexArray9.h new file mode 100644 index 0000000000..66a6c64d81 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexArray9.h @@ -0,0 +1,43 @@ +// +// Copyright 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// VertexArray9.h: Defines the rx::VertexArray9 class which implements rx::VertexArrayImpl. + +#ifndef LIBGLESV2_RENDERER_VERTEXARRAY9_H_ +#define LIBGLESV2_RENDERER_VERTEXARRAY9_H_ + +#include "libGLESv2/renderer/VertexArrayImpl.h" +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" + +namespace rx +{ +class Renderer9; + +class VertexArray9 : public VertexArrayImpl +{ + public: + VertexArray9(rx::Renderer9 *renderer) + : VertexArrayImpl(), + mRenderer(renderer) + { + } + + virtual ~VertexArray9() { } + + virtual void setElementArrayBuffer(const gl::Buffer *buffer) { } + virtual void setAttribute(size_t idx, const gl::VertexAttribute &attr) { } + virtual void setAttributeDivisor(size_t idx, GLuint divisor) { } + virtual void enableAttribute(size_t idx, bool enabledState) { } + + private: + DISALLOW_COPY_AND_ASSIGN(VertexArray9); + + rx::Renderer9 *mRenderer; +}; + +} + +#endif // LIBGLESV2_RENDERER_VERTEXARRAY9_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexBuffer9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexBuffer9.cpp new file mode 100644 index 0000000000..4cf7779118 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexBuffer9.cpp @@ -0,0 +1,233 @@ +// +// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// VertexBuffer9.cpp: Defines the D3D9 VertexBuffer implementation. + +#include "libGLESv2/renderer/d3d/d3d9/VertexBuffer9.h" +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" +#include "libGLESv2/renderer/d3d/d3d9/formatutils9.h" +#include "libGLESv2/renderer/vertexconversion.h" +#include "libGLESv2/renderer/BufferImpl.h" +#include "libGLESv2/VertexAttribute.h" +#include "libGLESv2/Buffer.h" + +namespace rx +{ + +VertexBuffer9::VertexBuffer9(rx::Renderer9 *renderer) : mRenderer(renderer) +{ + mVertexBuffer = NULL; + mBufferSize = 0; + mDynamicUsage = false; +} + +VertexBuffer9::~VertexBuffer9() +{ + SafeRelease(mVertexBuffer); +} + +gl::Error VertexBuffer9::initialize(unsigned int size, bool dynamicUsage) +{ + SafeRelease(mVertexBuffer); + + updateSerial(); + + if (size > 0) + { + DWORD flags = D3DUSAGE_WRITEONLY; + if (dynamicUsage) + { + flags |= D3DUSAGE_DYNAMIC; + } + + HRESULT result = mRenderer->createVertexBuffer(size, flags, &mVertexBuffer); + + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to allocate internal vertex buffer of size, %lu.", size); + } + } + + mBufferSize = size; + mDynamicUsage = dynamicUsage; + return gl::Error(GL_NO_ERROR); +} + +VertexBuffer9 *VertexBuffer9::makeVertexBuffer9(VertexBuffer *vertexBuffer) +{ + ASSERT(HAS_DYNAMIC_TYPE(VertexBuffer9*, vertexBuffer)); + return static_cast<VertexBuffer9*>(vertexBuffer); +} + +gl::Error VertexBuffer9::storeVertexAttributes(const gl::VertexAttribute &attrib, const gl::VertexAttribCurrentValueData ¤tValue, + GLint start, GLsizei count, GLsizei instances, unsigned int offset) +{ + if (!mVertexBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal vertex buffer is not initialized."); + } + + gl::Buffer *buffer = attrib.buffer.get(); + + int inputStride = gl::ComputeVertexAttributeStride(attrib); + int elementSize = gl::ComputeVertexAttributeTypeSize(attrib); + + DWORD lockFlags = mDynamicUsage ? D3DLOCK_NOOVERWRITE : 0; + + uint8_t *mapPtr = NULL; + + unsigned int mapSize; + gl::Error error = spaceRequired(attrib, count, instances, &mapSize); + if (error.isError()) + { + return error; + } + + HRESULT result = mVertexBuffer->Lock(offset, mapSize, reinterpret_cast<void**>(&mapPtr), lockFlags); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to lock internal vertex buffer, HRESULT: 0x%08x.", result); + } + + const uint8_t *input = NULL; + if (attrib.enabled) + { + if (buffer) + { + BufferImpl *storage = buffer->getImplementation(); + input = static_cast<const uint8_t*>(storage->getData()) + static_cast<int>(attrib.offset); + } + else + { + input = static_cast<const uint8_t*>(attrib.pointer); + } + } + else + { + input = reinterpret_cast<const uint8_t*>(currentValue.FloatValues); + } + + if (instances == 0 || attrib.divisor == 0) + { + input += inputStride * start; + } + + gl::VertexFormat vertexFormat(attrib, currentValue.Type); + const d3d9::VertexFormat &d3dVertexInfo = d3d9::GetVertexFormatInfo(mRenderer->getCapsDeclTypes(), vertexFormat); + bool needsConversion = (d3dVertexInfo.conversionType & VERTEX_CONVERT_CPU) > 0; + + if (!needsConversion && inputStride == elementSize) + { + size_t copySize = static_cast<size_t>(count) * static_cast<size_t>(inputStride); + memcpy(mapPtr, input, copySize); + } + else + { + d3dVertexInfo.copyFunction(input, inputStride, count, mapPtr); + } + + mVertexBuffer->Unlock(); + + return gl::Error(GL_NO_ERROR); +} + +gl::Error VertexBuffer9::getSpaceRequired(const gl::VertexAttribute &attrib, GLsizei count, GLsizei instances, + unsigned int *outSpaceRequired) const +{ + return spaceRequired(attrib, count, instances, outSpaceRequired); +} + +unsigned int VertexBuffer9::getBufferSize() const +{ + return mBufferSize; +} + +gl::Error VertexBuffer9::setBufferSize(unsigned int size) +{ + if (size > mBufferSize) + { + return initialize(size, mDynamicUsage); + } + else + { + return gl::Error(GL_NO_ERROR); + } +} + +gl::Error VertexBuffer9::discard() +{ + if (!mVertexBuffer) + { + return gl::Error(GL_OUT_OF_MEMORY, "Internal vertex buffer is not initialized."); + } + + void *dummy; + HRESULT result; + + result = mVertexBuffer->Lock(0, 1, &dummy, D3DLOCK_DISCARD); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to lock internal buffer for discarding, HRESULT: 0x%08x", result); + } + + result = mVertexBuffer->Unlock(); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to unlock internal buffer for discarding, HRESULT: 0x%08x", result); + } + + return gl::Error(GL_NO_ERROR); +} + +IDirect3DVertexBuffer9 * VertexBuffer9::getBuffer() const +{ + return mVertexBuffer; +} + +gl::Error VertexBuffer9::spaceRequired(const gl::VertexAttribute &attrib, std::size_t count, GLsizei instances, + unsigned int *outSpaceRequired) const +{ + gl::VertexFormat vertexFormat(attrib, GL_FLOAT); + const d3d9::VertexFormat &d3d9VertexInfo = d3d9::GetVertexFormatInfo(mRenderer->getCapsDeclTypes(), vertexFormat); + + if (attrib.enabled) + { + unsigned int elementCount = 0; + if (instances == 0 || attrib.divisor == 0) + { + elementCount = count; + } + else + { + // Round up to divisor, if possible + elementCount = rx::UnsignedCeilDivide(static_cast<unsigned int>(instances), attrib.divisor); + } + + if (d3d9VertexInfo.outputElementSize <= std::numeric_limits<unsigned int>::max() / elementCount) + { + if (outSpaceRequired) + { + *outSpaceRequired = d3d9VertexInfo.outputElementSize * elementCount; + } + return gl::Error(GL_NO_ERROR); + } + else + { + return gl::Error(GL_OUT_OF_MEMORY, "New vertex buffer size would result in an overflow."); + } + } + else + { + const unsigned int elementSize = 4; + if (outSpaceRequired) + { + *outSpaceRequired = elementSize * 4; + } + return gl::Error(GL_NO_ERROR); + } +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexBuffer9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexBuffer9.h new file mode 100644 index 0000000000..bdcf4bb64a --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexBuffer9.h @@ -0,0 +1,54 @@ +// +// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// VertexBuffer9.h: Defines the D3D9 VertexBuffer implementation. + +#ifndef LIBGLESV2_RENDERER_VERTEXBUFFER9_H_ +#define LIBGLESV2_RENDERER_VERTEXBUFFER9_H_ + +#include "libGLESv2/renderer/d3d/VertexBuffer.h" + +namespace rx +{ +class Renderer9; + +class VertexBuffer9 : public VertexBuffer +{ + public: + explicit VertexBuffer9(rx::Renderer9 *renderer); + virtual ~VertexBuffer9(); + + virtual gl::Error initialize(unsigned int size, bool dynamicUsage); + + static VertexBuffer9 *makeVertexBuffer9(VertexBuffer *vertexBuffer); + + virtual gl::Error storeVertexAttributes(const gl::VertexAttribute &attrib, const gl::VertexAttribCurrentValueData ¤tValue, + GLint start, GLsizei count, GLsizei instances, unsigned int offset); + + virtual gl::Error getSpaceRequired(const gl::VertexAttribute &attrib, GLsizei count, GLsizei instances, unsigned int *outSpaceRequired) const; + + virtual unsigned int getBufferSize() const; + virtual gl::Error setBufferSize(unsigned int size); + virtual gl::Error discard(); + + IDirect3DVertexBuffer9 *getBuffer() const; + + private: + DISALLOW_COPY_AND_ASSIGN(VertexBuffer9); + + rx::Renderer9 *mRenderer; + + IDirect3DVertexBuffer9 *mVertexBuffer; + unsigned int mBufferSize; + bool mDynamicUsage; + + gl::Error spaceRequired(const gl::VertexAttribute &attrib, std::size_t count, GLsizei instances, + unsigned int *outSpaceRequired) const; +}; + +} + +#endif // LIBGLESV2_RENDERER_VERTEXBUFFER9_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/VertexDeclarationCache.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexDeclarationCache.cpp index e5c8a14232..cefd786f11 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/VertexDeclarationCache.cpp +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexDeclarationCache.cpp @@ -1,4 +1,3 @@ -#include "precompiled.h" // // Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be @@ -7,10 +6,11 @@ // VertexDeclarationCache.cpp: Implements a helper class to construct and cache vertex declarations. +#include "libGLESv2/renderer/d3d/d3d9/VertexDeclarationCache.h" +#include "libGLESv2/renderer/d3d/d3d9/VertexBuffer9.h" +#include "libGLESv2/renderer/d3d/d3d9/formatutils9.h" #include "libGLESv2/ProgramBinary.h" -#include "libGLESv2/Context.h" -#include "libGLESv2/renderer/d3d9/VertexBuffer9.h" -#include "libGLESv2/renderer/d3d9/VertexDeclarationCache.h" +#include "libGLESv2/VertexAttribute.h" namespace rx { @@ -36,20 +36,31 @@ VertexDeclarationCache::~VertexDeclarationCache() { for (int i = 0; i < NUM_VERTEX_DECL_CACHE_ENTRIES; i++) { - if (mVertexDeclCache[i].vertexDeclaration) - { - mVertexDeclCache[i].vertexDeclaration->Release(); - } + SafeRelease(mVertexDeclCache[i].vertexDeclaration); } } -GLenum VertexDeclarationCache::applyDeclaration(IDirect3DDevice9 *device, TranslatedAttribute attributes[], gl::ProgramBinary *programBinary, GLsizei instances, GLsizei *repeatDraw) +gl::Error VertexDeclarationCache::applyDeclaration(IDirect3DDevice9 *device, TranslatedAttribute attributes[], gl::ProgramBinary *programBinary, GLsizei instances, GLsizei *repeatDraw) { *repeatDraw = 1; int indexedAttribute = gl::MAX_VERTEX_ATTRIBS; int instancedAttribute = gl::MAX_VERTEX_ATTRIBS; + if (instances == 0) + { + for (int i = 0; i < gl::MAX_VERTEX_ATTRIBS; ++i) + { + if (attributes[i].divisor != 0) + { + // If a divisor is set, it still applies even if an instanced draw was not used, so treat + // as a single-instance draw. + instances = 1; + break; + } + } + } + if (instances > 0) { // Find an indexed attribute to be mapped to D3D stream 0 @@ -70,12 +81,14 @@ GLenum VertexDeclarationCache::applyDeclaration(IDirect3DDevice9 *device, Transl } } - if (indexedAttribute == gl::MAX_VERTEX_ATTRIBS) - { - return GL_INVALID_OPERATION; - } + // The validation layer checks that there is at least one active attribute with a zero divisor as per + // the GL_ANGLE_instanced_arrays spec. + ASSERT(indexedAttribute != gl::MAX_VERTEX_ATTRIBS); } + D3DCAPS9 caps; + device->GetDeviceCaps(&caps); + D3DVERTEXELEMENT9 elements[gl::MAX_VERTEX_ATTRIBS + 1]; D3DVERTEXELEMENT9 *element = &elements[0]; @@ -134,9 +147,12 @@ GLenum VertexDeclarationCache::applyDeclaration(IDirect3DDevice9 *device, Transl mAppliedVBs[stream].offset = attributes[i].offset; } + gl::VertexFormat vertexFormat(*attributes[i].attribute, GL_FLOAT); + const d3d9::VertexFormat &d3d9VertexInfo = d3d9::GetVertexFormatInfo(caps.DeclTypes, vertexFormat); + element->Stream = stream; element->Offset = 0; - element->Type = attributes[i].attribute->mArrayEnabled ? vertexBuffer->getDeclType(*attributes[i].attribute) : D3DDECLTYPE_FLOAT4; + element->Type = d3d9VertexInfo.nativeFormat; element->Method = D3DDECLMETHOD_DEFAULT; element->Usage = D3DDECLUSAGE_TEXCOORD; element->UsageIndex = programBinary->getSemanticIndex(i); @@ -172,7 +188,7 @@ GLenum VertexDeclarationCache::applyDeclaration(IDirect3DDevice9 *device, Transl mLastSetVDecl = entry->vertexDeclaration; } - return GL_NO_ERROR; + return gl::Error(GL_NO_ERROR); } } @@ -188,19 +204,23 @@ GLenum VertexDeclarationCache::applyDeclaration(IDirect3DDevice9 *device, Transl if (lastCache->vertexDeclaration != NULL) { - lastCache->vertexDeclaration->Release(); - lastCache->vertexDeclaration = NULL; + SafeRelease(lastCache->vertexDeclaration); // mLastSetVDecl is set to the replacement, so we don't have to worry // about it. } memcpy(lastCache->cachedElements, elements, (element - elements) * sizeof(D3DVERTEXELEMENT9)); - device->CreateVertexDeclaration(elements, &lastCache->vertexDeclaration); + HRESULT result = device->CreateVertexDeclaration(elements, &lastCache->vertexDeclaration); + if (FAILED(result)) + { + return gl::Error(GL_OUT_OF_MEMORY, "Failed to create internal vertex declaration, result: 0x%X.", result); + } + device->SetVertexDeclaration(lastCache->vertexDeclaration); mLastSetVDecl = lastCache->vertexDeclaration; lastCache->lruCount = ++mMaxLru; - return GL_NO_ERROR; + return gl::Error(GL_NO_ERROR); } void VertexDeclarationCache::markStateDirty() diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/VertexDeclarationCache.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexDeclarationCache.h index 3fc024a9ba..9af36e0d7a 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/VertexDeclarationCache.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/VertexDeclarationCache.h @@ -9,7 +9,8 @@ #ifndef LIBGLESV2_RENDERER_VERTEXDECLARATIONCACHE_H_ #define LIBGLESV2_RENDERER_VERTEXDECLARATIONCACHE_H_ -#include "libGLESv2/renderer/VertexDataManager.h" +#include "libGLESv2/Error.h" +#include "libGLESv2/renderer/d3d/VertexDataManager.h" namespace gl { @@ -25,7 +26,7 @@ class VertexDeclarationCache VertexDeclarationCache(); ~VertexDeclarationCache(); - GLenum applyDeclaration(IDirect3DDevice9 *device, TranslatedAttribute attributes[], gl::ProgramBinary *programBinary, GLsizei instances, GLsizei *repeatDraw); + gl::Error applyDeclaration(IDirect3DDevice9 *device, TranslatedAttribute attributes[], gl::ProgramBinary *programBinary, GLsizei instances, GLsizei *repeatDraw); void markStateDirty(); diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/formatutils9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/formatutils9.cpp new file mode 100644 index 0000000000..f3acaf7987 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/formatutils9.cpp @@ -0,0 +1,588 @@ +// +// Copyright (c) 2013-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// formatutils9.cpp: Queries for GL image formats and their translations to D3D9 +// formats. + +#include "libGLESv2/renderer/d3d/d3d9/formatutils9.h" +#include "libGLESv2/renderer/d3d/d3d9/Renderer9.h" +#include "libGLESv2/renderer/generatemip.h" +#include "libGLESv2/renderer/loadimage.h" +#include "libGLESv2/renderer/copyimage.h" +#include "libGLESv2/renderer/vertexconversion.h" + +namespace rx +{ + +namespace d3d9 +{ + +const D3DFORMAT D3DFMT_INTZ = ((D3DFORMAT)(MAKEFOURCC('I', 'N', 'T', 'Z'))); +const D3DFORMAT D3DFMT_NULL = ((D3DFORMAT)(MAKEFOURCC('N', 'U', 'L', 'L'))); + +struct D3D9FastCopyFormat +{ + GLenum destFormat; + GLenum destType; + ColorCopyFunction copyFunction; + + D3D9FastCopyFormat(GLenum destFormat, GLenum destType, ColorCopyFunction copyFunction) + : destFormat(destFormat), destType(destType), copyFunction(copyFunction) + { } + + bool operator<(const D3D9FastCopyFormat& other) const + { + return memcmp(this, &other, sizeof(D3D9FastCopyFormat)) < 0; + } +}; + +typedef std::multimap<D3DFORMAT, D3D9FastCopyFormat> D3D9FastCopyMap; + +static D3D9FastCopyMap BuildFastCopyMap() +{ + D3D9FastCopyMap map; + + map.insert(std::make_pair(D3DFMT_A8R8G8B8, D3D9FastCopyFormat(GL_RGBA, GL_UNSIGNED_BYTE, CopyBGRA8ToRGBA8))); + + return map; +} + +// A map to determine the pixel size and mip generation function of a given D3D format +typedef std::map<D3DFORMAT, D3DFormat> D3D9FormatInfoMap; + +D3DFormat::D3DFormat() + : pixelBytes(0), + blockWidth(0), + blockHeight(0), + internalFormat(GL_NONE), + mipGenerationFunction(NULL), + colorReadFunction(NULL), + fastCopyFunctions() +{ +} + +ColorCopyFunction D3DFormat::getFastCopyFunction(GLenum format, GLenum type) const +{ + FastCopyFunctionMap::const_iterator iter = fastCopyFunctions.find(std::make_pair(format, type)); + return (iter != fastCopyFunctions.end()) ? iter->second : NULL; +} + +static inline void InsertD3DFormatInfo(D3D9FormatInfoMap *map, D3DFORMAT format, GLuint bits, GLuint blockWidth, + GLuint blockHeight, GLenum internalFormat, MipGenerationFunction mipFunc, + ColorReadFunction colorReadFunc) +{ + D3DFormat info; + info.pixelBytes = bits / 8; + info.blockWidth = blockWidth; + info.blockHeight = blockHeight; + info.internalFormat = internalFormat; + info.mipGenerationFunction = mipFunc; + info.colorReadFunction = colorReadFunc; + + static const D3D9FastCopyMap fastCopyMap = BuildFastCopyMap(); + std::pair<D3D9FastCopyMap::const_iterator, D3D9FastCopyMap::const_iterator> fastCopyIter = fastCopyMap.equal_range(format); + for (D3D9FastCopyMap::const_iterator i = fastCopyIter.first; i != fastCopyIter.second; i++) + { + info.fastCopyFunctions.insert(std::make_pair(std::make_pair(i->second.destFormat, i->second.destType), i->second.copyFunction)); + } + + map->insert(std::make_pair(format, info)); +} + +static D3D9FormatInfoMap BuildD3D9FormatInfoMap() +{ + D3D9FormatInfoMap map; + + // | D3DFORMAT | S |W |H | Internal format | Mip generation function | Color read function | + InsertD3DFormatInfo(&map, D3DFMT_NULL, 0, 0, 0, GL_NONE, NULL, NULL ); + InsertD3DFormatInfo(&map, D3DFMT_UNKNOWN, 0, 0, 0, GL_NONE, NULL, NULL ); + + InsertD3DFormatInfo(&map, D3DFMT_L8, 8, 1, 1, GL_LUMINANCE8_EXT, GenerateMip<L8>, ReadColor<L8, GLfloat> ); + InsertD3DFormatInfo(&map, D3DFMT_A8, 8, 1, 1, GL_ALPHA8_EXT, GenerateMip<A8>, ReadColor<A8, GLfloat> ); + InsertD3DFormatInfo(&map, D3DFMT_A8L8, 16, 1, 1, GL_LUMINANCE8_ALPHA8_EXT, GenerateMip<A8L8>, ReadColor<A8L8, GLfloat> ); + InsertD3DFormatInfo(&map, D3DFMT_A4R4G4B4, 16, 1, 1, GL_BGRA4_ANGLEX, GenerateMip<B4G4R4A4>, ReadColor<B4G4R4A4, GLfloat> ); + InsertD3DFormatInfo(&map, D3DFMT_A1R5G5B5, 16, 1, 1, GL_BGR5_A1_ANGLEX, GenerateMip<B5G5R5A1>, ReadColor<B5G5R5A1, GLfloat> ); + InsertD3DFormatInfo(&map, D3DFMT_R5G6B5, 16, 1, 1, GL_RGB565, GenerateMip<R5G6B5>, ReadColor<R5G6B5, GLfloat> ); + InsertD3DFormatInfo(&map, D3DFMT_X8R8G8B8, 32, 1, 1, GL_BGRA8_EXT, GenerateMip<B8G8R8X8>, ReadColor<B8G8R8X8, GLfloat> ); + InsertD3DFormatInfo(&map, D3DFMT_A8R8G8B8, 32, 1, 1, GL_BGRA8_EXT, GenerateMip<B8G8R8A8>, ReadColor<B8G8R8A8, GLfloat> ); + InsertD3DFormatInfo(&map, D3DFMT_R16F, 16, 1, 1, GL_R16F_EXT, GenerateMip<R16F>, ReadColor<R16F, GLfloat> ); + InsertD3DFormatInfo(&map, D3DFMT_G16R16F, 32, 1, 1, GL_RG16F_EXT, GenerateMip<R16G16F>, ReadColor<R16G16F, GLfloat> ); + InsertD3DFormatInfo(&map, D3DFMT_A16B16G16R16F, 64, 1, 1, GL_RGBA16F_EXT, GenerateMip<R16G16B16A16F>, ReadColor<R16G16B16A16F, GLfloat>); + InsertD3DFormatInfo(&map, D3DFMT_R32F, 32, 1, 1, GL_R32F_EXT, GenerateMip<R32F>, ReadColor<R32F, GLfloat> ); + InsertD3DFormatInfo(&map, D3DFMT_G32R32F, 64, 1, 1, GL_RG32F_EXT, GenerateMip<R32G32F>, ReadColor<R32G32F, GLfloat> ); + InsertD3DFormatInfo(&map, D3DFMT_A32B32G32R32F, 128, 1, 1, GL_RGBA32F_EXT, GenerateMip<R32G32B32A32F>, ReadColor<R32G32B32A32F, GLfloat>); + + InsertD3DFormatInfo(&map, D3DFMT_D16, 16, 1, 1, GL_DEPTH_COMPONENT16, NULL, NULL ); + InsertD3DFormatInfo(&map, D3DFMT_D24S8, 32, 1, 1, GL_DEPTH24_STENCIL8_OES, NULL, NULL ); + InsertD3DFormatInfo(&map, D3DFMT_D24X8, 32, 1, 1, GL_DEPTH_COMPONENT16, NULL, NULL ); + InsertD3DFormatInfo(&map, D3DFMT_D32, 32, 1, 1, GL_DEPTH_COMPONENT32_OES, NULL, NULL ); + + InsertD3DFormatInfo(&map, D3DFMT_INTZ, 32, 1, 1, GL_DEPTH24_STENCIL8_OES, NULL, NULL ); + + InsertD3DFormatInfo(&map, D3DFMT_DXT1, 64, 4, 4, GL_COMPRESSED_RGBA_S3TC_DXT1_EXT, NULL, NULL ); + InsertD3DFormatInfo(&map, D3DFMT_DXT3, 128, 4, 4, GL_COMPRESSED_RGBA_S3TC_DXT3_ANGLE, NULL, NULL ); + InsertD3DFormatInfo(&map, D3DFMT_DXT5, 128, 4, 4, GL_COMPRESSED_RGBA_S3TC_DXT5_ANGLE, NULL, NULL ); + + return map; +} + +const D3DFormat &GetD3DFormatInfo(D3DFORMAT format) +{ + static const D3D9FormatInfoMap infoMap = BuildD3D9FormatInfoMap(); + D3D9FormatInfoMap::const_iterator iter = infoMap.find(format); + if (iter != infoMap.end()) + { + return iter->second; + } + else + { + static const D3DFormat defaultInfo; + return defaultInfo; + } +} + + + +typedef std::pair<GLint, InitializeTextureDataFunction> InternalFormatInitialzerPair; +typedef std::map<GLint, InitializeTextureDataFunction> InternalFormatInitialzerMap; + +static InternalFormatInitialzerMap BuildInternalFormatInitialzerMap() +{ + InternalFormatInitialzerMap map; + + map.insert(InternalFormatInitialzerPair(GL_RGB16F, Initialize4ComponentData<GLhalf, 0x0000, 0x0000, 0x0000, gl::Float16One>)); + map.insert(InternalFormatInitialzerPair(GL_RGB32F, Initialize4ComponentData<GLfloat, 0x00000000, 0x00000000, 0x00000000, gl::Float32One>)); + + return map; +} + +// Each GL internal format corresponds to one D3D format and data loading function. +// Due to not all formats being available all the time, some of the function/format types are wrapped +// in templates that perform format support queries on a Renderer9 object which is supplied +// when requesting the function or format. + +typedef bool(*FallbackPredicateFunction)(); + +template <FallbackPredicateFunction pred, LoadImageFunction prefered, LoadImageFunction fallback> +static void FallbackLoad(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + if (pred()) + { + prefered(width, height, depth, input, inputRowPitch, inputDepthPitch, output, outputRowPitch, outputDepthPitch); + } + else + { + fallback(width, height, depth, input, inputRowPitch, inputDepthPitch, output, outputRowPitch, outputDepthPitch); + } +} + +static void UnreachableLoad(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + UNREACHABLE(); +} + +typedef std::pair<GLenum, TextureFormat> D3D9FormatPair; +typedef std::map<GLenum, TextureFormat> D3D9FormatMap; + +TextureFormat::TextureFormat() + : texFormat(D3DFMT_NULL), + renderFormat(D3DFMT_NULL), + dataInitializerFunction(NULL), + loadFunction(UnreachableLoad) +{ +} + +static inline void InsertD3D9FormatInfo(D3D9FormatMap *map, GLenum internalFormat, D3DFORMAT texFormat, + D3DFORMAT renderFormat, LoadImageFunction loadFunction) +{ + TextureFormat info; + info.texFormat = texFormat; + info.renderFormat = renderFormat; + + static const InternalFormatInitialzerMap dataInitializationMap = BuildInternalFormatInitialzerMap(); + InternalFormatInitialzerMap::const_iterator dataInitIter = dataInitializationMap.find(internalFormat); + info.dataInitializerFunction = (dataInitIter != dataInitializationMap.end()) ? dataInitIter->second : NULL; + + info.loadFunction = loadFunction; + + map->insert(std::make_pair(internalFormat, info)); +} + +static D3D9FormatMap BuildD3D9FormatMap() +{ + D3D9FormatMap map; + + // | Internal format | Texture format | Render format | Load function | + InsertD3D9FormatInfo(&map, GL_NONE, D3DFMT_NULL, D3DFMT_NULL, UnreachableLoad ); + + // We choose to downsample the GL_DEPTH_COMPONENT32_OES format to a 24-bit format because D3DFMT_D32 is not widely + // supported. We're allowed to do this because: + // - The ES spec 2.0.25 sec 3.7.1 states that we're allowed to store texture formats with internal format + // resolutions of our own choosing. + // - OES_depth_texture states that downsampling of the depth formats is allowed. + // - ANGLE_depth_texture does not state minimum required resolutions of the depth texture formats it + // introduces. + // In ES3 however, there are minimum resolutions for the texture formats and this would not be allowed. + + InsertD3D9FormatInfo(&map, GL_DEPTH_COMPONENT16, D3DFMT_INTZ, D3DFMT_D24S8, UnreachableLoad ); + InsertD3D9FormatInfo(&map, GL_DEPTH_COMPONENT32_OES, D3DFMT_INTZ, D3DFMT_D24X8, UnreachableLoad ); + InsertD3D9FormatInfo(&map, GL_DEPTH24_STENCIL8_OES, D3DFMT_INTZ, D3DFMT_D24S8, UnreachableLoad ); + InsertD3D9FormatInfo(&map, GL_STENCIL_INDEX8, D3DFMT_UNKNOWN, D3DFMT_D24S8, UnreachableLoad ); // TODO: What's the texture format? + + InsertD3D9FormatInfo(&map, GL_RGBA32F_EXT, D3DFMT_A32B32G32R32F, D3DFMT_A32B32G32R32F, LoadToNative<GLfloat, 4> ); + InsertD3D9FormatInfo(&map, GL_RGB32F_EXT, D3DFMT_A32B32G32R32F, D3DFMT_A32B32G32R32F, LoadToNative3To4<GLfloat, gl::Float32One>); + InsertD3D9FormatInfo(&map, GL_RG32F_EXT, D3DFMT_G32R32F, D3DFMT_G32R32F, LoadToNative<GLfloat, 2> ); + InsertD3D9FormatInfo(&map, GL_R32F_EXT, D3DFMT_R32F, D3DFMT_R32F, LoadToNative<GLfloat, 1> ); + InsertD3D9FormatInfo(&map, GL_ALPHA32F_EXT, D3DFMT_A32B32G32R32F, D3DFMT_UNKNOWN, LoadA32FToRGBA32F ); + InsertD3D9FormatInfo(&map, GL_LUMINANCE32F_EXT, D3DFMT_A32B32G32R32F, D3DFMT_UNKNOWN, LoadL32FToRGBA32F ); + InsertD3D9FormatInfo(&map, GL_LUMINANCE_ALPHA32F_EXT, D3DFMT_A32B32G32R32F, D3DFMT_UNKNOWN, LoadLA32FToRGBA32F ); + + InsertD3D9FormatInfo(&map, GL_RGBA16F_EXT, D3DFMT_A16B16G16R16F, D3DFMT_A16B16G16R16F, LoadToNative<GLhalf, 4> ); + InsertD3D9FormatInfo(&map, GL_RGB16F_EXT, D3DFMT_A16B16G16R16F, D3DFMT_A16B16G16R16F, LoadToNative3To4<GLhalf, gl::Float16One> ); + InsertD3D9FormatInfo(&map, GL_RG16F_EXT, D3DFMT_G16R16F, D3DFMT_G16R16F, LoadToNative<GLhalf, 2> ); + InsertD3D9FormatInfo(&map, GL_R16F_EXT, D3DFMT_R16F, D3DFMT_R16F, LoadToNative<GLhalf, 1> ); + InsertD3D9FormatInfo(&map, GL_ALPHA16F_EXT, D3DFMT_A16B16G16R16F, D3DFMT_UNKNOWN, LoadA16FToRGBA16F ); + InsertD3D9FormatInfo(&map, GL_LUMINANCE16F_EXT, D3DFMT_A16B16G16R16F, D3DFMT_UNKNOWN, LoadL16FToRGBA16F ); + InsertD3D9FormatInfo(&map, GL_LUMINANCE_ALPHA16F_EXT, D3DFMT_A16B16G16R16F, D3DFMT_UNKNOWN, LoadLA16FToRGBA16F ); + + InsertD3D9FormatInfo(&map, GL_ALPHA8_EXT, D3DFMT_A8R8G8B8, D3DFMT_A8R8G8B8, FallbackLoad<gl::supportsSSE2, LoadA8ToBGRA8_SSE2, LoadA8ToBGRA8>); + + InsertD3D9FormatInfo(&map, GL_RGB8_OES, D3DFMT_X8R8G8B8, D3DFMT_X8R8G8B8, LoadRGB8ToBGRX8 ); + InsertD3D9FormatInfo(&map, GL_RGB565, D3DFMT_X8R8G8B8, D3DFMT_X8R8G8B8, LoadR5G6B5ToBGRA8 ); + InsertD3D9FormatInfo(&map, GL_RGBA8_OES, D3DFMT_A8R8G8B8, D3DFMT_A8R8G8B8, FallbackLoad<gl::supportsSSE2, LoadRGBA8ToBGRA8_SSE2, LoadRGBA8ToBGRA8>); + InsertD3D9FormatInfo(&map, GL_RGBA4, D3DFMT_A8R8G8B8, D3DFMT_A8R8G8B8, LoadRGBA4ToBGRA8 ); + InsertD3D9FormatInfo(&map, GL_RGB5_A1, D3DFMT_A8R8G8B8, D3DFMT_A8R8G8B8, LoadRGB5A1ToBGRA8 ); + InsertD3D9FormatInfo(&map, GL_R8_EXT, D3DFMT_X8R8G8B8, D3DFMT_X8R8G8B8, LoadR8ToBGRX8 ); + InsertD3D9FormatInfo(&map, GL_RG8_EXT, D3DFMT_X8R8G8B8, D3DFMT_X8R8G8B8, LoadRG8ToBGRX8 ); + + InsertD3D9FormatInfo(&map, GL_BGRA8_EXT, D3DFMT_A8R8G8B8, D3DFMT_A8R8G8B8, LoadToNative<GLubyte, 4> ); + InsertD3D9FormatInfo(&map, GL_BGRA4_ANGLEX, D3DFMT_A8R8G8B8, D3DFMT_A8R8G8B8, LoadBGRA4ToBGRA8 ); + InsertD3D9FormatInfo(&map, GL_BGR5_A1_ANGLEX, D3DFMT_A8R8G8B8, D3DFMT_A8R8G8B8, LoadBGR5A1ToBGRA8 ); + + InsertD3D9FormatInfo(&map, GL_COMPRESSED_RGB_S3TC_DXT1_EXT, D3DFMT_DXT1, D3DFMT_UNKNOWN, LoadCompressedToNative<4, 4, 8> ); + InsertD3D9FormatInfo(&map, GL_COMPRESSED_RGBA_S3TC_DXT1_EXT, D3DFMT_DXT1, D3DFMT_UNKNOWN, LoadCompressedToNative<4, 4, 8> ); + InsertD3D9FormatInfo(&map, GL_COMPRESSED_RGBA_S3TC_DXT3_ANGLE, D3DFMT_DXT3, D3DFMT_UNKNOWN, LoadCompressedToNative<4, 4, 16> ); + InsertD3D9FormatInfo(&map, GL_COMPRESSED_RGBA_S3TC_DXT5_ANGLE, D3DFMT_DXT5, D3DFMT_UNKNOWN, LoadCompressedToNative<4, 4, 16> ); + + // These formats require checking if the renderer supports D3DFMT_L8 or D3DFMT_A8L8 and + // then changing the format and loading function appropriately. + InsertD3D9FormatInfo(&map, GL_LUMINANCE8_EXT, D3DFMT_L8, D3DFMT_UNKNOWN, LoadToNative<GLubyte, 1> ); + InsertD3D9FormatInfo(&map, GL_LUMINANCE8_ALPHA8_EXT, D3DFMT_A8L8, D3DFMT_UNKNOWN, LoadToNative<GLubyte, 2> ); + + return map; +} + +const TextureFormat &GetTextureFormatInfo(GLenum internalFormat) +{ + static const D3D9FormatMap formatMap = BuildD3D9FormatMap(); + D3D9FormatMap::const_iterator iter = formatMap.find(internalFormat); + if (iter != formatMap.end()) + { + return iter->second; + } + else + { + static const TextureFormat defaultInfo; + return defaultInfo; + } +} + +static GLenum GetDeclTypeComponentType(D3DDECLTYPE declType) +{ + switch (declType) + { + case D3DDECLTYPE_FLOAT1: return GL_FLOAT; + case D3DDECLTYPE_FLOAT2: return GL_FLOAT; + case D3DDECLTYPE_FLOAT3: return GL_FLOAT; + case D3DDECLTYPE_FLOAT4: return GL_FLOAT; + case D3DDECLTYPE_UBYTE4: return GL_UNSIGNED_INT; + case D3DDECLTYPE_SHORT2: return GL_INT; + case D3DDECLTYPE_SHORT4: return GL_INT; + case D3DDECLTYPE_UBYTE4N: return GL_UNSIGNED_NORMALIZED; + case D3DDECLTYPE_SHORT4N: return GL_SIGNED_NORMALIZED; + case D3DDECLTYPE_USHORT4N: return GL_UNSIGNED_NORMALIZED; + case D3DDECLTYPE_SHORT2N: return GL_SIGNED_NORMALIZED; + case D3DDECLTYPE_USHORT2N: return GL_UNSIGNED_NORMALIZED; + default: UNREACHABLE(); return GL_NONE; + } +} + +// Attribute format conversion +enum { NUM_GL_VERTEX_ATTRIB_TYPES = 6 }; + +struct TranslationDescription +{ + DWORD capsFlag; + VertexFormat preferredConversion; + VertexFormat fallbackConversion; +}; + +// Mapping from OpenGL-ES vertex attrib type to D3D decl type: +// +// BYTE SHORT (Cast) +// BYTE-norm FLOAT (Normalize) (can't be exactly represented as SHORT-norm) +// UNSIGNED_BYTE UBYTE4 (Identity) or SHORT (Cast) +// UNSIGNED_BYTE-norm UBYTE4N (Identity) or FLOAT (Normalize) +// SHORT SHORT (Identity) +// SHORT-norm SHORT-norm (Identity) or FLOAT (Normalize) +// UNSIGNED_SHORT FLOAT (Cast) +// UNSIGNED_SHORT-norm USHORT-norm (Identity) or FLOAT (Normalize) +// FIXED (not in WebGL) FLOAT (FixedToFloat) +// FLOAT FLOAT (Identity) + +// GLToCType maps from GL type (as GLenum) to the C typedef. +template <GLenum GLType> struct GLToCType { }; + +template <> struct GLToCType<GL_BYTE> { typedef GLbyte type; }; +template <> struct GLToCType<GL_UNSIGNED_BYTE> { typedef GLubyte type; }; +template <> struct GLToCType<GL_SHORT> { typedef GLshort type; }; +template <> struct GLToCType<GL_UNSIGNED_SHORT> { typedef GLushort type; }; +template <> struct GLToCType<GL_FIXED> { typedef GLuint type; }; +template <> struct GLToCType<GL_FLOAT> { typedef GLfloat type; }; + +// This differs from D3DDECLTYPE in that it is unsized. (Size expansion is applied last.) +enum D3DVertexType +{ + D3DVT_FLOAT, + D3DVT_SHORT, + D3DVT_SHORT_NORM, + D3DVT_UBYTE, + D3DVT_UBYTE_NORM, + D3DVT_USHORT_NORM +}; + +// D3DToCType maps from D3D vertex type (as enum D3DVertexType) to the corresponding C type. +template <unsigned int D3DType> struct D3DToCType { }; + +template <> struct D3DToCType<D3DVT_FLOAT> { typedef float type; }; +template <> struct D3DToCType<D3DVT_SHORT> { typedef short type; }; +template <> struct D3DToCType<D3DVT_SHORT_NORM> { typedef short type; }; +template <> struct D3DToCType<D3DVT_UBYTE> { typedef unsigned char type; }; +template <> struct D3DToCType<D3DVT_UBYTE_NORM> { typedef unsigned char type; }; +template <> struct D3DToCType<D3DVT_USHORT_NORM> { typedef unsigned short type; }; + +// Encode the type/size combinations that D3D permits. For each type/size it expands to a widener that will provide the appropriate final size. +template <unsigned int type, int size> struct WidenRule { }; + +template <int size> struct WidenRule<D3DVT_FLOAT, size> : NoWiden<size> { }; +template <int size> struct WidenRule<D3DVT_SHORT, size> : WidenToEven<size> { }; +template <int size> struct WidenRule<D3DVT_SHORT_NORM, size> : WidenToEven<size> { }; +template <int size> struct WidenRule<D3DVT_UBYTE, size> : WidenToFour<size> { }; +template <int size> struct WidenRule<D3DVT_UBYTE_NORM, size> : WidenToFour<size> { }; +template <int size> struct WidenRule<D3DVT_USHORT_NORM, size> : WidenToEven<size> { }; + +// VertexTypeFlags encodes the D3DCAPS9::DeclType flag and vertex declaration flag for each D3D vertex type & size combination. +template <unsigned int d3dtype, int size> struct VertexTypeFlags { }; + +template <unsigned int _capflag, unsigned int _declflag> +struct VertexTypeFlagsHelper +{ + enum { capflag = _capflag }; + enum { declflag = _declflag }; +}; + +template <> struct VertexTypeFlags<D3DVT_FLOAT, 1> : VertexTypeFlagsHelper<0, D3DDECLTYPE_FLOAT1> { }; +template <> struct VertexTypeFlags<D3DVT_FLOAT, 2> : VertexTypeFlagsHelper<0, D3DDECLTYPE_FLOAT2> { }; +template <> struct VertexTypeFlags<D3DVT_FLOAT, 3> : VertexTypeFlagsHelper<0, D3DDECLTYPE_FLOAT3> { }; +template <> struct VertexTypeFlags<D3DVT_FLOAT, 4> : VertexTypeFlagsHelper<0, D3DDECLTYPE_FLOAT4> { }; +template <> struct VertexTypeFlags<D3DVT_SHORT, 2> : VertexTypeFlagsHelper<0, D3DDECLTYPE_SHORT2> { }; +template <> struct VertexTypeFlags<D3DVT_SHORT, 4> : VertexTypeFlagsHelper<0, D3DDECLTYPE_SHORT4> { }; +template <> struct VertexTypeFlags<D3DVT_SHORT_NORM, 2> : VertexTypeFlagsHelper<D3DDTCAPS_SHORT2N, D3DDECLTYPE_SHORT2N> { }; +template <> struct VertexTypeFlags<D3DVT_SHORT_NORM, 4> : VertexTypeFlagsHelper<D3DDTCAPS_SHORT4N, D3DDECLTYPE_SHORT4N> { }; +template <> struct VertexTypeFlags<D3DVT_UBYTE, 4> : VertexTypeFlagsHelper<D3DDTCAPS_UBYTE4, D3DDECLTYPE_UBYTE4> { }; +template <> struct VertexTypeFlags<D3DVT_UBYTE_NORM, 4> : VertexTypeFlagsHelper<D3DDTCAPS_UBYTE4N, D3DDECLTYPE_UBYTE4N> { }; +template <> struct VertexTypeFlags<D3DVT_USHORT_NORM, 2> : VertexTypeFlagsHelper<D3DDTCAPS_USHORT2N, D3DDECLTYPE_USHORT2N> { }; +template <> struct VertexTypeFlags<D3DVT_USHORT_NORM, 4> : VertexTypeFlagsHelper<D3DDTCAPS_USHORT4N, D3DDECLTYPE_USHORT4N> { }; + + +// VertexTypeMapping maps GL type & normalized flag to preferred and fallback D3D vertex types (as D3DVertexType enums). +template <GLenum GLtype, bool normalized> struct VertexTypeMapping { }; + +template <D3DVertexType Preferred, D3DVertexType Fallback = Preferred> +struct VertexTypeMappingBase +{ + enum { preferred = Preferred }; + enum { fallback = Fallback }; +}; + +template <> struct VertexTypeMapping<GL_BYTE, false> : VertexTypeMappingBase<D3DVT_SHORT> { }; // Cast +template <> struct VertexTypeMapping<GL_BYTE, true> : VertexTypeMappingBase<D3DVT_FLOAT> { }; // Normalize +template <> struct VertexTypeMapping<GL_UNSIGNED_BYTE, false> : VertexTypeMappingBase<D3DVT_UBYTE, D3DVT_FLOAT> { }; // Identity, Cast +template <> struct VertexTypeMapping<GL_UNSIGNED_BYTE, true> : VertexTypeMappingBase<D3DVT_UBYTE_NORM, D3DVT_FLOAT> { }; // Identity, Normalize +template <> struct VertexTypeMapping<GL_SHORT, false> : VertexTypeMappingBase<D3DVT_SHORT> { }; // Identity +template <> struct VertexTypeMapping<GL_SHORT, true> : VertexTypeMappingBase<D3DVT_SHORT_NORM, D3DVT_FLOAT> { }; // Cast, Normalize +template <> struct VertexTypeMapping<GL_UNSIGNED_SHORT, false> : VertexTypeMappingBase<D3DVT_FLOAT> { }; // Cast +template <> struct VertexTypeMapping<GL_UNSIGNED_SHORT, true> : VertexTypeMappingBase<D3DVT_USHORT_NORM, D3DVT_FLOAT> { }; // Cast, Normalize +template <bool normalized> struct VertexTypeMapping<GL_FIXED, normalized> : VertexTypeMappingBase<D3DVT_FLOAT> { }; // FixedToFloat +template <bool normalized> struct VertexTypeMapping<GL_FLOAT, normalized> : VertexTypeMappingBase<D3DVT_FLOAT> { }; // Identity + + +// Given a GL type & norm flag and a D3D type, ConversionRule provides the type conversion rule (Cast, Normalize, Identity, FixedToFloat). +// The conversion rules themselves are defined in vertexconversion.h. + +// Almost all cases are covered by Cast (including those that are actually Identity since Cast<T,T> knows it's an identity mapping). +template <GLenum fromType, bool normalized, unsigned int toType> +struct ConversionRule : Cast<typename GLToCType<fromType>::type, typename D3DToCType<toType>::type> { }; + +// All conversions from normalized types to float use the Normalize operator. +template <GLenum fromType> struct ConversionRule<fromType, true, D3DVT_FLOAT> : Normalize<typename GLToCType<fromType>::type> { }; + +// Use a full specialization for this so that it preferentially matches ahead of the generic normalize-to-float rules. +template <> struct ConversionRule<GL_FIXED, true, D3DVT_FLOAT> : FixedToFloat<GLint, 16> { }; +template <> struct ConversionRule<GL_FIXED, false, D3DVT_FLOAT> : FixedToFloat<GLint, 16> { }; + +// A 2-stage construction is used for DefaultVertexValues because float must use SimpleDefaultValues (i.e. 0/1) +// whether it is normalized or not. +template <class T, bool normalized> struct DefaultVertexValuesStage2 { }; + +template <class T> struct DefaultVertexValuesStage2<T, true> : NormalizedDefaultValues<T> { }; +template <class T> struct DefaultVertexValuesStage2<T, false> : SimpleDefaultValues<T> { }; + +// Work out the default value rule for a D3D type (expressed as the C type) and +template <class T, bool normalized> struct DefaultVertexValues : DefaultVertexValuesStage2<T, normalized> { }; +template <bool normalized> struct DefaultVertexValues<float, normalized> : SimpleDefaultValues<float> { }; + +// Policy rules for use with Converter, to choose whether to use the preferred or fallback conversion. +// The fallback conversion produces an output that all D3D9 devices must support. +template <class T> struct UsePreferred { enum { type = T::preferred }; }; +template <class T> struct UseFallback { enum { type = T::fallback }; }; + +// Converter ties it all together. Given an OpenGL type/norm/size and choice of preferred/fallback conversion, +// it provides all the members of the appropriate VertexDataConverter, the D3DCAPS9::DeclTypes flag in cap flag +// and the D3DDECLTYPE member needed for the vertex declaration in declflag. +template <GLenum fromType, bool normalized, int size, template <class T> class PreferenceRule> +struct Converter + : VertexDataConverter<typename GLToCType<fromType>::type, + WidenRule<PreferenceRule< VertexTypeMapping<fromType, normalized> >::type, size>, + ConversionRule<fromType, + normalized, + PreferenceRule< VertexTypeMapping<fromType, normalized> >::type>, + DefaultVertexValues<typename D3DToCType<PreferenceRule< VertexTypeMapping<fromType, normalized> >::type>::type, normalized > > +{ +private: + enum { d3dtype = PreferenceRule< VertexTypeMapping<fromType, normalized> >::type }; + enum { d3dsize = WidenRule<d3dtype, size>::finalWidth }; + +public: + enum { capflag = VertexTypeFlags<d3dtype, d3dsize>::capflag }; + enum { declflag = VertexTypeFlags<d3dtype, d3dsize>::declflag }; +}; + +VertexFormat::VertexFormat() + : conversionType(VERTEX_CONVERT_NONE), + outputElementSize(0), + copyFunction(NULL), + nativeFormat(D3DDECLTYPE_UNUSED), + componentType(GL_NONE) +{ +} + +// Initialize a TranslationInfo +VertexFormat CreateVertexFormatInfo(bool identity, size_t elementSize, VertexCopyFunction copyFunc, D3DDECLTYPE nativeFormat) +{ + VertexFormat formatInfo; + formatInfo.conversionType = identity ? VERTEX_CONVERT_NONE : VERTEX_CONVERT_CPU; + formatInfo.outputElementSize = elementSize; + formatInfo.copyFunction = copyFunc; + formatInfo.nativeFormat = nativeFormat; + formatInfo.componentType = GetDeclTypeComponentType(nativeFormat); + return formatInfo; +} + +#define TRANSLATION(type, norm, size, preferred) \ + CreateVertexFormatInfo \ + ( \ + Converter<type, norm, size, preferred>::identity, \ + Converter<type, norm, size, preferred>::finalSize, \ + Converter<type, norm, size, preferred>::convertArray, \ + static_cast<D3DDECLTYPE>(Converter<type, norm, size, preferred>::declflag) \ + ) + +#define TRANSLATION_FOR_TYPE_NORM_SIZE(type, norm, size) \ + { \ + Converter<type, norm, size, UsePreferred>::capflag, \ + TRANSLATION(type, norm, size, UsePreferred), \ + TRANSLATION(type, norm, size, UseFallback) \ + } + +#define TRANSLATIONS_FOR_TYPE(type) \ + { \ + { TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 1), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 2), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 3), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 4) }, \ + { TRANSLATION_FOR_TYPE_NORM_SIZE(type, true, 1), TRANSLATION_FOR_TYPE_NORM_SIZE(type, true, 2), TRANSLATION_FOR_TYPE_NORM_SIZE(type, true, 3), TRANSLATION_FOR_TYPE_NORM_SIZE(type, true, 4) }, \ + } + +#define TRANSLATIONS_FOR_TYPE_NO_NORM(type) \ + { \ + { TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 1), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 2), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 3), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 4) }, \ + { TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 1), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 2), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 3), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 4) }, \ + } + +static inline unsigned int ComputeTypeIndex(GLenum type) +{ + switch (type) + { + case GL_BYTE: return 0; + case GL_UNSIGNED_BYTE: return 1; + case GL_SHORT: return 2; + case GL_UNSIGNED_SHORT: return 3; + case GL_FIXED: return 4; + case GL_FLOAT: return 5; + + default: UNREACHABLE(); return 5; + } +} + +const VertexFormat &GetVertexFormatInfo(DWORD supportedDeclTypes, const gl::VertexFormat &vertexFormat) +{ + static bool initialized = false; + static DWORD intializedDeclTypes = 0; + static VertexFormat formatConverters[NUM_GL_VERTEX_ATTRIB_TYPES][2][4]; + if (!initialized) + { + const TranslationDescription translations[NUM_GL_VERTEX_ATTRIB_TYPES][2][4] = // [GL types as enumerated by typeIndex()][normalized][size-1] + { + TRANSLATIONS_FOR_TYPE(GL_BYTE), + TRANSLATIONS_FOR_TYPE(GL_UNSIGNED_BYTE), + TRANSLATIONS_FOR_TYPE(GL_SHORT), + TRANSLATIONS_FOR_TYPE(GL_UNSIGNED_SHORT), + TRANSLATIONS_FOR_TYPE_NO_NORM(GL_FIXED), + TRANSLATIONS_FOR_TYPE_NO_NORM(GL_FLOAT) + }; + for (unsigned int i = 0; i < NUM_GL_VERTEX_ATTRIB_TYPES; i++) + { + for (unsigned int j = 0; j < 2; j++) + { + for (unsigned int k = 0; k < 4; k++) + { + if (translations[i][j][k].capsFlag == 0 || (supportedDeclTypes & translations[i][j][k].capsFlag) != 0) + { + formatConverters[i][j][k] = translations[i][j][k].preferredConversion; + } + else + { + formatConverters[i][j][k] = translations[i][j][k].fallbackConversion; + } + } + } + } + initialized = true; + intializedDeclTypes = supportedDeclTypes; + } + + ASSERT(intializedDeclTypes == supportedDeclTypes); + + // Pure integer attributes only supported in ES3.0 + ASSERT(!vertexFormat.mPureInteger); + return formatConverters[ComputeTypeIndex(vertexFormat.mType)][vertexFormat.mNormalized][vertexFormat.mComponents - 1]; +} + +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/formatutils9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/formatutils9.h new file mode 100644 index 0000000000..f26fe43b36 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/formatutils9.h @@ -0,0 +1,74 @@ +// +// Copyright (c) 2013-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// formatutils9.h: Queries for GL image formats and their translations to D3D9 +// formats. + +#ifndef LIBGLESV2_RENDERER_FORMATUTILS9_H_ +#define LIBGLESV2_RENDERER_FORMATUTILS9_H_ + +#include "libGLESv2/formatutils.h" + +#include <map> + +namespace rx +{ + +class Renderer9; + +namespace d3d9 +{ + +typedef std::map<std::pair<GLenum, GLenum>, ColorCopyFunction> FastCopyFunctionMap; + +struct D3DFormat +{ + D3DFormat(); + + GLuint pixelBytes; + GLuint blockWidth; + GLuint blockHeight; + + GLenum internalFormat; + + MipGenerationFunction mipGenerationFunction; + ColorReadFunction colorReadFunction; + + FastCopyFunctionMap fastCopyFunctions; + ColorCopyFunction getFastCopyFunction(GLenum format, GLenum type) const; +}; +const D3DFormat &GetD3DFormatInfo(D3DFORMAT format); + +struct VertexFormat +{ + VertexFormat(); + + VertexConversionType conversionType; + size_t outputElementSize; + VertexCopyFunction copyFunction; + D3DDECLTYPE nativeFormat; + GLenum componentType; +}; +const VertexFormat &GetVertexFormatInfo(DWORD supportedDeclTypes, const gl::VertexFormat &vertexFormat); + +struct TextureFormat +{ + TextureFormat(); + + D3DFORMAT texFormat; + D3DFORMAT renderFormat; + + InitializeTextureDataFunction dataInitializerFunction; + + LoadImageFunction loadFunction; +}; +const TextureFormat &GetTextureFormatInfo(GLenum internalFormat); + +} + +} + +#endif // LIBGLESV2_RENDERER_FORMATUTILS9_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/renderer9_utils.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/renderer9_utils.cpp new file mode 100644 index 0000000000..e7a91e62d6 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/renderer9_utils.cpp @@ -0,0 +1,542 @@ +// +// Copyright (c) 2002-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// renderer9_utils.cpp: Conversion functions and other utility routines +// specific to the D3D9 renderer. + +#include "libGLESv2/renderer/d3d/d3d9/renderer9_utils.h" +#include "libGLESv2/renderer/d3d/d3d9/formatutils9.h" +#include "libGLESv2/formatutils.h" +#include "libGLESv2/Framebuffer.h" +#include "libGLESv2/renderer/d3d/d3d9/RenderTarget9.h" + +#include "common/mathutil.h" +#include "common/debug.h" + +#include "third_party/systeminfo/SystemInfo.h" + +namespace rx +{ + +namespace gl_d3d9 +{ + +D3DCMPFUNC ConvertComparison(GLenum comparison) +{ + D3DCMPFUNC d3dComp = D3DCMP_ALWAYS; + switch (comparison) + { + case GL_NEVER: d3dComp = D3DCMP_NEVER; break; + case GL_ALWAYS: d3dComp = D3DCMP_ALWAYS; break; + case GL_LESS: d3dComp = D3DCMP_LESS; break; + case GL_LEQUAL: d3dComp = D3DCMP_LESSEQUAL; break; + case GL_EQUAL: d3dComp = D3DCMP_EQUAL; break; + case GL_GREATER: d3dComp = D3DCMP_GREATER; break; + case GL_GEQUAL: d3dComp = D3DCMP_GREATEREQUAL; break; + case GL_NOTEQUAL: d3dComp = D3DCMP_NOTEQUAL; break; + default: UNREACHABLE(); + } + + return d3dComp; +} + +D3DCOLOR ConvertColor(gl::ColorF color) +{ + return D3DCOLOR_RGBA(gl::unorm<8>(color.red), + gl::unorm<8>(color.green), + gl::unorm<8>(color.blue), + gl::unorm<8>(color.alpha)); +} + +D3DBLEND ConvertBlendFunc(GLenum blend) +{ + D3DBLEND d3dBlend = D3DBLEND_ZERO; + + switch (blend) + { + case GL_ZERO: d3dBlend = D3DBLEND_ZERO; break; + case GL_ONE: d3dBlend = D3DBLEND_ONE; break; + case GL_SRC_COLOR: d3dBlend = D3DBLEND_SRCCOLOR; break; + case GL_ONE_MINUS_SRC_COLOR: d3dBlend = D3DBLEND_INVSRCCOLOR; break; + case GL_DST_COLOR: d3dBlend = D3DBLEND_DESTCOLOR; break; + case GL_ONE_MINUS_DST_COLOR: d3dBlend = D3DBLEND_INVDESTCOLOR; break; + case GL_SRC_ALPHA: d3dBlend = D3DBLEND_SRCALPHA; break; + case GL_ONE_MINUS_SRC_ALPHA: d3dBlend = D3DBLEND_INVSRCALPHA; break; + case GL_DST_ALPHA: d3dBlend = D3DBLEND_DESTALPHA; break; + case GL_ONE_MINUS_DST_ALPHA: d3dBlend = D3DBLEND_INVDESTALPHA; break; + case GL_CONSTANT_COLOR: d3dBlend = D3DBLEND_BLENDFACTOR; break; + case GL_ONE_MINUS_CONSTANT_COLOR: d3dBlend = D3DBLEND_INVBLENDFACTOR; break; + case GL_CONSTANT_ALPHA: d3dBlend = D3DBLEND_BLENDFACTOR; break; + case GL_ONE_MINUS_CONSTANT_ALPHA: d3dBlend = D3DBLEND_INVBLENDFACTOR; break; + case GL_SRC_ALPHA_SATURATE: d3dBlend = D3DBLEND_SRCALPHASAT; break; + default: UNREACHABLE(); + } + + return d3dBlend; +} + +D3DBLENDOP ConvertBlendOp(GLenum blendOp) +{ + D3DBLENDOP d3dBlendOp = D3DBLENDOP_ADD; + + switch (blendOp) + { + case GL_FUNC_ADD: d3dBlendOp = D3DBLENDOP_ADD; break; + case GL_FUNC_SUBTRACT: d3dBlendOp = D3DBLENDOP_SUBTRACT; break; + case GL_FUNC_REVERSE_SUBTRACT: d3dBlendOp = D3DBLENDOP_REVSUBTRACT; break; + case GL_MIN_EXT: d3dBlendOp = D3DBLENDOP_MIN; break; + case GL_MAX_EXT: d3dBlendOp = D3DBLENDOP_MAX; break; + default: UNREACHABLE(); + } + + return d3dBlendOp; +} + +D3DSTENCILOP ConvertStencilOp(GLenum stencilOp) +{ + D3DSTENCILOP d3dStencilOp = D3DSTENCILOP_KEEP; + + switch (stencilOp) + { + case GL_ZERO: d3dStencilOp = D3DSTENCILOP_ZERO; break; + case GL_KEEP: d3dStencilOp = D3DSTENCILOP_KEEP; break; + case GL_REPLACE: d3dStencilOp = D3DSTENCILOP_REPLACE; break; + case GL_INCR: d3dStencilOp = D3DSTENCILOP_INCRSAT; break; + case GL_DECR: d3dStencilOp = D3DSTENCILOP_DECRSAT; break; + case GL_INVERT: d3dStencilOp = D3DSTENCILOP_INVERT; break; + case GL_INCR_WRAP: d3dStencilOp = D3DSTENCILOP_INCR; break; + case GL_DECR_WRAP: d3dStencilOp = D3DSTENCILOP_DECR; break; + default: UNREACHABLE(); + } + + return d3dStencilOp; +} + +D3DTEXTUREADDRESS ConvertTextureWrap(GLenum wrap) +{ + D3DTEXTUREADDRESS d3dWrap = D3DTADDRESS_WRAP; + + switch (wrap) + { + case GL_REPEAT: d3dWrap = D3DTADDRESS_WRAP; break; + case GL_CLAMP_TO_EDGE: d3dWrap = D3DTADDRESS_CLAMP; break; + case GL_MIRRORED_REPEAT: d3dWrap = D3DTADDRESS_MIRROR; break; + default: UNREACHABLE(); + } + + return d3dWrap; +} + +D3DCULL ConvertCullMode(GLenum cullFace, GLenum frontFace) +{ + D3DCULL cull = D3DCULL_CCW; + switch (cullFace) + { + case GL_FRONT: + cull = (frontFace == GL_CCW ? D3DCULL_CW : D3DCULL_CCW); + break; + case GL_BACK: + cull = (frontFace == GL_CCW ? D3DCULL_CCW : D3DCULL_CW); + break; + case GL_FRONT_AND_BACK: + cull = D3DCULL_NONE; // culling will be handled during draw + break; + default: UNREACHABLE(); + } + + return cull; +} + +D3DCUBEMAP_FACES ConvertCubeFace(GLenum cubeFace) +{ + D3DCUBEMAP_FACES face = D3DCUBEMAP_FACE_POSITIVE_X; + + switch (cubeFace) + { + case GL_TEXTURE_CUBE_MAP_POSITIVE_X: + face = D3DCUBEMAP_FACE_POSITIVE_X; + break; + case GL_TEXTURE_CUBE_MAP_NEGATIVE_X: + face = D3DCUBEMAP_FACE_NEGATIVE_X; + break; + case GL_TEXTURE_CUBE_MAP_POSITIVE_Y: + face = D3DCUBEMAP_FACE_POSITIVE_Y; + break; + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y: + face = D3DCUBEMAP_FACE_NEGATIVE_Y; + break; + case GL_TEXTURE_CUBE_MAP_POSITIVE_Z: + face = D3DCUBEMAP_FACE_POSITIVE_Z; + break; + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z: + face = D3DCUBEMAP_FACE_NEGATIVE_Z; + break; + default: UNREACHABLE(); + } + + return face; +} + +DWORD ConvertColorMask(bool red, bool green, bool blue, bool alpha) +{ + return (red ? D3DCOLORWRITEENABLE_RED : 0) | + (green ? D3DCOLORWRITEENABLE_GREEN : 0) | + (blue ? D3DCOLORWRITEENABLE_BLUE : 0) | + (alpha ? D3DCOLORWRITEENABLE_ALPHA : 0); +} + +D3DTEXTUREFILTERTYPE ConvertMagFilter(GLenum magFilter, float maxAnisotropy) +{ + if (maxAnisotropy > 1.0f) + { + return D3DTEXF_ANISOTROPIC; + } + + D3DTEXTUREFILTERTYPE d3dMagFilter = D3DTEXF_POINT; + switch (magFilter) + { + case GL_NEAREST: d3dMagFilter = D3DTEXF_POINT; break; + case GL_LINEAR: d3dMagFilter = D3DTEXF_LINEAR; break; + default: UNREACHABLE(); + } + + return d3dMagFilter; +} + +void ConvertMinFilter(GLenum minFilter, D3DTEXTUREFILTERTYPE *d3dMinFilter, D3DTEXTUREFILTERTYPE *d3dMipFilter, float maxAnisotropy) +{ + switch (minFilter) + { + case GL_NEAREST: + *d3dMinFilter = D3DTEXF_POINT; + *d3dMipFilter = D3DTEXF_NONE; + break; + case GL_LINEAR: + *d3dMinFilter = D3DTEXF_LINEAR; + *d3dMipFilter = D3DTEXF_NONE; + break; + case GL_NEAREST_MIPMAP_NEAREST: + *d3dMinFilter = D3DTEXF_POINT; + *d3dMipFilter = D3DTEXF_POINT; + break; + case GL_LINEAR_MIPMAP_NEAREST: + *d3dMinFilter = D3DTEXF_LINEAR; + *d3dMipFilter = D3DTEXF_POINT; + break; + case GL_NEAREST_MIPMAP_LINEAR: + *d3dMinFilter = D3DTEXF_POINT; + *d3dMipFilter = D3DTEXF_LINEAR; + break; + case GL_LINEAR_MIPMAP_LINEAR: + *d3dMinFilter = D3DTEXF_LINEAR; + *d3dMipFilter = D3DTEXF_LINEAR; + break; + default: + *d3dMinFilter = D3DTEXF_POINT; + *d3dMipFilter = D3DTEXF_NONE; + UNREACHABLE(); + } + + if (maxAnisotropy > 1.0f) + { + *d3dMinFilter = D3DTEXF_ANISOTROPIC; + } +} + +D3DMULTISAMPLE_TYPE GetMultisampleType(GLuint samples) +{ + return (samples > 1) ? static_cast<D3DMULTISAMPLE_TYPE>(samples) : D3DMULTISAMPLE_NONE; +} + +} + +namespace d3d9_gl +{ + +GLsizei GetSamplesCount(D3DMULTISAMPLE_TYPE type) +{ + return (type != D3DMULTISAMPLE_NONMASKABLE) ? type : 0; +} + +bool IsFormatChannelEquivalent(D3DFORMAT d3dformat, GLenum format) +{ + GLenum internalFormat = d3d9::GetD3DFormatInfo(d3dformat).internalFormat; + GLenum convertedFormat = gl::GetInternalFormatInfo(internalFormat).format; + return convertedFormat == format; +} + +static gl::TextureCaps GenerateTextureFormatCaps(GLenum internalFormat, IDirect3D9 *d3d9, D3DDEVTYPE deviceType, + UINT adapter, D3DFORMAT adapterFormat) +{ + gl::TextureCaps textureCaps; + + const d3d9::TextureFormat &d3dFormatInfo = d3d9::GetTextureFormatInfo(internalFormat); + const gl::InternalFormat &formatInfo = gl::GetInternalFormatInfo(internalFormat); + if (formatInfo.depthBits > 0 || formatInfo.stencilBits > 0) + { + textureCaps.texturable = SUCCEEDED(d3d9->CheckDeviceFormat(adapter, deviceType, adapterFormat, 0, D3DRTYPE_TEXTURE, d3dFormatInfo.texFormat)); + } + else + { + textureCaps.texturable = SUCCEEDED(d3d9->CheckDeviceFormat(adapter, deviceType, adapterFormat, 0, D3DRTYPE_TEXTURE, d3dFormatInfo.texFormat)) && + SUCCEEDED(d3d9->CheckDeviceFormat(adapter, deviceType, adapterFormat, 0, D3DRTYPE_CUBETEXTURE, d3dFormatInfo.texFormat)); + } + + textureCaps.filterable = SUCCEEDED(d3d9->CheckDeviceFormat(adapter, deviceType, adapterFormat, D3DUSAGE_QUERY_FILTER, D3DRTYPE_TEXTURE, d3dFormatInfo.texFormat)); + textureCaps.renderable = SUCCEEDED(d3d9->CheckDeviceFormat(adapter, deviceType, adapterFormat, D3DUSAGE_DEPTHSTENCIL, D3DRTYPE_TEXTURE, d3dFormatInfo.renderFormat)) || + SUCCEEDED(d3d9->CheckDeviceFormat(adapter, deviceType, adapterFormat, D3DUSAGE_RENDERTARGET, D3DRTYPE_TEXTURE, d3dFormatInfo.renderFormat)); + + textureCaps.sampleCounts.insert(1); + for (size_t i = D3DMULTISAMPLE_2_SAMPLES; i <= D3DMULTISAMPLE_16_SAMPLES; i++) + { + D3DMULTISAMPLE_TYPE multisampleType = D3DMULTISAMPLE_TYPE(i); + + HRESULT result = d3d9->CheckDeviceMultiSampleType(adapter, deviceType, d3dFormatInfo.renderFormat, TRUE, multisampleType, NULL); + if (SUCCEEDED(result)) + { + textureCaps.sampleCounts.insert(i); + } + } + + return textureCaps; +} + +void GenerateCaps(IDirect3D9 *d3d9, IDirect3DDevice9 *device, D3DDEVTYPE deviceType, UINT adapter, gl::Caps *caps, + gl::TextureCapsMap *textureCapsMap, gl::Extensions *extensions) +{ + D3DCAPS9 deviceCaps; + if (FAILED(d3d9->GetDeviceCaps(adapter, deviceType, &deviceCaps))) + { + // Can't continue with out device caps + return; + } + + D3DDISPLAYMODE currentDisplayMode; + d3d9->GetAdapterDisplayMode(adapter, ¤tDisplayMode); + + GLuint maxSamples = 0; + const gl::FormatSet &allFormats = gl::GetAllSizedInternalFormats(); + for (gl::FormatSet::const_iterator internalFormat = allFormats.begin(); internalFormat != allFormats.end(); ++internalFormat) + { + gl::TextureCaps textureCaps = GenerateTextureFormatCaps(*internalFormat, d3d9, deviceType, adapter, + currentDisplayMode.Format); + textureCapsMap->insert(*internalFormat, textureCaps); + + maxSamples = std::max(maxSamples, textureCaps.getMaxSamples()); + + if (gl::GetInternalFormatInfo(*internalFormat).compressed) + { + caps->compressedTextureFormats.push_back(*internalFormat); + } + } + + // GL core feature limits + caps->maxElementIndex = static_cast<GLint64>(std::numeric_limits<unsigned int>::max()); + + // 3D textures are unimplemented in D3D9 + caps->max3DTextureSize = 1; + + // Only one limit in GL, use the minimum dimension + caps->max2DTextureSize = std::min(deviceCaps.MaxTextureWidth, deviceCaps.MaxTextureHeight); + + // D3D treats cube maps as a special case of 2D textures + caps->maxCubeMapTextureSize = caps->max2DTextureSize; + + // Array textures are not available in D3D9 + caps->maxArrayTextureLayers = 1; + + // ES3-only feature + caps->maxLODBias = 0.0f; + + // No specific limits on render target size, maximum 2D texture size is equivalent + caps->maxRenderbufferSize = caps->max2DTextureSize; + + // Draw buffers are not supported in D3D9 + caps->maxDrawBuffers = 1; + caps->maxColorAttachments = 1; + + // No specific limits on viewport size, maximum 2D texture size is equivalent + caps->maxViewportWidth = caps->max2DTextureSize; + caps->maxViewportHeight = caps->maxViewportWidth; + + // Point size is clamped to 1.0f when the shader model is less than 3 + caps->minAliasedPointSize = 1.0f; + caps->maxAliasedPointSize = ((D3DSHADER_VERSION_MAJOR(deviceCaps.PixelShaderVersion) >= 3) ? deviceCaps.MaxPointSize : 1.0f); + + // Wide lines not supported + caps->minAliasedLineWidth = 1.0f; + caps->maxAliasedLineWidth = 1.0f; + + // Primitive count limits (unused in ES2) + caps->maxElementsIndices = 0; + caps->maxElementsVertices = 0; + + // Program and shader binary formats (no supported shader binary formats) + caps->programBinaryFormats.push_back(GL_PROGRAM_BINARY_ANGLE); + + // WaitSync is ES3-only, set to zero + caps->maxServerWaitTimeout = 0; + + // Vertex shader limits + caps->maxVertexAttributes = 16; + + const size_t reservedVertexUniformVectors = 2; // dx_ViewAdjust and dx_DepthRange. + const size_t MAX_VERTEX_CONSTANT_VECTORS_D3D9 = 256; + caps->maxVertexUniformVectors = MAX_VERTEX_CONSTANT_VECTORS_D3D9 - reservedVertexUniformVectors; + caps->maxVertexUniformComponents = caps->maxVertexUniformVectors * 4; + + caps->maxVertexUniformBlocks = 0; + + const size_t MAX_VERTEX_OUTPUT_VECTORS_SM3 = 10; + const size_t MAX_VERTEX_OUTPUT_VECTORS_SM2 = 8; + caps->maxVertexOutputComponents = ((deviceCaps.VertexShaderVersion >= D3DVS_VERSION(3, 0)) ? MAX_VERTEX_OUTPUT_VECTORS_SM3 + : MAX_VERTEX_OUTPUT_VECTORS_SM2) * 4; + + // Only Direct3D 10 ready devices support all the necessary vertex texture formats. + // We test this using D3D9 by checking support for the R16F format. + if (deviceCaps.VertexShaderVersion >= D3DVS_VERSION(3, 0) && + SUCCEEDED(d3d9->CheckDeviceFormat(adapter, deviceType, currentDisplayMode.Format, + D3DUSAGE_QUERY_VERTEXTEXTURE, D3DRTYPE_TEXTURE, D3DFMT_R16F))) + { + const size_t MAX_TEXTURE_IMAGE_UNITS_VTF_SM3 = 4; + caps->maxVertexTextureImageUnits = MAX_TEXTURE_IMAGE_UNITS_VTF_SM3; + } + else + { + caps->maxVertexTextureImageUnits = 0; + } + + // Fragment shader limits + const size_t reservedPixelUniformVectors = 3; // dx_ViewCoords, dx_DepthFront and dx_DepthRange. + + const size_t MAX_PIXEL_CONSTANT_VECTORS_SM3 = 224; + const size_t MAX_PIXEL_CONSTANT_VECTORS_SM2 = 32; + caps->maxFragmentUniformVectors = ((deviceCaps.PixelShaderVersion >= D3DPS_VERSION(3, 0)) ? MAX_PIXEL_CONSTANT_VECTORS_SM3 + : MAX_PIXEL_CONSTANT_VECTORS_SM2) - reservedPixelUniformVectors; + caps->maxFragmentUniformComponents = caps->maxFragmentUniformVectors * 4; + caps->maxFragmentUniformBlocks = 0; + caps->maxFragmentInputComponents = caps->maxVertexOutputComponents; + caps->maxTextureImageUnits = 16; + caps->minProgramTexelOffset = 0; + caps->maxProgramTexelOffset = 0; + + // Aggregate shader limits (unused in ES2) + caps->maxUniformBufferBindings = 0; + caps->maxUniformBlockSize = 0; + caps->uniformBufferOffsetAlignment = 0; + caps->maxCombinedUniformBlocks = 0; + caps->maxCombinedVertexUniformComponents = 0; + caps->maxCombinedFragmentUniformComponents = 0; + caps->maxVaryingComponents = 0; + + // Aggregate shader limits + caps->maxVaryingVectors = caps->maxVertexOutputComponents / 4; + caps->maxCombinedTextureImageUnits = caps->maxVertexTextureImageUnits + caps->maxTextureImageUnits; + + // Transform feedback limits + caps->maxTransformFeedbackInterleavedComponents = 0; + caps->maxTransformFeedbackSeparateAttributes = 0; + caps->maxTransformFeedbackSeparateComponents = 0; + + // GL extension support + extensions->setTextureExtensionSupport(*textureCapsMap); + extensions->elementIndexUint = deviceCaps.MaxVertexIndex >= (1 << 16); + extensions->packedDepthStencil = true; + extensions->getProgramBinary = true; + extensions->rgb8rgba8 = true; + extensions->readFormatBGRA = true; + extensions->pixelBufferObject = false; + extensions->mapBuffer = false; + extensions->mapBufferRange = false; + + // ATI cards on XP have problems with non-power-of-two textures. + D3DADAPTER_IDENTIFIER9 adapterId = { 0 }; + if (SUCCEEDED(d3d9->GetAdapterIdentifier(adapter, 0, &adapterId))) + { + extensions->textureNPOT = !(deviceCaps.TextureCaps & D3DPTEXTURECAPS_POW2) && + !(deviceCaps.TextureCaps & D3DPTEXTURECAPS_CUBEMAP_POW2) && + !(deviceCaps.TextureCaps & D3DPTEXTURECAPS_NONPOW2CONDITIONAL) && + !(isWindowsVistaOrGreater() && adapterId.VendorId == VENDOR_ID_AMD); + } + else + { + extensions->textureNPOT = false; + } + + extensions->drawBuffers = false; + extensions->textureStorage = true; + + // Must support a minimum of 2:1 anisotropy for max anisotropy to be considered supported, per the spec + extensions->textureFilterAnisotropic = (deviceCaps.RasterCaps & D3DPRASTERCAPS_ANISOTROPY) != 0 && deviceCaps.MaxAnisotropy >= 2; + extensions->maxTextureAnisotropy = static_cast<GLfloat>(deviceCaps.MaxAnisotropy); + + // Check occlusion query support by trying to create one + IDirect3DQuery9 *occlusionQuery = NULL; + extensions->occlusionQueryBoolean = SUCCEEDED(device->CreateQuery(D3DQUERYTYPE_OCCLUSION, &occlusionQuery)) && occlusionQuery; + SafeRelease(occlusionQuery); + + // Check event query support by trying to create one + IDirect3DQuery9 *eventQuery = NULL; + extensions->fence = SUCCEEDED(device->CreateQuery(D3DQUERYTYPE_EVENT, &eventQuery)) && eventQuery; + SafeRelease(eventQuery); + + extensions->timerQuery = false; // Unimplemented + extensions->robustness = true; + extensions->blendMinMax = true; + extensions->framebufferBlit = true; + extensions->framebufferMultisample = true; + extensions->maxSamples = maxSamples; + extensions->instancedArrays = deviceCaps.PixelShaderVersion >= D3DPS_VERSION(3, 0); + extensions->packReverseRowOrder = true; + extensions->standardDerivatives = (deviceCaps.PS20Caps.Caps & D3DPS20CAPS_GRADIENTINSTRUCTIONS) != 0; + extensions->shaderTextureLOD = true; + extensions->fragDepth = true; + extensions->textureUsage = true; + extensions->translatedShaderSource = true; + extensions->colorBufferFloat = false; +} + +} + +namespace d3d9 +{ + +GLuint ComputeBlockSize(D3DFORMAT format, GLuint width, GLuint height) +{ + const D3DFormat &d3dFormatInfo = d3d9::GetD3DFormatInfo(format); + GLuint numBlocksWide = (width + d3dFormatInfo.blockWidth - 1) / d3dFormatInfo.blockWidth; + GLuint numBlocksHight = (height + d3dFormatInfo.blockHeight - 1) / d3dFormatInfo.blockHeight; + return (d3dFormatInfo.pixelBytes * numBlocksWide * numBlocksHight); +} + +void MakeValidSize(bool isImage, D3DFORMAT format, GLsizei *requestWidth, GLsizei *requestHeight, int *levelOffset) +{ + const D3DFormat &d3dFormatInfo = d3d9::GetD3DFormatInfo(format); + + int upsampleCount = 0; + // Don't expand the size of full textures that are at least (blockWidth x blockHeight) already. + if (isImage || *requestWidth < static_cast<GLsizei>(d3dFormatInfo.blockWidth) || + *requestHeight < static_cast<GLsizei>(d3dFormatInfo.blockHeight)) + { + while (*requestWidth % d3dFormatInfo.blockWidth != 0 || *requestHeight % d3dFormatInfo.blockHeight != 0) + { + *requestWidth <<= 1; + *requestHeight <<= 1; + upsampleCount++; + } + } + *levelOffset = upsampleCount; +} + +RenderTarget9 *GetAttachmentRenderTarget(gl::FramebufferAttachment *attachment) +{ + RenderTarget *renderTarget = rx::GetAttachmentRenderTarget(attachment); + return RenderTarget9::makeRenderTarget9(renderTarget); +} + +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/renderer9_utils.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/renderer9_utils.h index bf6cdf1ea6..b0a940e60a 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/renderer9_utils.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/renderer9_utils.h @@ -1,25 +1,32 @@ // -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2002-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // -// renderer9_utils.h: Conversion functions and other utility routines +// renderer9_utils.h: Conversion functions and other utility routines // specific to the D3D9 renderer #ifndef LIBGLESV2_RENDERER_RENDERER9_UTILS_H #define LIBGLESV2_RENDERER_RENDERER9_UTILS_H -#include "libGLESv2/utilities.h" +#include "libGLESv2/angletypes.h" +#include "libGLESv2/Caps.h" -const D3DFORMAT D3DFMT_INTZ = ((D3DFORMAT)(MAKEFOURCC('I','N','T','Z'))); -const D3DFORMAT D3DFMT_NULL = ((D3DFORMAT)(MAKEFOURCC('N','U','L','L'))); +namespace gl +{ +class FramebufferAttachment; +} + +namespace rx +{ +class RenderTarget9; namespace gl_d3d9 { D3DCMPFUNC ConvertComparison(GLenum comparison); -D3DCOLOR ConvertColor(gl::Color color); +D3DCOLOR ConvertColor(gl::ColorF color); D3DBLEND ConvertBlendFunc(GLenum blend); D3DBLENDOP ConvertBlendOp(GLenum blendOp); D3DSTENCILOP ConvertStencilOp(GLenum stencilOp); @@ -29,31 +36,29 @@ D3DCUBEMAP_FACES ConvertCubeFace(GLenum cubeFace); DWORD ConvertColorMask(bool red, bool green, bool blue, bool alpha); D3DTEXTUREFILTERTYPE ConvertMagFilter(GLenum magFilter, float maxAnisotropy); void ConvertMinFilter(GLenum minFilter, D3DTEXTUREFILTERTYPE *d3dMinFilter, D3DTEXTUREFILTERTYPE *d3dMipFilter, float maxAnisotropy); -D3DFORMAT ConvertRenderbufferFormat(GLenum format); -D3DMULTISAMPLE_TYPE GetMultisampleTypeFromSamples(GLsizei samples); + +D3DMULTISAMPLE_TYPE GetMultisampleType(GLuint samples); } namespace d3d9_gl { -GLuint GetAlphaSize(D3DFORMAT colorFormat); -GLuint GetStencilSize(D3DFORMAT stencilFormat); - -GLsizei GetSamplesFromMultisampleType(D3DMULTISAMPLE_TYPE type); +GLsizei GetSamplesCount(D3DMULTISAMPLE_TYPE type); bool IsFormatChannelEquivalent(D3DFORMAT d3dformat, GLenum format); -GLenum ConvertBackBufferFormat(D3DFORMAT format); -GLenum ConvertDepthStencilFormat(D3DFORMAT format); -GLenum ConvertRenderTargetFormat(D3DFORMAT format); -GLenum GetEquivalentFormat(D3DFORMAT format); + +void GenerateCaps(IDirect3D9 *d3d9, IDirect3DDevice9 *device, D3DDEVTYPE deviceType, UINT adapter, gl::Caps *caps, + gl::TextureCapsMap *textureCapsMap, gl::Extensions *extensions); } namespace d3d9 { -bool IsCompressedFormat(D3DFORMAT format); -size_t ComputeRowSize(D3DFORMAT format, unsigned int width); + +GLuint ComputeBlockSize(D3DFORMAT format, GLuint width, GLuint height); + +void MakeValidSize(bool isImage, D3DFORMAT format, GLsizei *requestWidth, GLsizei *requestHeight, int *levelOffset); inline bool isDeviceLostError(HRESULT errorCode) { @@ -69,6 +74,10 @@ inline bool isDeviceLostError(HRESULT errorCode) } } +RenderTarget9 *GetAttachmentRenderTarget(gl::FramebufferAttachment *attachment); + +} + } #endif // LIBGLESV2_RENDERER_RENDERER9_UTILS_H diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/shaders/Blit.ps b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/shaders/Blit.ps new file mode 100644 index 0000000000..eb43eb3e7a --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/shaders/Blit.ps @@ -0,0 +1,33 @@ +// +// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +sampler2D tex : s0; + +uniform float4 mult : c0; +uniform float4 add : c1; + +// Passthrough Pixel Shader +// Outputs texture 0 sampled at texcoord 0. +float4 PS_passthrough(float4 texcoord : TEXCOORD0) : COLOR +{ + return tex2D(tex, texcoord.xy); +}; + +// Luminance Conversion Pixel Shader +// Performs a mad operation using the LA data from the texture with mult.xw and add.xw. +// Returns data in the form of llla +float4 PS_luminance(float4 texcoord : TEXCOORD0) : COLOR +{ + return (tex2D(tex, texcoord.xy).xw * mult.xw + add.xw).xxxy; +}; + +// RGB/A Component Mask Pixel Shader +// Performs a mad operation using the texture's RGBA data with mult.xyzw and add.xyzw. +// Returns data in the form of rgba +float4 PS_componentmask(float4 texcoord : TEXCOORD0) : COLOR +{ + return tex2D(tex, texcoord.xy) * mult + add; +}; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/shaders/Blit.vs b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/shaders/Blit.vs index 3a36980b93..3bd611ba5d 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/shaders/Blit.vs +++ b/src/3rdparty/angle/src/libGLESv2/renderer/d3d/d3d9/shaders/Blit.vs @@ -17,7 +17,7 @@ uniform float4 halfPixelSize : c0; // Outputs the homogenous position as-is. // Outputs a tex coord with (0,0) in the upper-left corner of the screen and (1,1) in the bottom right. // C0.X must be negative half-pixel width, C0.Y must be half-pixel height. C0.ZW must be 0. -VS_OUTPUT standardvs(in float4 position : POSITION) +VS_OUTPUT VS_standard(in float4 position : POSITION) { VS_OUTPUT Out; @@ -32,7 +32,7 @@ VS_OUTPUT standardvs(in float4 position : POSITION) // Outputs the homogenous position as-is. // Outputs a tex coord with (0,1) in the upper-left corner of the screen and (1,0) in the bottom right. // C0.XY must be the half-pixel width and height. C0.ZW must be 0. -VS_OUTPUT flipyvs(in float4 position : POSITION) +VS_OUTPUT VS_flipy(in float4 position : POSITION) { VS_OUTPUT Out; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/BufferStorage11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/BufferStorage11.cpp deleted file mode 100644 index 31d5b8b886..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/BufferStorage11.cpp +++ /dev/null @@ -1,366 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// BufferStorage11.cpp Defines the BufferStorage11 class. - -#include "libGLESv2/renderer/d3d11/BufferStorage11.h" -#include "libGLESv2/main.h" -#include "libGLESv2/renderer/d3d11/Renderer11.h" - -namespace rx -{ - -BufferStorage11::BufferStorage11(Renderer11 *renderer) -{ - mRenderer = renderer; - - mStagingBuffer = NULL; - mStagingBufferSize = 0; - - mSize = 0; - - mResolvedData = NULL; - mResolvedDataSize = 0; - mResolvedDataValid = false; - - mReadUsageCount = 0; - mWriteUsageCount = 0; -} - -BufferStorage11::~BufferStorage11() -{ - SafeRelease(mStagingBuffer); - - if (mResolvedData) - { - free(mResolvedData); - mResolvedData = NULL; - } - - for (auto it = mDirectBuffers.begin(); it != mDirectBuffers.end(); it++) - { - SafeDelete(it->second); - } -} - -BufferStorage11 *BufferStorage11::makeBufferStorage11(BufferStorage *bufferStorage) -{ - ASSERT(HAS_DYNAMIC_TYPE(BufferStorage11*, bufferStorage)); - return static_cast<BufferStorage11*>(bufferStorage); -} - -void *BufferStorage11::getData() -{ - ASSERT(mStagingBuffer); - - if (!mResolvedDataValid) - { - ID3D11Device *device = mRenderer->getDevice(); - ID3D11DeviceContext *context = mRenderer->getDeviceContext(); - HRESULT result; - - if (!mResolvedData || mResolvedDataSize < mStagingBufferSize) - { - free(mResolvedData); - mResolvedData = malloc(mSize); - mResolvedDataSize = mSize; - } - - D3D11_MAPPED_SUBRESOURCE mappedResource; - result = context->Map(mStagingBuffer, 0, D3D11_MAP_READ, 0, &mappedResource); - if (FAILED(result)) - { - return gl::error(GL_OUT_OF_MEMORY, (void*)NULL); - } - - memcpy(mResolvedData, mappedResource.pData, mSize); - - context->Unmap(mStagingBuffer, 0); - - mResolvedDataValid = true; - } - - mReadUsageCount = 0; - - return mResolvedData; -} - -void BufferStorage11::setData(const void* data, unsigned int size, unsigned int offset) -{ - ID3D11Device *device = mRenderer->getDevice(); - ID3D11DeviceContext *context = mRenderer->getDeviceContext(); - HRESULT result; - - const unsigned int requiredStagingBufferSize = size + offset; - const bool createStagingBuffer = !mStagingBuffer || mStagingBufferSize < requiredStagingBufferSize; - - if (createStagingBuffer) - { - D3D11_BUFFER_DESC bufferDesc; - bufferDesc.ByteWidth = requiredStagingBufferSize; - bufferDesc.Usage = D3D11_USAGE_STAGING; - bufferDesc.BindFlags = 0; - bufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; - bufferDesc.MiscFlags = 0; - bufferDesc.StructureByteStride = 0; - - HRESULT result; - ID3D11Device *device = mRenderer->getDevice(); - ID3D11DeviceContext *context = mRenderer->getDeviceContext(); - ID3D11Buffer *newStagingBuffer; - - if (data && offset == 0) - { - D3D11_SUBRESOURCE_DATA initialData; - initialData.pSysMem = data; - initialData.SysMemPitch = requiredStagingBufferSize; - initialData.SysMemSlicePitch = 0; - - result = device->CreateBuffer(&bufferDesc, &initialData, &newStagingBuffer); - } - else - { - result = device->CreateBuffer(&bufferDesc, NULL, &newStagingBuffer); - } - - if (FAILED(result)) - { - mStagingBufferSize = 0; - return gl::error(GL_OUT_OF_MEMORY); - } - - mStagingBufferSize = requiredStagingBufferSize; - - if (mStagingBuffer && offset > 0) - { - // If offset is greater than zero and the buffer is non-null, need to preserve the data from - // the old buffer up to offset - D3D11_BOX srcBox; - srcBox.left = 0; - srcBox.right = std::min(offset, requiredStagingBufferSize); - srcBox.top = 0; - srcBox.bottom = 1; - srcBox.front = 0; - srcBox.back = 1; - - context->CopySubresourceRegion(newStagingBuffer, 0, 0, 0, 0, mStagingBuffer, 0, &srcBox); - } - - SafeRelease(mStagingBuffer); - mStagingBuffer = newStagingBuffer; - } - - if (data && (offset != 0 || !createStagingBuffer)) - { - D3D11_MAPPED_SUBRESOURCE mappedResource; - result = context->Map(mStagingBuffer, 0, D3D11_MAP_WRITE, 0, &mappedResource); - if (FAILED(result)) - { - return gl::error(GL_OUT_OF_MEMORY); - } - - unsigned char *offsetBufferPointer = reinterpret_cast<unsigned char *>(mappedResource.pData) + offset; - memcpy(offsetBufferPointer, data, size); - - context->Unmap(mStagingBuffer, 0); - } - - for (auto it = mDirectBuffers.begin(); it != mDirectBuffers.end(); it++) - { - it->second->markDirty(); - } - - mSize = std::max(mSize, requiredStagingBufferSize); - mWriteUsageCount = 0; - - mResolvedDataValid = false; -} - -void BufferStorage11::copyData(BufferStorage* sourceStorage, unsigned int size, - unsigned int sourceOffset, unsigned int destOffset) -{ - BufferStorage11* source = makeBufferStorage11(sourceStorage); - if (source) - { - ID3D11DeviceContext *context = mRenderer->getDeviceContext(); - - D3D11_BOX srcBox; - srcBox.left = sourceOffset; - srcBox.right = sourceOffset + size; - srcBox.top = 0; - srcBox.bottom = 1; - srcBox.front = 0; - srcBox.back = 1; - - ASSERT(mStagingBuffer && source->mStagingBuffer); - context->CopySubresourceRegion(mStagingBuffer, 0, destOffset, 0, 0, source->mStagingBuffer, 0, &srcBox); - } -} - -void BufferStorage11::clear() -{ - mResolvedDataValid = false; - mSize = 0; -} - -unsigned int BufferStorage11::getSize() const -{ - return mSize; -} - -bool BufferStorage11::supportsDirectBinding() const -{ - return true; -} - -void BufferStorage11::markBufferUsage() -{ - mReadUsageCount++; - mWriteUsageCount++; - - const unsigned int usageLimit = 5; - - if (mReadUsageCount > usageLimit && mResolvedData) - { - free(mResolvedData); - mResolvedData = NULL; - mResolvedDataSize = 0; - mResolvedDataValid = false; - } -} - -ID3D11Buffer *BufferStorage11::getBuffer(BufferUsage usage) -{ - markBufferUsage(); - - DirectBufferStorage11 *directBuffer = NULL; - - auto directBufferIt = mDirectBuffers.find(usage); - if (directBufferIt != mDirectBuffers.end()) - { - directBuffer = directBufferIt->second; - } - - if (directBuffer) - { - if (directBuffer->isDirty()) - { - // if updateFromStagingBuffer returns true, the D3D buffer has been recreated - // and we should update our serial - if (directBuffer->updateFromStagingBuffer(mStagingBuffer, mSize, 0)) - { - updateSerial(); - } - } - } - else - { - // buffer is not allocated, create it - directBuffer = new DirectBufferStorage11(mRenderer, usage); - directBuffer->updateFromStagingBuffer(mStagingBuffer, mSize, 0); - - mDirectBuffers.insert(std::make_pair(usage, directBuffer)); - updateSerial(); - } - - return directBuffer->getD3DBuffer(); -} - -DirectBufferStorage11::DirectBufferStorage11(Renderer11 *renderer, BufferUsage usage) - : mRenderer(renderer), - mUsage(usage), - mDirectBuffer(NULL), - mBufferSize(0), - mDirty(false) -{ -} - -DirectBufferStorage11::~DirectBufferStorage11() -{ - SafeRelease(mDirectBuffer); -} - -BufferUsage DirectBufferStorage11::getUsage() const -{ - return mUsage; -} - -// Returns true if it recreates the direct buffer -bool DirectBufferStorage11::updateFromStagingBuffer(ID3D11Buffer *stagingBuffer, size_t size, size_t offset) -{ - ID3D11Device *device = mRenderer->getDevice(); - ID3D11DeviceContext *context = mRenderer->getDeviceContext(); - - // unused for now - ASSERT(offset == 0); - - unsigned int requiredBufferSize = size + offset; - bool createBuffer = !mDirectBuffer || mBufferSize < requiredBufferSize; - - // (Re)initialize D3D buffer if needed - if (createBuffer) - { - D3D11_BUFFER_DESC bufferDesc; - fillBufferDesc(&bufferDesc, mRenderer, mUsage, requiredBufferSize); - - ID3D11Buffer *newBuffer; - HRESULT result = device->CreateBuffer(&bufferDesc, NULL, &newBuffer); - - if (FAILED(result)) - { - return gl::error(GL_OUT_OF_MEMORY, false); - } - - // No longer need the old buffer - SafeRelease(mDirectBuffer); - mDirectBuffer = newBuffer; - - mBufferSize = bufferDesc.ByteWidth; - } - - // Copy data via staging buffer - D3D11_BOX srcBox; - srcBox.left = 0; - srcBox.right = size; - srcBox.top = 0; - srcBox.bottom = 1; - srcBox.front = 0; - srcBox.back = 1; - - context->CopySubresourceRegion(mDirectBuffer, 0, offset, 0, 0, stagingBuffer, 0, &srcBox); - - mDirty = false; - - return createBuffer; -} - -void DirectBufferStorage11::fillBufferDesc(D3D11_BUFFER_DESC* bufferDesc, Renderer *renderer, BufferUsage usage, unsigned int bufferSize) -{ - bufferDesc->ByteWidth = bufferSize; - bufferDesc->MiscFlags = 0; - bufferDesc->StructureByteStride = 0; - - switch (usage) - { - case BUFFER_USAGE_VERTEX: - bufferDesc->Usage = D3D11_USAGE_DEFAULT; - bufferDesc->BindFlags = D3D11_BIND_VERTEX_BUFFER; - bufferDesc->CPUAccessFlags = 0; - break; - - case BUFFER_USAGE_INDEX: - bufferDesc->Usage = D3D11_USAGE_DEFAULT; - bufferDesc->BindFlags = D3D11_BIND_INDEX_BUFFER; - bufferDesc->CPUAccessFlags = 0; - break; - - default: - UNREACHABLE(); - } -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/BufferStorage11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/BufferStorage11.h deleted file mode 100644 index a6afafe1b4..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/BufferStorage11.h +++ /dev/null @@ -1,92 +0,0 @@ -// -// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// BufferStorage11.h Defines the BufferStorage11 class. - -#ifndef LIBGLESV2_RENDERER_BUFFERSTORAGE11_H_ -#define LIBGLESV2_RENDERER_BUFFERSTORAGE11_H_ - -#include "libGLESv2/renderer/BufferStorage.h" - -namespace rx -{ -class Renderer; -class Renderer11; -class DirectBufferStorage11; - -enum BufferUsage -{ - BUFFER_USAGE_VERTEX, - BUFFER_USAGE_INDEX, -}; - -class BufferStorage11 : public BufferStorage -{ - public: - explicit BufferStorage11(Renderer11 *renderer); - virtual ~BufferStorage11(); - - static BufferStorage11 *makeBufferStorage11(BufferStorage *bufferStorage); - - virtual void *getData(); - virtual void setData(const void* data, unsigned int size, unsigned int offset); - virtual void copyData(BufferStorage* sourceStorage, unsigned int size, - unsigned int sourceOffset, unsigned int destOffset); - virtual void clear(); - virtual unsigned int getSize() const; - virtual bool supportsDirectBinding() const; - - ID3D11Buffer *getBuffer(BufferUsage usage); - - private: - Renderer11 *mRenderer; - - ID3D11Buffer *mStagingBuffer; - unsigned int mStagingBufferSize; - - std::map<BufferUsage, DirectBufferStorage11*> mDirectBuffers; - - unsigned int mSize; - - void *mResolvedData; - unsigned int mResolvedDataSize; - bool mResolvedDataValid; - - unsigned int mReadUsageCount; - unsigned int mWriteUsageCount; - - void markBufferUsage(); -}; - -// Each instance of BufferStorageD3DBuffer11 is specialized for a class of D3D binding points -// - vertex buffers -// - index buffers -class DirectBufferStorage11 -{ - public: - DirectBufferStorage11(Renderer11 *renderer, BufferUsage usage); - ~DirectBufferStorage11(); - - BufferUsage getUsage() const; - bool updateFromStagingBuffer(ID3D11Buffer *stagingBuffer, size_t size, size_t offset); - - ID3D11Buffer *getD3DBuffer() { return mDirectBuffer; } - bool isDirty() const { return mDirty; } - void markDirty() { mDirty = true; } - - private: - Renderer11 *mRenderer; - const BufferUsage mUsage; - ID3D11Buffer *mDirectBuffer; - size_t mBufferSize; - bool mDirty; - - static void fillBufferDesc(D3D11_BUFFER_DESC* bufferDesc, Renderer *renderer, BufferUsage usage, unsigned int bufferSize); -}; - -} - -#endif // LIBGLESV2_RENDERER_BUFFERSTORAGE11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Fence11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Fence11.cpp deleted file mode 100644 index 2a7d4d43ef..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Fence11.cpp +++ /dev/null @@ -1,134 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// Fence11.cpp: Defines the rx::Fence11 class which implements rx::FenceImpl. - -#include "libGLESv2/renderer/d3d11/Fence11.h" -#include "libGLESv2/main.h" -#include "libGLESv2/renderer/d3d11/Renderer11.h" - -namespace rx -{ - -Fence11::Fence11(rx::Renderer11 *renderer) -{ - mRenderer = renderer; - mQuery = NULL; -} - -Fence11::~Fence11() -{ - if (mQuery) - { - mQuery->Release(); - mQuery = NULL; - } -} - -GLboolean Fence11::isFence() -{ - // GL_NV_fence spec: - // A name returned by GenFencesNV, but not yet set via SetFenceNV, is not the name of an existing fence. - return mQuery != NULL; -} - -void Fence11::setFence(GLenum condition) -{ - if (!mQuery) - { - D3D11_QUERY_DESC queryDesc; - queryDesc.Query = D3D11_QUERY_EVENT; - queryDesc.MiscFlags = 0; - - if (FAILED(mRenderer->getDevice()->CreateQuery(&queryDesc, &mQuery))) - { - return gl::error(GL_OUT_OF_MEMORY); - } - } - - mRenderer->getDeviceContext()->End(mQuery); - - setCondition(condition); - setStatus(GL_FALSE); -} - -GLboolean Fence11::testFence() -{ - if (mQuery == NULL) - { - return gl::error(GL_INVALID_OPERATION, GL_TRUE); - } - - HRESULT result = mRenderer->getDeviceContext()->GetData(mQuery, NULL, 0, 0); - - if (mRenderer->isDeviceLost()) - { - return gl::error(GL_OUT_OF_MEMORY, GL_TRUE); - } - - ASSERT(result == S_OK || result == S_FALSE); - setStatus(result == S_OK); - return getStatus(); -} - -void Fence11::finishFence() -{ - if (mQuery == NULL) - { - return gl::error(GL_INVALID_OPERATION); - } - - while (!testFence()) - { - Sleep(0); - } -} - -void Fence11::getFenceiv(GLenum pname, GLint *params) -{ - if (mQuery == NULL) - { - return gl::error(GL_INVALID_OPERATION); - } - - switch (pname) - { - case GL_FENCE_STATUS_NV: - { - // GL_NV_fence spec: - // Once the status of a fence has been finished (via FinishFenceNV) or tested and the returned status is TRUE (via either TestFenceNV - // or GetFenceivNV querying the FENCE_STATUS_NV), the status remains TRUE until the next SetFenceNV of the fence. - if (getStatus()) - { - params[0] = GL_TRUE; - return; - } - - HRESULT result = mRenderer->getDeviceContext()->GetData(mQuery, NULL, 0, D3D11_ASYNC_GETDATA_DONOTFLUSH); - - if (mRenderer->isDeviceLost()) - { - params[0] = GL_TRUE; - return gl::error(GL_OUT_OF_MEMORY); - } - - ASSERT(result == S_OK || result == S_FALSE); - setStatus(result == S_OK); - params[0] = getStatus(); - - break; - } - case GL_FENCE_CONDITION_NV: - params[0] = getCondition(); - break; - default: - return gl::error(GL_INVALID_ENUM); - break; - } -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Image11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Image11.cpp deleted file mode 100644 index 5d039a35e8..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Image11.cpp +++ /dev/null @@ -1,498 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// Image11.h: Implements the rx::Image11 class, which acts as the interface to -// the actual underlying resources of a Texture - -#include "libGLESv2/renderer/d3d11/Renderer11.h" -#include "libGLESv2/renderer/d3d11/Image11.h" -#include "libGLESv2/renderer/d3d11/TextureStorage11.h" -#include "libGLESv2/Framebuffer.h" -#include "libGLESv2/Renderbuffer.h" - -#include "libGLESv2/main.h" -#include "libGLESv2/utilities.h" -#include "libGLESv2/renderer/d3d11/renderer11_utils.h" -#include "libGLESv2/renderer/generatemip.h" - -namespace rx -{ - -Image11::Image11() -{ - mStagingTexture = NULL; - mRenderer = NULL; - mDXGIFormat = DXGI_FORMAT_UNKNOWN; -} - -Image11::~Image11() -{ - if (mStagingTexture) - { - mStagingTexture->Release(); - } -} - -Image11 *Image11::makeImage11(Image *img) -{ - ASSERT(HAS_DYNAMIC_TYPE(rx::Image11*, img)); - return static_cast<rx::Image11*>(img); -} - -void Image11::generateMipmap(Image11 *dest, Image11 *src) -{ - ASSERT(src->getDXGIFormat() == dest->getDXGIFormat()); - ASSERT(src->getWidth() == 1 || src->getWidth() / 2 == dest->getWidth()); - ASSERT(src->getHeight() == 1 || src->getHeight() / 2 == dest->getHeight()); - - D3D11_MAPPED_SUBRESOURCE destMapped, srcMapped; - dest->map(D3D11_MAP_WRITE, &destMapped); - src->map(D3D11_MAP_READ, &srcMapped); - - const unsigned char *sourceData = reinterpret_cast<const unsigned char*>(srcMapped.pData); - unsigned char *destData = reinterpret_cast<unsigned char*>(destMapped.pData); - - if (sourceData && destData) - { - switch (src->getDXGIFormat()) - { - case DXGI_FORMAT_R8G8B8A8_UNORM: - case DXGI_FORMAT_B8G8R8A8_UNORM: - GenerateMip<R8G8B8A8>(src->getWidth(), src->getHeight(), sourceData, srcMapped.RowPitch, destData, destMapped.RowPitch); - break; - case DXGI_FORMAT_A8_UNORM: - GenerateMip<A8>(src->getWidth(), src->getHeight(), sourceData, srcMapped.RowPitch, destData, destMapped.RowPitch); - break; - case DXGI_FORMAT_R8_UNORM: - GenerateMip<R8>(src->getWidth(), src->getHeight(), sourceData, srcMapped.RowPitch, destData, destMapped.RowPitch); - break; - case DXGI_FORMAT_R32G32B32A32_FLOAT: - GenerateMip<A32B32G32R32F>(src->getWidth(), src->getHeight(), sourceData, srcMapped.RowPitch, destData, destMapped.RowPitch); - break; - case DXGI_FORMAT_R32G32B32_FLOAT: - GenerateMip<R32G32B32F>(src->getWidth(), src->getHeight(), sourceData, srcMapped.RowPitch, destData, destMapped.RowPitch); - break; - case DXGI_FORMAT_R16G16B16A16_FLOAT: - GenerateMip<A16B16G16R16F>(src->getWidth(), src->getHeight(), sourceData, srcMapped.RowPitch, destData, destMapped.RowPitch); - break; - case DXGI_FORMAT_R8G8_UNORM: - GenerateMip<R8G8>(src->getWidth(), src->getHeight(), sourceData, srcMapped.RowPitch, destData, destMapped.RowPitch); - break; - case DXGI_FORMAT_R16_FLOAT: - GenerateMip<R16F>(src->getWidth(), src->getHeight(), sourceData, srcMapped.RowPitch, destData, destMapped.RowPitch); - break; - case DXGI_FORMAT_R16G16_FLOAT: - GenerateMip<R16G16F>(src->getWidth(), src->getHeight(), sourceData, srcMapped.RowPitch, destData, destMapped.RowPitch); - break; - case DXGI_FORMAT_R32_FLOAT: - GenerateMip<R32F>(src->getWidth(), src->getHeight(), sourceData, srcMapped.RowPitch, destData, destMapped.RowPitch); - break; - case DXGI_FORMAT_R32G32_FLOAT: - GenerateMip<R32G32F>(src->getWidth(), src->getHeight(), sourceData, srcMapped.RowPitch, destData, destMapped.RowPitch); - break; - default: - UNREACHABLE(); - break; - } - - dest->unmap(); - src->unmap(); - } - - dest->markDirty(); -} - -static bool FormatRequiresInitialization(DXGI_FORMAT dxgiFormat, GLenum internalFormat) -{ - return (dxgiFormat == DXGI_FORMAT_R8G8B8A8_UNORM && gl::GetAlphaSize(internalFormat) == 0) || - (dxgiFormat == DXGI_FORMAT_R32G32B32A32_FLOAT && gl::GetAlphaSize(internalFormat) == 0); -} - -bool Image11::isDirty() const -{ - return ((mStagingTexture || FormatRequiresInitialization(mDXGIFormat, mInternalFormat)) && mDirty); -} - -bool Image11::updateSurface(TextureStorageInterface2D *storage, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) -{ - TextureStorage11_2D *storage11 = TextureStorage11_2D::makeTextureStorage11_2D(storage->getStorageInstance()); - return storage11->updateSubresourceLevel(getStagingTexture(), getStagingSubresource(), level, 0, xoffset, yoffset, width, height); -} - -bool Image11::updateSurface(TextureStorageInterfaceCube *storage, int face, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height) -{ - TextureStorage11_Cube *storage11 = TextureStorage11_Cube::makeTextureStorage11_Cube(storage->getStorageInstance()); - return storage11->updateSubresourceLevel(getStagingTexture(), getStagingSubresource(), level, face, xoffset, yoffset, width, height); -} - -bool Image11::redefine(Renderer *renderer, GLint internalformat, GLsizei width, GLsizei height, bool forceRelease) -{ - if (mWidth != width || - mHeight != height || - mInternalFormat != internalformat || - forceRelease) - { - mRenderer = Renderer11::makeRenderer11(renderer); - - mWidth = width; - mHeight = height; - mInternalFormat = internalformat; - // compute the d3d format that will be used - mDXGIFormat = gl_d3d11::ConvertTextureFormat(internalformat, mRenderer->getFeatureLevel()); - mActualFormat = d3d11_gl::ConvertTextureInternalFormat(mDXGIFormat); - - if (mStagingTexture) - { - mStagingTexture->Release(); - mStagingTexture = NULL; - } - - return true; - } - - return false; -} - -bool Image11::isRenderableFormat() const -{ - return TextureStorage11::IsTextureFormatRenderable(mDXGIFormat); -} - -DXGI_FORMAT Image11::getDXGIFormat() const -{ - // this should only happen if the image hasn't been redefined first - // which would be a bug by the caller - ASSERT(mDXGIFormat != DXGI_FORMAT_UNKNOWN); - - return mDXGIFormat; -} - -// Store the pixel rectangle designated by xoffset,yoffset,width,height with pixels stored as format/type at input -// into the target pixel rectangle. -void Image11::loadData(GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, - GLint unpackAlignment, const void *input) -{ - D3D11_MAPPED_SUBRESOURCE mappedImage; - HRESULT result = map(D3D11_MAP_WRITE, &mappedImage); - if (FAILED(result)) - { - ERR("Could not map image for loading."); - return; - } - - GLsizei inputPitch = gl::ComputePitch(width, mInternalFormat, unpackAlignment); - size_t pixelSize = d3d11::ComputePixelSizeBits(mDXGIFormat) / 8; - void* offsetMappedData = (void*)((BYTE *)mappedImage.pData + (yoffset * mappedImage.RowPitch + xoffset * pixelSize)); - - switch (mInternalFormat) - { - case GL_ALPHA8_EXT: - if (mRenderer->getFeatureLevel() >= D3D_FEATURE_LEVEL_10_0) - loadAlphaDataToNative(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - else - loadAlphaDataToBGRA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_LUMINANCE8_EXT: - loadLuminanceDataToNativeOrBGRA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData, false); - break; - case GL_ALPHA32F_EXT: - loadAlphaFloatDataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_LUMINANCE32F_EXT: - loadLuminanceFloatDataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_ALPHA16F_EXT: - loadAlphaHalfFloatDataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_LUMINANCE16F_EXT: - loadLuminanceHalfFloatDataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_LUMINANCE8_ALPHA8_EXT: - loadLuminanceAlphaDataToNativeOrBGRA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData, false); - break; - case GL_LUMINANCE_ALPHA32F_EXT: - loadLuminanceAlphaFloatDataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_LUMINANCE_ALPHA16F_EXT: - loadLuminanceAlphaHalfFloatDataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_RGB8_OES: - loadRGBUByteDataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_RGB565: - loadRGB565DataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_RGBA8_OES: - loadRGBAUByteDataToNative(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_RGBA4: - loadRGBA4444DataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_RGB5_A1: - loadRGBA5551DataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_BGRA8_EXT: - loadBGRADataToBGRA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_RGB32F_EXT: - loadRGBFloatDataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_RGB16F_EXT: - loadRGBHalfFloatDataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_RGBA32F_EXT: - loadRGBAFloatDataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - case GL_RGBA16F_EXT: - loadRGBAHalfFloatDataToRGBA(width, height, inputPitch, input, mappedImage.RowPitch, offsetMappedData); - break; - default: UNREACHABLE(); - } - - unmap(); -} - -void Image11::loadCompressedData(GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, - const void *input) -{ - ASSERT(xoffset % 4 == 0); - ASSERT(yoffset % 4 == 0); - - D3D11_MAPPED_SUBRESOURCE mappedImage; - HRESULT result = map(D3D11_MAP_WRITE, &mappedImage); - if (FAILED(result)) - { - ERR("Could not map image for loading."); - return; - } - - // Size computation assumes a 4x4 block compressed texture format - size_t blockSize = d3d11::ComputeBlockSizeBits(mDXGIFormat) / 8; - void* offsetMappedData = (void*)((BYTE *)mappedImage.pData + ((yoffset / 4) * mappedImage.RowPitch + (xoffset / 4) * blockSize)); - - GLsizei inputSize = gl::ComputeCompressedSize(width, height, mInternalFormat); - GLsizei inputPitch = gl::ComputeCompressedPitch(width, mInternalFormat); - int rows = inputSize / inputPitch; - for (int i = 0; i < rows; ++i) - { - memcpy((void*)((BYTE*)offsetMappedData + i * mappedImage.RowPitch), (void*)((BYTE*)input + i * inputPitch), inputPitch); - } - - unmap(); -} - -void Image11::copy(GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source) -{ - gl::Renderbuffer *colorbuffer = source->getReadColorbuffer(); - - if (colorbuffer && colorbuffer->getActualFormat() == (GLuint)mActualFormat) - { - // No conversion needed-- use copyback fastpath - ID3D11Texture2D *colorBufferTexture = NULL; - unsigned int subresourceIndex = 0; - - if (mRenderer->getRenderTargetResource(colorbuffer, &subresourceIndex, &colorBufferTexture)) - { - D3D11_TEXTURE2D_DESC textureDesc; - colorBufferTexture->GetDesc(&textureDesc); - - ID3D11Device *device = mRenderer->getDevice(); - ID3D11DeviceContext *deviceContext = mRenderer->getDeviceContext(); - - ID3D11Texture2D* srcTex = NULL; - if (textureDesc.SampleDesc.Count > 1) - { - D3D11_TEXTURE2D_DESC resolveDesc; - resolveDesc.Width = textureDesc.Width; - resolveDesc.Height = textureDesc.Height; - resolveDesc.MipLevels = 1; - resolveDesc.ArraySize = 1; - resolveDesc.Format = textureDesc.Format; - resolveDesc.SampleDesc.Count = 1; - resolveDesc.SampleDesc.Quality = 0; - resolveDesc.Usage = D3D11_USAGE_DEFAULT; - resolveDesc.BindFlags = 0; - resolveDesc.CPUAccessFlags = 0; - resolveDesc.MiscFlags = 0; - - HRESULT result = device->CreateTexture2D(&resolveDesc, NULL, &srcTex); - if (FAILED(result)) - { - ERR("Failed to create resolve texture for Image11::copy, HRESULT: 0x%X.", result); - return; - } - - deviceContext->ResolveSubresource(srcTex, 0, colorBufferTexture, subresourceIndex, textureDesc.Format); - subresourceIndex = 0; - } - else - { - srcTex = colorBufferTexture; - srcTex->AddRef(); - } - - D3D11_BOX srcBox; - srcBox.left = x; - srcBox.right = x + width; - srcBox.top = y; - srcBox.bottom = y + height; - srcBox.front = 0; - srcBox.back = 1; - - deviceContext->CopySubresourceRegion(mStagingTexture, 0, xoffset, yoffset, 0, srcTex, subresourceIndex, &srcBox); - - srcTex->Release(); - colorBufferTexture->Release(); - } - } - else - { - // This format requires conversion, so we must copy the texture to staging and manually convert via readPixels - D3D11_MAPPED_SUBRESOURCE mappedImage; - HRESULT result = map(D3D11_MAP_WRITE, &mappedImage); - - // determine the offset coordinate into the destination buffer - GLsizei rowOffset = gl::ComputePixelSize(mActualFormat) * xoffset; - void *dataOffset = static_cast<unsigned char*>(mappedImage.pData) + mappedImage.RowPitch * yoffset + rowOffset; - - mRenderer->readPixels(source, x, y, width, height, gl::ExtractFormat(mInternalFormat), - gl::ExtractType(mInternalFormat), mappedImage.RowPitch, false, 4, dataOffset); - - unmap(); - } -} - -ID3D11Texture2D *Image11::getStagingTexture() -{ - createStagingTexture(); - - return mStagingTexture; -} - -unsigned int Image11::getStagingSubresource() -{ - createStagingTexture(); - - return mStagingSubresource; -} - -template <typename T, size_t N> -static void setDefaultData(ID3D11DeviceContext *deviceContext, ID3D11Texture2D *texture, UINT subresource, - GLsizei width, GLsizei height, const T (&defaultData)[N]) -{ - D3D11_MAPPED_SUBRESOURCE map; - deviceContext->Map(texture, subresource, D3D11_MAP_WRITE, 0, &map); - - unsigned char* ptr = reinterpret_cast<unsigned char*>(map.pData); - size_t pixelSize = sizeof(T) * N; - - for (GLsizei y = 0; y < height; y++) - { - for (GLsizei x = 0; x < width; x++) - { - memcpy(ptr + (y * map.RowPitch) + (x * pixelSize), defaultData, pixelSize); - } - } - - deviceContext->Unmap(texture, subresource); -} - -void Image11::createStagingTexture() -{ - if (mStagingTexture) - { - return; - } - - ID3D11Texture2D *newTexture = NULL; - int lodOffset = 1; - const DXGI_FORMAT dxgiFormat = getDXGIFormat(); - ASSERT(!d3d11::IsDepthStencilFormat(dxgiFormat)); // We should never get here for depth textures - - if (mWidth != 0 && mHeight != 0) - { - GLsizei width = mWidth; - GLsizei height = mHeight; - - // adjust size if needed for compressed textures - gl::MakeValidSize(false, d3d11::IsCompressed(dxgiFormat), &width, &height, &lodOffset); - ID3D11Device *device = mRenderer->getDevice(); - - D3D11_TEXTURE2D_DESC desc; - desc.Width = width; - desc.Height = height; - desc.MipLevels = lodOffset + 1; - desc.ArraySize = 1; - desc.Format = dxgiFormat; - desc.SampleDesc.Count = 1; - desc.SampleDesc.Quality = 0; - desc.Usage = D3D11_USAGE_STAGING; - desc.BindFlags = 0; - desc.CPUAccessFlags = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; - desc.MiscFlags = 0; - - HRESULT result = device->CreateTexture2D(&desc, NULL, &newTexture); - - if (FAILED(result)) - { - ASSERT(result == E_OUTOFMEMORY); - ERR("Creating image failed."); - return gl::error(GL_OUT_OF_MEMORY); - } - } - - mStagingTexture = newTexture; - mStagingSubresource = D3D11CalcSubresource(lodOffset, 0, lodOffset + 1); - mDirty = false; - - if (mDXGIFormat == DXGI_FORMAT_R8G8B8A8_UNORM && gl::GetAlphaSize(mInternalFormat) == 0) - { - unsigned char defaultPixel[4] = { 0, 0, 0, 255 }; - setDefaultData(mRenderer->getDeviceContext(), mStagingTexture, mStagingSubresource, mWidth, mHeight, defaultPixel); - } - else if (mDXGIFormat == DXGI_FORMAT_R32G32B32A32_FLOAT && gl::GetAlphaSize(mInternalFormat) == 0) - { - float defaultPixel[4] = { 0.0f, 0.0f, 0.0f, 1.0f }; - setDefaultData(mRenderer->getDeviceContext(), mStagingTexture, mStagingSubresource, mWidth, mHeight, defaultPixel); - } -} - -HRESULT Image11::map(D3D11_MAP mapType, D3D11_MAPPED_SUBRESOURCE *map) -{ - createStagingTexture(); - - HRESULT result = E_FAIL; - - if (mStagingTexture) - { - ID3D11DeviceContext *deviceContext = mRenderer->getDeviceContext(); - result = deviceContext->Map(mStagingTexture, mStagingSubresource, mapType, 0, map); - - // this can fail if the device is removed (from TDR) - if (d3d11::isDeviceLostError(result)) - { - mRenderer->notifyDeviceLost(); - } - else if (SUCCEEDED(result)) - { - mDirty = true; - } - } - - return result; -} - -void Image11::unmap() -{ - if (mStagingTexture) - { - ID3D11DeviceContext *deviceContext = mRenderer->getDeviceContext(); - deviceContext->Unmap(mStagingTexture, mStagingSubresource); - } -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Image11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Image11.h deleted file mode 100644 index 11a6492dc8..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Image11.h +++ /dev/null @@ -1,76 +0,0 @@ -// -// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// Image11.h: Defines the rx::Image11 class, which acts as the interface to -// the actual underlying resources of a Texture - -#ifndef LIBGLESV2_RENDERER_IMAGE11_H_ -#define LIBGLESV2_RENDERER_IMAGE11_H_ - -#include "libGLESv2/renderer/Image.h" - -#include "common/debug.h" - -namespace gl -{ -class Framebuffer; -} - -namespace rx -{ -class Renderer; -class Renderer11; -class TextureStorageInterface2D; -class TextureStorageInterfaceCube; - -class Image11 : public Image -{ - public: - Image11(); - virtual ~Image11(); - - static Image11 *makeImage11(Image *img); - - static void generateMipmap(Image11 *dest, Image11 *src); - - virtual bool isDirty() const; - - virtual bool updateSurface(TextureStorageInterface2D *storage, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); - virtual bool updateSurface(TextureStorageInterfaceCube *storage, int face, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); - - virtual bool redefine(Renderer *renderer, GLint internalformat, GLsizei width, GLsizei height, bool forceRelease); - - virtual bool isRenderableFormat() const; - DXGI_FORMAT getDXGIFormat() const; - - virtual void loadData(GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, - GLint unpackAlignment, const void *input); - virtual void loadCompressedData(GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, - const void *input); - - virtual void copy(GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source); - - protected: - HRESULT map(D3D11_MAP mapType, D3D11_MAPPED_SUBRESOURCE *map); - void unmap(); - - private: - DISALLOW_COPY_AND_ASSIGN(Image11); - - ID3D11Texture2D *getStagingTexture(); - unsigned int getStagingSubresource(); - void createStagingTexture(); - - Renderer11 *mRenderer; - - DXGI_FORMAT mDXGIFormat; - ID3D11Texture2D *mStagingTexture; - unsigned int mStagingSubresource; -}; - -} - -#endif // LIBGLESV2_RENDERER_IMAGE11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/IndexBuffer11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/IndexBuffer11.cpp deleted file mode 100644 index 44f9976d43..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/IndexBuffer11.cpp +++ /dev/null @@ -1,183 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// IndexBuffer11.cpp: Defines the D3D11 IndexBuffer implementation. - -#include "libGLESv2/renderer/d3d11/IndexBuffer11.h" -#include "libGLESv2/renderer/d3d11/Renderer11.h" - -namespace rx -{ - -IndexBuffer11::IndexBuffer11(Renderer11 *const renderer) : mRenderer(renderer) -{ - mBuffer = NULL; - mBufferSize = 0; - mDynamicUsage = false; -} - -IndexBuffer11::~IndexBuffer11() -{ - if (mBuffer) - { - mBuffer->Release(); - mBuffer = NULL; - } -} - -bool IndexBuffer11::initialize(unsigned int bufferSize, GLenum indexType, bool dynamic) -{ - if (mBuffer) - { - mBuffer->Release(); - mBuffer = NULL; - } - - updateSerial(); - - if (bufferSize > 0) - { - ID3D11Device* dxDevice = mRenderer->getDevice(); - - D3D11_BUFFER_DESC bufferDesc; - bufferDesc.ByteWidth = bufferSize; - bufferDesc.Usage = D3D11_USAGE_DYNAMIC; - bufferDesc.BindFlags = D3D11_BIND_INDEX_BUFFER; - bufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; - bufferDesc.MiscFlags = 0; - bufferDesc.StructureByteStride = 0; - - HRESULT result = dxDevice->CreateBuffer(&bufferDesc, NULL, &mBuffer); - if (FAILED(result)) - { - return false; - } - } - - mBufferSize = bufferSize; - mIndexType = indexType; - mDynamicUsage = dynamic; - - return true; -} - -IndexBuffer11 *IndexBuffer11::makeIndexBuffer11(IndexBuffer *indexBuffer) -{ - ASSERT(HAS_DYNAMIC_TYPE(IndexBuffer11*, indexBuffer)); - return static_cast<IndexBuffer11*>(indexBuffer); -} - -bool IndexBuffer11::mapBuffer(unsigned int offset, unsigned int size, void** outMappedMemory) -{ - if (mBuffer) - { - // Check for integer overflows and out-out-bounds map requests - if (offset + size < offset || offset + size > mBufferSize) - { - ERR("Index buffer map range is not inside the buffer."); - return false; - } - - ID3D11DeviceContext *dxContext = mRenderer->getDeviceContext(); - - D3D11_MAPPED_SUBRESOURCE mappedResource; - HRESULT result = dxContext->Map(mBuffer, 0, D3D11_MAP_WRITE_NO_OVERWRITE, 0, &mappedResource); - if (FAILED(result)) - { - ERR("Index buffer map failed with error 0x%08x", result); - return false; - } - - *outMappedMemory = reinterpret_cast<char*>(mappedResource.pData) + offset; - return true; - } - else - { - ERR("Index buffer not initialized."); - return false; - } -} - -bool IndexBuffer11::unmapBuffer() -{ - if (mBuffer) - { - ID3D11DeviceContext *dxContext = mRenderer->getDeviceContext(); - dxContext->Unmap(mBuffer, 0); - return true; - } - else - { - ERR("Index buffer not initialized."); - return false; - } -} - -GLenum IndexBuffer11::getIndexType() const -{ - return mIndexType; -} - -unsigned int IndexBuffer11::getBufferSize() const -{ - return mBufferSize; -} - -bool IndexBuffer11::setSize(unsigned int bufferSize, GLenum indexType) -{ - if (bufferSize > mBufferSize || indexType != mIndexType) - { - return initialize(bufferSize, indexType, mDynamicUsage); - } - else - { - return true; - } -} - -bool IndexBuffer11::discard() -{ - if (mBuffer) - { - ID3D11DeviceContext *dxContext = mRenderer->getDeviceContext(); - - D3D11_MAPPED_SUBRESOURCE mappedResource; - HRESULT result = dxContext->Map(mBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); - if (FAILED(result)) - { - ERR("Index buffer map failed with error 0x%08x", result); - return false; - } - - dxContext->Unmap(mBuffer, 0); - - return true; - } - else - { - ERR("Index buffer not initialized."); - return false; - } -} - -DXGI_FORMAT IndexBuffer11::getIndexFormat() const -{ - switch (mIndexType) - { - case GL_UNSIGNED_BYTE: return DXGI_FORMAT_R16_UINT; - case GL_UNSIGNED_SHORT: return DXGI_FORMAT_R16_UINT; - case GL_UNSIGNED_INT: return DXGI_FORMAT_R32_UINT; - default: UNREACHABLE(); return DXGI_FORMAT_UNKNOWN; - } -} - -ID3D11Buffer *IndexBuffer11::getBuffer() const -{ - return mBuffer; -} - -}
\ No newline at end of file diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Query11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Query11.cpp deleted file mode 100644 index 24c0330a1e..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Query11.cpp +++ /dev/null @@ -1,122 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// Query11.cpp: Defines the rx::Query11 class which implements rx::QueryImpl. - -#include "libGLESv2/renderer/d3d11/Query11.h" -#include "libGLESv2/renderer/d3d11/Renderer11.h" -#include "libGLESv2/main.h" - -namespace rx -{ - -Query11::Query11(rx::Renderer11 *renderer, GLenum type) : QueryImpl(type) -{ - mRenderer = renderer; - mQuery = NULL; -} - -Query11::~Query11() -{ - if (mQuery) - { - mQuery->Release(); - mQuery = NULL; - } -} - -void Query11::begin() -{ - if (mQuery == NULL) - { - D3D11_QUERY_DESC queryDesc; - queryDesc.Query = D3D11_QUERY_OCCLUSION; - queryDesc.MiscFlags = 0; - - if (FAILED(mRenderer->getDevice()->CreateQuery(&queryDesc, &mQuery))) - { - return gl::error(GL_OUT_OF_MEMORY); - } - } - - mRenderer->getDeviceContext()->Begin(mQuery); -} - -void Query11::end() -{ - if (mQuery == NULL) - { - return gl::error(GL_INVALID_OPERATION); - } - - mRenderer->getDeviceContext()->End(mQuery); - - mStatus = GL_FALSE; - mResult = GL_FALSE; -} - -GLuint Query11::getResult() -{ - if (mQuery != NULL) - { - while (!testQuery()) - { - Sleep(0); - // explicitly check for device loss, some drivers seem to return S_FALSE - // if the device is lost - if (mRenderer->testDeviceLost(true)) - { - return gl::error(GL_OUT_OF_MEMORY, 0); - } - } - } - - return mResult; -} - -GLboolean Query11::isResultAvailable() -{ - if (mQuery != NULL) - { - testQuery(); - } - - return mStatus; -} - -GLboolean Query11::testQuery() -{ - if (mQuery != NULL && mStatus != GL_TRUE) - { - UINT64 numPixels = 0; - HRESULT result = mRenderer->getDeviceContext()->GetData(mQuery, &numPixels, sizeof(UINT64), 0); - if (result == S_OK) - { - mStatus = GL_TRUE; - - switch (getType()) - { - case GL_ANY_SAMPLES_PASSED_EXT: - case GL_ANY_SAMPLES_PASSED_CONSERVATIVE_EXT: - mResult = (numPixels > 0) ? GL_TRUE : GL_FALSE; - break; - default: - UNREACHABLE(); - } - } - else if (mRenderer->testDeviceLost(true)) - { - return gl::error(GL_OUT_OF_MEMORY, GL_TRUE); - } - - return mStatus; - } - - return GL_TRUE; // prevent blocking when query is null -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Renderer11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Renderer11.cpp deleted file mode 100644 index e70727c65e..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Renderer11.cpp +++ /dev/null @@ -1,3736 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// Renderer11.cpp: Implements a back-end specific class for the D3D11 renderer. - -#include "libGLESv2/main.h" -#include "libGLESv2/utilities.h" -#include "libGLESv2/Buffer.h" -#include "libGLESv2/ProgramBinary.h" -#include "libGLESv2/Framebuffer.h" -#include "libGLESv2/Renderbuffer.h" -#include "libGLESv2/renderer/d3d11/Renderer11.h" -#include "libGLESv2/renderer/d3d11/RenderTarget11.h" -#include "libGLESv2/renderer/d3d11/renderer11_utils.h" -#include "libGLESv2/renderer/d3d11/ShaderExecutable11.h" -#include "libGLESv2/renderer/d3d11/SwapChain11.h" -#include "libGLESv2/renderer/d3d11/Image11.h" -#include "libGLESv2/renderer/d3d11/VertexBuffer11.h" -#include "libGLESv2/renderer/d3d11/IndexBuffer11.h" -#include "libGLESv2/renderer/d3d11/BufferStorage11.h" -#include "libGLESv2/renderer/VertexDataManager.h" -#include "libGLESv2/renderer/IndexDataManager.h" -#include "libGLESv2/renderer/d3d11/TextureStorage11.h" -#include "libGLESv2/renderer/d3d11/Query11.h" -#include "libGLESv2/renderer/d3d11/Fence11.h" - -#include "libGLESv2/renderer/d3d11/shaders/compiled/passthrough11vs.h" -#include "libGLESv2/renderer/d3d11/shaders/compiled/passthroughrgba11ps.h" -#include "libGLESv2/renderer/d3d11/shaders/compiled/passthroughrgb11ps.h" -#include "libGLESv2/renderer/d3d11/shaders/compiled/passthroughlum11ps.h" -#include "libGLESv2/renderer/d3d11/shaders/compiled/passthroughlumalpha11ps.h" - -#include "libGLESv2/renderer/d3d11/shaders/compiled/clear11vs.h" -#include "libGLESv2/renderer/d3d11/shaders/compiled/clearsingle11ps.h" -#include "libGLESv2/renderer/d3d11/shaders/compiled/clearmultiple11ps.h" - -#include "libEGL/Display.h" - -#if defined(ANGLE_OS_WINRT) && !defined(ANGLE_OS_WINPHONE) -# include <dxgi1_3.h> -# include <wrl.h> -# include <windows.applicationmodel.core.h> -typedef ABI::Windows::Foundation::IEventHandler<ABI::Windows::ApplicationModel::SuspendingEventArgs *> SuspendEventHandler; -#endif - -#ifdef _DEBUG -// this flag enables suppressing some spurious warnings that pop up in certain WebGL samples -// and conformance tests. to enable all warnings, remove this define. -#define ANGLE_SUPPRESS_D3D11_HAZARD_WARNINGS 1 -#endif - -namespace rx -{ -static const DXGI_FORMAT RenderTargetFormats[] = - { - DXGI_FORMAT_B8G8R8A8_UNORM, - DXGI_FORMAT_R8G8B8A8_UNORM - }; - -static const DXGI_FORMAT DepthStencilFormats[] = - { - DXGI_FORMAT_UNKNOWN, - DXGI_FORMAT_D24_UNORM_S8_UINT, - DXGI_FORMAT_D16_UNORM - }; - -enum -{ - MAX_TEXTURE_IMAGE_UNITS_VTF_SM4 = 16 -}; - -Renderer11::Renderer11(egl::Display *display) : Renderer(display) -{ - mVertexDataManager = NULL; - mIndexDataManager = NULL; - - mLineLoopIB = NULL; - mTriangleFanIB = NULL; - - mCopyResourcesInitialized = false; - mCopyVB = NULL; - mCopySampler = NULL; - mCopyIL = NULL; - mCopyVS = NULL; - mCopyRGBAPS = NULL; - mCopyRGBPS = NULL; - mCopyLumPS = NULL; - mCopyLumAlphaPS = NULL; - - mClearResourcesInitialized = false; - mClearVB = NULL; - mClearIL = NULL; - mClearVS = NULL; - mClearSinglePS = NULL; - mClearMultiplePS = NULL; - mClearScissorRS = NULL; - mClearNoScissorRS = NULL; - - mSyncQuery = NULL; - - mD3d11Module = NULL; - mDxgiModule = NULL; - - mDeviceLost = false; - - mMaxSupportedSamples = 0; - - mDevice = NULL; - mDeviceContext = NULL; - mDxgiAdapter = NULL; - mDxgiFactory = NULL; - - mDriverConstantBufferVS = NULL; - mDriverConstantBufferPS = NULL; - - mBGRATextureSupport = false; - - mIsGeometryShaderActive = false; -} - -Renderer11::~Renderer11() -{ - release(); -} - -Renderer11 *Renderer11::makeRenderer11(Renderer *renderer) -{ - ASSERT(HAS_DYNAMIC_TYPE(rx::Renderer11*, renderer)); - return static_cast<rx::Renderer11*>(renderer); -} - -#ifndef __d3d11_1_h__ -#define D3D11_MESSAGE_ID_DEVICE_DRAW_RENDERTARGETVIEW_NOT_SET ((D3D11_MESSAGE_ID)3146081) -#endif - -EGLint Renderer11::initialize() -{ - if (!initializeCompiler()) - { - return EGL_NOT_INITIALIZED; - } - -#if !defined(ANGLE_OS_WINRT) - mDxgiModule = LoadLibrary(TEXT("dxgi.dll")); - mD3d11Module = LoadLibrary(TEXT("d3d11.dll")); - - if (mD3d11Module == NULL || mDxgiModule == NULL) - { - ERR("Could not load D3D11 or DXGI library - aborting!\n"); - return EGL_NOT_INITIALIZED; - } - - - // create the D3D11 device - ASSERT(mDevice == NULL); - PFN_D3D11_CREATE_DEVICE D3D11CreateDevice = (PFN_D3D11_CREATE_DEVICE)GetProcAddress(mD3d11Module, "D3D11CreateDevice"); - - if (D3D11CreateDevice == NULL) - { - ERR("Could not retrieve D3D11CreateDevice address - aborting!\n"); - return EGL_NOT_INITIALIZED; - } -#endif - - D3D_FEATURE_LEVEL featureLevels[] = - { - D3D_FEATURE_LEVEL_11_0, - D3D_FEATURE_LEVEL_10_1, - D3D_FEATURE_LEVEL_10_0, -#if !defined(ANGLE_ENABLE_D3D9) - D3D_FEATURE_LEVEL_9_3, - D3D_FEATURE_LEVEL_9_2, - D3D_FEATURE_LEVEL_9_1, -#endif - }; - - HRESULT result = S_OK; - -#ifdef _DEBUG - result = D3D11CreateDevice(NULL, - D3D_DRIVER_TYPE_HARDWARE, - NULL, - D3D11_CREATE_DEVICE_DEBUG, - featureLevels, - ArraySize(featureLevels), - D3D11_SDK_VERSION, - &mDevice, - &mFeatureLevel, - &mDeviceContext); - - if (!mDevice || FAILED(result)) - { - ERR("Failed creating Debug D3D11 device - falling back to release runtime.\n"); - } - - if (!mDevice || FAILED(result)) -#endif - { - result = D3D11CreateDevice(NULL, - D3D_DRIVER_TYPE_HARDWARE, - NULL, - 0, - featureLevels, - ArraySize(featureLevels), - D3D11_SDK_VERSION, - &mDevice, - &mFeatureLevel, - &mDeviceContext); - - if (!mDevice || FAILED(result)) - { - ERR("Could not create D3D11 device - aborting!\n"); - return EGL_NOT_INITIALIZED; // Cleanup done by destructor through glDestroyRenderer - } - } - -#if !defined(ANGLE_OS_WINRT) - IDXGIDevice *dxgiDevice = NULL; -#else - IDXGIDevice1 *dxgiDevice = NULL; -#endif - result = mDevice->QueryInterface(IID_PPV_ARGS(&dxgiDevice)); - - if (FAILED(result)) - { - ERR("Could not query DXGI device - aborting!\n"); - return EGL_NOT_INITIALIZED; - } - - result = dxgiDevice->GetParent(__uuidof(IDXGIAdapter), (void**)&mDxgiAdapter); - - if (FAILED(result)) - { - ERR("Could not retrieve DXGI adapter - aborting!\n"); - return EGL_NOT_INITIALIZED; - } - - dxgiDevice->Release(); - - mDxgiAdapter->GetDesc(&mAdapterDescription); - memset(mDescription, 0, sizeof(mDescription)); - wcstombs(mDescription, mAdapterDescription.Description, sizeof(mDescription) - 1); - - result = mDxgiAdapter->GetParent(__uuidof(IDXGIFactory), (void**)&mDxgiFactory); - - if (!mDxgiFactory || FAILED(result)) - { - ERR("Could not create DXGI factory - aborting!\n"); - return EGL_NOT_INITIALIZED; - } - - // Disable some spurious D3D11 debug warnings to prevent them from flooding the output log -#if defined(ANGLE_SUPPRESS_D3D11_HAZARD_WARNINGS) && defined(_DEBUG) - ID3D11InfoQueue *infoQueue; - result = mDevice->QueryInterface(__uuidof(ID3D11InfoQueue), (void **)&infoQueue); - - if (SUCCEEDED(result)) - { - D3D11_MESSAGE_ID hideMessages[] = - { - D3D11_MESSAGE_ID_DEVICE_DRAW_RENDERTARGETVIEW_NOT_SET - }; - - D3D11_INFO_QUEUE_FILTER filter = {0}; - filter.DenyList.NumIDs = ArraySize(hideMessages); - filter.DenyList.pIDList = hideMessages; - - infoQueue->AddStorageFilterEntries(&filter); - - infoQueue->Release(); - } -#endif - - unsigned int maxSupportedSamples = 0; - unsigned int rtFormatCount = ArraySize(RenderTargetFormats); - unsigned int dsFormatCount = ArraySize(DepthStencilFormats); - for (unsigned int i = 0; i < rtFormatCount + dsFormatCount; ++i) - { - DXGI_FORMAT format = (i < rtFormatCount) ? RenderTargetFormats[i] : DepthStencilFormats[i - rtFormatCount]; - if (format != DXGI_FORMAT_UNKNOWN) - { - UINT formatSupport; - result = mDevice->CheckFormatSupport(format, &formatSupport); - if (SUCCEEDED(result) && (formatSupport & D3D11_FORMAT_SUPPORT_MULTISAMPLE_RENDERTARGET)) - { - MultisampleSupportInfo supportInfo; - - for (unsigned int j = 1; j <= D3D11_MAX_MULTISAMPLE_SAMPLE_COUNT; j++) - { - result = mDevice->CheckMultisampleQualityLevels(format, j, &supportInfo.qualityLevels[j - 1]); - if (SUCCEEDED(result) && supportInfo.qualityLevels[j - 1] > 0) - { - maxSupportedSamples = std::max(j, maxSupportedSamples); - } - else - { - supportInfo.qualityLevels[j - 1] = 0; - } - } - - mMultisampleSupportMap.insert(std::make_pair(format, supportInfo)); - } - } - } - mMaxSupportedSamples = maxSupportedSamples; - - initializeDevice(); - - // BGRA texture support is optional in feature levels 10 and 10_1 - UINT formatSupport; - result = mDevice->CheckFormatSupport(DXGI_FORMAT_B8G8R8A8_UNORM, &formatSupport); - if (FAILED(result)) - { - ERR("Error checking BGRA format support: 0x%08X", result); - } - else - { - const int flags = (D3D11_FORMAT_SUPPORT_TEXTURE2D | D3D11_FORMAT_SUPPORT_RENDER_TARGET); - mBGRATextureSupport = (formatSupport & flags) == flags; - } - - // Check floating point texture support - static const unsigned int requiredTextureFlags = D3D11_FORMAT_SUPPORT_TEXTURE2D | D3D11_FORMAT_SUPPORT_TEXTURECUBE; - static const unsigned int requiredRenderableFlags = D3D11_FORMAT_SUPPORT_RENDER_TARGET; - static const unsigned int requiredFilterFlags = D3D11_FORMAT_SUPPORT_SHADER_SAMPLE; - - DXGI_FORMAT float16Formats[] = - { - DXGI_FORMAT_R16_FLOAT, - DXGI_FORMAT_R16G16_FLOAT, - DXGI_FORMAT_R16G16B16A16_FLOAT, - }; - - DXGI_FORMAT float32Formats[] = - { - DXGI_FORMAT_R32_FLOAT, - DXGI_FORMAT_R32G32_FLOAT, - DXGI_FORMAT_R32G32B32A32_FLOAT, - }; - - mFloat16TextureSupport = true; - mFloat16FilterSupport = true; - mFloat16RenderSupport = true; - for (unsigned int i = 0; i < ArraySize(float16Formats); i++) - { - if (SUCCEEDED(mDevice->CheckFormatSupport(float16Formats[i], &formatSupport))) - { - mFloat16TextureSupport = mFloat16TextureSupport && (formatSupport & requiredTextureFlags) == requiredTextureFlags; - mFloat16FilterSupport = mFloat16FilterSupport && (formatSupport & requiredFilterFlags) == requiredFilterFlags; - mFloat16RenderSupport = mFloat16RenderSupport && (formatSupport & requiredRenderableFlags) == requiredRenderableFlags; - } - else - { - mFloat16TextureSupport = false; - mFloat16RenderSupport = false; - mFloat16FilterSupport = false; - } - } - - mFloat32TextureSupport = true; - mFloat32FilterSupport = true; - mFloat32RenderSupport = true; - for (unsigned int i = 0; i < ArraySize(float32Formats); i++) - { - if (SUCCEEDED(mDevice->CheckFormatSupport(float32Formats[i], &formatSupport))) - { - mFloat32TextureSupport = mFloat32TextureSupport && (formatSupport & requiredTextureFlags) == requiredTextureFlags; - mFloat32FilterSupport = mFloat32FilterSupport && (formatSupport & requiredFilterFlags) == requiredFilterFlags; - mFloat32RenderSupport = mFloat32RenderSupport && (formatSupport & requiredRenderableFlags) == requiredRenderableFlags; - } - else - { - mFloat32TextureSupport = false; - mFloat32FilterSupport = false; - mFloat32RenderSupport = false; - } - } - - // Check compressed texture support - const unsigned int requiredCompressedTextureFlags = D3D11_FORMAT_SUPPORT_TEXTURE2D; - - if (SUCCEEDED(mDevice->CheckFormatSupport(DXGI_FORMAT_BC1_UNORM, &formatSupport))) - { - mDXT1TextureSupport = (formatSupport & requiredCompressedTextureFlags) == requiredCompressedTextureFlags; - } - else - { - mDXT1TextureSupport = false; - } - - if (SUCCEEDED(mDevice->CheckFormatSupport(DXGI_FORMAT_BC3_UNORM, &formatSupport))) - { - mDXT3TextureSupport = (formatSupport & requiredCompressedTextureFlags) == requiredCompressedTextureFlags; - } - else - { - mDXT3TextureSupport = false; - } - - if (SUCCEEDED(mDevice->CheckFormatSupport(DXGI_FORMAT_BC5_UNORM, &formatSupport))) - { - mDXT5TextureSupport = (formatSupport & requiredCompressedTextureFlags) == requiredCompressedTextureFlags; - } - else - { - mDXT5TextureSupport = false; - } - - // Check depth texture support - DXGI_FORMAT depthTextureFormats[] = - { - DXGI_FORMAT_D16_UNORM, - DXGI_FORMAT_D24_UNORM_S8_UINT, - }; - - static const unsigned int requiredDepthTextureFlags = D3D11_FORMAT_SUPPORT_DEPTH_STENCIL | - D3D11_FORMAT_SUPPORT_TEXTURE2D; - - mDepthTextureSupport = true; - for (unsigned int i = 0; i < ArraySize(depthTextureFormats); i++) - { - if (SUCCEEDED(mDevice->CheckFormatSupport(depthTextureFormats[i], &formatSupport))) - { - mDepthTextureSupport = mDepthTextureSupport && ((formatSupport & requiredDepthTextureFlags) == requiredDepthTextureFlags); - } - else - { - mDepthTextureSupport = false; - } - } - -#if defined(ANGLE_OS_WINRT) && !defined(ANGLE_OS_WINPHONE) - // Monitor when the application suspends so that Trim() can be called - Microsoft::WRL::ComPtr<ABI::Windows::ApplicationModel::Core::ICoreApplication> application; - result = RoGetActivationFactory(Microsoft::WRL::Wrappers::HString::MakeReference(RuntimeClass_Windows_ApplicationModel_Core_CoreApplication).Get(), - IID_PPV_ARGS(&application)); - if (FAILED(result)) - { - ERR("Error obtaining CoreApplication: 0x%08X", result); - return EGL_NOT_INITIALIZED; - } - - EventRegistrationToken cookie; - result = application->add_Suspending(Microsoft::WRL::Callback<SuspendEventHandler>(this, &Renderer11::onSuspend).Get(), &cookie); - if (FAILED(result)) - { - ERR("Error setting suspend callback: 0x%08X", result); - return EGL_NOT_INITIALIZED; - } -#endif - - return EGL_SUCCESS; -} - -#if defined(ANGLE_OS_WINRT) && !defined(ANGLE_OS_WINPHONE) -HRESULT Renderer11::onSuspend(IInspectable *, ABI::Windows::ApplicationModel::ISuspendingEventArgs *) -{ - if (!mDevice) - return S_OK; - - Microsoft::WRL::ComPtr<IDXGIDevice3> dxgiDevice; - HRESULT result = mDevice->QueryInterface(IID_PPV_ARGS(&dxgiDevice)); - if (FAILED(result)) - { - ERR("Error obtaining DXGIDevice3 on suspend: 0x%08X", result); - return S_OK; - } - - dxgiDevice->Trim(); - - return S_OK; -} -#endif - -// do any one-time device initialization -// NOTE: this is also needed after a device lost/reset -// to reset the scene status and ensure the default states are reset. -void Renderer11::initializeDevice() -{ - mStateCache.initialize(mDevice); - mInputLayoutCache.initialize(mDevice, mDeviceContext); - - ASSERT(!mVertexDataManager && !mIndexDataManager); - mVertexDataManager = new VertexDataManager(this); - mIndexDataManager = new IndexDataManager(this); - - markAllStateDirty(); -} - -int Renderer11::generateConfigs(ConfigDesc **configDescList) -{ - unsigned int numRenderFormats = ArraySize(RenderTargetFormats); - unsigned int numDepthFormats = ArraySize(DepthStencilFormats); - (*configDescList) = new ConfigDesc[numRenderFormats * numDepthFormats]; - int numConfigs = 0; - - for (unsigned int formatIndex = 0; formatIndex < numRenderFormats; formatIndex++) - { - for (unsigned int depthStencilIndex = 0; depthStencilIndex < numDepthFormats; depthStencilIndex++) - { - DXGI_FORMAT renderTargetFormat = RenderTargetFormats[formatIndex]; - - UINT formatSupport = 0; - HRESULT result = mDevice->CheckFormatSupport(renderTargetFormat, &formatSupport); - - if (SUCCEEDED(result) && (formatSupport & D3D11_FORMAT_SUPPORT_RENDER_TARGET)) - { - DXGI_FORMAT depthStencilFormat = DepthStencilFormats[depthStencilIndex]; - - bool depthStencilFormatOK = true; - - if (depthStencilFormat != DXGI_FORMAT_UNKNOWN) - { - UINT formatSupport = 0; - result = mDevice->CheckFormatSupport(depthStencilFormat, &formatSupport); - depthStencilFormatOK = SUCCEEDED(result) && (formatSupport & D3D11_FORMAT_SUPPORT_DEPTH_STENCIL); - } - - if (depthStencilFormatOK) - { - ConfigDesc newConfig; - newConfig.renderTargetFormat = d3d11_gl::ConvertBackBufferFormat(renderTargetFormat); - newConfig.depthStencilFormat = d3d11_gl::ConvertDepthStencilFormat(depthStencilFormat); - newConfig.multiSample = 0; // FIXME: enumerate multi-sampling - newConfig.fastConfig = true; // Assume all DX11 format conversions to be fast - - (*configDescList)[numConfigs++] = newConfig; - } - } - } - } - - return numConfigs; -} - -void Renderer11::deleteConfigs(ConfigDesc *configDescList) -{ - delete [] (configDescList); -} - -void Renderer11::sync(bool block) -{ - if (block) - { - HRESULT result; - - if (!mSyncQuery) - { - D3D11_QUERY_DESC queryDesc; - queryDesc.Query = D3D11_QUERY_EVENT; - queryDesc.MiscFlags = 0; - - result = mDevice->CreateQuery(&queryDesc, &mSyncQuery); - ASSERT(SUCCEEDED(result)); - } - - mDeviceContext->End(mSyncQuery); - mDeviceContext->Flush(); - - do - { - result = mDeviceContext->GetData(mSyncQuery, NULL, 0, D3D11_ASYNC_GETDATA_DONOTFLUSH); - - // Keep polling, but allow other threads to do something useful first - Sleep(0); - - if (testDeviceLost(true)) - { - return; - } - } - while (result == S_FALSE); - } - else - { - mDeviceContext->Flush(); - } -} - -SwapChain *Renderer11::createSwapChain(EGLNativeWindowType window, HANDLE shareHandle, GLenum backBufferFormat, GLenum depthBufferFormat) -{ - return new rx::SwapChain11(this, window, shareHandle, backBufferFormat, depthBufferFormat); -} - -void Renderer11::setSamplerState(gl::SamplerType type, int index, const gl::SamplerState &samplerState) -{ - if (type == gl::SAMPLER_PIXEL) - { - if (index < 0 || index >= gl::MAX_TEXTURE_IMAGE_UNITS) - { - ERR("Pixel shader sampler index %i is not valid.", index); - return; - } - - if (mForceSetPixelSamplerStates[index] || memcmp(&samplerState, &mCurPixelSamplerStates[index], sizeof(gl::SamplerState)) != 0) - { - ID3D11SamplerState *dxSamplerState = mStateCache.getSamplerState(samplerState); - - if (!dxSamplerState) - { - ERR("NULL sampler state returned by RenderStateCache::getSamplerState, setting the default" - "sampler state for pixel shaders at slot %i.", index); - } - - mDeviceContext->PSSetSamplers(index, 1, &dxSamplerState); - - mCurPixelSamplerStates[index] = samplerState; - } - - mForceSetPixelSamplerStates[index] = false; - } - else if (type == gl::SAMPLER_VERTEX) - { - if (index < 0 || index >= (int)getMaxVertexTextureImageUnits()) - { - ERR("Vertex shader sampler index %i is not valid.", index); - return; - } - - if (mForceSetVertexSamplerStates[index] || memcmp(&samplerState, &mCurVertexSamplerStates[index], sizeof(gl::SamplerState)) != 0) - { - ID3D11SamplerState *dxSamplerState = mStateCache.getSamplerState(samplerState); - - if (!dxSamplerState) - { - ERR("NULL sampler state returned by RenderStateCache::getSamplerState, setting the default" - "sampler state for vertex shaders at slot %i.", index); - } - - mDeviceContext->VSSetSamplers(index, 1, &dxSamplerState); - - mCurVertexSamplerStates[index] = samplerState; - } - - mForceSetVertexSamplerStates[index] = false; - } - else UNREACHABLE(); -} - -void Renderer11::setTexture(gl::SamplerType type, int index, gl::Texture *texture) -{ - ID3D11ShaderResourceView *textureSRV = NULL; - unsigned int serial = 0; - bool forceSetTexture = false; - - if (texture) - { - TextureStorageInterface *texStorage = texture->getNativeTexture(); - if (texStorage) - { - TextureStorage11 *storage11 = TextureStorage11::makeTextureStorage11(texStorage->getStorageInstance()); - textureSRV = storage11->getSRV(); - } - - // If we get NULL back from getSRV here, something went wrong in the texture class and we're unexpectedly - // missing the shader resource view - ASSERT(textureSRV != NULL); - - serial = texture->getTextureSerial(); - forceSetTexture = texture->hasDirtyImages(); - } - - if (type == gl::SAMPLER_PIXEL) - { - if (index < 0 || index >= gl::MAX_TEXTURE_IMAGE_UNITS) - { - ERR("Pixel shader sampler index %i is not valid.", index); - return; - } - - if (forceSetTexture || mCurPixelTextureSerials[index] != serial) - { - mDeviceContext->PSSetShaderResources(index, 1, &textureSRV); - } - - mCurPixelTextureSerials[index] = serial; - } - else if (type == gl::SAMPLER_VERTEX) - { - if (index < 0 || index >= (int)getMaxVertexTextureImageUnits()) - { - ERR("Vertex shader sampler index %i is not valid.", index); - return; - } - - if (forceSetTexture || mCurVertexTextureSerials[index] != serial) - { - mDeviceContext->VSSetShaderResources(index, 1, &textureSRV); - } - - mCurVertexTextureSerials[index] = serial; - } - else UNREACHABLE(); -} - -void Renderer11::setRasterizerState(const gl::RasterizerState &rasterState) -{ - if (mForceSetRasterState || memcmp(&rasterState, &mCurRasterState, sizeof(gl::RasterizerState)) != 0) - { - ID3D11RasterizerState *dxRasterState = mStateCache.getRasterizerState(rasterState, mScissorEnabled, - mCurDepthSize); - if (!dxRasterState) - { - ERR("NULL rasterizer state returned by RenderStateCache::getRasterizerState, setting the default" - "rasterizer state."); - } - - mDeviceContext->RSSetState(dxRasterState); - - mCurRasterState = rasterState; - } - - mForceSetRasterState = false; -} - -void Renderer11::setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState &blendState, const gl::Color &blendColor, - unsigned int sampleMask) -{ - if (mForceSetBlendState || - memcmp(&blendState, &mCurBlendState, sizeof(gl::BlendState)) != 0 || - memcmp(&blendColor, &mCurBlendColor, sizeof(gl::Color)) != 0 || - sampleMask != mCurSampleMask) - { - ID3D11BlendState *dxBlendState = mStateCache.getBlendState(framebuffer, blendState); - if (!dxBlendState) - { - ERR("NULL blend state returned by RenderStateCache::getBlendState, setting the default " - "blend state."); - } - - float blendColors[4] = {0.0f}; - if (blendState.sourceBlendRGB != GL_CONSTANT_ALPHA && blendState.sourceBlendRGB != GL_ONE_MINUS_CONSTANT_ALPHA && - blendState.destBlendRGB != GL_CONSTANT_ALPHA && blendState.destBlendRGB != GL_ONE_MINUS_CONSTANT_ALPHA) - { - blendColors[0] = blendColor.red; - blendColors[1] = blendColor.green; - blendColors[2] = blendColor.blue; - blendColors[3] = blendColor.alpha; - } - else - { - blendColors[0] = blendColor.alpha; - blendColors[1] = blendColor.alpha; - blendColors[2] = blendColor.alpha; - blendColors[3] = blendColor.alpha; - } - - mDeviceContext->OMSetBlendState(dxBlendState, blendColors, sampleMask); - - mCurBlendState = blendState; - mCurBlendColor = blendColor; - mCurSampleMask = sampleMask; - } - - mForceSetBlendState = false; -} - -void Renderer11::setDepthStencilState(const gl::DepthStencilState &depthStencilState, int stencilRef, - int stencilBackRef, bool frontFaceCCW) -{ - if (mForceSetDepthStencilState || - memcmp(&depthStencilState, &mCurDepthStencilState, sizeof(gl::DepthStencilState)) != 0 || - stencilRef != mCurStencilRef || stencilBackRef != mCurStencilBackRef) - { - if (depthStencilState.stencilWritemask != depthStencilState.stencilBackWritemask || - stencilRef != stencilBackRef || - depthStencilState.stencilMask != depthStencilState.stencilBackMask) - { - ERR("Separate front/back stencil writemasks, reference values, or stencil mask values are " - "invalid under WebGL."); - return gl::error(GL_INVALID_OPERATION); - } - - ID3D11DepthStencilState *dxDepthStencilState = mStateCache.getDepthStencilState(depthStencilState); - if (!dxDepthStencilState) - { - ERR("NULL depth stencil state returned by RenderStateCache::getDepthStencilState, " - "setting the default depth stencil state."); - } - - mDeviceContext->OMSetDepthStencilState(dxDepthStencilState, static_cast<UINT>(stencilRef)); - - mCurDepthStencilState = depthStencilState; - mCurStencilRef = stencilRef; - mCurStencilBackRef = stencilBackRef; - } - - mForceSetDepthStencilState = false; -} - -void Renderer11::setScissorRectangle(const gl::Rectangle &scissor, bool enabled) -{ - if (mForceSetScissor || memcmp(&scissor, &mCurScissor, sizeof(gl::Rectangle)) != 0 || - enabled != mScissorEnabled) - { - if (enabled) - { - D3D11_RECT rect; - rect.left = std::max(0, scissor.x); - rect.top = std::max(0, scissor.y); - rect.right = scissor.x + std::max(0, scissor.width); - rect.bottom = scissor.y + std::max(0, scissor.height); - - mDeviceContext->RSSetScissorRects(1, &rect); - } - - if (enabled != mScissorEnabled) - { - mForceSetRasterState = true; - } - - mCurScissor = scissor; - mScissorEnabled = enabled; - } - - mForceSetScissor = false; -} - -bool Renderer11::setViewport(const gl::Rectangle &viewport, float zNear, float zFar, GLenum drawMode, GLenum frontFace, - bool ignoreViewport) -{ - gl::Rectangle actualViewport = viewport; - float actualZNear = gl::clamp01(zNear); - float actualZFar = gl::clamp01(zFar); - if (ignoreViewport) - { - actualViewport.x = 0; - actualViewport.y = 0; - actualViewport.width = mRenderTargetDesc.width; - actualViewport.height = mRenderTargetDesc.height; - actualZNear = 0.0f; - actualZFar = 1.0f; - } - - // Get D3D viewport bounds, which depends on the feature level - const Range& viewportBounds = getViewportBounds(); - - // Clamp width and height first to the gl maximum, then clamp further if we extend past the D3D maximum bounds - D3D11_VIEWPORT dxViewport; - dxViewport.TopLeftX = gl::clamp(actualViewport.x, viewportBounds.start, viewportBounds.end); - dxViewport.TopLeftY = gl::clamp(actualViewport.y, viewportBounds.start, viewportBounds.end); - dxViewport.Width = gl::clamp(actualViewport.width, 0, getMaxViewportDimension()); - dxViewport.Height = gl::clamp(actualViewport.height, 0, getMaxViewportDimension()); - dxViewport.Width = std::min((int)dxViewport.Width, viewportBounds.end - static_cast<int>(dxViewport.TopLeftX)); - dxViewport.Height = std::min((int)dxViewport.Height, viewportBounds.end - static_cast<int>(dxViewport.TopLeftY)); - dxViewport.MinDepth = actualZNear; - dxViewport.MaxDepth = actualZFar; - - if (dxViewport.Width <= 0 || dxViewport.Height <= 0) - { - return false; // Nothing to render - } - - bool viewportChanged = mForceSetViewport || memcmp(&actualViewport, &mCurViewport, sizeof(gl::Rectangle)) != 0 || - actualZNear != mCurNear || actualZFar != mCurFar; - - if (viewportChanged) - { - mDeviceContext->RSSetViewports(1, &dxViewport); - - mCurViewport = actualViewport; - mCurNear = actualZNear; - mCurFar = actualZFar; - - mPixelConstants.viewCoords[0] = actualViewport.width * 0.5f; - mPixelConstants.viewCoords[1] = actualViewport.height * 0.5f; - mPixelConstants.viewCoords[2] = actualViewport.x + (actualViewport.width * 0.5f); - mPixelConstants.viewCoords[3] = actualViewport.y + (actualViewport.height * 0.5f); - - mPixelConstants.depthFront[0] = (actualZFar - actualZNear) * 0.5f; - mPixelConstants.depthFront[1] = (actualZNear + actualZFar) * 0.5f; - - mVertexConstants.depthRange[0] = actualZNear; - mVertexConstants.depthRange[1] = actualZFar; - mVertexConstants.depthRange[2] = actualZFar - actualZNear; - - mPixelConstants.depthRange[0] = actualZNear; - mPixelConstants.depthRange[1] = actualZFar; - mPixelConstants.depthRange[2] = actualZFar - actualZNear; - } - - mForceSetViewport = false; - return true; -} - -bool Renderer11::applyPrimitiveType(GLenum mode, GLsizei count) -{ - D3D11_PRIMITIVE_TOPOLOGY primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_UNDEFINED; - - GLsizei minCount = 0; - - switch (mode) - { - case GL_POINTS: primitiveTopology = D3D11_PRIMITIVE_TOPOLOGY_POINTLIST; minCount = 1; break; - case GL_LINES: primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_LINELIST; minCount = 2; break; - case GL_LINE_LOOP: primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_LINESTRIP; minCount = 2; break; - case GL_LINE_STRIP: primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_LINESTRIP; minCount = 2; break; - case GL_TRIANGLES: primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_TRIANGLELIST; minCount = 3; break; - case GL_TRIANGLE_STRIP: primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_TRIANGLESTRIP; minCount = 3; break; - // emulate fans via rewriting index buffer - case GL_TRIANGLE_FAN: primitiveTopology = D3D_PRIMITIVE_TOPOLOGY_TRIANGLELIST; minCount = 3; break; - default: - return gl::error(GL_INVALID_ENUM, false); - } - - if (primitiveTopology != mCurrentPrimitiveTopology) - { - mDeviceContext->IASetPrimitiveTopology(primitiveTopology); - mCurrentPrimitiveTopology = primitiveTopology; - } - - return count >= minCount; -} - -bool Renderer11::applyRenderTarget(gl::Framebuffer *framebuffer) -{ - // Get the color render buffer and serial - // Also extract the render target dimensions and view - unsigned int renderTargetWidth = 0; - unsigned int renderTargetHeight = 0; - GLenum renderTargetFormat = 0; - unsigned int renderTargetSerials[gl::IMPLEMENTATION_MAX_DRAW_BUFFERS] = {0}; - ID3D11RenderTargetView* framebufferRTVs[gl::IMPLEMENTATION_MAX_DRAW_BUFFERS] = {NULL}; - bool missingColorRenderTarget = true; - - for (unsigned int colorAttachment = 0; colorAttachment < gl::IMPLEMENTATION_MAX_DRAW_BUFFERS; colorAttachment++) - { - const GLenum drawBufferState = framebuffer->getDrawBufferState(colorAttachment); - - if (framebuffer->getColorbufferType(colorAttachment) != GL_NONE && drawBufferState != GL_NONE) - { - // the draw buffer must be either "none", "back" for the default buffer or the same index as this color (in order) - ASSERT(drawBufferState == GL_BACK || drawBufferState == (GL_COLOR_ATTACHMENT0_EXT + colorAttachment)); - - gl::Renderbuffer *colorbuffer = framebuffer->getColorbuffer(colorAttachment); - - if (!colorbuffer) - { - ERR("render target pointer unexpectedly null."); - return false; - } - - // check for zero-sized default framebuffer, which is a special case. - // in this case we do not wish to modify any state and just silently return false. - // this will not report any gl error but will cause the calling method to return. - if (colorbuffer->getWidth() == 0 || colorbuffer->getHeight() == 0) - { - return false; - } - - renderTargetSerials[colorAttachment] = colorbuffer->getSerial(); - - // Extract the render target dimensions and view - RenderTarget11 *renderTarget = RenderTarget11::makeRenderTarget11(colorbuffer->getRenderTarget()); - if (!renderTarget) - { - ERR("render target pointer unexpectedly null."); - return false; - } - - framebufferRTVs[colorAttachment] = renderTarget->getRenderTargetView(); - if (!framebufferRTVs[colorAttachment]) - { - ERR("render target view pointer unexpectedly null."); - return false; - } - - if (missingColorRenderTarget) - { - renderTargetWidth = colorbuffer->getWidth(); - renderTargetHeight = colorbuffer->getHeight(); - renderTargetFormat = colorbuffer->getActualFormat(); - missingColorRenderTarget = false; - } - -#ifdef _DEBUG - // Workaround for Debug SETSHADERRESOURCES_HAZARD D3D11 warnings - for (unsigned int vertexSerialIndex = 0; vertexSerialIndex < gl::IMPLEMENTATION_MAX_VERTEX_TEXTURE_IMAGE_UNITS; vertexSerialIndex++) - { - if (colorbuffer->getTextureSerial() != 0 && mCurVertexTextureSerials[vertexSerialIndex] == colorbuffer->getTextureSerial()) - { - setTexture(gl::SAMPLER_VERTEX, vertexSerialIndex, NULL); - } - } - - for (unsigned int pixelSerialIndex = 0; pixelSerialIndex < gl::MAX_TEXTURE_IMAGE_UNITS; pixelSerialIndex++) - { - if (colorbuffer->getTextureSerial() != 0 && mCurPixelTextureSerials[pixelSerialIndex] == colorbuffer->getTextureSerial()) - { - setTexture(gl::SAMPLER_PIXEL, pixelSerialIndex, NULL); - } - } -#endif - } - } - - // Get the depth stencil render buffer and serials - gl::Renderbuffer *depthStencil = NULL; - unsigned int depthbufferSerial = 0; - unsigned int stencilbufferSerial = 0; - if (framebuffer->getDepthbufferType() != GL_NONE) - { - depthStencil = framebuffer->getDepthbuffer(); - if (!depthStencil) - { - ERR("Depth stencil pointer unexpectedly null."); - SafeRelease(framebufferRTVs); - return false; - } - - depthbufferSerial = depthStencil->getSerial(); - } - else if (framebuffer->getStencilbufferType() != GL_NONE) - { - depthStencil = framebuffer->getStencilbuffer(); - if (!depthStencil) - { - ERR("Depth stencil pointer unexpectedly null."); - SafeRelease(framebufferRTVs); - return false; - } - - stencilbufferSerial = depthStencil->getSerial(); - } - - // Extract the depth stencil sizes and view - unsigned int depthSize = 0; - unsigned int stencilSize = 0; - ID3D11DepthStencilView* framebufferDSV = NULL; - if (depthStencil) - { - RenderTarget11 *depthStencilRenderTarget = RenderTarget11::makeRenderTarget11(depthStencil->getDepthStencil()); - if (!depthStencilRenderTarget) - { - ERR("render target pointer unexpectedly null."); - SafeRelease(framebufferRTVs); - return false; - } - - framebufferDSV = depthStencilRenderTarget->getDepthStencilView(); - if (!framebufferDSV) - { - ERR("depth stencil view pointer unexpectedly null."); - SafeRelease(framebufferRTVs); - return false; - } - - // If there is no render buffer, the width, height and format values come from - // the depth stencil - if (missingColorRenderTarget) - { - renderTargetWidth = depthStencil->getWidth(); - renderTargetHeight = depthStencil->getHeight(); - renderTargetFormat = depthStencil->getActualFormat(); - } - - depthSize = depthStencil->getDepthSize(); - stencilSize = depthStencil->getStencilSize(); - } - - // Apply the render target and depth stencil - if (!mRenderTargetDescInitialized || !mDepthStencilInitialized || - memcmp(renderTargetSerials, mAppliedRenderTargetSerials, sizeof(renderTargetSerials)) != 0 || - depthbufferSerial != mAppliedDepthbufferSerial || - stencilbufferSerial != mAppliedStencilbufferSerial) - { - mDeviceContext->OMSetRenderTargets(getMaxRenderTargets(), framebufferRTVs, framebufferDSV); - - mRenderTargetDesc.width = renderTargetWidth; - mRenderTargetDesc.height = renderTargetHeight; - mRenderTargetDesc.format = renderTargetFormat; - mForceSetViewport = true; - mForceSetScissor = true; - - if (!mDepthStencilInitialized || depthSize != mCurDepthSize) - { - mCurDepthSize = depthSize; - mForceSetRasterState = true; - } - - mCurStencilSize = stencilSize; - - for (unsigned int rtIndex = 0; rtIndex < gl::IMPLEMENTATION_MAX_DRAW_BUFFERS; rtIndex++) - { - mAppliedRenderTargetSerials[rtIndex] = renderTargetSerials[rtIndex]; - } - mAppliedDepthbufferSerial = depthbufferSerial; - mAppliedStencilbufferSerial = stencilbufferSerial; - mRenderTargetDescInitialized = true; - mDepthStencilInitialized = true; - } - - return true; -} - -GLenum Renderer11::applyVertexBuffer(gl::ProgramBinary *programBinary, gl::VertexAttribute vertexAttributes[], GLint first, GLsizei count, GLsizei instances) -{ - TranslatedAttribute attributes[gl::MAX_VERTEX_ATTRIBS]; - GLenum err = mVertexDataManager->prepareVertexData(vertexAttributes, programBinary, first, count, attributes, instances); - if (err != GL_NO_ERROR) - { - return err; - } - - return mInputLayoutCache.applyVertexBuffers(attributes, programBinary); -} - -GLenum Renderer11::applyIndexBuffer(const GLvoid *indices, gl::Buffer *elementArrayBuffer, GLsizei count, GLenum mode, GLenum type, TranslatedIndexData *indexInfo) -{ - GLenum err = mIndexDataManager->prepareIndexData(type, count, elementArrayBuffer, indices, indexInfo); - - if (err == GL_NO_ERROR) - { - if (indexInfo->storage) - { - if (indexInfo->serial != mAppliedStorageIBSerial || indexInfo->startOffset != mAppliedIBOffset) - { - BufferStorage11 *storage = BufferStorage11::makeBufferStorage11(indexInfo->storage); - IndexBuffer11* indexBuffer = IndexBuffer11::makeIndexBuffer11(indexInfo->indexBuffer); - - mDeviceContext->IASetIndexBuffer(storage->getBuffer(BUFFER_USAGE_INDEX), indexBuffer->getIndexFormat(), indexInfo->startOffset); - - mAppliedIBSerial = 0; - mAppliedStorageIBSerial = storage->getSerial(); - mAppliedIBOffset = indexInfo->startOffset; - } - } - else if (indexInfo->serial != mAppliedIBSerial || indexInfo->startOffset != mAppliedIBOffset) - { - IndexBuffer11* indexBuffer = IndexBuffer11::makeIndexBuffer11(indexInfo->indexBuffer); - - mDeviceContext->IASetIndexBuffer(indexBuffer->getBuffer(), indexBuffer->getIndexFormat(), indexInfo->startOffset); - - mAppliedIBSerial = indexInfo->serial; - mAppliedStorageIBSerial = 0; - mAppliedIBOffset = indexInfo->startOffset; - } - } - - return err; -} - -void Renderer11::drawArrays(GLenum mode, GLsizei count, GLsizei instances) -{ - if (mode == GL_LINE_LOOP) - { - drawLineLoop(count, GL_NONE, NULL, 0, NULL); - } - else if (mode == GL_TRIANGLE_FAN) - { - drawTriangleFan(count, GL_NONE, NULL, 0, NULL, instances); - } - else if (instances > 0) - { - mDeviceContext->DrawInstanced(count, instances, 0, 0); - } - else - { - mDeviceContext->Draw(count, 0); - } -} - -void Renderer11::drawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, gl::Buffer *elementArrayBuffer, const TranslatedIndexData &indexInfo, GLsizei instances) -{ - if (mode == GL_LINE_LOOP) - { - drawLineLoop(count, type, indices, indexInfo.minIndex, elementArrayBuffer); - } - else if (mode == GL_TRIANGLE_FAN) - { - drawTriangleFan(count, type, indices, indexInfo.minIndex, elementArrayBuffer, instances); - } - else if (instances > 0) - { - mDeviceContext->DrawIndexedInstanced(count, instances, 0, -static_cast<int>(indexInfo.minIndex), 0); - } - else - { - mDeviceContext->DrawIndexed(count, 0, -static_cast<int>(indexInfo.minIndex)); - } -} - -template<typename T> -static void fillLineLoopIndices(GLenum type, GLsizei count, const GLvoid *indices, T *data) -{ - switch (type) - { - case GL_NONE: // Non-indexed draw - for (int i = 0; i < count; i++) - { - data[i] = i; - } - data[count] = 0; - break; - case GL_UNSIGNED_BYTE: - for (int i = 0; i < count; i++) - { - data[i] = static_cast<const GLubyte*>(indices)[i]; - } - data[count] = static_cast<const GLubyte*>(indices)[0]; - break; - case GL_UNSIGNED_SHORT: - for (int i = 0; i < count; i++) - { - data[i] = static_cast<const GLushort*>(indices)[i]; - } - data[count] = static_cast<const GLushort*>(indices)[0]; - break; - case GL_UNSIGNED_INT: - for (int i = 0; i < count; i++) - { - data[i] = static_cast<const GLuint*>(indices)[i]; - } - data[count] = static_cast<const GLuint*>(indices)[0]; - break; - default: UNREACHABLE(); - } -} - -template<typename T> -static void fillTriangleFanIndices(GLenum type, unsigned int numTris, const GLvoid *indices, T *data) -{ - switch (type) - { - case GL_NONE: // Non-indexed draw - for (unsigned int i = 0; i < numTris; i++) - { - data[i*3 + 0] = 0; - data[i*3 + 1] = i + 1; - data[i*3 + 2] = i + 2; - } - break; - case GL_UNSIGNED_BYTE: - for (unsigned int i = 0; i < numTris; i++) - { - data[i*3 + 0] = static_cast<const GLubyte*>(indices)[0]; - data[i*3 + 1] = static_cast<const GLubyte*>(indices)[i + 1]; - data[i*3 + 2] = static_cast<const GLubyte*>(indices)[i + 2]; - } - break; - case GL_UNSIGNED_SHORT: - for (unsigned int i = 0; i < numTris; i++) - { - data[i*3 + 0] = static_cast<const GLushort*>(indices)[0]; - data[i*3 + 1] = static_cast<const GLushort*>(indices)[i + 1]; - data[i*3 + 2] = static_cast<const GLushort*>(indices)[i + 2]; - } - break; - case GL_UNSIGNED_INT: - for (unsigned int i = 0; i < numTris; i++) - { - data[i*3 + 0] = static_cast<const GLuint*>(indices)[0]; - data[i*3 + 1] = static_cast<const GLuint*>(indices)[i + 1]; - data[i*3 + 2] = static_cast<const GLuint*>(indices)[i + 2]; - } - break; - default: UNREACHABLE(); - } -} - -void Renderer11::drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer) -{ - // Get the raw indices for an indexed draw - if (type != GL_NONE && elementArrayBuffer) - { - gl::Buffer *indexBuffer = elementArrayBuffer; - BufferStorage *storage = indexBuffer->getStorage(); - intptr_t offset = reinterpret_cast<intptr_t>(indices); - indices = static_cast<const GLubyte*>(storage->getData()) + offset; - } - - const int indexType = get32BitIndexSupport() ? GL_UNSIGNED_INT : GL_UNSIGNED_SHORT; - - if (!mLineLoopIB) - { - mLineLoopIB = new StreamingIndexBufferInterface(this); - if (!mLineLoopIB->reserveBufferSpace(INITIAL_INDEX_BUFFER_SIZE, indexType)) - { - delete mLineLoopIB; - mLineLoopIB = NULL; - - ERR("Could not create a 32-bit looping index buffer for GL_LINE_LOOP."); - return gl::error(GL_OUT_OF_MEMORY); - } - } - - // Checked by Renderer11::applyPrimitiveType - ASSERT(count >= 0); - - if (static_cast<unsigned int>(count) + 1 > (std::numeric_limits<unsigned int>::max() / sizeof(unsigned int))) - { - ERR("Could not create a 32-bit looping index buffer for GL_LINE_LOOP, too many indices required."); - return gl::error(GL_OUT_OF_MEMORY); - } - - const unsigned int spaceNeeded = (static_cast<unsigned int>(count) + 1) * sizeof(unsigned int); - if (!mLineLoopIB->reserveBufferSpace(spaceNeeded, indexType)) - { - ERR("Could not reserve enough space in looping index buffer for GL_LINE_LOOP."); - return gl::error(GL_OUT_OF_MEMORY); - } - - void* mappedMemory = NULL; - unsigned int offset; - if (!mLineLoopIB->mapBuffer(spaceNeeded, &mappedMemory, &offset)) - { - ERR("Could not map index buffer for GL_LINE_LOOP."); - return gl::error(GL_OUT_OF_MEMORY); - } - - if (indexType == GL_UNSIGNED_SHORT) - fillLineLoopIndices(type, count, indices, reinterpret_cast<unsigned short*>(mappedMemory)); - else - fillLineLoopIndices(type, count, indices, reinterpret_cast<unsigned int*>(mappedMemory)); - unsigned int indexBufferOffset = offset; - - if (!mLineLoopIB->unmapBuffer()) - { - ERR("Could not unmap index buffer for GL_LINE_LOOP."); - return gl::error(GL_OUT_OF_MEMORY); - } - - if (mAppliedIBSerial != mLineLoopIB->getSerial() || mAppliedIBOffset != indexBufferOffset) - { - IndexBuffer11 *indexBuffer = IndexBuffer11::makeIndexBuffer11(mLineLoopIB->getIndexBuffer()); - - mDeviceContext->IASetIndexBuffer(indexBuffer->getBuffer(), indexBuffer->getIndexFormat(), indexBufferOffset); - mAppliedIBSerial = mLineLoopIB->getSerial(); - mAppliedStorageIBSerial = 0; - mAppliedIBOffset = indexBufferOffset; - } - - mDeviceContext->DrawIndexed(count + 1, 0, -minIndex); -} - -void Renderer11::drawTriangleFan(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer, int instances) -{ - // Get the raw indices for an indexed draw - if (type != GL_NONE && elementArrayBuffer) - { - gl::Buffer *indexBuffer = elementArrayBuffer; - BufferStorage *storage = indexBuffer->getStorage(); - intptr_t offset = reinterpret_cast<intptr_t>(indices); - indices = static_cast<const GLubyte*>(storage->getData()) + offset; - } - - const int indexType = get32BitIndexSupport() ? GL_UNSIGNED_INT : GL_UNSIGNED_SHORT; - - if (!mTriangleFanIB) - { - mTriangleFanIB = new StreamingIndexBufferInterface(this); - if (!mTriangleFanIB->reserveBufferSpace(INITIAL_INDEX_BUFFER_SIZE, indexType)) - { - delete mTriangleFanIB; - mTriangleFanIB = NULL; - - ERR("Could not create a scratch index buffer for GL_TRIANGLE_FAN."); - return gl::error(GL_OUT_OF_MEMORY); - } - } - - // Checked by Renderer11::applyPrimitiveType - ASSERT(count >= 3); - - const unsigned int numTris = count - 2; - - if (numTris > (std::numeric_limits<unsigned int>::max() / (sizeof(unsigned int) * 3))) - { - ERR("Could not create a scratch index buffer for GL_TRIANGLE_FAN, too many indices required."); - return gl::error(GL_OUT_OF_MEMORY); - } - - const unsigned int spaceNeeded = (numTris * 3) * sizeof(unsigned int); - if (!mTriangleFanIB->reserveBufferSpace(spaceNeeded, indexType)) - { - ERR("Could not reserve enough space in scratch index buffer for GL_TRIANGLE_FAN."); - return gl::error(GL_OUT_OF_MEMORY); - } - - void* mappedMemory = NULL; - unsigned int offset; - if (!mTriangleFanIB->mapBuffer(spaceNeeded, &mappedMemory, &offset)) - { - ERR("Could not map scratch index buffer for GL_TRIANGLE_FAN."); - return gl::error(GL_OUT_OF_MEMORY); - } - - if (indexType == GL_UNSIGNED_SHORT) - fillTriangleFanIndices(type, numTris, indices, reinterpret_cast<unsigned short*>(mappedMemory)); - else - fillTriangleFanIndices(type, numTris, indices, reinterpret_cast<unsigned int*>(mappedMemory)); - unsigned int indexBufferOffset = offset; - - - if (!mTriangleFanIB->unmapBuffer()) - { - ERR("Could not unmap scratch index buffer for GL_TRIANGLE_FAN."); - return gl::error(GL_OUT_OF_MEMORY); - } - - if (mAppliedIBSerial != mTriangleFanIB->getSerial() || mAppliedIBOffset != indexBufferOffset) - { - IndexBuffer11 *indexBuffer = IndexBuffer11::makeIndexBuffer11(mTriangleFanIB->getIndexBuffer()); - - mDeviceContext->IASetIndexBuffer(indexBuffer->getBuffer(), indexBuffer->getIndexFormat(), indexBufferOffset); - mAppliedIBSerial = mTriangleFanIB->getSerial(); - mAppliedStorageIBSerial = 0; - mAppliedIBOffset = indexBufferOffset; - } - - if (instances > 0) - { - mDeviceContext->DrawIndexedInstanced(numTris * 3, instances, 0, -minIndex, 0); - } - else - { - mDeviceContext->DrawIndexed(numTris * 3, 0, -minIndex); - } -} - -void Renderer11::applyShaders(gl::ProgramBinary *programBinary) -{ - unsigned int programBinarySerial = programBinary->getSerial(); - const bool updateProgramState = (programBinarySerial != mAppliedProgramBinarySerial); - - if (updateProgramState) - { - ShaderExecutable *vertexExe = programBinary->getVertexExecutable(); - ShaderExecutable *pixelExe = programBinary->getPixelExecutable(); - - ID3D11VertexShader *vertexShader = NULL; - if (vertexExe) vertexShader = ShaderExecutable11::makeShaderExecutable11(vertexExe)->getVertexShader(); - - ID3D11PixelShader *pixelShader = NULL; - if (pixelExe) pixelShader = ShaderExecutable11::makeShaderExecutable11(pixelExe)->getPixelShader(); - - mDeviceContext->PSSetShader(pixelShader, NULL, 0); - mDeviceContext->VSSetShader(vertexShader, NULL, 0); - - programBinary->dirtyAllUniforms(); - - mAppliedProgramBinarySerial = programBinarySerial; - } - - // Only use the geometry shader currently for point sprite drawing - const bool usesGeometryShader = (programBinary->usesGeometryShader() && mCurRasterState.pointDrawMode); - - if (updateProgramState || usesGeometryShader != mIsGeometryShaderActive) - { - if (usesGeometryShader) - { - ShaderExecutable *geometryExe = programBinary->getGeometryExecutable(); - ID3D11GeometryShader *geometryShader = ShaderExecutable11::makeShaderExecutable11(geometryExe)->getGeometryShader(); - mDeviceContext->GSSetShader(geometryShader, NULL, 0); - } - else - { - mDeviceContext->GSSetShader(NULL, NULL, 0); - } - - mIsGeometryShaderActive = usesGeometryShader; - } -} - -void Renderer11::applyUniforms(gl::ProgramBinary *programBinary, gl::UniformArray *uniformArray) -{ - ShaderExecutable11 *vertexExecutable = ShaderExecutable11::makeShaderExecutable11(programBinary->getVertexExecutable()); - ShaderExecutable11 *pixelExecutable = ShaderExecutable11::makeShaderExecutable11(programBinary->getPixelExecutable()); - - unsigned int totalRegisterCountVS = 0; - unsigned int totalRegisterCountPS = 0; - - bool vertexUniformsDirty = false; - bool pixelUniformsDirty = false; - - for (gl::UniformArray::const_iterator uniform_iterator = uniformArray->begin(); uniform_iterator != uniformArray->end(); uniform_iterator++) - { - const gl::Uniform *uniform = *uniform_iterator; - - if (uniform->vsRegisterIndex >= 0) - { - totalRegisterCountVS += uniform->registerCount; - vertexUniformsDirty = vertexUniformsDirty || uniform->dirty; - } - - if (uniform->psRegisterIndex >= 0) - { - totalRegisterCountPS += uniform->registerCount; - pixelUniformsDirty = pixelUniformsDirty || uniform->dirty; - } - } - - ID3D11Buffer *vertexConstantBuffer = vertexExecutable->getConstantBuffer(mDevice, totalRegisterCountVS); - ID3D11Buffer *pixelConstantBuffer = pixelExecutable->getConstantBuffer(mDevice, totalRegisterCountPS); - - float (*mapVS)[4] = NULL; - float (*mapPS)[4] = NULL; - - if (totalRegisterCountVS > 0 && vertexUniformsDirty) - { - D3D11_MAPPED_SUBRESOURCE map = {0}; - HRESULT result = mDeviceContext->Map(vertexConstantBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &map); - ASSERT(SUCCEEDED(result)); - mapVS = (float(*)[4])map.pData; - } - - if (totalRegisterCountPS > 0 && pixelUniformsDirty) - { - D3D11_MAPPED_SUBRESOURCE map = {0}; - HRESULT result = mDeviceContext->Map(pixelConstantBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &map); - ASSERT(SUCCEEDED(result)); - mapPS = (float(*)[4])map.pData; - } - - for (gl::UniformArray::iterator uniform_iterator = uniformArray->begin(); uniform_iterator != uniformArray->end(); uniform_iterator++) - { - gl::Uniform *uniform = *uniform_iterator; - - if (uniform->type != GL_SAMPLER_2D && uniform->type != GL_SAMPLER_CUBE) - { - if (uniform->vsRegisterIndex >= 0 && mapVS) - { - memcpy(mapVS + uniform->vsRegisterIndex, uniform->data, uniform->registerCount * sizeof(float[4])); - } - - if (uniform->psRegisterIndex >= 0 && mapPS) - { - memcpy(mapPS + uniform->psRegisterIndex, uniform->data, uniform->registerCount * sizeof(float[4])); - } - } - - uniform->dirty = false; - } - - if (mapVS) - { - mDeviceContext->Unmap(vertexConstantBuffer, 0); - } - - if (mapPS) - { - mDeviceContext->Unmap(pixelConstantBuffer, 0); - } - - if (mCurrentVertexConstantBuffer != vertexConstantBuffer) - { - mDeviceContext->VSSetConstantBuffers(0, 1, &vertexConstantBuffer); - mCurrentVertexConstantBuffer = vertexConstantBuffer; - } - - if (mCurrentPixelConstantBuffer != pixelConstantBuffer) - { - mDeviceContext->PSSetConstantBuffers(0, 1, &pixelConstantBuffer); - mCurrentPixelConstantBuffer = pixelConstantBuffer; - } - - // Driver uniforms - if (!mDriverConstantBufferVS) - { - D3D11_BUFFER_DESC constantBufferDescription = {0}; - constantBufferDescription.ByteWidth = sizeof(dx_VertexConstants); - constantBufferDescription.Usage = D3D11_USAGE_DEFAULT; - constantBufferDescription.BindFlags = D3D11_BIND_CONSTANT_BUFFER; - constantBufferDescription.CPUAccessFlags = 0; - constantBufferDescription.MiscFlags = 0; - constantBufferDescription.StructureByteStride = 0; - - HRESULT result = mDevice->CreateBuffer(&constantBufferDescription, NULL, &mDriverConstantBufferVS); - ASSERT(SUCCEEDED(result)); - - mDeviceContext->VSSetConstantBuffers(1, 1, &mDriverConstantBufferVS); - } - - if (!mDriverConstantBufferPS) - { - D3D11_BUFFER_DESC constantBufferDescription = {0}; - constantBufferDescription.ByteWidth = sizeof(dx_PixelConstants); - constantBufferDescription.Usage = D3D11_USAGE_DEFAULT; - constantBufferDescription.BindFlags = D3D11_BIND_CONSTANT_BUFFER; - constantBufferDescription.CPUAccessFlags = 0; - constantBufferDescription.MiscFlags = 0; - constantBufferDescription.StructureByteStride = 0; - - HRESULT result = mDevice->CreateBuffer(&constantBufferDescription, NULL, &mDriverConstantBufferPS); - ASSERT(SUCCEEDED(result)); - - mDeviceContext->PSSetConstantBuffers(1, 1, &mDriverConstantBufferPS); - } - - if (memcmp(&mVertexConstants, &mAppliedVertexConstants, sizeof(dx_VertexConstants)) != 0) - { - mDeviceContext->UpdateSubresource(mDriverConstantBufferVS, 0, NULL, &mVertexConstants, 16, 0); - memcpy(&mAppliedVertexConstants, &mVertexConstants, sizeof(dx_VertexConstants)); - } - - if (memcmp(&mPixelConstants, &mAppliedPixelConstants, sizeof(dx_PixelConstants)) != 0) - { - mDeviceContext->UpdateSubresource(mDriverConstantBufferPS, 0, NULL, &mPixelConstants, 16, 0); - memcpy(&mAppliedPixelConstants, &mPixelConstants, sizeof(dx_PixelConstants)); - } - - // needed for the point sprite geometry shader - if (mFeatureLevel >= D3D_FEATURE_LEVEL_10_0 && mCurrentGeometryConstantBuffer != mDriverConstantBufferPS) - { - mDeviceContext->GSSetConstantBuffers(0, 1, &mDriverConstantBufferPS); - mCurrentGeometryConstantBuffer = mDriverConstantBufferPS; - } -} - -void Renderer11::clear(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer) -{ - gl::Renderbuffer *firstRenderbuffer = frameBuffer->getFirstColorbuffer(); - GLenum internalFormat = firstRenderbuffer ? firstRenderbuffer->getInternalFormat() : GL_NONE; - - bool needMaskedColorClear = (clearParams.mask & GL_COLOR_BUFFER_BIT) && - ((!clearParams.colorMaskRed && gl::GetRedSize(internalFormat) > 0) || - (!clearParams.colorMaskGreen && gl::GetGreenSize(internalFormat) > 0) || - (!clearParams.colorMaskBlue && gl::GetBlueSize(internalFormat) > 0) || - (!clearParams.colorMaskAlpha && gl::GetAlphaSize(internalFormat) > 0)); - - unsigned int stencilUnmasked = 0x0; - if (frameBuffer->hasStencil()) - { - unsigned int stencilSize = gl::GetStencilSize(frameBuffer->getStencilbuffer()->getActualFormat()); - stencilUnmasked = (0x1 << stencilSize) - 1; - } - bool needMaskedStencilClear = (clearParams.mask & GL_STENCIL_BUFFER_BIT) && - (clearParams.stencilWriteMask & stencilUnmasked) != stencilUnmasked; - - bool needScissoredClear = mScissorEnabled && (mCurScissor.x > 0 || mCurScissor.y > 0 || - mCurScissor.x + mCurScissor.width < mRenderTargetDesc.width || - mCurScissor.y + mCurScissor.height < mRenderTargetDesc.height); - - if (needMaskedColorClear || needMaskedStencilClear || needScissoredClear) - { - maskedClear(clearParams, frameBuffer); - } - else - { - if (clearParams.mask & GL_COLOR_BUFFER_BIT) - { - for (unsigned int colorAttachment = 0; colorAttachment < gl::IMPLEMENTATION_MAX_DRAW_BUFFERS; colorAttachment++) - { - if (frameBuffer->isEnabledColorAttachment(colorAttachment)) - { - gl::Renderbuffer *renderbufferObject = frameBuffer->getColorbuffer(colorAttachment); - if (renderbufferObject) - { - RenderTarget11 *renderTarget = RenderTarget11::makeRenderTarget11(renderbufferObject->getRenderTarget()); - if (!renderTarget) - { - ERR("render target pointer unexpectedly null."); - return; - } - - ID3D11RenderTargetView *framebufferRTV = renderTarget->getRenderTargetView(); - if (!framebufferRTV) - { - ERR("render target view pointer unexpectedly null."); - return; - } - - GLenum format = renderbufferObject->getInternalFormat(); - - const float clearValues[4] = { (gl::GetRedSize(format) > 0) ? clearParams.colorClearValue.red : 0.0f, - (gl::GetGreenSize(format) > 0) ? clearParams.colorClearValue.green : 0.0f, - (gl::GetBlueSize(format) > 0) ? clearParams.colorClearValue.blue : 0.0f, - (gl::GetAlphaSize(format) > 0) ? clearParams.colorClearValue.alpha : 1.0f }; - mDeviceContext->ClearRenderTargetView(framebufferRTV, clearValues); - } - } - } - } - if (clearParams.mask & GL_DEPTH_BUFFER_BIT || clearParams.mask & GL_STENCIL_BUFFER_BIT) - { - gl::Renderbuffer *renderbufferObject = frameBuffer->getDepthOrStencilbuffer(); - if (renderbufferObject) - { - RenderTarget11 *renderTarget = RenderTarget11::makeRenderTarget11(renderbufferObject->getDepthStencil()); - if (!renderTarget) - { - ERR("render target pointer unexpectedly null."); - return; - } - - ID3D11DepthStencilView *framebufferDSV = renderTarget->getDepthStencilView(); - if (!framebufferDSV) - { - ERR("depth stencil view pointer unexpectedly null."); - return; - } - - UINT clearFlags = 0; - if (clearParams.mask & GL_DEPTH_BUFFER_BIT) - { - clearFlags |= D3D11_CLEAR_DEPTH; - } - if (clearParams.mask & GL_STENCIL_BUFFER_BIT) - { - clearFlags |= D3D11_CLEAR_STENCIL; - } - - float depthClear = gl::clamp01(clearParams.depthClearValue); - UINT8 stencilClear = clearParams.stencilClearValue & 0x000000FF; - - mDeviceContext->ClearDepthStencilView(framebufferDSV, clearFlags, depthClear, stencilClear); - } - } - } -} - -void Renderer11::maskedClear(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer) -{ - HRESULT result; - - if (!mClearResourcesInitialized) - { - ASSERT(!mClearVB && !mClearVS && !mClearSinglePS && !mClearMultiplePS && !mClearScissorRS && !mClearNoScissorRS); - - D3D11_BUFFER_DESC vbDesc; - vbDesc.ByteWidth = sizeof(d3d11::PositionDepthColorVertex) * 4; - vbDesc.Usage = D3D11_USAGE_DYNAMIC; - vbDesc.BindFlags = D3D11_BIND_VERTEX_BUFFER; - vbDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; - vbDesc.MiscFlags = 0; - vbDesc.StructureByteStride = 0; - - result = mDevice->CreateBuffer(&vbDesc, NULL, &mClearVB); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mClearVB, "Renderer11 masked clear vertex buffer"); - - D3D11_INPUT_ELEMENT_DESC quadLayout[] = - { - { "POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0 }, - { "COLOR", 0, DXGI_FORMAT_R32G32B32A32_FLOAT, 0, 12, D3D11_INPUT_PER_VERTEX_DATA, 0 }, - }; - - result = mDevice->CreateInputLayout(quadLayout, 2, g_VS_Clear, sizeof(g_VS_Clear), &mClearIL); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mClearIL, "Renderer11 masked clear input layout"); - - result = mDevice->CreateVertexShader(g_VS_Clear, sizeof(g_VS_Clear), NULL, &mClearVS); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mClearVS, "Renderer11 masked clear vertex shader"); - - result = mDevice->CreatePixelShader(g_PS_ClearSingle, sizeof(g_PS_ClearSingle), NULL, &mClearSinglePS); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mClearSinglePS, "Renderer11 masked clear pixel shader (1 RT)"); - - result = mDevice->CreatePixelShader(g_PS_ClearMultiple, sizeof(g_PS_ClearMultiple), NULL, &mClearMultiplePS); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mClearMultiplePS, "Renderer11 masked clear pixel shader (MRT)"); - - D3D11_RASTERIZER_DESC rsScissorDesc; - rsScissorDesc.FillMode = D3D11_FILL_SOLID; - rsScissorDesc.CullMode = D3D11_CULL_NONE; - rsScissorDesc.FrontCounterClockwise = FALSE; - rsScissorDesc.DepthBias = 0; - rsScissorDesc.DepthBiasClamp = 0.0f; - rsScissorDesc.SlopeScaledDepthBias = 0.0f; - rsScissorDesc.DepthClipEnable = FALSE; - rsScissorDesc.ScissorEnable = TRUE; - rsScissorDesc.MultisampleEnable = FALSE; - rsScissorDesc.AntialiasedLineEnable = FALSE; - - result = mDevice->CreateRasterizerState(&rsScissorDesc, &mClearScissorRS); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mClearScissorRS, "Renderer11 masked clear scissor rasterizer state"); - - D3D11_RASTERIZER_DESC rsNoScissorDesc; - rsNoScissorDesc.FillMode = D3D11_FILL_SOLID; - rsNoScissorDesc.CullMode = D3D11_CULL_NONE; - rsNoScissorDesc.FrontCounterClockwise = FALSE; - rsNoScissorDesc.DepthBias = 0; - rsNoScissorDesc.DepthBiasClamp = 0.0f; - rsNoScissorDesc.SlopeScaledDepthBias = 0.0f; - rsNoScissorDesc.DepthClipEnable = FALSE; - rsNoScissorDesc.ScissorEnable = FALSE; - rsNoScissorDesc.MultisampleEnable = FALSE; - rsNoScissorDesc.AntialiasedLineEnable = FALSE; - - result = mDevice->CreateRasterizerState(&rsNoScissorDesc, &mClearNoScissorRS); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mClearNoScissorRS, "Renderer11 masked clear no scissor rasterizer state"); - - mClearResourcesInitialized = true; - } - - // Prepare the depth stencil state to write depth values if the depth should be cleared - // and stencil values if the stencil should be cleared - gl::DepthStencilState glDSState; - glDSState.depthTest = (clearParams.mask & GL_DEPTH_BUFFER_BIT) != 0; - glDSState.depthFunc = GL_ALWAYS; - glDSState.depthMask = (clearParams.mask & GL_DEPTH_BUFFER_BIT) != 0; - glDSState.stencilTest = (clearParams.mask & GL_STENCIL_BUFFER_BIT) != 0; - glDSState.stencilFunc = GL_ALWAYS; - glDSState.stencilMask = 0; - glDSState.stencilFail = GL_REPLACE; - glDSState.stencilPassDepthFail = GL_REPLACE; - glDSState.stencilPassDepthPass = GL_REPLACE; - glDSState.stencilWritemask = clearParams.stencilWriteMask; - glDSState.stencilBackFunc = GL_ALWAYS; - glDSState.stencilBackMask = 0; - glDSState.stencilBackFail = GL_REPLACE; - glDSState.stencilBackPassDepthFail = GL_REPLACE; - glDSState.stencilBackPassDepthPass = GL_REPLACE; - glDSState.stencilBackWritemask = clearParams.stencilWriteMask; - - int stencilClear = clearParams.stencilClearValue & 0x000000FF; - - ID3D11DepthStencilState *dsState = mStateCache.getDepthStencilState(glDSState); - - // Prepare the blend state to use a write mask if the color buffer should be cleared - gl::BlendState glBlendState; - glBlendState.blend = false; - glBlendState.sourceBlendRGB = GL_ONE; - glBlendState.destBlendRGB = GL_ZERO; - glBlendState.sourceBlendAlpha = GL_ONE; - glBlendState.destBlendAlpha = GL_ZERO; - glBlendState.blendEquationRGB = GL_FUNC_ADD; - glBlendState.blendEquationAlpha = GL_FUNC_ADD; - glBlendState.colorMaskRed = (clearParams.mask & GL_COLOR_BUFFER_BIT) ? clearParams.colorMaskRed : false; - glBlendState.colorMaskGreen = (clearParams.mask & GL_COLOR_BUFFER_BIT) ? clearParams.colorMaskGreen : false; - glBlendState.colorMaskBlue = (clearParams.mask & GL_COLOR_BUFFER_BIT) ? clearParams.colorMaskBlue : false; - glBlendState.colorMaskAlpha = (clearParams.mask & GL_COLOR_BUFFER_BIT) ? clearParams.colorMaskAlpha : false; - glBlendState.sampleAlphaToCoverage = false; - glBlendState.dither = false; - - static const float blendFactors[4] = { 1.0f, 1.0f, 1.0f, 1.0f }; - static const UINT sampleMask = 0xFFFFFFFF; - - ID3D11BlendState *blendState = mStateCache.getBlendState(frameBuffer, glBlendState); - - // Set the vertices - D3D11_MAPPED_SUBRESOURCE mappedResource; - result = mDeviceContext->Map(mClearVB, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); - if (FAILED(result)) - { - ERR("Failed to map masked clear vertex buffer, HRESULT: 0x%X.", result); - return; - } - - d3d11::PositionDepthColorVertex *vertices = reinterpret_cast<d3d11::PositionDepthColorVertex*>(mappedResource.pData); - - float depthClear = gl::clamp01(clearParams.depthClearValue); - d3d11::SetPositionDepthColorVertex(&vertices[0], -1.0f, 1.0f, depthClear, clearParams.colorClearValue); - d3d11::SetPositionDepthColorVertex(&vertices[1], -1.0f, -1.0f, depthClear, clearParams.colorClearValue); - d3d11::SetPositionDepthColorVertex(&vertices[2], 1.0f, 1.0f, depthClear, clearParams.colorClearValue); - d3d11::SetPositionDepthColorVertex(&vertices[3], 1.0f, -1.0f, depthClear, clearParams.colorClearValue); - - mDeviceContext->Unmap(mClearVB, 0); - - // Apply state - mDeviceContext->OMSetBlendState(blendState, blendFactors, sampleMask); - mDeviceContext->OMSetDepthStencilState(dsState, stencilClear); - mDeviceContext->RSSetState(mScissorEnabled ? mClearScissorRS : mClearNoScissorRS); - - // Apply shaders - ID3D11PixelShader *pixelShader = frameBuffer->usingExtendedDrawBuffers() ? mClearMultiplePS : mClearSinglePS; - - mDeviceContext->IASetInputLayout(mClearIL); - mDeviceContext->VSSetShader(mClearVS, NULL, 0); - mDeviceContext->PSSetShader(pixelShader, NULL, 0); - mDeviceContext->GSSetShader(NULL, NULL, 0); - - // Apply vertex buffer - static UINT stride = sizeof(d3d11::PositionDepthColorVertex); - static UINT startIdx = 0; - mDeviceContext->IASetVertexBuffers(0, 1, &mClearVB, &stride, &startIdx); - mDeviceContext->IASetPrimitiveTopology(D3D_PRIMITIVE_TOPOLOGY_TRIANGLESTRIP); - - // Draw the clear quad - mDeviceContext->Draw(4, 0); - - // Clean up - markAllStateDirty(); -} - -void Renderer11::markAllStateDirty() -{ - for (unsigned int rtIndex = 0; rtIndex < gl::IMPLEMENTATION_MAX_DRAW_BUFFERS; rtIndex++) - { - mAppliedRenderTargetSerials[rtIndex] = 0; - } - mAppliedDepthbufferSerial = 0; - mAppliedStencilbufferSerial = 0; - mDepthStencilInitialized = false; - mRenderTargetDescInitialized = false; - - for (int i = 0; i < gl::IMPLEMENTATION_MAX_VERTEX_TEXTURE_IMAGE_UNITS; i++) - { - mForceSetVertexSamplerStates[i] = true; - mCurVertexTextureSerials[i] = 0; - } - for (int i = 0; i < gl::MAX_TEXTURE_IMAGE_UNITS; i++) - { - mForceSetPixelSamplerStates[i] = true; - mCurPixelTextureSerials[i] = 0; - } - - mForceSetBlendState = true; - mForceSetRasterState = true; - mForceSetDepthStencilState = true; - mForceSetScissor = true; - mForceSetViewport = true; - - mAppliedIBSerial = 0; - mAppliedStorageIBSerial = 0; - mAppliedIBOffset = 0; - - mAppliedProgramBinarySerial = 0; - memset(&mAppliedVertexConstants, 0, sizeof(dx_VertexConstants)); - memset(&mAppliedPixelConstants, 0, sizeof(dx_PixelConstants)); - - mInputLayoutCache.markDirty(); - - mCurrentVertexConstantBuffer = NULL; - mCurrentPixelConstantBuffer = NULL; - mCurrentGeometryConstantBuffer = NULL; - - mCurrentPrimitiveTopology = D3D_PRIMITIVE_TOPOLOGY_UNDEFINED; -} - -void Renderer11::releaseDeviceResources() -{ - mStateCache.clear(); - mInputLayoutCache.clear(); - - delete mVertexDataManager; - mVertexDataManager = NULL; - - delete mIndexDataManager; - mIndexDataManager = NULL; - - delete mLineLoopIB; - mLineLoopIB = NULL; - - delete mTriangleFanIB; - mTriangleFanIB = NULL; - - SafeRelease(mCopyVB); - SafeRelease(mCopySampler); - SafeRelease(mCopyIL); - SafeRelease(mCopyIL); - SafeRelease(mCopyVS); - SafeRelease(mCopyRGBAPS); - SafeRelease(mCopyRGBPS); - SafeRelease(mCopyLumPS); - SafeRelease(mCopyLumAlphaPS); - - mCopyResourcesInitialized = false; - - SafeRelease(mClearVB); - SafeRelease(mClearIL); - SafeRelease(mClearVS); - SafeRelease(mClearSinglePS); - SafeRelease(mClearMultiplePS); - SafeRelease(mClearScissorRS); - SafeRelease(mClearNoScissorRS); - - mClearResourcesInitialized = false; - - SafeRelease(mDriverConstantBufferVS); - SafeRelease(mDriverConstantBufferPS); - SafeRelease(mSyncQuery); -} - -void Renderer11::notifyDeviceLost() -{ - mDeviceLost = true; - mDisplay->notifyDeviceLost(); -} - -bool Renderer11::isDeviceLost() -{ - return mDeviceLost; -} - -// set notify to true to broadcast a message to all contexts of the device loss -bool Renderer11::testDeviceLost(bool notify) -{ - bool isLost = false; - - // GetRemovedReason is used to test if the device is removed - HRESULT result = mDevice->GetDeviceRemovedReason(); - isLost = d3d11::isDeviceLostError(result); - - if (isLost) - { - // Log error if this is a new device lost event - if (mDeviceLost == false) - { - ERR("The D3D11 device was removed: 0x%08X", result); - } - - // ensure we note the device loss -- - // we'll probably get this done again by notifyDeviceLost - // but best to remember it! - // Note that we don't want to clear the device loss status here - // -- this needs to be done by resetDevice - mDeviceLost = true; - if (notify) - { - notifyDeviceLost(); - } - } - - return isLost; -} - -bool Renderer11::testDeviceResettable() -{ - // determine if the device is resettable by creating a dummy device - PFN_D3D11_CREATE_DEVICE D3D11CreateDevice = (PFN_D3D11_CREATE_DEVICE)GetProcAddress(mD3d11Module, "D3D11CreateDevice"); - - if (D3D11CreateDevice == NULL) - { - return false; - } - - D3D_FEATURE_LEVEL featureLevels[] = - { - D3D_FEATURE_LEVEL_11_0, - D3D_FEATURE_LEVEL_10_1, - D3D_FEATURE_LEVEL_10_0, -#if !defined(ANGLE_ENABLE_D3D9) - D3D_FEATURE_LEVEL_9_3, - D3D_FEATURE_LEVEL_9_2, - D3D_FEATURE_LEVEL_9_1, -#endif - }; - - ID3D11Device* dummyDevice; - D3D_FEATURE_LEVEL dummyFeatureLevel; - ID3D11DeviceContext* dummyContext; - - HRESULT result = D3D11CreateDevice(NULL, - D3D_DRIVER_TYPE_HARDWARE, - NULL, - #if defined(_DEBUG) - D3D11_CREATE_DEVICE_DEBUG, - #else - 0, - #endif - featureLevels, - ArraySize(featureLevels), - D3D11_SDK_VERSION, - &dummyDevice, - &dummyFeatureLevel, - &dummyContext); - - if (!mDevice || FAILED(result)) - { - return false; - } - - dummyContext->Release(); - dummyDevice->Release(); - - return true; -} - -void Renderer11::release() -{ - releaseDeviceResources(); - - if (mDxgiFactory) - { - mDxgiFactory->Release(); - mDxgiFactory = NULL; - } - - if (mDxgiAdapter) - { - mDxgiAdapter->Release(); - mDxgiAdapter = NULL; - } - - if (mDeviceContext) - { - mDeviceContext->ClearState(); - mDeviceContext->Flush(); - mDeviceContext->Release(); - mDeviceContext = NULL; - } - - if (mDevice) - { - mDevice->Release(); - mDevice = NULL; - } - - if (mD3d11Module) - { - FreeLibrary(mD3d11Module); - mD3d11Module = NULL; - } - - if (mDxgiModule) - { - FreeLibrary(mDxgiModule); - mDxgiModule = NULL; - } -} - -bool Renderer11::resetDevice() -{ - // recreate everything - release(); - EGLint result = initialize(); - - if (result != EGL_SUCCESS) - { - ERR("Could not reinitialize D3D11 device: %08X", result); - return false; - } - - mDeviceLost = false; - - return true; -} - -DWORD Renderer11::getAdapterVendor() const -{ - return mAdapterDescription.VendorId; -} - -std::string Renderer11::getRendererDescription() const -{ - std::ostringstream rendererString; - - rendererString << mDescription; - rendererString << " Direct3D11"; - - rendererString << " vs_" << getMajorShaderModel() << "_" << getMinorShaderModel(); - rendererString << " ps_" << getMajorShaderModel() << "_" << getMinorShaderModel(); - - return rendererString.str(); -} - -GUID Renderer11::getAdapterIdentifier() const -{ - // Use the adapter LUID as our adapter ID - // This number is local to a machine is only guaranteed to be unique between restarts - META_ASSERT(sizeof(LUID) <= sizeof(GUID)); - GUID adapterId = {0}; - memcpy(&adapterId, &mAdapterDescription.AdapterLuid, sizeof(LUID)); - return adapterId; -} - -bool Renderer11::getBGRATextureSupport() const -{ - return mBGRATextureSupport; -} - -bool Renderer11::getDXT1TextureSupport() -{ - return mDXT1TextureSupport; -} - -bool Renderer11::getDXT3TextureSupport() -{ - return mDXT3TextureSupport; -} - -bool Renderer11::getDXT5TextureSupport() -{ - return mDXT5TextureSupport; -} - -bool Renderer11::getDepthTextureSupport() const -{ - return mDepthTextureSupport; -} - -bool Renderer11::getFloat32TextureSupport(bool *filtering, bool *renderable) -{ - *renderable = mFloat32RenderSupport; - *filtering = mFloat32FilterSupport; - return mFloat32TextureSupport; -} - -bool Renderer11::getFloat16TextureSupport(bool *filtering, bool *renderable) -{ - *renderable = mFloat16RenderSupport; - *filtering = mFloat16FilterSupport; - return mFloat16TextureSupport; -} - -bool Renderer11::getLuminanceTextureSupport() -{ - return false; -} - -bool Renderer11::getLuminanceAlphaTextureSupport() -{ - return false; -} - -bool Renderer11::getTextureFilterAnisotropySupport() const -{ - return true; -} - -float Renderer11::getTextureMaxAnisotropy() const -{ - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: - return D3D11_MAX_MAXANISOTROPY; - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: - return D3D10_MAX_MAXANISOTROPY; - case D3D_FEATURE_LEVEL_9_3: - case D3D_FEATURE_LEVEL_9_2: - return 16; - case D3D_FEATURE_LEVEL_9_1: - return D3D_FL9_1_DEFAULT_MAX_ANISOTROPY; - default: UNREACHABLE(); - return 0; - } -} - -bool Renderer11::getEventQuerySupport() -{ - return true; -} - -Range Renderer11::getViewportBounds() const -{ - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: - return Range(D3D11_VIEWPORT_BOUNDS_MIN, D3D11_VIEWPORT_BOUNDS_MAX); - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: - return Range(D3D10_VIEWPORT_BOUNDS_MIN, D3D10_VIEWPORT_BOUNDS_MAX); - case D3D_FEATURE_LEVEL_9_3: - return Range(D3D_FL9_3_REQ_TEXTURE2D_U_OR_V_DIMENSION * -2, D3D_FL9_3_REQ_TEXTURE2D_U_OR_V_DIMENSION * 2); - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: - return Range(D3D_FL9_1_REQ_TEXTURE2D_U_OR_V_DIMENSION * -2, D3D_FL9_1_REQ_TEXTURE2D_U_OR_V_DIMENSION * 2); - default: UNREACHABLE(); - return Range(0, 0); - } -} - -unsigned int Renderer11::getMaxVertexTextureImageUnits() const -{ - META_ASSERT(MAX_TEXTURE_IMAGE_UNITS_VTF_SM4 <= gl::IMPLEMENTATION_MAX_VERTEX_TEXTURE_IMAGE_UNITS); - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: - return MAX_TEXTURE_IMAGE_UNITS_VTF_SM4; - case D3D_FEATURE_LEVEL_9_3: - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: - return 0; - default: UNREACHABLE(); - return 0; - } -} - -unsigned int Renderer11::getMaxCombinedTextureImageUnits() const -{ - return gl::MAX_TEXTURE_IMAGE_UNITS + getMaxVertexTextureImageUnits(); -} - -unsigned int Renderer11::getReservedVertexUniformVectors() const -{ - return 0; // Driver uniforms are stored in a separate constant buffer -} - -unsigned int Renderer11::getReservedFragmentUniformVectors() const -{ - return 0; // Driver uniforms are stored in a separate constant buffer -} - -unsigned int Renderer11::getMaxVertexUniformVectors() const -{ - META_ASSERT(MAX_VERTEX_UNIFORM_VECTORS_D3D11 <= D3D10_REQ_CONSTANT_BUFFER_ELEMENT_COUNT); - ASSERT(mFeatureLevel >= D3D_FEATURE_LEVEL_9_1); - return MAX_VERTEX_UNIFORM_VECTORS_D3D11; -} - -unsigned int Renderer11::getMaxFragmentUniformVectors() const -{ - META_ASSERT(MAX_FRAGMENT_UNIFORM_VECTORS_D3D11 <= D3D10_REQ_CONSTANT_BUFFER_ELEMENT_COUNT); - ASSERT(mFeatureLevel >= D3D_FEATURE_LEVEL_9_1); - return MAX_FRAGMENT_UNIFORM_VECTORS_D3D11; -} - -unsigned int Renderer11::getMaxVaryingVectors() const -{ - META_ASSERT(gl::IMPLEMENTATION_MAX_VARYING_VECTORS == D3D11_VS_OUTPUT_REGISTER_COUNT); - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: - return D3D11_VS_OUTPUT_REGISTER_COUNT; - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: - return D3D10_VS_OUTPUT_REGISTER_COUNT; - case D3D_FEATURE_LEVEL_9_3: - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: - return 8; - default: UNREACHABLE(); - return 0; - } -} - -bool Renderer11::getNonPower2TextureSupport() const -{ - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: - return true; - case D3D_FEATURE_LEVEL_9_3: - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: - return false; - default: UNREACHABLE(); - return false; - } -} - -bool Renderer11::getOcclusionQuerySupport() const -{ - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: - return true; - case D3D_FEATURE_LEVEL_9_3: - case D3D_FEATURE_LEVEL_9_2: - return true; - case D3D_FEATURE_LEVEL_9_1: - return false; - default: UNREACHABLE(); - return false; - } -} - -bool Renderer11::getInstancingSupport() const -{ - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: - case D3D_FEATURE_LEVEL_9_3: - return true; - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: - return false; - default: UNREACHABLE(); - return false; - } -} - -bool Renderer11::getShareHandleSupport() const -{ - // We only currently support share handles with BGRA surfaces, because - // chrome needs BGRA. Once chrome fixes this, we should always support them. - // PIX doesn't seem to support using share handles, so disable them. - return getBGRATextureSupport() && !gl::perfActive(); -} - -bool Renderer11::getDerivativeInstructionSupport() const -{ - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: - return true; - case D3D_FEATURE_LEVEL_9_3: - return true; - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: - return false; - default: UNREACHABLE(); - return false; - } -} - -bool Renderer11::getPostSubBufferSupport() const -{ - // D3D11 does not support present with dirty rectangles until D3D11.1 and DXGI 1.2. - return false; -} - -int Renderer11::getMajorShaderModel() const -{ - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: return D3D11_SHADER_MAJOR_VERSION; // 5 - case D3D_FEATURE_LEVEL_10_1: return D3D10_1_SHADER_MAJOR_VERSION; // 4 - case D3D_FEATURE_LEVEL_10_0: return D3D10_SHADER_MAJOR_VERSION; // 4 - case D3D_FEATURE_LEVEL_9_3: - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: return D3D10_SHADER_MAJOR_VERSION; // 4 (level 9) - default: UNREACHABLE(); return 0; - } -} - -int Renderer11::getMinorShaderModel() const -{ - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: return D3D11_SHADER_MINOR_VERSION; // 0 - case D3D_FEATURE_LEVEL_10_1: return D3D10_1_SHADER_MINOR_VERSION; // 1 - case D3D_FEATURE_LEVEL_10_0: return D3D10_SHADER_MINOR_VERSION; // 0 - case D3D_FEATURE_LEVEL_9_3: - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: return D3D10_SHADER_MINOR_VERSION; // 0 (level 9) - default: UNREACHABLE(); return 0; - } -} - -float Renderer11::getMaxPointSize() const -{ - // choose a reasonable maximum. we enforce this in the shader. - // (nb: on a Radeon 2600xt, DX9 reports a 256 max point size) - return 1024.0f; -} - -int Renderer11::getMaxViewportDimension() const -{ - // Maximum viewport size must be at least as large as the largest render buffer (or larger). - // In our case return the maximum texture size, which is the maximum render buffer size. - META_ASSERT(D3D11_REQ_TEXTURE2D_U_OR_V_DIMENSION * 2 - 1 <= D3D11_VIEWPORT_BOUNDS_MAX); - META_ASSERT(D3D10_REQ_TEXTURE2D_U_OR_V_DIMENSION * 2 - 1 <= D3D10_VIEWPORT_BOUNDS_MAX); - - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: - return D3D11_REQ_TEXTURE2D_U_OR_V_DIMENSION; // 16384 - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: - return D3D10_REQ_TEXTURE2D_U_OR_V_DIMENSION; // 8192 - case D3D_FEATURE_LEVEL_9_3: - return D3D_FL9_3_REQ_TEXTURE2D_U_OR_V_DIMENSION; // 4096 - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: - return D3D_FL9_1_REQ_TEXTURE2D_U_OR_V_DIMENSION; // 2048 - default: UNREACHABLE(); - return 0; - } -} - -int Renderer11::getMaxTextureWidth() const -{ - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: return D3D11_REQ_TEXTURE2D_U_OR_V_DIMENSION; // 16384 - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: return D3D10_REQ_TEXTURE2D_U_OR_V_DIMENSION; // 8192 - case D3D_FEATURE_LEVEL_9_3: return D3D_FL9_3_REQ_TEXTURE2D_U_OR_V_DIMENSION; // 4096 - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: return D3D_FL9_1_REQ_TEXTURE2D_U_OR_V_DIMENSION; // 2048 - default: UNREACHABLE(); return 0; - } -} - -int Renderer11::getMaxTextureHeight() const -{ - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: return D3D11_REQ_TEXTURE2D_U_OR_V_DIMENSION; // 16384 - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: return D3D10_REQ_TEXTURE2D_U_OR_V_DIMENSION; // 8192 - case D3D_FEATURE_LEVEL_9_3: return D3D_FL9_3_REQ_TEXTURE2D_U_OR_V_DIMENSION; // 4096 - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: return D3D_FL9_1_REQ_TEXTURE2D_U_OR_V_DIMENSION; // 2048 - default: UNREACHABLE(); return 0; - } -} - -bool Renderer11::get32BitIndexSupport() const -{ - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: return D3D10_REQ_DRAWINDEXED_INDEX_COUNT_2_TO_EXP >= 32; // true - case D3D_FEATURE_LEVEL_9_3: - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: return false; - default: UNREACHABLE(); return false; - } -} - -int Renderer11::getMinSwapInterval() const -{ - return 0; -} - -int Renderer11::getMaxSwapInterval() const -{ - return 4; -} - -int Renderer11::getMaxSupportedSamples() const -{ - return mMaxSupportedSamples; -} - -int Renderer11::getNearestSupportedSamples(DXGI_FORMAT format, unsigned int requested) const -{ - if (requested == 0) - { - return 0; - } - - MultisampleSupportMap::const_iterator iter = mMultisampleSupportMap.find(format); - if (iter != mMultisampleSupportMap.end()) - { - const MultisampleSupportInfo& info = iter->second; - for (unsigned int i = requested - 1; i < D3D11_MAX_MULTISAMPLE_SAMPLE_COUNT; i++) - { - if (info.qualityLevels[i] > 0) - { - return i + 1; - } - } - } - - return -1; -} - -unsigned int Renderer11::getMaxRenderTargets() const -{ - META_ASSERT(D3D11_SIMULTANEOUS_RENDER_TARGET_COUNT <= gl::IMPLEMENTATION_MAX_DRAW_BUFFERS); - META_ASSERT(D3D10_SIMULTANEOUS_RENDER_TARGET_COUNT <= gl::IMPLEMENTATION_MAX_DRAW_BUFFERS); - META_ASSERT(D3D_FL9_3_SIMULTANEOUS_RENDER_TARGET_COUNT <= gl::IMPLEMENTATION_MAX_DRAW_BUFFERS); - META_ASSERT(D3D_FL9_1_SIMULTANEOUS_RENDER_TARGET_COUNT <= gl::IMPLEMENTATION_MAX_DRAW_BUFFERS); - - switch (mFeatureLevel) - { - case D3D_FEATURE_LEVEL_11_0: - return D3D11_SIMULTANEOUS_RENDER_TARGET_COUNT; // 8 - case D3D_FEATURE_LEVEL_10_1: - case D3D_FEATURE_LEVEL_10_0: - case D3D_FEATURE_LEVEL_9_3: // return D3D_FL9_3_SIMULTANEOUS_RENDER_TARGET_COUNT; // 4 - case D3D_FEATURE_LEVEL_9_2: - case D3D_FEATURE_LEVEL_9_1: // return D3D_FL9_1_SIMULTANEOUS_RENDER_TARGET_COUNT; // 1 - // Feature level 10.0 and 10.1 cards perform very poorly when the pixel shader - // outputs to multiple RTs that are not bound. - // TODO: Remove pixel shader outputs for render targets that are not bound. - return 1; - default: - UNREACHABLE(); - return 1; - } -} - -bool Renderer11::copyToRenderTarget(TextureStorageInterface2D *dest, TextureStorageInterface2D *source) -{ - if (source && dest) - { - TextureStorage11_2D *source11 = TextureStorage11_2D::makeTextureStorage11_2D(source->getStorageInstance()); - TextureStorage11_2D *dest11 = TextureStorage11_2D::makeTextureStorage11_2D(dest->getStorageInstance()); - - mDeviceContext->CopyResource(dest11->getBaseTexture(), source11->getBaseTexture()); - return true; - } - - return false; -} - -bool Renderer11::copyToRenderTarget(TextureStorageInterfaceCube *dest, TextureStorageInterfaceCube *source) -{ - if (source && dest) - { - TextureStorage11_Cube *source11 = TextureStorage11_Cube::makeTextureStorage11_Cube(source->getStorageInstance()); - TextureStorage11_Cube *dest11 = TextureStorage11_Cube::makeTextureStorage11_Cube(dest->getStorageInstance()); - - mDeviceContext->CopyResource(dest11->getBaseTexture(), source11->getBaseTexture()); - return true; - } - - return false; -} - -bool Renderer11::copyImage(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, - GLint xoffset, GLint yoffset, TextureStorageInterface2D *storage, GLint level) -{ - gl::Renderbuffer *colorbuffer = framebuffer->getReadColorbuffer(); - if (!colorbuffer) - { - ERR("Failed to retrieve the color buffer from the frame buffer."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - RenderTarget11 *sourceRenderTarget = RenderTarget11::makeRenderTarget11(colorbuffer->getRenderTarget()); - if (!sourceRenderTarget) - { - ERR("Failed to retrieve the render target from the frame buffer."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - ID3D11ShaderResourceView *source = sourceRenderTarget->getShaderResourceView(); - if (!source) - { - ERR("Failed to retrieve the render target view from the render target."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - TextureStorage11_2D *storage11 = TextureStorage11_2D::makeTextureStorage11_2D(storage->getStorageInstance()); - if (!storage11) - { - ERR("Failed to retrieve the texture storage from the destination."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - RenderTarget11 *destRenderTarget = RenderTarget11::makeRenderTarget11(storage11->getRenderTarget(level)); - if (!destRenderTarget) - { - ERR("Failed to retrieve the render target from the destination storage."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - ID3D11RenderTargetView *dest = destRenderTarget->getRenderTargetView(); - if (!dest) - { - ERR("Failed to retrieve the render target view from the destination render target."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - gl::Rectangle destRect; - destRect.x = xoffset; - destRect.y = yoffset; - destRect.width = sourceRect.width; - destRect.height = sourceRect.height; - - bool ret = copyTexture(source, sourceRect, sourceRenderTarget->getWidth(), sourceRenderTarget->getHeight(), - dest, destRect, destRenderTarget->getWidth(), destRenderTarget->getHeight(), destFormat); - - return ret; -} - -bool Renderer11::copyImage(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, - GLint xoffset, GLint yoffset, TextureStorageInterfaceCube *storage, GLenum target, GLint level) -{ - gl::Renderbuffer *colorbuffer = framebuffer->getReadColorbuffer(); - if (!colorbuffer) - { - ERR("Failed to retrieve the color buffer from the frame buffer."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - RenderTarget11 *sourceRenderTarget = RenderTarget11::makeRenderTarget11(colorbuffer->getRenderTarget()); - if (!sourceRenderTarget) - { - ERR("Failed to retrieve the render target from the frame buffer."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - ID3D11ShaderResourceView *source = sourceRenderTarget->getShaderResourceView(); - if (!source) - { - ERR("Failed to retrieve the render target view from the render target."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - TextureStorage11_Cube *storage11 = TextureStorage11_Cube::makeTextureStorage11_Cube(storage->getStorageInstance()); - if (!storage11) - { - ERR("Failed to retrieve the texture storage from the destination."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - RenderTarget11 *destRenderTarget = RenderTarget11::makeRenderTarget11(storage11->getRenderTarget(target, level)); - if (!destRenderTarget) - { - ERR("Failed to retrieve the render target from the destination storage."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - ID3D11RenderTargetView *dest = destRenderTarget->getRenderTargetView(); - if (!dest) - { - ERR("Failed to retrieve the render target view from the destination render target."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - gl::Rectangle destRect; - destRect.x = xoffset; - destRect.y = yoffset; - destRect.width = sourceRect.width; - destRect.height = sourceRect.height; - - bool ret = copyTexture(source, sourceRect, sourceRenderTarget->getWidth(), sourceRenderTarget->getHeight(), - dest, destRect, destRenderTarget->getWidth(), destRenderTarget->getHeight(), destFormat); - - return ret; -} - -bool Renderer11::copyTexture(ID3D11ShaderResourceView *source, const gl::Rectangle &sourceArea, unsigned int sourceWidth, unsigned int sourceHeight, - ID3D11RenderTargetView *dest, const gl::Rectangle &destArea, unsigned int destWidth, unsigned int destHeight, GLenum destFormat) -{ - HRESULT result; - - if (!mCopyResourcesInitialized) - { - ASSERT(!mCopyVB && !mCopySampler && !mCopyIL && !mCopyVS && !mCopyRGBAPS && !mCopyRGBPS && !mCopyLumPS && !mCopyLumAlphaPS); - - D3D11_BUFFER_DESC vbDesc; - vbDesc.ByteWidth = sizeof(d3d11::PositionTexCoordVertex) * 4; - vbDesc.Usage = D3D11_USAGE_DYNAMIC; - vbDesc.BindFlags = D3D11_BIND_VERTEX_BUFFER; - vbDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; - vbDesc.MiscFlags = 0; - vbDesc.StructureByteStride = 0; - - result = mDevice->CreateBuffer(&vbDesc, NULL, &mCopyVB); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mCopyVB, "Renderer11 copy texture vertex buffer"); - - D3D11_SAMPLER_DESC samplerDesc; - samplerDesc.Filter = D3D11_FILTER_MIN_MAG_MIP_LINEAR; - samplerDesc.AddressU = D3D11_TEXTURE_ADDRESS_CLAMP; - samplerDesc.AddressV = D3D11_TEXTURE_ADDRESS_CLAMP; - samplerDesc.AddressW = D3D11_TEXTURE_ADDRESS_CLAMP; - samplerDesc.MipLODBias = 0.0f; - samplerDesc.MaxAnisotropy = 0; - samplerDesc.ComparisonFunc = D3D11_COMPARISON_NEVER; - samplerDesc.BorderColor[0] = 0.0f; - samplerDesc.BorderColor[1] = 0.0f; - samplerDesc.BorderColor[2] = 0.0f; - samplerDesc.BorderColor[3] = 0.0f; - samplerDesc.MinLOD = 0.0f; - samplerDesc.MaxLOD = mDevice->GetFeatureLevel() >= D3D_FEATURE_LEVEL_10_0 ? 0.0f : FLT_MAX; - - result = mDevice->CreateSamplerState(&samplerDesc, &mCopySampler); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mCopySampler, "Renderer11 copy sampler"); - - D3D11_INPUT_ELEMENT_DESC quadLayout[] = - { - { "POSITION", 0, DXGI_FORMAT_R32G32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0 }, - { "TEXCOORD", 0, DXGI_FORMAT_R32G32_FLOAT, 0, 8, D3D11_INPUT_PER_VERTEX_DATA, 0 }, - }; - - result = mDevice->CreateInputLayout(quadLayout, 2, g_VS_Passthrough, sizeof(g_VS_Passthrough), &mCopyIL); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mCopyIL, "Renderer11 copy texture input layout"); - - result = mDevice->CreateVertexShader(g_VS_Passthrough, sizeof(g_VS_Passthrough), NULL, &mCopyVS); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mCopyVS, "Renderer11 copy texture vertex shader"); - - result = mDevice->CreatePixelShader(g_PS_PassthroughRGBA, sizeof(g_PS_PassthroughRGBA), NULL, &mCopyRGBAPS); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mCopyRGBAPS, "Renderer11 copy texture RGBA pixel shader"); - - result = mDevice->CreatePixelShader(g_PS_PassthroughRGB, sizeof(g_PS_PassthroughRGB), NULL, &mCopyRGBPS); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mCopyRGBPS, "Renderer11 copy texture RGB pixel shader"); - - result = mDevice->CreatePixelShader(g_PS_PassthroughLum, sizeof(g_PS_PassthroughLum), NULL, &mCopyLumPS); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mCopyLumPS, "Renderer11 copy texture luminance pixel shader"); - - result = mDevice->CreatePixelShader(g_PS_PassthroughLumAlpha, sizeof(g_PS_PassthroughLumAlpha), NULL, &mCopyLumAlphaPS); - ASSERT(SUCCEEDED(result)); - d3d11::SetDebugName(mCopyLumAlphaPS, "Renderer11 copy texture luminance alpha pixel shader"); - - mCopyResourcesInitialized = true; - } - - // Verify the source and destination area sizes - if (sourceArea.x < 0 || sourceArea.x + sourceArea.width > static_cast<int>(sourceWidth) || - sourceArea.y < 0 || sourceArea.y + sourceArea.height > static_cast<int>(sourceHeight) || - destArea.x < 0 || destArea.x + destArea.width > static_cast<int>(destWidth) || - destArea.y < 0 || destArea.y + destArea.height > static_cast<int>(destHeight)) - { - return gl::error(GL_INVALID_VALUE, false); - } - - // Set vertices - D3D11_MAPPED_SUBRESOURCE mappedResource; - result = mDeviceContext->Map(mCopyVB, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); - if (FAILED(result)) - { - ERR("Failed to map vertex buffer for texture copy, HRESULT: 0x%X.", result); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - d3d11::PositionTexCoordVertex *vertices = static_cast<d3d11::PositionTexCoordVertex*>(mappedResource.pData); - - // Create a quad in homogeneous coordinates - float x1 = (destArea.x / float(destWidth)) * 2.0f - 1.0f; - float y1 = ((destHeight - destArea.y - destArea.height) / float(destHeight)) * 2.0f - 1.0f; - float x2 = ((destArea.x + destArea.width) / float(destWidth)) * 2.0f - 1.0f; - float y2 = ((destHeight - destArea.y) / float(destHeight)) * 2.0f - 1.0f; - - float u1 = sourceArea.x / float(sourceWidth); - float v1 = sourceArea.y / float(sourceHeight); - float u2 = (sourceArea.x + sourceArea.width) / float(sourceWidth); - float v2 = (sourceArea.y + sourceArea.height) / float(sourceHeight); - - d3d11::SetPositionTexCoordVertex(&vertices[0], x1, y1, u1, v2); - d3d11::SetPositionTexCoordVertex(&vertices[1], x1, y2, u1, v1); - d3d11::SetPositionTexCoordVertex(&vertices[2], x2, y1, u2, v2); - d3d11::SetPositionTexCoordVertex(&vertices[3], x2, y2, u2, v1); - - mDeviceContext->Unmap(mCopyVB, 0); - - static UINT stride = sizeof(d3d11::PositionTexCoordVertex); - static UINT startIdx = 0; - mDeviceContext->IASetVertexBuffers(0, 1, &mCopyVB, &stride, &startIdx); - - // Apply state - mDeviceContext->OMSetBlendState(NULL, NULL, 0xFFFFFFF); - mDeviceContext->OMSetDepthStencilState(NULL, 0xFFFFFFFF); - mDeviceContext->RSSetState(NULL); - - // Apply shaders - mDeviceContext->IASetInputLayout(mCopyIL); - mDeviceContext->IASetPrimitiveTopology(D3D_PRIMITIVE_TOPOLOGY_TRIANGLESTRIP); - mDeviceContext->VSSetShader(mCopyVS, NULL, 0); - - ID3D11PixelShader *ps = NULL; - switch(destFormat) - { - case GL_RGBA: ps = mCopyRGBAPS; break; - case GL_RGB: ps = mCopyRGBPS; break; - case GL_ALPHA: ps = mCopyRGBAPS; break; - case GL_BGRA_EXT: ps = mCopyRGBAPS; break; - case GL_LUMINANCE: ps = mCopyLumPS; break; - case GL_LUMINANCE_ALPHA: ps = mCopyLumAlphaPS; break; - default: UNREACHABLE(); ps = NULL; break; - } - - mDeviceContext->PSSetShader(ps, NULL, 0); - mDeviceContext->GSSetShader(NULL, NULL, 0); - - // Unset the currently bound shader resource to avoid conflicts - static ID3D11ShaderResourceView *const nullSRV = NULL; - mDeviceContext->PSSetShaderResources(0, 1, &nullSRV); - - // Apply render target - setOneTimeRenderTarget(dest); - - // Set the viewport - D3D11_VIEWPORT viewport; - viewport.TopLeftX = 0; - viewport.TopLeftY = 0; - viewport.Width = destWidth; - viewport.Height = destHeight; - viewport.MinDepth = 0.0f; - viewport.MaxDepth = 1.0f; - mDeviceContext->RSSetViewports(1, &viewport); - - // Apply textures - mDeviceContext->PSSetShaderResources(0, 1, &source); - mDeviceContext->PSSetSamplers(0, 1, &mCopySampler); - - // Draw the quad - mDeviceContext->Draw(4, 0); - - // Unbind textures and render targets and vertex buffer - mDeviceContext->PSSetShaderResources(0, 1, &nullSRV); - - unapplyRenderTargets(); - - UINT zero = 0; - ID3D11Buffer *const nullBuffer = NULL; - mDeviceContext->IASetVertexBuffers(0, 1, &nullBuffer, &zero, &zero); - - markAllStateDirty(); - - return true; -} - -void Renderer11::unapplyRenderTargets() -{ - setOneTimeRenderTarget(NULL); -} - -void Renderer11::setOneTimeRenderTarget(ID3D11RenderTargetView *renderTargetView) -{ - ID3D11RenderTargetView *rtvArray[gl::IMPLEMENTATION_MAX_DRAW_BUFFERS] = {NULL}; - - rtvArray[0] = renderTargetView; - - mDeviceContext->OMSetRenderTargets(getMaxRenderTargets(), rtvArray, NULL); - - // Do not preserve the serial for this one-time-use render target - for (unsigned int rtIndex = 0; rtIndex < gl::IMPLEMENTATION_MAX_DRAW_BUFFERS; rtIndex++) - { - mAppliedRenderTargetSerials[rtIndex] = 0; - } -} - -RenderTarget *Renderer11::createRenderTarget(SwapChain *swapChain, bool depth) -{ - SwapChain11 *swapChain11 = SwapChain11::makeSwapChain11(swapChain); - RenderTarget11 *renderTarget = NULL; - - if (depth) - { - // Note: depth stencil may be NULL for 0 sized surfaces - renderTarget = new RenderTarget11(this, swapChain11->getDepthStencil(), - swapChain11->getDepthStencilTexture(), NULL, - swapChain11->getWidth(), swapChain11->getHeight()); - } - else - { - // Note: render target may be NULL for 0 sized surfaces - renderTarget = new RenderTarget11(this, swapChain11->getRenderTarget(), - swapChain11->getOffscreenTexture(), - swapChain11->getRenderTargetShaderResource(), - swapChain11->getWidth(), swapChain11->getHeight()); - } - return renderTarget; -} - -RenderTarget *Renderer11::createRenderTarget(int width, int height, GLenum format, GLsizei samples, bool depth) -{ - RenderTarget11 *renderTarget = new RenderTarget11(this, width, height, format, samples, depth); - return renderTarget; -} - -ShaderExecutable *Renderer11::loadExecutable(const void *function, size_t length, rx::ShaderType type) -{ - ShaderExecutable11 *executable = NULL; - - switch (type) - { - case rx::SHADER_VERTEX: - { - ID3D11VertexShader *vshader = NULL; - HRESULT result = mDevice->CreateVertexShader(function, length, NULL, &vshader); - ASSERT(SUCCEEDED(result)); - - if (vshader) - { - executable = new ShaderExecutable11(function, length, vshader); - } - } - break; - case rx::SHADER_PIXEL: - { - ID3D11PixelShader *pshader = NULL; - HRESULT result = mDevice->CreatePixelShader(function, length, NULL, &pshader); - ASSERT(SUCCEEDED(result)); - - if (pshader) - { - executable = new ShaderExecutable11(function, length, pshader); - } - } - break; - case rx::SHADER_GEOMETRY: - { - ID3D11GeometryShader *gshader = NULL; - HRESULT result = mDevice->CreateGeometryShader(function, length, NULL, &gshader); - ASSERT(SUCCEEDED(result)); - - if (gshader) - { - executable = new ShaderExecutable11(function, length, gshader); - } - } - break; - default: - UNREACHABLE(); - break; - } - - return executable; -} - -ShaderExecutable *Renderer11::compileToExecutable(gl::InfoLog &infoLog, const char *shaderHLSL, rx::ShaderType type, D3DWorkaroundType workaround) -{ - std::string profile; - - switch (type) - { - case rx::SHADER_VERTEX: - profile = "vs_4_0"; - break; - case rx::SHADER_PIXEL: - profile = "ps_4_0"; - break; - case rx::SHADER_GEOMETRY: - profile = "gs_4_0"; - break; - default: - UNREACHABLE(); - return NULL; - } - - if (mFeatureLevel == D3D_FEATURE_LEVEL_9_3) - profile += "_level_9_3"; - else if (mFeatureLevel == D3D_FEATURE_LEVEL_9_2 || mFeatureLevel == D3D_FEATURE_LEVEL_9_1) - profile += "_level_9_1"; - - ID3DBlob *binary = (ID3DBlob*)compileToBinary(infoLog, shaderHLSL, profile.c_str(), D3DCOMPILE_OPTIMIZATION_LEVEL0, false); - if (!binary) - return NULL; - - ShaderExecutable *executable = loadExecutable((DWORD *)binary->GetBufferPointer(), binary->GetBufferSize(), type); - binary->Release(); - - return executable; -} - -VertexBuffer *Renderer11::createVertexBuffer() -{ - return new VertexBuffer11(this); -} - -IndexBuffer *Renderer11::createIndexBuffer() -{ - return new IndexBuffer11(this); -} - -BufferStorage *Renderer11::createBufferStorage() -{ - return new BufferStorage11(this); -} - -QueryImpl *Renderer11::createQuery(GLenum type) -{ - return new Query11(this, type); -} - -FenceImpl *Renderer11::createFence() -{ - return new Fence11(this); -} - -bool Renderer11::getRenderTargetResource(gl::Renderbuffer *colorbuffer, unsigned int *subresourceIndex, ID3D11Texture2D **resource) -{ - ASSERT(colorbuffer != NULL); - - RenderTarget11 *renderTarget = RenderTarget11::makeRenderTarget11(colorbuffer->getRenderTarget()); - if (renderTarget) - { - *subresourceIndex = renderTarget->getSubresourceIndex(); - - ID3D11RenderTargetView *colorBufferRTV = renderTarget->getRenderTargetView(); - if (colorBufferRTV) - { - ID3D11Resource *textureResource = NULL; - colorBufferRTV->GetResource(&textureResource); - - if (textureResource) - { - HRESULT result = textureResource->QueryInterface(IID_ID3D11Texture2D, (void**)resource); - textureResource->Release(); - - if (SUCCEEDED(result)) - { - return true; - } - else - { - ERR("Failed to extract the ID3D11Texture2D from the render target resource, " - "HRESULT: 0x%X.", result); - } - } - } - } - - return false; -} - -bool Renderer11::blitRect(gl::Framebuffer *readTarget, const gl::Rectangle &readRect, gl::Framebuffer *drawTarget, const gl::Rectangle &drawRect, - bool blitRenderTarget, bool blitDepthStencil) -{ - if (blitRenderTarget) - { - gl::Renderbuffer *readBuffer = readTarget->getReadColorbuffer(); - - if (!readBuffer) - { - ERR("Failed to retrieve the read buffer from the read framebuffer."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - RenderTarget *readRenderTarget = readBuffer->getRenderTarget(); - - for (unsigned int colorAttachment = 0; colorAttachment < gl::IMPLEMENTATION_MAX_DRAW_BUFFERS; colorAttachment++) - { - if (drawTarget->isEnabledColorAttachment(colorAttachment)) - { - gl::Renderbuffer *drawBuffer = drawTarget->getColorbuffer(colorAttachment); - - if (!drawBuffer) - { - ERR("Failed to retrieve the draw buffer from the draw framebuffer."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - RenderTarget *drawRenderTarget = drawBuffer->getRenderTarget(); - - if (!blitRenderbufferRect(readRect, drawRect, readRenderTarget, drawRenderTarget, false)) - { - return false; - } - } - } - } - - if (blitDepthStencil) - { - gl::Renderbuffer *readBuffer = readTarget->getDepthOrStencilbuffer(); - gl::Renderbuffer *drawBuffer = drawTarget->getDepthOrStencilbuffer(); - - if (!readBuffer) - { - ERR("Failed to retrieve the read depth-stencil buffer from the read framebuffer."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - if (!drawBuffer) - { - ERR("Failed to retrieve the draw depth-stencil buffer from the draw framebuffer."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - RenderTarget *readRenderTarget = readBuffer->getDepthStencil(); - RenderTarget *drawRenderTarget = drawBuffer->getDepthStencil(); - - if (!blitRenderbufferRect(readRect, drawRect, readRenderTarget, drawRenderTarget, true)) - { - return false; - } - } - - return true; -} - -void Renderer11::readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, - GLsizei outputPitch, bool packReverseRowOrder, GLint packAlignment, void* pixels) -{ - ID3D11Texture2D *colorBufferTexture = NULL; - unsigned int subresourceIndex = 0; - - gl::Renderbuffer *colorbuffer = framebuffer->getReadColorbuffer(); - - if (colorbuffer && getRenderTargetResource(colorbuffer, &subresourceIndex, &colorBufferTexture)) - { - gl::Rectangle area; - area.x = x; - area.y = y; - area.width = width; - area.height = height; - - readTextureData(colorBufferTexture, subresourceIndex, area, colorbuffer->getActualFormat(), format, type, outputPitch, - packReverseRowOrder, packAlignment, pixels); - - colorBufferTexture->Release(); - colorBufferTexture = NULL; - } -} - -Image *Renderer11::createImage() -{ - return new Image11(); -} - -void Renderer11::generateMipmap(Image *dest, Image *src) -{ - Image11 *dest11 = Image11::makeImage11(dest); - Image11 *src11 = Image11::makeImage11(src); - Image11::generateMipmap(dest11, src11); -} - -TextureStorage *Renderer11::createTextureStorage2D(SwapChain *swapChain) -{ - SwapChain11 *swapChain11 = SwapChain11::makeSwapChain11(swapChain); - return new TextureStorage11_2D(this, swapChain11); -} - -TextureStorage *Renderer11::createTextureStorage2D(int levels, GLenum internalformat, GLenum usage, bool forceRenderable, GLsizei width, GLsizei height) -{ - return new TextureStorage11_2D(this, levels, internalformat, usage, forceRenderable, width, height); -} - -TextureStorage *Renderer11::createTextureStorageCube(int levels, GLenum internalformat, GLenum usage, bool forceRenderable, int size) -{ - return new TextureStorage11_Cube(this, levels, internalformat, usage, forceRenderable, size); -} - -static inline unsigned int getFastPixelCopySize(DXGI_FORMAT sourceFormat, GLenum sourceGLFormat, GLenum destFormat, GLenum destType) -{ - if (sourceFormat == DXGI_FORMAT_A8_UNORM && - destFormat == GL_ALPHA && - destType == GL_UNSIGNED_BYTE) - { - return 1; - } - else if (sourceFormat == DXGI_FORMAT_R8G8B8A8_UNORM && - sourceGLFormat == GL_RGBA8_OES && - destFormat == GL_RGBA && - destType == GL_UNSIGNED_BYTE) - { - return 4; - } - else if (sourceFormat == DXGI_FORMAT_B8G8R8A8_UNORM && - destFormat == GL_BGRA_EXT && - destType == GL_UNSIGNED_BYTE) - { - return 4; - } - else if (sourceFormat == DXGI_FORMAT_R16G16B16A16_FLOAT && - sourceGLFormat == GL_RGBA16F_EXT && - destFormat == GL_RGBA && - destType == GL_HALF_FLOAT_OES) - { - return 8; - } - else if (sourceFormat == DXGI_FORMAT_R32G32B32_FLOAT && - destFormat == GL_RGB && - destType == GL_FLOAT) - { - return 12; - } - else if (sourceFormat == DXGI_FORMAT_R32G32B32A32_FLOAT && - sourceGLFormat == GL_RGBA32F_EXT && - destFormat == GL_RGBA && - destType == GL_FLOAT) - { - return 16; - } - else - { - return 0; - } -} - -static inline void readPixelColor(const unsigned char *data, DXGI_FORMAT format, GLenum glFormat, unsigned int x, - unsigned int y, int inputPitch, gl::Color *outColor) -{ - switch (format) - { - case DXGI_FORMAT_R8G8B8A8_UNORM: - { - unsigned int rgba = *reinterpret_cast<const unsigned int*>(data + 4 * x + y * inputPitch); - outColor->red = (rgba & 0x000000FF) * (1.0f / 0x000000FF); - outColor->green = (rgba & 0x0000FF00) * (1.0f / 0x0000FF00); - outColor->blue = (rgba & 0x00FF0000) * (1.0f / 0x00FF0000); - - if (gl::GetAlphaSize(glFormat) > 0) - { - outColor->alpha = (rgba & 0xFF000000) * (1.0f / 0xFF000000); - } - else - { - outColor->alpha = 1.0f; - } - } - break; - - case DXGI_FORMAT_A8_UNORM: - { - outColor->red = 0.0f; - outColor->green = 0.0f; - outColor->blue = 0.0f; - outColor->alpha = *(data + x + y * inputPitch) / 255.0f; - } - break; - - case DXGI_FORMAT_R32G32B32A32_FLOAT: - { - outColor->red = *(reinterpret_cast<const float*>(data + 16 * x + y * inputPitch) + 0); - outColor->green = *(reinterpret_cast<const float*>(data + 16 * x + y * inputPitch) + 1); - outColor->blue = *(reinterpret_cast<const float*>(data + 16 * x + y * inputPitch) + 2); - - if (gl::GetAlphaSize(glFormat) > 0) - { - outColor->alpha = *(reinterpret_cast<const float*>(data + 16 * x + y * inputPitch) + 3); - } - else - { - outColor->alpha = 1.0f; - } - } - break; - - case DXGI_FORMAT_R32G32B32_FLOAT: - { - outColor->red = *(reinterpret_cast<const float*>(data + 12 * x + y * inputPitch) + 0); - outColor->green = *(reinterpret_cast<const float*>(data + 12 * x + y * inputPitch) + 1); - outColor->blue = *(reinterpret_cast<const float*>(data + 12 * x + y * inputPitch) + 2); - outColor->alpha = 1.0f; - } - break; - - case DXGI_FORMAT_R16G16B16A16_FLOAT: - { - outColor->red = gl::float16ToFloat32(*(reinterpret_cast<const unsigned short*>(data + 8 * x + y * inputPitch) + 0)); - outColor->green = gl::float16ToFloat32(*(reinterpret_cast<const unsigned short*>(data + 8 * x + y * inputPitch) + 1)); - outColor->blue = gl::float16ToFloat32(*(reinterpret_cast<const unsigned short*>(data + 8 * x + y * inputPitch) + 2)); - - if (gl::GetAlphaSize(glFormat) > 0) - { - outColor->alpha = gl::float16ToFloat32(*(reinterpret_cast<const unsigned short*>(data + 8 * x + y * inputPitch) + 3)); - } - else - { - outColor->alpha = 1.0f; - } - } - break; - - case DXGI_FORMAT_B8G8R8A8_UNORM: - { - unsigned int bgra = *reinterpret_cast<const unsigned int*>(data + 4 * x + y * inputPitch); - outColor->red = (bgra & 0x00FF0000) * (1.0f / 0x00FF0000); - outColor->blue = (bgra & 0x000000FF) * (1.0f / 0x000000FF); - outColor->green = (bgra & 0x0000FF00) * (1.0f / 0x0000FF00); - outColor->alpha = (bgra & 0xFF000000) * (1.0f / 0xFF000000); - } - break; - - case DXGI_FORMAT_R8_UNORM: - { - outColor->red = *(data + x + y * inputPitch) / 255.0f; - outColor->green = 0.0f; - outColor->blue = 0.0f; - outColor->alpha = 1.0f; - } - break; - - case DXGI_FORMAT_R8G8_UNORM: - { - unsigned short rg = *reinterpret_cast<const unsigned short*>(data + 2 * x + y * inputPitch); - - outColor->red = (rg & 0xFF00) * (1.0f / 0xFF00); - outColor->green = (rg & 0x00FF) * (1.0f / 0x00FF); - outColor->blue = 0.0f; - outColor->alpha = 1.0f; - } - break; - - case DXGI_FORMAT_R16_FLOAT: - { - outColor->red = gl::float16ToFloat32(*reinterpret_cast<const unsigned short*>(data + 2 * x + y * inputPitch)); - outColor->green = 0.0f; - outColor->blue = 0.0f; - outColor->alpha = 1.0f; - } - break; - - case DXGI_FORMAT_R16G16_FLOAT: - { - outColor->red = gl::float16ToFloat32(*(reinterpret_cast<const unsigned short*>(data + 4 * x + y * inputPitch) + 0)); - outColor->green = gl::float16ToFloat32(*(reinterpret_cast<const unsigned short*>(data + 4 * x + y * inputPitch) + 1)); - outColor->blue = 0.0f; - outColor->alpha = 1.0f; - } - break; - - default: - ERR("ReadPixelColor not implemented for DXGI format %u.", format); - UNIMPLEMENTED(); - break; - } -} - -static inline void writePixelColor(const gl::Color &color, GLenum format, GLenum type, unsigned int x, - unsigned int y, int outputPitch, void *outData) -{ - unsigned char* byteData = reinterpret_cast<unsigned char*>(outData); - unsigned short* shortData = reinterpret_cast<unsigned short*>(outData); - - switch (format) - { - case GL_RGBA: - switch (type) - { - case GL_UNSIGNED_BYTE: - byteData[4 * x + y * outputPitch + 0] = static_cast<unsigned char>(255 * color.red + 0.5f); - byteData[4 * x + y * outputPitch + 1] = static_cast<unsigned char>(255 * color.green + 0.5f); - byteData[4 * x + y * outputPitch + 2] = static_cast<unsigned char>(255 * color.blue + 0.5f); - byteData[4 * x + y * outputPitch + 3] = static_cast<unsigned char>(255 * color.alpha + 0.5f); - break; - - default: - ERR("WritePixelColor not implemented for format GL_RGBA and type 0x%X.", type); - UNIMPLEMENTED(); - break; - } - break; - - case GL_BGRA_EXT: - switch (type) - { - case GL_UNSIGNED_BYTE: - byteData[4 * x + y * outputPitch + 0] = static_cast<unsigned char>(255 * color.blue + 0.5f); - byteData[4 * x + y * outputPitch + 1] = static_cast<unsigned char>(255 * color.green + 0.5f); - byteData[4 * x + y * outputPitch + 2] = static_cast<unsigned char>(255 * color.red + 0.5f); - byteData[4 * x + y * outputPitch + 3] = static_cast<unsigned char>(255 * color.alpha + 0.5f); - break; - - case GL_UNSIGNED_SHORT_4_4_4_4_REV_EXT: - // According to the desktop GL spec in the "Transfer of Pixel Rectangles" section - // this type is packed as follows: - // 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 - // -------------------------------------------------------------------------------- - // | 4th | 3rd | 2nd | 1st component | - // -------------------------------------------------------------------------------- - // in the case of BGRA_EXT, B is the first component, G the second, and so forth. - shortData[x + y * outputPitch / sizeof(unsigned short)] = - (static_cast<unsigned short>(15 * color.alpha + 0.5f) << 12) | - (static_cast<unsigned short>(15 * color.red + 0.5f) << 8) | - (static_cast<unsigned short>(15 * color.green + 0.5f) << 4) | - (static_cast<unsigned short>(15 * color.blue + 0.5f) << 0); - break; - - case GL_UNSIGNED_SHORT_1_5_5_5_REV_EXT: - // According to the desktop GL spec in the "Transfer of Pixel Rectangles" section - // this type is packed as follows: - // 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 - // -------------------------------------------------------------------------------- - // | 4th | 3rd | 2nd | 1st component | - // -------------------------------------------------------------------------------- - // in the case of BGRA_EXT, B is the first component, G the second, and so forth. - shortData[x + y * outputPitch / sizeof(unsigned short)] = - (static_cast<unsigned short>( color.alpha + 0.5f) << 15) | - (static_cast<unsigned short>(31 * color.red + 0.5f) << 10) | - (static_cast<unsigned short>(31 * color.green + 0.5f) << 5) | - (static_cast<unsigned short>(31 * color.blue + 0.5f) << 0); - break; - - default: - ERR("WritePixelColor not implemented for format GL_BGRA_EXT and type 0x%X.", type); - UNIMPLEMENTED(); - break; - } - break; - - case GL_RGB: - switch (type) - { - case GL_UNSIGNED_SHORT_5_6_5: - shortData[x + y * outputPitch / sizeof(unsigned short)] = - (static_cast<unsigned short>(31 * color.blue + 0.5f) << 0) | - (static_cast<unsigned short>(63 * color.green + 0.5f) << 5) | - (static_cast<unsigned short>(31 * color.red + 0.5f) << 11); - break; - - case GL_UNSIGNED_BYTE: - byteData[3 * x + y * outputPitch + 0] = static_cast<unsigned char>(255 * color.red + 0.5f); - byteData[3 * x + y * outputPitch + 1] = static_cast<unsigned char>(255 * color.green + 0.5f); - byteData[3 * x + y * outputPitch + 2] = static_cast<unsigned char>(255 * color.blue + 0.5f); - break; - - default: - ERR("WritePixelColor not implemented for format GL_RGB and type 0x%X.", type); - UNIMPLEMENTED(); - break; - } - break; - - default: - ERR("WritePixelColor not implemented for format 0x%X.", format); - UNIMPLEMENTED(); - break; - } -} - -void Renderer11::readTextureData(ID3D11Texture2D *texture, unsigned int subResource, const gl::Rectangle &area, - GLenum sourceFormat, GLenum format, GLenum type, GLsizei outputPitch, bool packReverseRowOrder, - GLint packAlignment, void *pixels) -{ - D3D11_TEXTURE2D_DESC textureDesc; - texture->GetDesc(&textureDesc); - - D3D11_TEXTURE2D_DESC stagingDesc; - stagingDesc.Width = area.width; - stagingDesc.Height = area.height; - stagingDesc.MipLevels = 1; - stagingDesc.ArraySize = 1; - stagingDesc.Format = textureDesc.Format; - stagingDesc.SampleDesc.Count = 1; - stagingDesc.SampleDesc.Quality = 0; - stagingDesc.Usage = D3D11_USAGE_STAGING; - stagingDesc.BindFlags = 0; - stagingDesc.CPUAccessFlags = D3D11_CPU_ACCESS_READ; - stagingDesc.MiscFlags = 0; - - ID3D11Texture2D* stagingTex = NULL; - HRESULT result = mDevice->CreateTexture2D(&stagingDesc, NULL, &stagingTex); - if (FAILED(result)) - { - ERR("Failed to create staging texture for readPixels, HRESULT: 0x%X.", result); - return; - } - - ID3D11Texture2D* srcTex = NULL; - if (textureDesc.SampleDesc.Count > 1) - { - D3D11_TEXTURE2D_DESC resolveDesc; - resolveDesc.Width = textureDesc.Width; - resolveDesc.Height = textureDesc.Height; - resolveDesc.MipLevels = 1; - resolveDesc.ArraySize = 1; - resolveDesc.Format = textureDesc.Format; - resolveDesc.SampleDesc.Count = 1; - resolveDesc.SampleDesc.Quality = 0; - resolveDesc.Usage = D3D11_USAGE_DEFAULT; - resolveDesc.BindFlags = 0; - resolveDesc.CPUAccessFlags = 0; - resolveDesc.MiscFlags = 0; - - result = mDevice->CreateTexture2D(&resolveDesc, NULL, &srcTex); - if (FAILED(result)) - { - ERR("Failed to create resolve texture for readPixels, HRESULT: 0x%X.", result); - stagingTex->Release(); - return; - } - - mDeviceContext->ResolveSubresource(srcTex, 0, texture, subResource, textureDesc.Format); - subResource = 0; - } - else - { - srcTex = texture; - srcTex->AddRef(); - } - - D3D11_BOX srcBox; - srcBox.left = area.x; - srcBox.right = area.x + area.width; - srcBox.top = area.y; - srcBox.bottom = area.y + area.height; - srcBox.front = 0; - srcBox.back = 1; - - mDeviceContext->CopySubresourceRegion(stagingTex, 0, 0, 0, 0, srcTex, subResource, &srcBox); - - srcTex->Release(); - srcTex = NULL; - - D3D11_MAPPED_SUBRESOURCE mapping; - mDeviceContext->Map(stagingTex, 0, D3D11_MAP_READ, 0, &mapping); - - unsigned char *source; - int inputPitch; - if (packReverseRowOrder) - { - source = static_cast<unsigned char*>(mapping.pData) + mapping.RowPitch * (area.height - 1); - inputPitch = -static_cast<int>(mapping.RowPitch); - } - else - { - source = static_cast<unsigned char*>(mapping.pData); - inputPitch = static_cast<int>(mapping.RowPitch); - } - - unsigned int fastPixelSize = getFastPixelCopySize(textureDesc.Format, sourceFormat, format, type); - if (fastPixelSize != 0) - { - unsigned char *dest = static_cast<unsigned char*>(pixels); - for (int j = 0; j < area.height; j++) - { - memcpy(dest + j * outputPitch, source + j * inputPitch, area.width * fastPixelSize); - } - } - else if (textureDesc.Format == DXGI_FORMAT_B8G8R8A8_UNORM && - format == GL_RGBA && - type == GL_UNSIGNED_BYTE) - { - // Fast path for swapping red with blue - unsigned char *dest = static_cast<unsigned char*>(pixels); - - for (int j = 0; j < area.height; j++) - { - for (int i = 0; i < area.width; i++) - { - unsigned int argb = *(unsigned int*)(source + 4 * i + j * inputPitch); - *(unsigned int*)(dest + 4 * i + j * outputPitch) = - (argb & 0xFF00FF00) | // Keep alpha and green - (argb & 0x00FF0000) >> 16 | // Move red to blue - (argb & 0x000000FF) << 16; // Move blue to red - } - } - } - else - { - gl::Color pixelColor; - for (int j = 0; j < area.height; j++) - { - for (int i = 0; i < area.width; i++) - { - readPixelColor(source, textureDesc.Format, sourceFormat, i, j, inputPitch, &pixelColor); - writePixelColor(pixelColor, format, type, i, j, outputPitch, pixels); - } - } - } - - mDeviceContext->Unmap(stagingTex, 0); - - stagingTex->Release(); - stagingTex = NULL; -} - -bool Renderer11::blitRenderbufferRect(const gl::Rectangle &readRect, const gl::Rectangle &drawRect, RenderTarget *readRenderTarget, - RenderTarget *drawRenderTarget, bool wholeBufferCopy) -{ - ASSERT(readRect.width == drawRect.width && readRect.height == drawRect.height); - - RenderTarget11 *drawRenderTarget11 = RenderTarget11::makeRenderTarget11(drawRenderTarget); - if (!drawRenderTarget) - { - ERR("Failed to retrieve the draw render target from the draw framebuffer."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - ID3D11Texture2D *drawTexture = drawRenderTarget11->getTexture(); - unsigned int drawSubresource = drawRenderTarget11->getSubresourceIndex(); - - RenderTarget11 *readRenderTarget11 = RenderTarget11::makeRenderTarget11(readRenderTarget); - if (!readRenderTarget) - { - ERR("Failed to retrieve the read render target from the read framebuffer."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - ID3D11Texture2D *readTexture = NULL; - unsigned int readSubresource = 0; - if (readRenderTarget->getSamples() > 0) - { - readTexture = resolveMultisampledTexture(readRenderTarget11->getTexture(), readRenderTarget11->getSubresourceIndex()); - readSubresource = 0; - } - else - { - readTexture = readRenderTarget11->getTexture(); - readTexture->AddRef(); - readSubresource = readRenderTarget11->getSubresourceIndex(); - } - - if (!readTexture) - { - ERR("Failed to retrieve the read render target view from the read render target."); - return gl::error(GL_OUT_OF_MEMORY, false); - } - - D3D11_BOX readBox; - readBox.left = readRect.x; - readBox.right = readRect.x + readRect.width; - readBox.top = readRect.y; - readBox.bottom = readRect.y + readRect.height; - readBox.front = 0; - readBox.back = 1; - - // D3D11 needs depth-stencil CopySubresourceRegions to have a NULL pSrcBox - // We also require complete framebuffer copies for depth-stencil blit. - D3D11_BOX *pSrcBox = wholeBufferCopy ? NULL : &readBox; - - mDeviceContext->CopySubresourceRegion(drawTexture, drawSubresource, drawRect.x, drawRect.y, 0, - readTexture, readSubresource, pSrcBox); - - SafeRelease(readTexture); - - return true; -} - -ID3D11Texture2D *Renderer11::resolveMultisampledTexture(ID3D11Texture2D *source, unsigned int subresource) -{ - D3D11_TEXTURE2D_DESC textureDesc; - source->GetDesc(&textureDesc); - - if (textureDesc.SampleDesc.Count > 1) - { - D3D11_TEXTURE2D_DESC resolveDesc; - resolveDesc.Width = textureDesc.Width; - resolveDesc.Height = textureDesc.Height; - resolveDesc.MipLevels = 1; - resolveDesc.ArraySize = 1; - resolveDesc.Format = textureDesc.Format; - resolveDesc.SampleDesc.Count = 1; - resolveDesc.SampleDesc.Quality = 0; - resolveDesc.Usage = textureDesc.Usage; - resolveDesc.BindFlags = textureDesc.BindFlags; - resolveDesc.CPUAccessFlags = 0; - resolveDesc.MiscFlags = 0; - - ID3D11Texture2D *resolveTexture = NULL; - HRESULT result = mDevice->CreateTexture2D(&resolveDesc, NULL, &resolveTexture); - if (FAILED(result)) - { - ERR("Failed to create a multisample resolve texture, HRESULT: 0x%X.", result); - return NULL; - } - - mDeviceContext->ResolveSubresource(resolveTexture, 0, source, subresource, textureDesc.Format); - return resolveTexture; - } - else - { - source->AddRef(); - return source; - } -} - -bool Renderer11::getLUID(LUID *adapterLuid) const -{ - adapterLuid->HighPart = 0; - adapterLuid->LowPart = 0; - - if (!mDxgiAdapter) - { - return false; - } - - DXGI_ADAPTER_DESC adapterDesc; - if (FAILED(mDxgiAdapter->GetDesc(&adapterDesc))) - { - return false; - } - - *adapterLuid = adapterDesc.AdapterLuid; - return true; -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Renderer11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Renderer11.h deleted file mode 100644 index 773fc26db1..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/Renderer11.h +++ /dev/null @@ -1,373 +0,0 @@ -// -// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// Renderer11.h: Defines a back-end specific class for the D3D11 renderer. - -#ifndef LIBGLESV2_RENDERER_RENDERER11_H_ -#define LIBGLESV2_RENDERER_RENDERER11_H_ - -#include "common/angleutils.h" -#include "libGLESv2/angletypes.h" -#include "libGLESv2/mathutil.h" - -#include "libGLESv2/renderer/Renderer.h" -#include "libGLESv2/renderer/d3d11/RenderStateCache.h" -#include "libGLESv2/renderer/d3d11/InputLayoutCache.h" -#include "libGLESv2/renderer/RenderTarget.h" - -#if defined(ANGLE_OS_WINRT) && !defined(ANGLE_OS_WINPHONE) -struct IInspectable; -namespace ABI { - namespace Windows { - namespace ApplicationModel { - struct ISuspendingEventArgs; - } - } -} -#endif - -namespace gl -{ -class Renderbuffer; -} - -namespace rx -{ - -class VertexDataManager; -class IndexDataManager; -class StreamingIndexBufferInterface; - -enum -{ - MAX_VERTEX_UNIFORM_VECTORS_D3D11 = 1024, - MAX_FRAGMENT_UNIFORM_VECTORS_D3D11 = 1024 -}; - -class Renderer11 : public Renderer -{ - public: - Renderer11(egl::Display *display); - virtual ~Renderer11(); - - static Renderer11 *makeRenderer11(Renderer *renderer); - - virtual EGLint initialize(); - virtual bool resetDevice(); - - virtual int generateConfigs(ConfigDesc **configDescList); - virtual void deleteConfigs(ConfigDesc *configDescList); - - virtual void sync(bool block); - - virtual SwapChain *createSwapChain(EGLNativeWindowType window, HANDLE shareHandle, GLenum backBufferFormat, GLenum depthBufferFormat); - - virtual void setSamplerState(gl::SamplerType type, int index, const gl::SamplerState &sampler); - virtual void setTexture(gl::SamplerType type, int index, gl::Texture *texture); - - virtual void setRasterizerState(const gl::RasterizerState &rasterState); - virtual void setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState &blendState, const gl::Color &blendColor, - unsigned int sampleMask); - virtual void setDepthStencilState(const gl::DepthStencilState &depthStencilState, int stencilRef, - int stencilBackRef, bool frontFaceCCW); - - virtual void setScissorRectangle(const gl::Rectangle &scissor, bool enabled); - virtual bool setViewport(const gl::Rectangle &viewport, float zNear, float zFar, GLenum drawMode, GLenum frontFace, - bool ignoreViewport); - - virtual bool applyPrimitiveType(GLenum mode, GLsizei count); - virtual bool applyRenderTarget(gl::Framebuffer *frameBuffer); - virtual void applyShaders(gl::ProgramBinary *programBinary); - virtual void applyUniforms(gl::ProgramBinary *programBinary, gl::UniformArray *uniformArray); - virtual GLenum applyVertexBuffer(gl::ProgramBinary *programBinary, gl::VertexAttribute vertexAttributes[], GLint first, GLsizei count, GLsizei instances); - virtual GLenum applyIndexBuffer(const GLvoid *indices, gl::Buffer *elementArrayBuffer, GLsizei count, GLenum mode, GLenum type, TranslatedIndexData *indexInfo); - - virtual void drawArrays(GLenum mode, GLsizei count, GLsizei instances); - virtual void drawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, gl::Buffer *elementArrayBuffer, const TranslatedIndexData &indexInfo, GLsizei instances); - - virtual void clear(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer); - - virtual void markAllStateDirty(); - - // lost device - void notifyDeviceLost(); - virtual bool isDeviceLost(); - virtual bool testDeviceLost(bool notify); - virtual bool testDeviceResettable(); - - // Renderer capabilities - virtual DWORD getAdapterVendor() const; - virtual std::string getRendererDescription() const; - virtual GUID getAdapterIdentifier() const; - - virtual bool getBGRATextureSupport() const; - virtual bool getDXT1TextureSupport(); - virtual bool getDXT3TextureSupport(); - virtual bool getDXT5TextureSupport(); - virtual bool getEventQuerySupport(); - virtual bool getFloat32TextureSupport(bool *filtering, bool *renderable); - virtual bool getFloat16TextureSupport(bool *filtering, bool *renderable); - virtual bool getLuminanceTextureSupport(); - virtual bool getLuminanceAlphaTextureSupport(); - virtual unsigned int getMaxVertexTextureImageUnits() const; - virtual unsigned int getMaxCombinedTextureImageUnits() const; - virtual unsigned int getReservedVertexUniformVectors() const; - virtual unsigned int getReservedFragmentUniformVectors() const; - virtual unsigned int getMaxVertexUniformVectors() const; - virtual unsigned int getMaxFragmentUniformVectors() const; - virtual unsigned int getMaxVaryingVectors() const; - virtual bool getNonPower2TextureSupport() const; - virtual bool getDepthTextureSupport() const; - virtual bool getOcclusionQuerySupport() const; - virtual bool getInstancingSupport() const; - virtual bool getTextureFilterAnisotropySupport() const; - virtual float getTextureMaxAnisotropy() const; - virtual bool getShareHandleSupport() const; - virtual bool getDerivativeInstructionSupport() const; - virtual bool getPostSubBufferSupport() const; - - virtual int getMajorShaderModel() const; - virtual float getMaxPointSize() const; - virtual int getMaxViewportDimension() const; - virtual int getMaxTextureWidth() const; - virtual int getMaxTextureHeight() const; - virtual bool get32BitIndexSupport() const; - virtual int getMinSwapInterval() const; - virtual int getMaxSwapInterval() const; - - virtual GLsizei getMaxSupportedSamples() const; - int getNearestSupportedSamples(DXGI_FORMAT format, unsigned int requested) const; - - virtual unsigned int getMaxRenderTargets() const; - - // Pixel operations - virtual bool copyToRenderTarget(TextureStorageInterface2D *dest, TextureStorageInterface2D *source); - virtual bool copyToRenderTarget(TextureStorageInterfaceCube *dest, TextureStorageInterfaceCube *source); - - virtual bool copyImage(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, - GLint xoffset, GLint yoffset, TextureStorageInterface2D *storage, GLint level); - virtual bool copyImage(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, - GLint xoffset, GLint yoffset, TextureStorageInterfaceCube *storage, GLenum target, GLint level); - - bool copyTexture(ID3D11ShaderResourceView *source, const gl::Rectangle &sourceArea, unsigned int sourceWidth, unsigned int sourceHeight, - ID3D11RenderTargetView *dest, const gl::Rectangle &destArea, unsigned int destWidth, unsigned int destHeight, GLenum destFormat); - - virtual bool blitRect(gl::Framebuffer *readTarget, const gl::Rectangle &readRect, gl::Framebuffer *drawTarget, const gl::Rectangle &drawRect, - bool blitRenderTarget, bool blitDepthStencil); - virtual void readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, - GLsizei outputPitch, bool packReverseRowOrder, GLint packAlignment, void* pixels); - - // RenderTarget creation - virtual RenderTarget *createRenderTarget(SwapChain *swapChain, bool depth); - virtual RenderTarget *createRenderTarget(int width, int height, GLenum format, GLsizei samples, bool depth); - - // Shader operations - virtual ShaderExecutable *loadExecutable(const void *function, size_t length, rx::ShaderType type); - virtual ShaderExecutable *compileToExecutable(gl::InfoLog &infoLog, const char *shaderHLSL, rx::ShaderType type, D3DWorkaroundType workaround); - - // Image operations - virtual Image *createImage(); - virtual void generateMipmap(Image *dest, Image *source); - virtual TextureStorage *createTextureStorage2D(SwapChain *swapChain); - virtual TextureStorage *createTextureStorage2D(int levels, GLenum internalformat, GLenum usage, bool forceRenderable, GLsizei width, GLsizei height); - virtual TextureStorage *createTextureStorageCube(int levels, GLenum internalformat, GLenum usage, bool forceRenderable, int size); - - // Buffer creation - virtual VertexBuffer *createVertexBuffer(); - virtual IndexBuffer *createIndexBuffer(); - virtual BufferStorage *createBufferStorage(); - - // Query and Fence creation - virtual QueryImpl *createQuery(GLenum type); - virtual FenceImpl *createFence(); - - // D3D11-renderer specific methods - ID3D11Device *getDevice() { return mDevice; } - ID3D11DeviceContext *getDeviceContext() { return mDeviceContext; }; - IDXGIFactory *getDxgiFactory() { return mDxgiFactory; }; - D3D_FEATURE_LEVEL getFeatureLevel() { return mFeatureLevel; } - - bool getRenderTargetResource(gl::Renderbuffer *colorbuffer, unsigned int *subresourceIndex, ID3D11Texture2D **resource); - void unapplyRenderTargets(); - void setOneTimeRenderTarget(ID3D11RenderTargetView *renderTargetView); - - virtual bool getLUID(LUID *adapterLuid) const; - - private: - DISALLOW_COPY_AND_ASSIGN(Renderer11); - - void drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer); - void drawTriangleFan(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer, int instances); - - void readTextureData(ID3D11Texture2D *texture, unsigned int subResource, const gl::Rectangle &area, - GLenum sourceFormat, GLenum format, GLenum type, GLsizei outputPitch, bool packReverseRowOrder, - GLint packAlignment, void *pixels); - - void maskedClear(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer); - rx::Range getViewportBounds() const; - - bool blitRenderbufferRect(const gl::Rectangle &readRect, const gl::Rectangle &drawRect, RenderTarget *readRenderTarget, - RenderTarget *drawRenderTarget, bool wholeBufferCopy); - ID3D11Texture2D *resolveMultisampledTexture(ID3D11Texture2D *source, unsigned int subresource); - -#if defined(ANGLE_OS_WINRT) && !defined(ANGLE_OS_WINPHONE) - HRESULT onSuspend(IInspectable *, ABI::Windows::ApplicationModel::ISuspendingEventArgs *); -#endif - - HMODULE mD3d11Module; - HMODULE mDxgiModule; - - bool mDeviceLost; - - void initializeDevice(); - void releaseDeviceResources(); - int getMinorShaderModel() const; - void release(); - - RenderStateCache mStateCache; - - // Support flags - bool mFloat16TextureSupport; - bool mFloat16FilterSupport; - bool mFloat16RenderSupport; - - bool mFloat32TextureSupport; - bool mFloat32FilterSupport; - bool mFloat32RenderSupport; - - bool mDXT1TextureSupport; - bool mDXT3TextureSupport; - bool mDXT5TextureSupport; - - bool mDepthTextureSupport; - - // Multisample format support - struct MultisampleSupportInfo - { - unsigned int qualityLevels[D3D11_MAX_MULTISAMPLE_SAMPLE_COUNT]; - }; - - typedef std::unordered_map<DXGI_FORMAT, MultisampleSupportInfo, std::hash<int> > MultisampleSupportMap; - MultisampleSupportMap mMultisampleSupportMap; - - unsigned int mMaxSupportedSamples; - - // current render target states - unsigned int mAppliedRenderTargetSerials[gl::IMPLEMENTATION_MAX_DRAW_BUFFERS]; - unsigned int mAppliedDepthbufferSerial; - unsigned int mAppliedStencilbufferSerial; - bool mDepthStencilInitialized; - bool mRenderTargetDescInitialized; - rx::RenderTarget::Desc mRenderTargetDesc; - unsigned int mCurDepthSize; - unsigned int mCurStencilSize; - - // Currently applied sampler states - bool mForceSetVertexSamplerStates[gl::IMPLEMENTATION_MAX_VERTEX_TEXTURE_IMAGE_UNITS]; - gl::SamplerState mCurVertexSamplerStates[gl::IMPLEMENTATION_MAX_VERTEX_TEXTURE_IMAGE_UNITS]; - - bool mForceSetPixelSamplerStates[gl::MAX_TEXTURE_IMAGE_UNITS]; - gl::SamplerState mCurPixelSamplerStates[gl::MAX_TEXTURE_IMAGE_UNITS]; - - // Currently applied textures - unsigned int mCurVertexTextureSerials[gl::IMPLEMENTATION_MAX_VERTEX_TEXTURE_IMAGE_UNITS]; - unsigned int mCurPixelTextureSerials[gl::MAX_TEXTURE_IMAGE_UNITS]; - - // Currently applied blend state - bool mForceSetBlendState; - gl::BlendState mCurBlendState; - gl::Color mCurBlendColor; - unsigned int mCurSampleMask; - - // Currently applied rasterizer state - bool mForceSetRasterState; - gl::RasterizerState mCurRasterState; - - // Currently applied depth stencil state - bool mForceSetDepthStencilState; - gl::DepthStencilState mCurDepthStencilState; - int mCurStencilRef; - int mCurStencilBackRef; - - // Currently applied scissor rectangle - bool mForceSetScissor; - bool mScissorEnabled; - gl::Rectangle mCurScissor; - - // Currently applied viewport - bool mForceSetViewport; - gl::Rectangle mCurViewport; - float mCurNear; - float mCurFar; - - // Currently applied primitive topology - D3D11_PRIMITIVE_TOPOLOGY mCurrentPrimitiveTopology; - - unsigned int mAppliedIBSerial; - unsigned int mAppliedStorageIBSerial; - unsigned int mAppliedIBOffset; - - unsigned int mAppliedProgramBinarySerial; - bool mIsGeometryShaderActive; - - dx_VertexConstants mVertexConstants; - dx_VertexConstants mAppliedVertexConstants; - ID3D11Buffer *mDriverConstantBufferVS; - ID3D11Buffer *mCurrentVertexConstantBuffer; - - dx_PixelConstants mPixelConstants; - dx_PixelConstants mAppliedPixelConstants; - ID3D11Buffer *mDriverConstantBufferPS; - ID3D11Buffer *mCurrentPixelConstantBuffer; - - ID3D11Buffer *mCurrentGeometryConstantBuffer; - - // Vertex, index and input layouts - VertexDataManager *mVertexDataManager; - IndexDataManager *mIndexDataManager; - InputLayoutCache mInputLayoutCache; - - StreamingIndexBufferInterface *mLineLoopIB; - StreamingIndexBufferInterface *mTriangleFanIB; - - // Texture copy resources - bool mCopyResourcesInitialized; - ID3D11Buffer *mCopyVB; - ID3D11SamplerState *mCopySampler; - ID3D11InputLayout *mCopyIL; - ID3D11VertexShader *mCopyVS; - ID3D11PixelShader *mCopyRGBAPS; - ID3D11PixelShader *mCopyRGBPS; - ID3D11PixelShader *mCopyLumPS; - ID3D11PixelShader *mCopyLumAlphaPS; - - // Masked clear resources - bool mClearResourcesInitialized; - ID3D11Buffer *mClearVB; - ID3D11InputLayout *mClearIL; - ID3D11VertexShader *mClearVS; - ID3D11PixelShader *mClearSinglePS; - ID3D11PixelShader *mClearMultiplePS; - ID3D11RasterizerState *mClearScissorRS; - ID3D11RasterizerState *mClearNoScissorRS; - - // Sync query - ID3D11Query *mSyncQuery; - - ID3D11Device *mDevice; - D3D_FEATURE_LEVEL mFeatureLevel; - ID3D11DeviceContext *mDeviceContext; - IDXGIAdapter *mDxgiAdapter; - DXGI_ADAPTER_DESC mAdapterDescription; - char mDescription[128]; - IDXGIFactory *mDxgiFactory; - - // Cached device caps - bool mBGRATextureSupport; -}; - -} -#endif // LIBGLESV2_RENDERER_RENDERER11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/TextureStorage11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/TextureStorage11.cpp deleted file mode 100644 index 0c981ac503..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/TextureStorage11.cpp +++ /dev/null @@ -1,667 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// TextureStorage11.cpp: Implements the abstract rx::TextureStorage11 class and its concrete derived -// classes TextureStorage11_2D and TextureStorage11_Cube, which act as the interface to the D3D11 texture. - -#include "libGLESv2/renderer/d3d11/TextureStorage11.h" - -#include "libGLESv2/renderer/d3d11/Renderer11.h" -#include "libGLESv2/renderer/d3d11/RenderTarget11.h" -#include "libGLESv2/renderer/d3d11/SwapChain11.h" -#include "libGLESv2/renderer/d3d11/renderer11_utils.h" - -#include "libGLESv2/utilities.h" -#include "libGLESv2/main.h" - -namespace rx -{ - -TextureStorage11::TextureStorage11(Renderer *renderer, UINT bindFlags) - : mBindFlags(bindFlags), - mLodOffset(0), - mMipLevels(0), - mTexture(NULL), - mTextureFormat(DXGI_FORMAT_UNKNOWN), - mShaderResourceFormat(DXGI_FORMAT_UNKNOWN), - mRenderTargetFormat(DXGI_FORMAT_UNKNOWN), - mDepthStencilFormat(DXGI_FORMAT_UNKNOWN), - mSRV(NULL), - mTextureWidth(0), - mTextureHeight(0) -{ - mRenderer = Renderer11::makeRenderer11(renderer); -} - -TextureStorage11::~TextureStorage11() -{ -} - -TextureStorage11 *TextureStorage11::makeTextureStorage11(TextureStorage *storage) -{ - ASSERT(HAS_DYNAMIC_TYPE(TextureStorage11*, storage)); - return static_cast<TextureStorage11*>(storage); -} - -DWORD TextureStorage11::GetTextureBindFlags(DXGI_FORMAT format, GLenum glusage, bool forceRenderable) -{ - UINT bindFlags = D3D11_BIND_SHADER_RESOURCE; - - if (d3d11::IsDepthStencilFormat(format)) - { - bindFlags |= D3D11_BIND_DEPTH_STENCIL; - } - else if(forceRenderable || (TextureStorage11::IsTextureFormatRenderable(format) && (glusage == GL_FRAMEBUFFER_ATTACHMENT_ANGLE))) - { - bindFlags |= D3D11_BIND_RENDER_TARGET; - } - return bindFlags; -} - -bool TextureStorage11::IsTextureFormatRenderable(DXGI_FORMAT format) -{ - switch(format) - { - case DXGI_FORMAT_R8G8B8A8_UNORM: - case DXGI_FORMAT_A8_UNORM: - case DXGI_FORMAT_R32G32B32A32_FLOAT: - case DXGI_FORMAT_R16G16B16A16_FLOAT: - case DXGI_FORMAT_B8G8R8A8_UNORM: - case DXGI_FORMAT_R8_UNORM: - case DXGI_FORMAT_R8G8_UNORM: - case DXGI_FORMAT_R16_FLOAT: - case DXGI_FORMAT_R16G16_FLOAT: - return true; - case DXGI_FORMAT_BC1_UNORM: - case DXGI_FORMAT_BC2_UNORM: - case DXGI_FORMAT_BC3_UNORM: - case DXGI_FORMAT_R32G32B32_FLOAT: // not renderable on all devices - return false; - default: - UNREACHABLE(); - return false; - } -} - -UINT TextureStorage11::getBindFlags() const -{ - return mBindFlags; -} - -ID3D11Texture2D *TextureStorage11::getBaseTexture() const -{ - return mTexture; -} - -int TextureStorage11::getLodOffset() const -{ - return mLodOffset; -} - -bool TextureStorage11::isRenderTarget() const -{ - return (mBindFlags & (D3D11_BIND_RENDER_TARGET | D3D11_BIND_DEPTH_STENCIL)) != 0; -} - -bool TextureStorage11::isManaged() const -{ - return false; -} - -int TextureStorage11::levelCount() -{ - int levels = 0; - if (getBaseTexture()) - { - levels = mMipLevels - getLodOffset(); - } - return levels; -} - -UINT TextureStorage11::getSubresourceIndex(int level, int faceIndex) -{ - UINT index = 0; - if (getBaseTexture()) - { - index = D3D11CalcSubresource(level, faceIndex, mMipLevels); - } - return index; -} - -bool TextureStorage11::updateSubresourceLevel(ID3D11Texture2D *srcTexture, unsigned int sourceSubresource, - int level, int face, GLint xoffset, GLint yoffset, - GLsizei width, GLsizei height) -{ - if (srcTexture) - { - // Round up the width and height to the nearest multiple of dimension alignment - unsigned int dimensionAlignment = d3d11::GetTextureFormatDimensionAlignment(mTextureFormat); - width = width + dimensionAlignment - 1 - (width - 1) % dimensionAlignment; - height = height + dimensionAlignment - 1 - (height - 1) % dimensionAlignment; - - D3D11_BOX srcBox; - srcBox.left = xoffset; - srcBox.top = yoffset; - srcBox.right = xoffset + width; - srcBox.bottom = yoffset + height; - srcBox.front = 0; - srcBox.back = 1; - - ID3D11DeviceContext *context = mRenderer->getDeviceContext(); - - ASSERT(getBaseTexture()); - context->CopySubresourceRegion(getBaseTexture(), getSubresourceIndex(level + mLodOffset, face), - xoffset, yoffset, 0, srcTexture, sourceSubresource, &srcBox); - return true; - } - - return false; -} - -void TextureStorage11::generateMipmapLayer(RenderTarget11 *source, RenderTarget11 *dest) -{ - if (source && dest) - { - ID3D11ShaderResourceView *sourceSRV = source->getShaderResourceView(); - ID3D11RenderTargetView *destRTV = dest->getRenderTargetView(); - - if (sourceSRV && destRTV) - { - gl::Rectangle sourceArea; - sourceArea.x = 0; - sourceArea.y = 0; - sourceArea.width = source->getWidth(); - sourceArea.height = source->getHeight(); - - gl::Rectangle destArea; - destArea.x = 0; - destArea.y = 0; - destArea.width = dest->getWidth(); - destArea.height = dest->getHeight(); - - mRenderer->copyTexture(sourceSRV, sourceArea, source->getWidth(), source->getHeight(), - destRTV, destArea, dest->getWidth(), dest->getHeight(), - GL_RGBA); - } - } -} - -TextureStorage11_2D::TextureStorage11_2D(Renderer *renderer, SwapChain11 *swapchain) - : TextureStorage11(renderer, D3D11_BIND_RENDER_TARGET | D3D11_BIND_SHADER_RESOURCE) -{ - mTexture = swapchain->getOffscreenTexture(); - mSRV = swapchain->getRenderTargetShaderResource(); - - for (unsigned int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) - { - mRenderTarget[i] = NULL; - } - - D3D11_TEXTURE2D_DESC texDesc; - mTexture->GetDesc(&texDesc); - mMipLevels = texDesc.MipLevels; - mTextureFormat = texDesc.Format; - mTextureWidth = texDesc.Width; - mTextureHeight = texDesc.Height; - - D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; - mSRV->GetDesc(&srvDesc); - mShaderResourceFormat = srvDesc.Format; - - ID3D11RenderTargetView* offscreenRTV = swapchain->getRenderTarget(); - D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; - offscreenRTV->GetDesc(&rtvDesc); - mRenderTargetFormat = rtvDesc.Format; - offscreenRTV->Release(); - - mDepthStencilFormat = DXGI_FORMAT_UNKNOWN; -} - -TextureStorage11_2D::TextureStorage11_2D(Renderer *renderer, int levels, GLenum internalformat, GLenum usage, bool forceRenderable, GLsizei width, GLsizei height) - : TextureStorage11(renderer, GetTextureBindFlags(gl_d3d11::ConvertTextureFormat(internalformat, Renderer11::makeRenderer11(renderer)->getFeatureLevel()), usage, forceRenderable)) -{ - for (unsigned int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) - { - mRenderTarget[i] = NULL; - } - - DXGI_FORMAT convertedFormat = gl_d3d11::ConvertTextureFormat(internalformat, mRenderer->getFeatureLevel()); - if (d3d11::IsDepthStencilFormat(convertedFormat)) - { - mTextureFormat = d3d11::GetDepthTextureFormat(convertedFormat); - mShaderResourceFormat = d3d11::GetDepthShaderResourceFormat(convertedFormat); - mDepthStencilFormat = convertedFormat; - mRenderTargetFormat = DXGI_FORMAT_UNKNOWN; - } - else - { - mTextureFormat = convertedFormat; - mShaderResourceFormat = convertedFormat; - mDepthStencilFormat = DXGI_FORMAT_UNKNOWN; - mRenderTargetFormat = convertedFormat; - } - - // if the width or height is not positive this should be treated as an incomplete texture - // we handle that here by skipping the d3d texture creation - if (width > 0 && height > 0) - { - // adjust size if needed for compressed textures - gl::MakeValidSize(false, gl::IsCompressed(internalformat), &width, &height, &mLodOffset); - - ID3D11Device *device = mRenderer->getDevice(); - - D3D11_TEXTURE2D_DESC desc; - desc.Width = width; // Compressed texture size constraints? - desc.Height = height; - desc.MipLevels = mRenderer->getFeatureLevel() >= D3D_FEATURE_LEVEL_10_0 ? ((levels > 0) ? levels + mLodOffset : 0) : 1; - desc.ArraySize = 1; - desc.Format = mTextureFormat; - desc.SampleDesc.Count = 1; - desc.SampleDesc.Quality = 0; - desc.Usage = D3D11_USAGE_DEFAULT; - desc.BindFlags = getBindFlags(); - desc.CPUAccessFlags = 0; - desc.MiscFlags = 0; - - HRESULT result = device->CreateTexture2D(&desc, NULL, &mTexture); - - // this can happen from windows TDR - if (d3d11::isDeviceLostError(result)) - { - mRenderer->notifyDeviceLost(); - gl::error(GL_OUT_OF_MEMORY); - } - else if (FAILED(result)) - { - ASSERT(result == E_OUTOFMEMORY); - ERR("Creating image failed."); - gl::error(GL_OUT_OF_MEMORY); - } - else - { - mTexture->GetDesc(&desc); - mMipLevels = desc.MipLevels; - mTextureWidth = desc.Width; - mTextureHeight = desc.Height; - } - } -} - -TextureStorage11_2D::~TextureStorage11_2D() -{ - if (mTexture) - { - mTexture->Release(); - mTexture = NULL; - } - - if (mSRV) - { - mSRV->Release(); - mSRV = NULL; - } - - for (unsigned int i = 0; i < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; i++) - { - delete mRenderTarget[i]; - mRenderTarget[i] = NULL; - } -} - -TextureStorage11_2D *TextureStorage11_2D::makeTextureStorage11_2D(TextureStorage *storage) -{ - ASSERT(HAS_DYNAMIC_TYPE(TextureStorage11_2D*, storage)); - return static_cast<TextureStorage11_2D*>(storage); -} - -RenderTarget *TextureStorage11_2D::getRenderTarget(int level) -{ - if (level >= 0 && level < static_cast<int>(mMipLevels)) - { - if (!mRenderTarget[level]) - { - ID3D11Device *device = mRenderer->getDevice(); - HRESULT result; - - D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; - srvDesc.Format = mShaderResourceFormat; - srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2D; - srvDesc.Texture2D.MostDetailedMip = level; - srvDesc.Texture2D.MipLevels = level ? 1 : -1; - - ID3D11ShaderResourceView *srv; - result = device->CreateShaderResourceView(mTexture, &srvDesc, &srv); - - if (result == E_OUTOFMEMORY) - { - return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); - } - ASSERT(SUCCEEDED(result)); - - if (mRenderTargetFormat != DXGI_FORMAT_UNKNOWN) - { - D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; - rtvDesc.Format = mRenderTargetFormat; - rtvDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2D; - rtvDesc.Texture2D.MipSlice = level; - - ID3D11RenderTargetView *rtv; - result = device->CreateRenderTargetView(mTexture, &rtvDesc, &rtv); - - if (result == E_OUTOFMEMORY) - { - srv->Release(); - return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); - } - ASSERT(SUCCEEDED(result)); - - // RenderTarget11 expects to be the owner of the resources it is given but TextureStorage11 - // also needs to keep a reference to the texture. - mTexture->AddRef(); - - mRenderTarget[level] = new RenderTarget11(mRenderer, rtv, mTexture, srv, - std::max(mTextureWidth >> level, 1U), - std::max(mTextureHeight >> level, 1U)); - } - else if (mDepthStencilFormat != DXGI_FORMAT_UNKNOWN) - { - D3D11_DEPTH_STENCIL_VIEW_DESC dsvDesc; - dsvDesc.Format = mDepthStencilFormat; - dsvDesc.ViewDimension = D3D11_DSV_DIMENSION_TEXTURE2D; - dsvDesc.Texture2D.MipSlice = level; - dsvDesc.Flags = 0; - - ID3D11DepthStencilView *dsv; - result = device->CreateDepthStencilView(mTexture, &dsvDesc, &dsv); - - if (result == E_OUTOFMEMORY) - { - srv->Release(); - return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); - } - ASSERT(SUCCEEDED(result)); - - // RenderTarget11 expects to be the owner of the resources it is given but TextureStorage11 - // also needs to keep a reference to the texture. - mTexture->AddRef(); - - mRenderTarget[level] = new RenderTarget11(mRenderer, dsv, mTexture, srv, - std::max(mTextureWidth >> level, 1U), - std::max(mTextureHeight >> level, 1U)); - } - else - { - UNREACHABLE(); - } - } - - return mRenderTarget[level]; - } - else - { - return NULL; - } -} - -ID3D11ShaderResourceView *TextureStorage11_2D::getSRV() -{ - if (!mSRV) - { - ID3D11Device *device = mRenderer->getDevice(); - - D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; - srvDesc.Format = mShaderResourceFormat; - srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2D; - srvDesc.Texture2D.MipLevels = (mMipLevels == 0 ? -1 : mMipLevels); - srvDesc.Texture2D.MostDetailedMip = 0; - - HRESULT result = device->CreateShaderResourceView(mTexture, &srvDesc, &mSRV); - - if (result == E_OUTOFMEMORY) - { - return gl::error(GL_OUT_OF_MEMORY, static_cast<ID3D11ShaderResourceView*>(NULL)); - } - ASSERT(SUCCEEDED(result)); - } - - return mSRV; -} - -void TextureStorage11_2D::generateMipmap(int level) -{ - RenderTarget11 *source = RenderTarget11::makeRenderTarget11(getRenderTarget(level - 1)); - RenderTarget11 *dest = RenderTarget11::makeRenderTarget11(getRenderTarget(level)); - - generateMipmapLayer(source, dest); -} - -TextureStorage11_Cube::TextureStorage11_Cube(Renderer *renderer, int levels, GLenum internalformat, GLenum usage, bool forceRenderable, int size) - : TextureStorage11(renderer, GetTextureBindFlags(gl_d3d11::ConvertTextureFormat(internalformat, Renderer11::makeRenderer11(renderer)->getFeatureLevel()), usage, forceRenderable)) -{ - for (unsigned int i = 0; i < 6; i++) - { - for (unsigned int j = 0; j < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; j++) - { - mRenderTarget[i][j] = NULL; - } - } - - DXGI_FORMAT convertedFormat = gl_d3d11::ConvertTextureFormat(internalformat, mRenderer->getFeatureLevel()); - if (d3d11::IsDepthStencilFormat(convertedFormat)) - { - mTextureFormat = d3d11::GetDepthTextureFormat(convertedFormat); - mShaderResourceFormat = d3d11::GetDepthShaderResourceFormat(convertedFormat); - mDepthStencilFormat = convertedFormat; - mRenderTargetFormat = DXGI_FORMAT_UNKNOWN; - } - else - { - mTextureFormat = convertedFormat; - mShaderResourceFormat = convertedFormat; - mDepthStencilFormat = DXGI_FORMAT_UNKNOWN; - mRenderTargetFormat = convertedFormat; - } - - // if the size is not positive this should be treated as an incomplete texture - // we handle that here by skipping the d3d texture creation - if (size > 0) - { - // adjust size if needed for compressed textures - int height = size; - gl::MakeValidSize(false, gl::IsCompressed(internalformat), &size, &height, &mLodOffset); - - ID3D11Device *device = mRenderer->getDevice(); - - D3D11_TEXTURE2D_DESC desc; - desc.Width = size; - desc.Height = size; - desc.MipLevels = (levels > 0) ? levels + mLodOffset : 0; - desc.ArraySize = 6; - desc.Format = mTextureFormat; - desc.SampleDesc.Count = 1; - desc.SampleDesc.Quality = 0; - desc.Usage = D3D11_USAGE_DEFAULT; - desc.BindFlags = getBindFlags(); - desc.CPUAccessFlags = 0; - desc.MiscFlags = D3D11_RESOURCE_MISC_TEXTURECUBE; - - HRESULT result = device->CreateTexture2D(&desc, NULL, &mTexture); - - if (FAILED(result)) - { - ASSERT(result == E_OUTOFMEMORY); - ERR("Creating image failed."); - gl::error(GL_OUT_OF_MEMORY); - } - else - { - mTexture->GetDesc(&desc); - mMipLevels = desc.MipLevels; - mTextureWidth = desc.Width; - mTextureHeight = desc.Height; - } - } -} - -TextureStorage11_Cube::~TextureStorage11_Cube() -{ - if (mTexture) - { - mTexture->Release(); - mTexture = NULL; - } - - if (mSRV) - { - mSRV->Release(); - mSRV = NULL; - } - - for (unsigned int i = 0; i < 6; i++) - { - for (unsigned int j = 0; j < gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS; j++) - { - delete mRenderTarget[i][j]; - mRenderTarget[i][j] = NULL; - } - } -} - -TextureStorage11_Cube *TextureStorage11_Cube::makeTextureStorage11_Cube(TextureStorage *storage) -{ - ASSERT(HAS_DYNAMIC_TYPE(TextureStorage11_Cube*, storage)); - return static_cast<TextureStorage11_Cube*>(storage); -} - -RenderTarget *TextureStorage11_Cube::getRenderTarget(GLenum faceTarget, int level) -{ - unsigned int faceIdx = gl::TextureCubeMap::faceIndex(faceTarget); - if (level >= 0 && level < static_cast<int>(mMipLevels)) - { - if (!mRenderTarget[faceIdx][level]) - { - ID3D11Device *device = mRenderer->getDevice(); - HRESULT result; - - D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; - srvDesc.Format = mShaderResourceFormat; - srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2DARRAY; // Will be used with Texture2D sampler, not TextureCube - srvDesc.Texture2DArray.MostDetailedMip = level; - srvDesc.Texture2DArray.MipLevels = 1; - srvDesc.Texture2DArray.FirstArraySlice = faceIdx; - srvDesc.Texture2DArray.ArraySize = 1; - - ID3D11ShaderResourceView *srv; - result = device->CreateShaderResourceView(mTexture, &srvDesc, &srv); - - if (result == E_OUTOFMEMORY) - { - return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); - } - ASSERT(SUCCEEDED(result)); - - if (mRenderTargetFormat != DXGI_FORMAT_UNKNOWN) - { - D3D11_RENDER_TARGET_VIEW_DESC rtvDesc; - rtvDesc.Format = mRenderTargetFormat; - rtvDesc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2DARRAY; - rtvDesc.Texture2DArray.MipSlice = level; - rtvDesc.Texture2DArray.FirstArraySlice = faceIdx; - rtvDesc.Texture2DArray.ArraySize = 1; - - ID3D11RenderTargetView *rtv; - result = device->CreateRenderTargetView(mTexture, &rtvDesc, &rtv); - - if (result == E_OUTOFMEMORY) - { - srv->Release(); - return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); - } - ASSERT(SUCCEEDED(result)); - - // RenderTarget11 expects to be the owner of the resources it is given but TextureStorage11 - // also needs to keep a reference to the texture. - mTexture->AddRef(); - - mRenderTarget[faceIdx][level] = new RenderTarget11(mRenderer, rtv, mTexture, srv, - std::max(mTextureWidth >> level, 1U), - std::max(mTextureHeight >> level, 1U)); - } - else if (mDepthStencilFormat != DXGI_FORMAT_UNKNOWN) - { - D3D11_DEPTH_STENCIL_VIEW_DESC dsvDesc; - dsvDesc.Format = mRenderTargetFormat; - dsvDesc.ViewDimension = D3D11_DSV_DIMENSION_TEXTURE2DARRAY; - dsvDesc.Texture2DArray.MipSlice = level; - dsvDesc.Texture2DArray.FirstArraySlice = faceIdx; - dsvDesc.Texture2DArray.ArraySize = 1; - - ID3D11DepthStencilView *dsv; - result = device->CreateDepthStencilView(mTexture, &dsvDesc, &dsv); - - if (result == E_OUTOFMEMORY) - { - srv->Release(); - return gl::error(GL_OUT_OF_MEMORY, static_cast<RenderTarget*>(NULL)); - } - ASSERT(SUCCEEDED(result)); - - // RenderTarget11 expects to be the owner of the resources it is given but TextureStorage11 - // also needs to keep a reference to the texture. - mTexture->AddRef(); - - mRenderTarget[faceIdx][level] = new RenderTarget11(mRenderer, dsv, mTexture, srv, - std::max(mTextureWidth >> level, 1U), - std::max(mTextureHeight >> level, 1U)); - } - else - { - UNREACHABLE(); - } - } - - return mRenderTarget[faceIdx][level]; - } - else - { - return NULL; - } -} - -ID3D11ShaderResourceView *TextureStorage11_Cube::getSRV() -{ - if (!mSRV) - { - ID3D11Device *device = mRenderer->getDevice(); - - D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc; - srvDesc.Format = mShaderResourceFormat; - srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURECUBE; - srvDesc.TextureCube.MipLevels = (mMipLevels == 0 ? -1 : mMipLevels); - srvDesc.TextureCube.MostDetailedMip = 0; - - HRESULT result = device->CreateShaderResourceView(mTexture, &srvDesc, &mSRV); - - if (result == E_OUTOFMEMORY) - { - return gl::error(GL_OUT_OF_MEMORY, static_cast<ID3D11ShaderResourceView*>(NULL)); - } - ASSERT(SUCCEEDED(result)); - } - - return mSRV; -} - -void TextureStorage11_Cube::generateMipmap(int face, int level) -{ - RenderTarget11 *source = RenderTarget11::makeRenderTarget11(getRenderTarget(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face, level - 1)); - RenderTarget11 *dest = RenderTarget11::makeRenderTarget11(getRenderTarget(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face, level)); - - generateMipmapLayer(source, dest); -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/TextureStorage11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/TextureStorage11.h deleted file mode 100644 index 3c5ded05b8..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/TextureStorage11.h +++ /dev/null @@ -1,120 +0,0 @@ -// -// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// TextureStorage11.h: Defines the abstract rx::TextureStorage11 class and its concrete derived -// classes TextureStorage11_2D and TextureStorage11_Cube, which act as the interface to the D3D11 texture. - -#ifndef LIBGLESV2_RENDERER_TEXTURESTORAGE11_H_ -#define LIBGLESV2_RENDERER_TEXTURESTORAGE11_H_ - -#include "libGLESv2/Texture.h" -#include "libGLESv2/renderer/TextureStorage.h" - -namespace rx -{ -class RenderTarget; -class RenderTarget11; -class Renderer; -class Renderer11; -class SwapChain11; - -class TextureStorage11 : public TextureStorage -{ - public: - TextureStorage11(Renderer *renderer, UINT bindFlags); - virtual ~TextureStorage11(); - - static TextureStorage11 *makeTextureStorage11(TextureStorage *storage); - - static DWORD GetTextureBindFlags(DXGI_FORMAT d3dfmt, GLenum glusage, bool forceRenderable); - static bool IsTextureFormatRenderable(DXGI_FORMAT format); - - UINT getBindFlags() const; - - virtual ID3D11Texture2D *getBaseTexture() const; - virtual ID3D11ShaderResourceView *getSRV() = 0; - virtual RenderTarget *getRenderTarget() { return getRenderTarget(0); } - virtual RenderTarget *getRenderTarget(int level) { return NULL; } - virtual RenderTarget *getRenderTarget(GLenum faceTarget) { return getRenderTarget(faceTarget, 0); } - virtual RenderTarget *getRenderTarget(GLenum faceTarget, int level) { return NULL; } - - virtual void generateMipmap(int level) {}; - virtual void generateMipmap(int face, int level) {}; - - virtual int getLodOffset() const; - virtual bool isRenderTarget() const; - virtual bool isManaged() const; - virtual int levelCount(); - UINT getSubresourceIndex(int level, int faceTarget); - - bool updateSubresourceLevel(ID3D11Texture2D *texture, unsigned int sourceSubresource, int level, - int faceTarget, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); - - protected: - void generateMipmapLayer(RenderTarget11 *source, RenderTarget11 *dest); - - Renderer11 *mRenderer; - int mLodOffset; - unsigned int mMipLevels; - - ID3D11Texture2D *mTexture; - DXGI_FORMAT mTextureFormat; - DXGI_FORMAT mShaderResourceFormat; - DXGI_FORMAT mRenderTargetFormat; - DXGI_FORMAT mDepthStencilFormat; - unsigned int mTextureWidth; - unsigned int mTextureHeight; - - ID3D11ShaderResourceView *mSRV; - - private: - DISALLOW_COPY_AND_ASSIGN(TextureStorage11); - - const UINT mBindFlags; -}; - -class TextureStorage11_2D : public TextureStorage11 -{ - public: - TextureStorage11_2D(Renderer *renderer, SwapChain11 *swapchain); - TextureStorage11_2D(Renderer *renderer, int levels, GLenum internalformat, GLenum usage, bool forceRenderable, GLsizei width, GLsizei height); - virtual ~TextureStorage11_2D(); - - static TextureStorage11_2D *makeTextureStorage11_2D(TextureStorage *storage); - - virtual ID3D11ShaderResourceView *getSRV(); - virtual RenderTarget *getRenderTarget(int level); - - virtual void generateMipmap(int level); - - private: - DISALLOW_COPY_AND_ASSIGN(TextureStorage11_2D); - - RenderTarget11 *mRenderTarget[gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; -}; - -class TextureStorage11_Cube : public TextureStorage11 -{ - public: - TextureStorage11_Cube(Renderer *renderer, int levels, GLenum internalformat, GLenum usage, bool forceRenderable, int size); - virtual ~TextureStorage11_Cube(); - - static TextureStorage11_Cube *makeTextureStorage11_Cube(TextureStorage *storage); - - virtual ID3D11ShaderResourceView *getSRV(); - virtual RenderTarget *getRenderTarget(GLenum faceTarget, int level); - - virtual void generateMipmap(int face, int level); - - private: - DISALLOW_COPY_AND_ASSIGN(TextureStorage11_Cube); - - RenderTarget11 *mRenderTarget[6][gl::IMPLEMENTATION_MAX_TEXTURE_LEVELS]; -}; - -} - -#endif // LIBGLESV2_RENDERER_TEXTURESTORAGE11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/VertexBuffer11.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/VertexBuffer11.cpp deleted file mode 100644 index 6f9b4181f1..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/VertexBuffer11.cpp +++ /dev/null @@ -1,440 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// VertexBuffer11.cpp: Defines the D3D11 VertexBuffer implementation. - -#include "libGLESv2/renderer/d3d11/VertexBuffer11.h" -#include "libGLESv2/renderer/BufferStorage.h" - -#include "libGLESv2/Buffer.h" -#include "libGLESv2/renderer/d3d11/Renderer11.h" -#include "libGLESv2/Context.h" - -namespace rx -{ - -VertexBuffer11::VertexBuffer11(rx::Renderer11 *const renderer) : mRenderer(renderer) -{ - mBuffer = NULL; - mBufferSize = 0; - mDynamicUsage = false; -} - -VertexBuffer11::~VertexBuffer11() -{ - if (mBuffer) - { - mBuffer->Release(); - mBuffer = NULL; - } -} - -bool VertexBuffer11::initialize(unsigned int size, bool dynamicUsage) -{ - if (mBuffer) - { - mBuffer->Release(); - mBuffer = NULL; - } - - updateSerial(); - - if (size > 0) - { - ID3D11Device* dxDevice = mRenderer->getDevice(); - - D3D11_BUFFER_DESC bufferDesc; - bufferDesc.ByteWidth = size; - bufferDesc.Usage = D3D11_USAGE_DYNAMIC; - bufferDesc.BindFlags = D3D11_BIND_VERTEX_BUFFER; - bufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; - bufferDesc.MiscFlags = 0; - bufferDesc.StructureByteStride = 0; - - HRESULT result = dxDevice->CreateBuffer(&bufferDesc, NULL, &mBuffer); - if (FAILED(result)) - { - return false; - } - } - - mBufferSize = size; - mDynamicUsage = dynamicUsage; - return true; -} - -VertexBuffer11 *VertexBuffer11::makeVertexBuffer11(VertexBuffer *vetexBuffer) -{ - ASSERT(HAS_DYNAMIC_TYPE(VertexBuffer11*, vetexBuffer)); - return static_cast<VertexBuffer11*>(vetexBuffer); -} - -bool VertexBuffer11::storeVertexAttributes(const gl::VertexAttribute &attrib, GLint start, GLsizei count, - GLsizei instances, unsigned int offset) -{ - if (mBuffer) - { - gl::Buffer *buffer = attrib.mBoundBuffer.get(); - - int inputStride = attrib.stride(); - const VertexConverter &converter = getVertexConversion(attrib); - - ID3D11DeviceContext *dxContext = mRenderer->getDeviceContext(); - - D3D11_MAPPED_SUBRESOURCE mappedResource; - HRESULT result = dxContext->Map(mBuffer, 0, D3D11_MAP_WRITE_NO_OVERWRITE, 0, &mappedResource); - if (FAILED(result)) - { - ERR("Vertex buffer map failed with error 0x%08x", result); - return false; - } - - char* output = reinterpret_cast<char*>(mappedResource.pData) + offset; - - const char *input = NULL; - if (buffer) - { - BufferStorage *storage = buffer->getStorage(); - input = static_cast<const char*>(storage->getData()) + static_cast<int>(attrib.mOffset); - } - else - { - input = static_cast<const char*>(attrib.mPointer); - } - - if (instances == 0 || attrib.mDivisor == 0) - { - input += inputStride * start; - } - - converter.conversionFunc(input, inputStride, count, output); - - dxContext->Unmap(mBuffer, 0); - - return true; - } - else - { - ERR("Vertex buffer not initialized."); - return false; - } -} - -bool VertexBuffer11::storeRawData(const void* data, unsigned int size, unsigned int offset) -{ - if (mBuffer) - { - ID3D11DeviceContext *dxContext = mRenderer->getDeviceContext(); - - D3D11_MAPPED_SUBRESOURCE mappedResource; - HRESULT result = dxContext->Map(mBuffer, 0, D3D11_MAP_WRITE_NO_OVERWRITE, 0, &mappedResource); - if (FAILED(result)) - { - ERR("Vertex buffer map failed with error 0x%08x", result); - return false; - } - - char* bufferData = static_cast<char*>(mappedResource.pData); - memcpy(bufferData + offset, data, size); - - dxContext->Unmap(mBuffer, 0); - - return true; - } - else - { - ERR("Vertex buffer not initialized."); - return false; - } -} - -bool VertexBuffer11::getSpaceRequired(const gl::VertexAttribute &attrib, GLsizei count, - GLsizei instances, unsigned int *outSpaceRequired) const -{ - unsigned int elementSize = getVertexConversion(attrib).outputElementSize; - - unsigned int elementCount = 0; - if (instances == 0 || attrib.mDivisor == 0) - { - elementCount = count; - } - else - { - if (static_cast<unsigned int>(instances) < std::numeric_limits<unsigned int>::max() - (attrib.mDivisor - 1)) - { - // Round up - elementCount = (static_cast<unsigned int>(instances) + (attrib.mDivisor - 1)) / attrib.mDivisor; - } - else - { - elementCount = instances / attrib.mDivisor; - } - } - - if (elementSize <= std::numeric_limits<unsigned int>::max() / elementCount) - { - if (outSpaceRequired) - { - *outSpaceRequired = elementSize * elementCount; - } - return true; - } - else - { - return false; - } -} - -bool VertexBuffer11::requiresConversion(const gl::VertexAttribute &attrib) const -{ - return !getVertexConversion(attrib).identity; -} - -unsigned int VertexBuffer11::getBufferSize() const -{ - return mBufferSize; -} - -bool VertexBuffer11::setBufferSize(unsigned int size) -{ - if (size > mBufferSize) - { - return initialize(size, mDynamicUsage); - } - else - { - return true; - } -} - -bool VertexBuffer11::discard() -{ - if (mBuffer) - { - ID3D11DeviceContext *dxContext = mRenderer->getDeviceContext(); - - D3D11_MAPPED_SUBRESOURCE mappedResource; - HRESULT result = dxContext->Map(mBuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); - if (FAILED(result)) - { - ERR("Vertex buffer map failed with error 0x%08x", result); - return false; - } - - dxContext->Unmap(mBuffer, 0); - - return true; - } - else - { - ERR("Vertex buffer not initialized."); - return false; - } -} - -unsigned int VertexBuffer11::getVertexSize(const gl::VertexAttribute &attrib) const -{ - return getVertexConversion(attrib).outputElementSize; -} - -DXGI_FORMAT VertexBuffer11::getDXGIFormat(const gl::VertexAttribute &attrib) const -{ - return getVertexConversion(attrib).dxgiFormat; -} - -ID3D11Buffer *VertexBuffer11::getBuffer() const -{ - return mBuffer; -} - -template <typename T, unsigned int componentCount, bool widen, bool normalized> -static void copyVertexData(const void *input, unsigned int stride, unsigned int count, void *output) -{ - unsigned int attribSize = sizeof(T) * componentCount; - - if (attribSize == stride && !widen) - { - memcpy(output, input, count * attribSize); - } - else - { - unsigned int outputStride = widen ? 4 : componentCount; - T defaultVal = normalized ? std::numeric_limits<T>::max() : T(1); - - for (unsigned int i = 0; i < count; i++) - { - const T *offsetInput = reinterpret_cast<const T*>(reinterpret_cast<const char*>(input) + i * stride); - T *offsetOutput = reinterpret_cast<T*>(output) + i * outputStride; - - for (unsigned int j = 0; j < componentCount; j++) - { - offsetOutput[j] = offsetInput[j]; - } - - if (widen) - { - offsetOutput[3] = defaultVal; - } - } - } -} - -template <unsigned int componentCount> -static void copyFixedVertexData(const void* input, unsigned int stride, unsigned int count, void* output) -{ - static const float divisor = 1.0f / (1 << 16); - - for (unsigned int i = 0; i < count; i++) - { - const GLfixed* offsetInput = reinterpret_cast<const GLfixed*>(reinterpret_cast<const char*>(input) + stride * i); - float* offsetOutput = reinterpret_cast<float*>(output) + i * componentCount; - - for (unsigned int j = 0; j < componentCount; j++) - { - offsetOutput[j] = static_cast<float>(offsetInput[j]) * divisor; - } - } -} - -template <typename T, unsigned int componentCount, bool normalized> -static void copyToFloatVertexData(const void* input, unsigned int stride, unsigned int count, void* output) -{ - typedef std::numeric_limits<T> NL; - - for (unsigned int i = 0; i < count; i++) - { - const T *offsetInput = reinterpret_cast<const T*>(reinterpret_cast<const char*>(input) + stride * i); - float *offsetOutput = reinterpret_cast<float*>(output) + i * componentCount; - - for (unsigned int j = 0; j < componentCount; j++) - { - if (normalized) - { - if (NL::is_signed) - { - const float divisor = 1.0f / (2 * static_cast<float>(NL::max()) + 1); - offsetOutput[j] = (2 * static_cast<float>(offsetInput[j]) + 1) * divisor; - } - else - { - offsetOutput[j] = static_cast<float>(offsetInput[j]) / NL::max(); - } - } - else - { - offsetOutput[j] = static_cast<float>(offsetInput[j]); - } - } - } -} - -const VertexBuffer11::VertexConverter VertexBuffer11::mPossibleTranslations[NUM_GL_VERTEX_ATTRIB_TYPES][2][4] = -{ - { // GL_BYTE - { // unnormalized - { ©ToFloatVertexData<GLbyte, 1, false>, false, DXGI_FORMAT_R32_FLOAT, 4 }, - { ©ToFloatVertexData<GLbyte, 2, false>, false, DXGI_FORMAT_R32G32_FLOAT, 8 }, - { ©ToFloatVertexData<GLbyte, 3, false>, false, DXGI_FORMAT_R32G32B32_FLOAT, 12 }, - { ©ToFloatVertexData<GLbyte, 4, false>, false, DXGI_FORMAT_R32G32B32A32_FLOAT, 16 }, - }, - { // normalized - { ©VertexData<GLbyte, 1, false, true>, true, DXGI_FORMAT_R8_SNORM, 1 }, - { ©VertexData<GLbyte, 2, false, true>, true, DXGI_FORMAT_R8G8_SNORM, 2 }, - { ©VertexData<GLbyte, 3, true, true>, false, DXGI_FORMAT_R8G8B8A8_SNORM, 4 }, - { ©VertexData<GLbyte, 4, false, true>, true, DXGI_FORMAT_R8G8B8A8_SNORM, 4 }, - }, - }, - { // GL_UNSIGNED_BYTE - { // unnormalized - { ©ToFloatVertexData<GLubyte, 1, false>, false, DXGI_FORMAT_R32_FLOAT, 4 }, - { ©ToFloatVertexData<GLubyte, 2, false>, false, DXGI_FORMAT_R32G32_FLOAT, 8 }, - { ©ToFloatVertexData<GLubyte, 3, false>, false, DXGI_FORMAT_R32G32B32_FLOAT, 12 }, - { ©ToFloatVertexData<GLubyte, 4, false>, false, DXGI_FORMAT_R32G32B32A32_FLOAT, 16 }, - }, - { // normalized - { ©VertexData<GLubyte, 1, false, true>, true, DXGI_FORMAT_R8_UNORM, 1 }, - { ©VertexData<GLubyte, 2, false, true>, true, DXGI_FORMAT_R8G8_UNORM, 2 }, - { ©VertexData<GLubyte, 3, true, true>, false, DXGI_FORMAT_R8G8B8A8_UNORM, 4 }, - { ©VertexData<GLubyte, 4, false, true>, true, DXGI_FORMAT_R8G8B8A8_UNORM, 4 }, - }, - }, - { // GL_SHORT - { // unnormalized - { ©ToFloatVertexData<GLshort, 1, false>, false, DXGI_FORMAT_R32_FLOAT, 4 }, - { ©ToFloatVertexData<GLshort, 2, false>, false, DXGI_FORMAT_R32G32_FLOAT, 8 }, - { ©ToFloatVertexData<GLshort, 3, false>, false, DXGI_FORMAT_R32G32B32_FLOAT, 12 }, - { ©ToFloatVertexData<GLshort, 4, false>, false, DXGI_FORMAT_R32G32B32A32_FLOAT, 16 }, - }, - { // normalized - { ©VertexData<GLshort, 1, false, true>, true, DXGI_FORMAT_R16_SNORM, 2 }, - { ©VertexData<GLshort, 2, false, true>, true, DXGI_FORMAT_R16G16_SNORM, 4 }, - { ©VertexData<GLshort, 3, true, true>, false, DXGI_FORMAT_R16G16B16A16_SNORM, 8 }, - { ©VertexData<GLshort, 4, false, true>, true, DXGI_FORMAT_R16G16B16A16_SNORM, 8 }, - }, - }, - { // GL_UNSIGNED_SHORT - { // unnormalized - { ©ToFloatVertexData<GLushort, 1, false>, false, DXGI_FORMAT_R32_FLOAT, 4 }, - { ©ToFloatVertexData<GLushort, 2, false>, false, DXGI_FORMAT_R32G32_FLOAT, 8 }, - { ©ToFloatVertexData<GLushort, 3, false>, false, DXGI_FORMAT_R32G32B32_FLOAT, 12 }, - { ©ToFloatVertexData<GLushort, 4, false>, false, DXGI_FORMAT_R32G32B32A32_FLOAT, 16 }, - }, - { // normalized - { ©VertexData<GLushort, 1, false, true>, true, DXGI_FORMAT_R16_UNORM, 2 }, - { ©VertexData<GLushort, 2, false, true>, true, DXGI_FORMAT_R16G16_UNORM, 4 }, - { ©VertexData<GLushort, 3, true, true>, false, DXGI_FORMAT_R16G16B16A16_UNORM, 8 }, - { ©VertexData<GLushort, 4, false, true>, true, DXGI_FORMAT_R16G16B16A16_UNORM, 8 }, - }, - }, - { // GL_FIXED - { // unnormalized - { ©FixedVertexData<1>, false, DXGI_FORMAT_R32_FLOAT, 4 }, - { ©FixedVertexData<2>, false, DXGI_FORMAT_R32G32_FLOAT, 8 }, - { ©FixedVertexData<3>, false, DXGI_FORMAT_R32G32B32_FLOAT, 12 }, - { ©FixedVertexData<4>, false, DXGI_FORMAT_R32G32B32A32_FLOAT, 16 }, - }, - { // normalized - { ©FixedVertexData<1>, false, DXGI_FORMAT_R32_FLOAT, 4 }, - { ©FixedVertexData<2>, false, DXGI_FORMAT_R32G32_FLOAT, 8 }, - { ©FixedVertexData<3>, false, DXGI_FORMAT_R32G32B32_FLOAT, 12 }, - { ©FixedVertexData<4>, false, DXGI_FORMAT_R32G32B32A32_FLOAT, 16 }, - }, - }, - { // GL_FLOAT - { // unnormalized - { ©VertexData<GLfloat, 1, false, false>, true, DXGI_FORMAT_R32_FLOAT, 4 }, - { ©VertexData<GLfloat, 2, false, false>, true, DXGI_FORMAT_R32G32_FLOAT, 8 }, - { ©VertexData<GLfloat, 3, false, false>, true, DXGI_FORMAT_R32G32B32_FLOAT, 12 }, - { ©VertexData<GLfloat, 4, false, false>, true, DXGI_FORMAT_R32G32B32A32_FLOAT, 16 }, - }, - { // normalized - { ©VertexData<GLfloat, 1, false, false>, true, DXGI_FORMAT_R32_FLOAT, 4 }, - { ©VertexData<GLfloat, 2, false, false>, true, DXGI_FORMAT_R32G32_FLOAT, 8 }, - { ©VertexData<GLfloat, 3, false, false>, true, DXGI_FORMAT_R32G32B32_FLOAT, 12 }, - { ©VertexData<GLfloat, 4, false, false>, true, DXGI_FORMAT_R32G32B32A32_FLOAT, 16 }, - }, - }, -}; - -const VertexBuffer11::VertexConverter &VertexBuffer11::getVertexConversion(const gl::VertexAttribute &attribute) -{ - unsigned int typeIndex = 0; - switch (attribute.mType) - { - case GL_BYTE: typeIndex = 0; break; - case GL_UNSIGNED_BYTE: typeIndex = 1; break; - case GL_SHORT: typeIndex = 2; break; - case GL_UNSIGNED_SHORT: typeIndex = 3; break; - case GL_FIXED: typeIndex = 4; break; - case GL_FLOAT: typeIndex = 5; break; - default: UNREACHABLE(); break; - } - - return mPossibleTranslations[typeIndex][attribute.mNormalized ? 1 : 0][attribute.mSize - 1]; -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/VertexBuffer11.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/VertexBuffer11.h deleted file mode 100644 index eceb426e82..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/VertexBuffer11.h +++ /dev/null @@ -1,74 +0,0 @@ -// -// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// VertexBuffer11.h: Defines the D3D11 VertexBuffer implementation. - -#ifndef LIBGLESV2_RENDERER_VERTEXBUFFER11_H_ -#define LIBGLESV2_RENDERER_VERTEXBUFFER11_H_ - -#include "libGLESv2/renderer/VertexBuffer.h" - -namespace rx -{ -class Renderer11; - -class VertexBuffer11 : public VertexBuffer -{ - public: - explicit VertexBuffer11(rx::Renderer11 *const renderer); - virtual ~VertexBuffer11(); - - virtual bool initialize(unsigned int size, bool dynamicUsage); - - static VertexBuffer11 *makeVertexBuffer11(VertexBuffer *vetexBuffer); - - virtual bool storeVertexAttributes(const gl::VertexAttribute &attrib, GLint start, GLsizei count, GLsizei instances, - unsigned int offset); - virtual bool storeRawData(const void* data, unsigned int size, unsigned int offset); - - virtual bool getSpaceRequired(const gl::VertexAttribute &attrib, GLsizei count, GLsizei instances, - unsigned int *outSpaceRequired) const; - - virtual bool requiresConversion(const gl::VertexAttribute &attrib) const; - - virtual unsigned int getBufferSize() const; - virtual bool setBufferSize(unsigned int size); - virtual bool discard(); - - unsigned int getVertexSize(const gl::VertexAttribute &attrib) const; - DXGI_FORMAT getDXGIFormat(const gl::VertexAttribute &attrib) const; - - ID3D11Buffer *getBuffer() const; - - private: - DISALLOW_COPY_AND_ASSIGN(VertexBuffer11); - - rx::Renderer11 *const mRenderer; - - ID3D11Buffer *mBuffer; - unsigned int mBufferSize; - bool mDynamicUsage; - - typedef void (*VertexConversionFunction)(const void *, unsigned int, unsigned int, void *); - struct VertexConverter - { - VertexConversionFunction conversionFunc; - bool identity; - DXGI_FORMAT dxgiFormat; - unsigned int outputElementSize; - }; - - enum { NUM_GL_VERTEX_ATTRIB_TYPES = 6 }; - - // This table is used to generate mAttributeTypes. - static const VertexConverter mPossibleTranslations[NUM_GL_VERTEX_ATTRIB_TYPES][2][4]; // [GL types as enumerated by typeIndex()][normalized][size - 1] - - static const VertexConverter &getVertexConversion(const gl::VertexAttribute &attribute); -}; - -} - -#endif // LIBGLESV2_RENDERER_VERTEXBUFFER11_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/renderer11_utils.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/renderer11_utils.cpp deleted file mode 100644 index 34b8259a80..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/renderer11_utils.cpp +++ /dev/null @@ -1,688 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// renderer11_utils.cpp: Conversion functions and other utility routines -// specific to the D3D11 renderer. - -#include "libGLESv2/renderer/d3d11/renderer11_utils.h" - -#include "common/debug.h" - -namespace gl_d3d11 -{ - -D3D11_BLEND ConvertBlendFunc(GLenum glBlend, bool isAlpha) -{ - D3D11_BLEND d3dBlend = D3D11_BLEND_ZERO; - - switch (glBlend) - { - case GL_ZERO: d3dBlend = D3D11_BLEND_ZERO; break; - case GL_ONE: d3dBlend = D3D11_BLEND_ONE; break; - case GL_SRC_COLOR: d3dBlend = (isAlpha ? D3D11_BLEND_SRC_ALPHA : D3D11_BLEND_SRC_COLOR); break; - case GL_ONE_MINUS_SRC_COLOR: d3dBlend = (isAlpha ? D3D11_BLEND_INV_SRC_ALPHA : D3D11_BLEND_INV_SRC_COLOR); break; - case GL_DST_COLOR: d3dBlend = (isAlpha ? D3D11_BLEND_DEST_ALPHA : D3D11_BLEND_DEST_COLOR); break; - case GL_ONE_MINUS_DST_COLOR: d3dBlend = (isAlpha ? D3D11_BLEND_INV_DEST_ALPHA : D3D11_BLEND_INV_DEST_COLOR); break; - case GL_SRC_ALPHA: d3dBlend = D3D11_BLEND_SRC_ALPHA; break; - case GL_ONE_MINUS_SRC_ALPHA: d3dBlend = D3D11_BLEND_INV_SRC_ALPHA; break; - case GL_DST_ALPHA: d3dBlend = D3D11_BLEND_DEST_ALPHA; break; - case GL_ONE_MINUS_DST_ALPHA: d3dBlend = D3D11_BLEND_INV_DEST_ALPHA; break; - case GL_CONSTANT_COLOR: d3dBlend = D3D11_BLEND_BLEND_FACTOR; break; - case GL_ONE_MINUS_CONSTANT_COLOR: d3dBlend = D3D11_BLEND_INV_BLEND_FACTOR; break; - case GL_CONSTANT_ALPHA: d3dBlend = D3D11_BLEND_BLEND_FACTOR; break; - case GL_ONE_MINUS_CONSTANT_ALPHA: d3dBlend = D3D11_BLEND_INV_BLEND_FACTOR; break; - case GL_SRC_ALPHA_SATURATE: d3dBlend = D3D11_BLEND_SRC_ALPHA_SAT; break; - default: UNREACHABLE(); - } - - return d3dBlend; -} - -D3D11_BLEND_OP ConvertBlendOp(GLenum glBlendOp) -{ - D3D11_BLEND_OP d3dBlendOp = D3D11_BLEND_OP_ADD; - - switch (glBlendOp) - { - case GL_FUNC_ADD: d3dBlendOp = D3D11_BLEND_OP_ADD; break; - case GL_FUNC_SUBTRACT: d3dBlendOp = D3D11_BLEND_OP_SUBTRACT; break; - case GL_FUNC_REVERSE_SUBTRACT: d3dBlendOp = D3D11_BLEND_OP_REV_SUBTRACT; break; - default: UNREACHABLE(); - } - - return d3dBlendOp; -} - -UINT8 ConvertColorMask(bool red, bool green, bool blue, bool alpha) -{ - UINT8 mask = 0; - if (red) - { - mask |= D3D11_COLOR_WRITE_ENABLE_RED; - } - if (green) - { - mask |= D3D11_COLOR_WRITE_ENABLE_GREEN; - } - if (blue) - { - mask |= D3D11_COLOR_WRITE_ENABLE_BLUE; - } - if (alpha) - { - mask |= D3D11_COLOR_WRITE_ENABLE_ALPHA; - } - return mask; -} - -D3D11_CULL_MODE ConvertCullMode(bool cullEnabled, GLenum cullMode) -{ - D3D11_CULL_MODE cull = D3D11_CULL_NONE; - - if (cullEnabled) - { - switch (cullMode) - { - case GL_FRONT: cull = D3D11_CULL_FRONT; break; - case GL_BACK: cull = D3D11_CULL_BACK; break; - case GL_FRONT_AND_BACK: cull = D3D11_CULL_NONE; break; - default: UNREACHABLE(); - } - } - else - { - cull = D3D11_CULL_NONE; - } - - return cull; -} - -D3D11_COMPARISON_FUNC ConvertComparison(GLenum comparison) -{ - D3D11_COMPARISON_FUNC d3dComp = D3D11_COMPARISON_NEVER; - switch (comparison) - { - case GL_NEVER: d3dComp = D3D11_COMPARISON_NEVER; break; - case GL_ALWAYS: d3dComp = D3D11_COMPARISON_ALWAYS; break; - case GL_LESS: d3dComp = D3D11_COMPARISON_LESS; break; - case GL_LEQUAL: d3dComp = D3D11_COMPARISON_LESS_EQUAL; break; - case GL_EQUAL: d3dComp = D3D11_COMPARISON_EQUAL; break; - case GL_GREATER: d3dComp = D3D11_COMPARISON_GREATER; break; - case GL_GEQUAL: d3dComp = D3D11_COMPARISON_GREATER_EQUAL; break; - case GL_NOTEQUAL: d3dComp = D3D11_COMPARISON_NOT_EQUAL; break; - default: UNREACHABLE(); - } - - return d3dComp; -} - -D3D11_DEPTH_WRITE_MASK ConvertDepthMask(bool depthWriteEnabled) -{ - return depthWriteEnabled ? D3D11_DEPTH_WRITE_MASK_ALL : D3D11_DEPTH_WRITE_MASK_ZERO; -} - -UINT8 ConvertStencilMask(GLuint stencilmask) -{ - return static_cast<UINT8>(stencilmask); -} - -D3D11_STENCIL_OP ConvertStencilOp(GLenum stencilOp) -{ - D3D11_STENCIL_OP d3dStencilOp = D3D11_STENCIL_OP_KEEP; - - switch (stencilOp) - { - case GL_ZERO: d3dStencilOp = D3D11_STENCIL_OP_ZERO; break; - case GL_KEEP: d3dStencilOp = D3D11_STENCIL_OP_KEEP; break; - case GL_REPLACE: d3dStencilOp = D3D11_STENCIL_OP_REPLACE; break; - case GL_INCR: d3dStencilOp = D3D11_STENCIL_OP_INCR_SAT; break; - case GL_DECR: d3dStencilOp = D3D11_STENCIL_OP_DECR_SAT; break; - case GL_INVERT: d3dStencilOp = D3D11_STENCIL_OP_INVERT; break; - case GL_INCR_WRAP: d3dStencilOp = D3D11_STENCIL_OP_INCR; break; - case GL_DECR_WRAP: d3dStencilOp = D3D11_STENCIL_OP_DECR; break; - default: UNREACHABLE(); - } - - return d3dStencilOp; -} - -D3D11_FILTER ConvertFilter(GLenum minFilter, GLenum magFilter, float maxAnisotropy) -{ - if (maxAnisotropy > 1.0f) - { - return D3D11_ENCODE_ANISOTROPIC_FILTER(false); - } - else - { - D3D11_FILTER_TYPE dxMin = D3D11_FILTER_TYPE_POINT; - D3D11_FILTER_TYPE dxMip = D3D11_FILTER_TYPE_POINT; - switch (minFilter) - { - case GL_NEAREST: dxMin = D3D11_FILTER_TYPE_POINT; dxMip = D3D11_FILTER_TYPE_POINT; break; - case GL_LINEAR: dxMin = D3D11_FILTER_TYPE_LINEAR; dxMip = D3D11_FILTER_TYPE_POINT; break; - case GL_NEAREST_MIPMAP_NEAREST: dxMin = D3D11_FILTER_TYPE_POINT; dxMip = D3D11_FILTER_TYPE_POINT; break; - case GL_LINEAR_MIPMAP_NEAREST: dxMin = D3D11_FILTER_TYPE_LINEAR; dxMip = D3D11_FILTER_TYPE_POINT; break; - case GL_NEAREST_MIPMAP_LINEAR: dxMin = D3D11_FILTER_TYPE_POINT; dxMip = D3D11_FILTER_TYPE_LINEAR; break; - case GL_LINEAR_MIPMAP_LINEAR: dxMin = D3D11_FILTER_TYPE_LINEAR; dxMip = D3D11_FILTER_TYPE_LINEAR; break; - default: UNREACHABLE(); - } - - D3D11_FILTER_TYPE dxMag = D3D11_FILTER_TYPE_POINT; - switch (magFilter) - { - case GL_NEAREST: dxMag = D3D11_FILTER_TYPE_POINT; break; - case GL_LINEAR: dxMag = D3D11_FILTER_TYPE_LINEAR; break; - default: UNREACHABLE(); - } - - return D3D11_ENCODE_BASIC_FILTER(dxMin, dxMag, dxMip, false); - } -} - -D3D11_TEXTURE_ADDRESS_MODE ConvertTextureWrap(GLenum wrap) -{ - switch (wrap) - { - case GL_REPEAT: return D3D11_TEXTURE_ADDRESS_WRAP; - case GL_CLAMP_TO_EDGE: return D3D11_TEXTURE_ADDRESS_CLAMP; - case GL_MIRRORED_REPEAT: return D3D11_TEXTURE_ADDRESS_MIRROR; - default: UNREACHABLE(); - } - - return D3D11_TEXTURE_ADDRESS_WRAP; -} - -FLOAT ConvertMinLOD(GLenum minFilter, unsigned int lodOffset) -{ - return (minFilter == GL_NEAREST || minFilter == GL_LINEAR) ? static_cast<float>(lodOffset) : -FLT_MAX; -} - -FLOAT ConvertMaxLOD(GLenum minFilter, unsigned int lodOffset) -{ - return (minFilter == GL_NEAREST || minFilter == GL_LINEAR) ? static_cast<float>(lodOffset) : FLT_MAX; -} - -} - -namespace d3d11_gl -{ - -GLenum ConvertBackBufferFormat(DXGI_FORMAT format) -{ - switch (format) - { - case DXGI_FORMAT_R8G8B8A8_UNORM: return GL_RGBA8_OES; - case DXGI_FORMAT_B8G8R8A8_UNORM: return GL_BGRA8_EXT; - default: - UNREACHABLE(); - } - - return GL_RGBA8_OES; -} - -GLenum ConvertDepthStencilFormat(DXGI_FORMAT format) -{ - switch (format) - { - case DXGI_FORMAT_UNKNOWN: return GL_NONE; - case DXGI_FORMAT_D16_UNORM: return GL_DEPTH_COMPONENT16; - case DXGI_FORMAT_D24_UNORM_S8_UINT: return GL_DEPTH24_STENCIL8_OES; - default: - UNREACHABLE(); - } - - return GL_DEPTH24_STENCIL8_OES; -} - -GLenum ConvertRenderbufferFormat(DXGI_FORMAT format) -{ - switch (format) - { - case DXGI_FORMAT_B8G8R8A8_UNORM: - return GL_BGRA8_EXT; - case DXGI_FORMAT_R8G8B8A8_UNORM: - return GL_RGBA8_OES; - case DXGI_FORMAT_D16_UNORM: - return GL_DEPTH_COMPONENT16; - case DXGI_FORMAT_D24_UNORM_S8_UINT: - return GL_DEPTH24_STENCIL8_OES; - default: - UNREACHABLE(); - } - - return GL_RGBA8_OES; -} - -GLenum ConvertTextureInternalFormat(DXGI_FORMAT format) -{ - switch (format) - { - case DXGI_FORMAT_R8G8B8A8_UNORM: - return GL_RGBA8_OES; - case DXGI_FORMAT_A8_UNORM: - return GL_ALPHA8_EXT; - case DXGI_FORMAT_BC1_UNORM: - return GL_COMPRESSED_RGBA_S3TC_DXT1_EXT; - case DXGI_FORMAT_BC2_UNORM: - return GL_COMPRESSED_RGBA_S3TC_DXT3_ANGLE; - case DXGI_FORMAT_BC3_UNORM: - return GL_COMPRESSED_RGBA_S3TC_DXT5_ANGLE; - case DXGI_FORMAT_R32G32B32A32_FLOAT: - return GL_RGBA32F_EXT; - case DXGI_FORMAT_R32G32B32_FLOAT: - return GL_RGB32F_EXT; - case DXGI_FORMAT_R16G16B16A16_FLOAT: - return GL_RGBA16F_EXT; - case DXGI_FORMAT_B8G8R8A8_UNORM: - return GL_BGRA8_EXT; - case DXGI_FORMAT_R8_UNORM: - return GL_R8_EXT; - case DXGI_FORMAT_R8G8_UNORM: - return GL_RG8_EXT; - case DXGI_FORMAT_R16_FLOAT: - return GL_R16F_EXT; - case DXGI_FORMAT_R16G16_FLOAT: - return GL_RG16F_EXT; - case DXGI_FORMAT_D16_UNORM: - return GL_DEPTH_COMPONENT16; - case DXGI_FORMAT_D24_UNORM_S8_UINT: - return GL_DEPTH24_STENCIL8_OES; - case DXGI_FORMAT_UNKNOWN: - return GL_NONE; - default: - UNREACHABLE(); - } - - return GL_RGBA8_OES; -} - -} - -namespace gl_d3d11 -{ - -DXGI_FORMAT ConvertRenderbufferFormat(GLenum format) -{ - switch (format) - { - case GL_RGBA4: - case GL_RGB5_A1: - case GL_RGBA8_OES: - case GL_RGB565: - case GL_RGB8_OES: - return DXGI_FORMAT_R8G8B8A8_UNORM; - case GL_BGRA8_EXT: - return DXGI_FORMAT_B8G8R8A8_UNORM; - case GL_DEPTH_COMPONENT16: - return DXGI_FORMAT_D16_UNORM; - case GL_STENCIL_INDEX8: - case GL_DEPTH24_STENCIL8_OES: - return DXGI_FORMAT_D24_UNORM_S8_UINT; - default: - UNREACHABLE(); - } - - return DXGI_FORMAT_R8G8B8A8_UNORM; -} - -DXGI_FORMAT ConvertTextureFormat(GLenum internalformat, D3D_FEATURE_LEVEL featureLevel) -{ - switch (internalformat) - { - case GL_RGB565: - case GL_RGBA4: - case GL_RGB5_A1: - case GL_RGB8_OES: - case GL_RGBA8_OES: - case GL_LUMINANCE8_EXT: - case GL_LUMINANCE8_ALPHA8_EXT: - return DXGI_FORMAT_R8G8B8A8_UNORM; - case GL_ALPHA8_EXT: - return featureLevel >= D3D_FEATURE_LEVEL_10_0 ? DXGI_FORMAT_A8_UNORM : DXGI_FORMAT_B8G8R8A8_UNORM; - case GL_COMPRESSED_RGB_S3TC_DXT1_EXT: - case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT: - return DXGI_FORMAT_BC1_UNORM; - case GL_COMPRESSED_RGBA_S3TC_DXT3_ANGLE: - return DXGI_FORMAT_BC2_UNORM; - case GL_COMPRESSED_RGBA_S3TC_DXT5_ANGLE: - return DXGI_FORMAT_BC3_UNORM; - case GL_RGBA32F_EXT: - case GL_ALPHA32F_EXT: - case GL_LUMINANCE_ALPHA32F_EXT: - return DXGI_FORMAT_R32G32B32A32_FLOAT; - case GL_RGB32F_EXT: - case GL_LUMINANCE32F_EXT: - return DXGI_FORMAT_R32G32B32A32_FLOAT; - case GL_RGBA16F_EXT: - case GL_ALPHA16F_EXT: - case GL_LUMINANCE_ALPHA16F_EXT: - case GL_RGB16F_EXT: - case GL_LUMINANCE16F_EXT: - return DXGI_FORMAT_R16G16B16A16_FLOAT; - case GL_BGRA8_EXT: - return DXGI_FORMAT_B8G8R8A8_UNORM; - case GL_R8_EXT: - return DXGI_FORMAT_R8_UNORM; - case GL_RG8_EXT: - return DXGI_FORMAT_R8G8_UNORM; - case GL_R16F_EXT: - return DXGI_FORMAT_R16_FLOAT; - case GL_RG16F_EXT: - return DXGI_FORMAT_R16G16_FLOAT; - case GL_DEPTH_COMPONENT16: - return DXGI_FORMAT_D16_UNORM; - case GL_DEPTH_COMPONENT32_OES: - case GL_DEPTH24_STENCIL8_OES: - return DXGI_FORMAT_D24_UNORM_S8_UINT; - case GL_NONE: - return DXGI_FORMAT_UNKNOWN; - default: - UNREACHABLE(); - } - - return DXGI_FORMAT_R8G8B8A8_UNORM; -} - -} - -namespace d3d11 -{ - -void SetPositionTexCoordVertex(PositionTexCoordVertex* vertex, float x, float y, float u, float v) -{ - vertex->x = x; - vertex->y = y; - vertex->u = u; - vertex->v = v; -} - -void SetPositionDepthColorVertex(PositionDepthColorVertex* vertex, float x, float y, float z, - const gl::Color &color) -{ - vertex->x = x; - vertex->y = y; - vertex->z = z; - vertex->r = color.red; - vertex->g = color.green; - vertex->b = color.blue; - vertex->a = color.alpha; -} - -size_t ComputePixelSizeBits(DXGI_FORMAT format) -{ - switch (format) - { - case DXGI_FORMAT_R1_UNORM: - return 1; - - case DXGI_FORMAT_A8_UNORM: - case DXGI_FORMAT_R8_SINT: - case DXGI_FORMAT_R8_SNORM: - case DXGI_FORMAT_R8_TYPELESS: - case DXGI_FORMAT_R8_UINT: - case DXGI_FORMAT_R8_UNORM: - return 8; - - case DXGI_FORMAT_B5G5R5A1_UNORM: - case DXGI_FORMAT_B5G6R5_UNORM: - case DXGI_FORMAT_D16_UNORM: - case DXGI_FORMAT_R16_FLOAT: - case DXGI_FORMAT_R16_SINT: - case DXGI_FORMAT_R16_SNORM: - case DXGI_FORMAT_R16_TYPELESS: - case DXGI_FORMAT_R16_UINT: - case DXGI_FORMAT_R16_UNORM: - case DXGI_FORMAT_R8G8_SINT: - case DXGI_FORMAT_R8G8_SNORM: - case DXGI_FORMAT_R8G8_TYPELESS: - case DXGI_FORMAT_R8G8_UINT: - case DXGI_FORMAT_R8G8_UNORM: - return 16; - - case DXGI_FORMAT_B8G8R8X8_TYPELESS: - case DXGI_FORMAT_B8G8R8X8_UNORM: - case DXGI_FORMAT_B8G8R8X8_UNORM_SRGB: - case DXGI_FORMAT_D24_UNORM_S8_UINT: - case DXGI_FORMAT_D32_FLOAT: - case DXGI_FORMAT_D32_FLOAT_S8X24_UINT: - case DXGI_FORMAT_G8R8_G8B8_UNORM: - case DXGI_FORMAT_R10G10B10_XR_BIAS_A2_UNORM: - case DXGI_FORMAT_R10G10B10A2_TYPELESS: - case DXGI_FORMAT_R10G10B10A2_UINT: - case DXGI_FORMAT_R10G10B10A2_UNORM: - case DXGI_FORMAT_R11G11B10_FLOAT: - case DXGI_FORMAT_R16G16_FLOAT: - case DXGI_FORMAT_R16G16_SINT: - case DXGI_FORMAT_R16G16_SNORM: - case DXGI_FORMAT_R16G16_TYPELESS: - case DXGI_FORMAT_R16G16_UINT: - case DXGI_FORMAT_R16G16_UNORM: - case DXGI_FORMAT_R24_UNORM_X8_TYPELESS: - case DXGI_FORMAT_R24G8_TYPELESS: - case DXGI_FORMAT_R32_FLOAT: - case DXGI_FORMAT_R32_FLOAT_X8X24_TYPELESS: - case DXGI_FORMAT_R32_SINT: - case DXGI_FORMAT_R32_TYPELESS: - case DXGI_FORMAT_R32_UINT: - case DXGI_FORMAT_R8G8_B8G8_UNORM: - case DXGI_FORMAT_R8G8B8A8_SINT: - case DXGI_FORMAT_R8G8B8A8_SNORM: - case DXGI_FORMAT_R8G8B8A8_TYPELESS: - case DXGI_FORMAT_R8G8B8A8_UINT: - case DXGI_FORMAT_R8G8B8A8_UNORM: - case DXGI_FORMAT_R8G8B8A8_UNORM_SRGB: - case DXGI_FORMAT_B8G8R8A8_TYPELESS: - case DXGI_FORMAT_B8G8R8A8_UNORM: - case DXGI_FORMAT_B8G8R8A8_UNORM_SRGB: - case DXGI_FORMAT_R9G9B9E5_SHAREDEXP: - case DXGI_FORMAT_X24_TYPELESS_G8_UINT: - case DXGI_FORMAT_X32_TYPELESS_G8X24_UINT: - return 32; - - case DXGI_FORMAT_R16G16B16A16_FLOAT: - case DXGI_FORMAT_R16G16B16A16_SINT: - case DXGI_FORMAT_R16G16B16A16_SNORM: - case DXGI_FORMAT_R16G16B16A16_TYPELESS: - case DXGI_FORMAT_R16G16B16A16_UINT: - case DXGI_FORMAT_R16G16B16A16_UNORM: - case DXGI_FORMAT_R32G32_FLOAT: - case DXGI_FORMAT_R32G32_SINT: - case DXGI_FORMAT_R32G32_TYPELESS: - case DXGI_FORMAT_R32G32_UINT: - case DXGI_FORMAT_R32G8X24_TYPELESS: - return 64; - - case DXGI_FORMAT_R32G32B32_FLOAT: - case DXGI_FORMAT_R32G32B32_SINT: - case DXGI_FORMAT_R32G32B32_TYPELESS: - case DXGI_FORMAT_R32G32B32_UINT: - return 96; - - case DXGI_FORMAT_R32G32B32A32_FLOAT: - case DXGI_FORMAT_R32G32B32A32_SINT: - case DXGI_FORMAT_R32G32B32A32_TYPELESS: - case DXGI_FORMAT_R32G32B32A32_UINT: - return 128; - - case DXGI_FORMAT_BC1_TYPELESS: - case DXGI_FORMAT_BC1_UNORM: - case DXGI_FORMAT_BC1_UNORM_SRGB: - case DXGI_FORMAT_BC4_SNORM: - case DXGI_FORMAT_BC4_TYPELESS: - case DXGI_FORMAT_BC4_UNORM: - return 4; - - case DXGI_FORMAT_BC2_TYPELESS: - case DXGI_FORMAT_BC2_UNORM: - case DXGI_FORMAT_BC2_UNORM_SRGB: - case DXGI_FORMAT_BC3_TYPELESS: - case DXGI_FORMAT_BC3_UNORM: - case DXGI_FORMAT_BC3_UNORM_SRGB: - case DXGI_FORMAT_BC5_SNORM: - case DXGI_FORMAT_BC5_TYPELESS: - case DXGI_FORMAT_BC5_UNORM: - case DXGI_FORMAT_BC6H_SF16: - case DXGI_FORMAT_BC6H_TYPELESS: - case DXGI_FORMAT_BC6H_UF16: - case DXGI_FORMAT_BC7_TYPELESS: - case DXGI_FORMAT_BC7_UNORM: - case DXGI_FORMAT_BC7_UNORM_SRGB: - return 8; - - default: - return 0; - } -} - -size_t ComputeBlockSizeBits(DXGI_FORMAT format) -{ - switch (format) - { - case DXGI_FORMAT_BC1_TYPELESS: - case DXGI_FORMAT_BC1_UNORM: - case DXGI_FORMAT_BC1_UNORM_SRGB: - case DXGI_FORMAT_BC4_SNORM: - case DXGI_FORMAT_BC4_TYPELESS: - case DXGI_FORMAT_BC4_UNORM: - case DXGI_FORMAT_BC2_TYPELESS: - case DXGI_FORMAT_BC2_UNORM: - case DXGI_FORMAT_BC2_UNORM_SRGB: - case DXGI_FORMAT_BC3_TYPELESS: - case DXGI_FORMAT_BC3_UNORM: - case DXGI_FORMAT_BC3_UNORM_SRGB: - case DXGI_FORMAT_BC5_SNORM: - case DXGI_FORMAT_BC5_TYPELESS: - case DXGI_FORMAT_BC5_UNORM: - case DXGI_FORMAT_BC6H_SF16: - case DXGI_FORMAT_BC6H_TYPELESS: - case DXGI_FORMAT_BC6H_UF16: - case DXGI_FORMAT_BC7_TYPELESS: - case DXGI_FORMAT_BC7_UNORM: - case DXGI_FORMAT_BC7_UNORM_SRGB: - return ComputePixelSizeBits(format) * 16; - default: - UNREACHABLE(); - return 0; - } -} - -bool IsCompressed(DXGI_FORMAT format) -{ - switch (format) - { - case DXGI_FORMAT_BC1_TYPELESS: - case DXGI_FORMAT_BC1_UNORM: - case DXGI_FORMAT_BC1_UNORM_SRGB: - case DXGI_FORMAT_BC4_SNORM: - case DXGI_FORMAT_BC4_TYPELESS: - case DXGI_FORMAT_BC4_UNORM: - case DXGI_FORMAT_BC2_TYPELESS: - case DXGI_FORMAT_BC2_UNORM: - case DXGI_FORMAT_BC2_UNORM_SRGB: - case DXGI_FORMAT_BC3_TYPELESS: - case DXGI_FORMAT_BC3_UNORM: - case DXGI_FORMAT_BC3_UNORM_SRGB: - case DXGI_FORMAT_BC5_SNORM: - case DXGI_FORMAT_BC5_TYPELESS: - case DXGI_FORMAT_BC5_UNORM: - case DXGI_FORMAT_BC6H_SF16: - case DXGI_FORMAT_BC6H_TYPELESS: - case DXGI_FORMAT_BC6H_UF16: - case DXGI_FORMAT_BC7_TYPELESS: - case DXGI_FORMAT_BC7_UNORM: - case DXGI_FORMAT_BC7_UNORM_SRGB: - return true; - case DXGI_FORMAT_UNKNOWN: - UNREACHABLE(); - return false; - default: - return false; - } -} - -unsigned int GetTextureFormatDimensionAlignment(DXGI_FORMAT format) -{ - switch (format) - { - case DXGI_FORMAT_BC1_TYPELESS: - case DXGI_FORMAT_BC1_UNORM: - case DXGI_FORMAT_BC1_UNORM_SRGB: - case DXGI_FORMAT_BC4_SNORM: - case DXGI_FORMAT_BC4_TYPELESS: - case DXGI_FORMAT_BC4_UNORM: - case DXGI_FORMAT_BC2_TYPELESS: - case DXGI_FORMAT_BC2_UNORM: - case DXGI_FORMAT_BC2_UNORM_SRGB: - case DXGI_FORMAT_BC3_TYPELESS: - case DXGI_FORMAT_BC3_UNORM: - case DXGI_FORMAT_BC3_UNORM_SRGB: - case DXGI_FORMAT_BC5_SNORM: - case DXGI_FORMAT_BC5_TYPELESS: - case DXGI_FORMAT_BC5_UNORM: - case DXGI_FORMAT_BC6H_SF16: - case DXGI_FORMAT_BC6H_TYPELESS: - case DXGI_FORMAT_BC6H_UF16: - case DXGI_FORMAT_BC7_TYPELESS: - case DXGI_FORMAT_BC7_UNORM: - case DXGI_FORMAT_BC7_UNORM_SRGB: - return 4; - case DXGI_FORMAT_UNKNOWN: - UNREACHABLE(); - return 1; - default: - return 1; - } -} - -bool IsDepthStencilFormat(DXGI_FORMAT format) -{ - switch (format) - { - case DXGI_FORMAT_D32_FLOAT_S8X24_UINT: - case DXGI_FORMAT_D32_FLOAT: - case DXGI_FORMAT_D24_UNORM_S8_UINT: - case DXGI_FORMAT_D16_UNORM: - return true; - default: - return false; - } -} - -DXGI_FORMAT GetDepthTextureFormat(DXGI_FORMAT format) -{ - switch (format) - { - case DXGI_FORMAT_D32_FLOAT_S8X24_UINT: return DXGI_FORMAT_R32G8X24_TYPELESS; - case DXGI_FORMAT_D32_FLOAT: return DXGI_FORMAT_R32_TYPELESS; - case DXGI_FORMAT_D24_UNORM_S8_UINT: return DXGI_FORMAT_R24G8_TYPELESS; - case DXGI_FORMAT_D16_UNORM: return DXGI_FORMAT_R16_TYPELESS; - default: UNREACHABLE(); return DXGI_FORMAT_UNKNOWN; - } -} - -DXGI_FORMAT GetDepthShaderResourceFormat(DXGI_FORMAT format) -{ - switch (format) - { - case DXGI_FORMAT_D32_FLOAT_S8X24_UINT: return DXGI_FORMAT_R32_FLOAT_X8X24_TYPELESS; - case DXGI_FORMAT_D32_FLOAT: return DXGI_FORMAT_R32_UINT; - case DXGI_FORMAT_D24_UNORM_S8_UINT: return DXGI_FORMAT_R24_UNORM_X8_TYPELESS; - case DXGI_FORMAT_D16_UNORM: return DXGI_FORMAT_R16_UNORM; - default: UNREACHABLE(); return DXGI_FORMAT_UNKNOWN; - } -} - -HRESULT SetDebugName(ID3D11DeviceChild *resource, const char *name) -{ -#if defined(_DEBUG) - return resource->SetPrivateData(WKPDID_D3DDebugObjectName, strlen(name), name); -#else - return S_OK; -#endif -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/renderer11_utils.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/renderer11_utils.h deleted file mode 100644 index 70ad4fea2b..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/renderer11_utils.h +++ /dev/null @@ -1,95 +0,0 @@ -// -// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// renderer11_utils.h: Conversion functions and other utility routines -// specific to the D3D11 renderer. - -#ifndef LIBGLESV2_RENDERER_RENDERER11_UTILS_H -#define LIBGLESV2_RENDERER_RENDERER11_UTILS_H - -#include "libGLESv2/angletypes.h" - -namespace gl_d3d11 -{ - -D3D11_BLEND ConvertBlendFunc(GLenum glBlend, bool isAlpha); -D3D11_BLEND_OP ConvertBlendOp(GLenum glBlendOp); -UINT8 ConvertColorMask(bool maskRed, bool maskGreen, bool maskBlue, bool maskAlpha); - -D3D11_CULL_MODE ConvertCullMode(bool cullEnabled, GLenum cullMode); - -D3D11_COMPARISON_FUNC ConvertComparison(GLenum comparison); -D3D11_DEPTH_WRITE_MASK ConvertDepthMask(bool depthWriteEnabled); -UINT8 ConvertStencilMask(GLuint stencilmask); -D3D11_STENCIL_OP ConvertStencilOp(GLenum stencilOp); - -D3D11_FILTER ConvertFilter(GLenum minFilter, GLenum magFilter, float maxAnisotropy); -D3D11_TEXTURE_ADDRESS_MODE ConvertTextureWrap(GLenum wrap); -FLOAT ConvertMinLOD(GLenum minFilter, unsigned int lodOffset); -FLOAT ConvertMaxLOD(GLenum minFilter, unsigned int lodOffset); - -DXGI_FORMAT ConvertRenderbufferFormat(GLenum format); -DXGI_FORMAT ConvertTextureFormat(GLenum format, D3D_FEATURE_LEVEL featureLevel); -} - -namespace d3d11_gl -{ - -GLenum ConvertBackBufferFormat(DXGI_FORMAT format); -GLenum ConvertDepthStencilFormat(DXGI_FORMAT format); -GLenum ConvertRenderbufferFormat(DXGI_FORMAT format); -GLenum ConvertTextureInternalFormat(DXGI_FORMAT format); - -} - -namespace d3d11 -{ - -struct PositionTexCoordVertex -{ - float x, y; - float u, v; -}; -void SetPositionTexCoordVertex(PositionTexCoordVertex* vertex, float x, float y, float u, float v); - -struct PositionDepthColorVertex -{ - float x, y, z; - float r, g, b, a; -}; -void SetPositionDepthColorVertex(PositionDepthColorVertex* vertex, float x, float y, float z, - const gl::Color &color); - -size_t ComputePixelSizeBits(DXGI_FORMAT format); -size_t ComputeBlockSizeBits(DXGI_FORMAT format); - -bool IsCompressed(DXGI_FORMAT format); -unsigned int GetTextureFormatDimensionAlignment(DXGI_FORMAT format); - -bool IsDepthStencilFormat(DXGI_FORMAT format); -DXGI_FORMAT GetDepthTextureFormat(DXGI_FORMAT format); -DXGI_FORMAT GetDepthShaderResourceFormat(DXGI_FORMAT format); - -HRESULT SetDebugName(ID3D11DeviceChild *resource, const char *name); - -inline bool isDeviceLostError(HRESULT errorCode) -{ - switch (errorCode) - { - case DXGI_ERROR_DEVICE_HUNG: - case DXGI_ERROR_DEVICE_REMOVED: - case DXGI_ERROR_DEVICE_RESET: - case DXGI_ERROR_DRIVER_INTERNAL_ERROR: - case DXGI_ERROR_NOT_CURRENTLY_AVAILABLE: - return true; - default: - return false; - } -} - -} - -#endif // LIBGLESV2_RENDERER_RENDERER11_UTILS_H diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/shaders/Clear11.hlsl b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/shaders/Clear11.hlsl deleted file mode 100644 index cb132dc99c..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/shaders/Clear11.hlsl +++ /dev/null @@ -1,42 +0,0 @@ -void VS_Clear( in float3 inPosition : POSITION, in float4 inColor : COLOR, - out float4 outPosition : SV_POSITION, out float4 outColor : COLOR) -{ - outPosition = float4(inPosition, 1.0f); - outColor = inColor; -} - -// Assume we are in SM4+, which has 8 color outputs -struct PS_OutputMultiple -{ - float4 color0 : SV_TARGET0; - float4 color1 : SV_TARGET1; - float4 color2 : SV_TARGET2; - float4 color3 : SV_TARGET3; -#ifdef SM4 - float4 color4 : SV_TARGET4; - float4 color5 : SV_TARGET5; - float4 color6 : SV_TARGET6; - float4 color7 : SV_TARGET7; -#endif -}; - -PS_OutputMultiple PS_ClearMultiple(in float4 inPosition : SV_POSITION, in float4 inColor : COLOR) -{ - PS_OutputMultiple outColor; - outColor.color0 = inColor; - outColor.color1 = inColor; - outColor.color2 = inColor; - outColor.color3 = inColor; -#ifdef SM4 - outColor.color4 = inColor; - outColor.color5 = inColor; - outColor.color6 = inColor; - outColor.color7 = inColor; -#endif - return outColor; -} - -float4 PS_ClearSingle(in float4 inPosition : SV_Position, in float4 inColor : COLOR) : SV_Target0 -{ - return inColor; -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/shaders/Passthrough11.hlsl b/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/shaders/Passthrough11.hlsl deleted file mode 100644 index 43b7801efc..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d11/shaders/Passthrough11.hlsl +++ /dev/null @@ -1,29 +0,0 @@ -Texture2D Texture : register(t0); -SamplerState Sampler : register(s0); - -void VS_Passthrough( in float2 inPosition : POSITION, in float2 inTexCoord : TEXCOORD0, - out float4 outPosition : SV_POSITION, out float2 outTexCoord : TEXCOORD0) -{ - outPosition = float4(inPosition, 0.0f, 1.0f); - outTexCoord = inTexCoord; -} - -float4 PS_PassthroughRGBA(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 -{ - return Texture.Sample(Sampler, inTexCoord).rgba; -} - -float4 PS_PassthroughRGB(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 -{ - return float4(Texture.Sample(Sampler, inTexCoord).rgb, 1.0f); -} - -float4 PS_PassthroughLum(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 -{ - return float4(Texture.Sample(Sampler, inTexCoord).rrr, 1.0f); -} - -float4 PS_PassthroughLumAlpha(in float4 inPosition : SV_POSITION, in float2 inTexCoord : TEXCOORD0) : SV_TARGET0 -{ - return Texture.Sample(Sampler, inTexCoord).rrra; -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/BufferStorage9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/BufferStorage9.cpp deleted file mode 100644 index 9fdc1246f1..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/BufferStorage9.cpp +++ /dev/null @@ -1,78 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// BufferStorage9.cpp Defines the BufferStorage9 class. - -#include "libGLESv2/renderer/d3d9/BufferStorage9.h" -#include "common/debug.h" - -namespace rx -{ - -BufferStorage9::BufferStorage9() -{ - mMemory = NULL; - mAllocatedSize = 0; - mSize = 0; -} - -BufferStorage9::~BufferStorage9() -{ - delete[] mMemory; -} - -BufferStorage9 *BufferStorage9::makeBufferStorage9(BufferStorage *bufferStorage) -{ - ASSERT(HAS_DYNAMIC_TYPE(BufferStorage9*, bufferStorage)); - return static_cast<BufferStorage9*>(bufferStorage); -} - -void *BufferStorage9::getData() -{ - return mMemory; -} - -void BufferStorage9::setData(const void* data, unsigned int size, unsigned int offset) -{ - if (!mMemory || offset + size > mAllocatedSize) - { - unsigned int newAllocatedSize = offset + size; - void *newMemory = new char[newAllocatedSize]; - - if (offset > 0 && mMemory && mAllocatedSize > 0) - { - memcpy(newMemory, mMemory, std::min(offset, mAllocatedSize)); - } - - delete[] mMemory; - mMemory = newMemory; - mAllocatedSize = newAllocatedSize; - } - - mSize = std::max(mSize, offset + size); - if (data) - { - memcpy(reinterpret_cast<char*>(mMemory) + offset, data, size); - } -} - -void BufferStorage9::clear() -{ - mSize = 0; -} - -unsigned int BufferStorage9::getSize() const -{ - return mSize; -} - -bool BufferStorage9::supportsDirectBinding() const -{ - return false; -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/BufferStorage9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/BufferStorage9.h deleted file mode 100644 index 3e803969bc..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/BufferStorage9.h +++ /dev/null @@ -1,42 +0,0 @@ -// -// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// BufferStorage9.h Defines the BufferStorage9 class. - -#ifndef LIBGLESV2_RENDERER_BUFFERSTORAGE9_H_ -#define LIBGLESV2_RENDERER_BUFFERSTORAGE9_H_ - -#include "libGLESv2/renderer/BufferStorage.h" - -namespace rx -{ - -class BufferStorage9 : public BufferStorage -{ - public: - BufferStorage9(); - virtual ~BufferStorage9(); - - static BufferStorage9 *makeBufferStorage9(BufferStorage *bufferStorage); - - virtual void *getData(); - virtual void setData(const void* data, unsigned int size, unsigned int offset); - virtual void clear(); - virtual unsigned int getSize() const; - virtual bool supportsDirectBinding() const; - - private: - DISALLOW_COPY_AND_ASSIGN(BufferStorage9); - - void *mMemory; - unsigned int mAllocatedSize; - - unsigned int mSize; -}; - -} - -#endif // LIBGLESV2_RENDERER_BUFFERSTORAGE9_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Fence9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Fence9.cpp deleted file mode 100644 index 639c37b4e4..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Fence9.cpp +++ /dev/null @@ -1,135 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// Fence9.cpp: Defines the rx::Fence9 class. - -#include "libGLESv2/renderer/d3d9/Fence9.h" -#include "libGLESv2/main.h" -#include "libGLESv2/renderer/d3d9/renderer9_utils.h" -#include "libGLESv2/renderer/d3d9/Renderer9.h" - -namespace rx -{ - -Fence9::Fence9(rx::Renderer9 *renderer) -{ - mRenderer = renderer; - mQuery = NULL; -} - -Fence9::~Fence9() -{ - if (mQuery) - { - mRenderer->freeEventQuery(mQuery); - mQuery = NULL; - } -} - -GLboolean Fence9::isFence() -{ - // GL_NV_fence spec: - // A name returned by GenFencesNV, but not yet set via SetFenceNV, is not the name of an existing fence. - return mQuery != NULL; -} - -void Fence9::setFence(GLenum condition) -{ - if (!mQuery) - { - mQuery = mRenderer->allocateEventQuery(); - if (!mQuery) - { - return gl::error(GL_OUT_OF_MEMORY); - } - } - - HRESULT result = mQuery->Issue(D3DISSUE_END); - ASSERT(SUCCEEDED(result)); - - setCondition(condition); - setStatus(GL_FALSE); -} - -GLboolean Fence9::testFence() -{ - if (mQuery == NULL) - { - return gl::error(GL_INVALID_OPERATION, GL_TRUE); - } - - HRESULT result = mQuery->GetData(NULL, 0, D3DGETDATA_FLUSH); - - if (d3d9::isDeviceLostError(result)) - { - mRenderer->notifyDeviceLost(); - return gl::error(GL_OUT_OF_MEMORY, GL_TRUE); - } - - ASSERT(result == S_OK || result == S_FALSE); - setStatus(result == S_OK); - return getStatus(); -} - -void Fence9::finishFence() -{ - if (mQuery == NULL) - { - return gl::error(GL_INVALID_OPERATION); - } - - while (!testFence()) - { - Sleep(0); - } -} - -void Fence9::getFenceiv(GLenum pname, GLint *params) -{ - if (mQuery == NULL) - { - return gl::error(GL_INVALID_OPERATION); - } - - switch (pname) - { - case GL_FENCE_STATUS_NV: - { - // GL_NV_fence spec: - // Once the status of a fence has been finished (via FinishFenceNV) or tested and the returned status is TRUE (via either TestFenceNV - // or GetFenceivNV querying the FENCE_STATUS_NV), the status remains TRUE until the next SetFenceNV of the fence. - if (getStatus()) - { - params[0] = GL_TRUE; - return; - } - - HRESULT result = mQuery->GetData(NULL, 0, 0); - - if (d3d9::isDeviceLostError(result)) - { - params[0] = GL_TRUE; - mRenderer->notifyDeviceLost(); - return gl::error(GL_OUT_OF_MEMORY); - } - - ASSERT(result == S_OK || result == S_FALSE); - setStatus(result == S_OK); - params[0] = getStatus(); - - break; - } - case GL_FENCE_CONDITION_NV: - params[0] = getCondition(); - break; - default: - return gl::error(GL_INVALID_ENUM); - break; - } -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Image9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Image9.h deleted file mode 100644 index 2fbbca3124..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Image9.h +++ /dev/null @@ -1,79 +0,0 @@ -// -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// Image9.h: Defines the rx::Image9 class, which acts as the interface to -// the actual underlying surfaces of a Texture. - -#ifndef LIBGLESV2_RENDERER_IMAGE9_H_ -#define LIBGLESV2_RENDERER_IMAGE9_H_ - -#include "libGLESv2/renderer/Image.h" -#include "common/debug.h" - -namespace gl -{ -class Framebuffer; -} - -namespace rx -{ -class Renderer; -class Renderer9; -class TextureStorageInterface2D; -class TextureStorageInterfaceCube; - -class Image9 : public Image -{ - public: - Image9(); - ~Image9(); - - static Image9 *makeImage9(Image *img); - - static void generateMipmap(Image9 *dest, Image9 *source); - static void generateMip(IDirect3DSurface9 *destSurface, IDirect3DSurface9 *sourceSurface); - static void copyLockableSurfaces(IDirect3DSurface9 *dest, IDirect3DSurface9 *source); - - virtual bool redefine(Renderer *renderer, GLint internalformat, GLsizei width, GLsizei height, bool forceRelease); - - virtual bool isRenderableFormat() const; - D3DFORMAT getD3DFormat() const; - - virtual bool isDirty() const {return mSurface && mDirty;} - IDirect3DSurface9 *getSurface(); - - virtual void setManagedSurface(TextureStorageInterface2D *storage, int level); - virtual void setManagedSurface(TextureStorageInterfaceCube *storage, int face, int level); - virtual bool updateSurface(TextureStorageInterface2D *storage, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); - virtual bool updateSurface(TextureStorageInterfaceCube *storage, int face, int level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); - - virtual void loadData(GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, - GLint unpackAlignment, const void *input); - virtual void loadCompressedData(GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, - const void *input); - - virtual void copy(GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height, gl::Framebuffer *source); - - private: - DISALLOW_COPY_AND_ASSIGN(Image9); - - void createSurface(); - void setManagedSurface(IDirect3DSurface9 *surface); - bool updateSurface(IDirect3DSurface9 *dest, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height); - - HRESULT lock(D3DLOCKED_RECT *lockedRect, const RECT *rect); - void unlock(); - - Renderer9 *mRenderer; - - D3DPOOL mD3DPool; // can only be D3DPOOL_SYSTEMMEM or D3DPOOL_MANAGED since it needs to be lockable. - D3DFORMAT mD3DFormat; - - IDirect3DSurface9 *mSurface; -}; -} - -#endif // LIBGLESV2_RENDERER_IMAGE9_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/IndexBuffer9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/IndexBuffer9.cpp deleted file mode 100644 index 7cb5d13a18..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/IndexBuffer9.cpp +++ /dev/null @@ -1,207 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// Indexffer9.cpp: Defines the D3D9 IndexBuffer implementation. - -#include "libGLESv2/renderer/d3d9/IndexBuffer9.h" -#include "libGLESv2/renderer/d3d9/Renderer9.h" - -namespace rx -{ - -IndexBuffer9::IndexBuffer9(Renderer9 *const renderer) : mRenderer(renderer) -{ - mIndexBuffer = NULL; - mBufferSize = 0; - mIndexType = 0; - mDynamic = false; -} - -IndexBuffer9::~IndexBuffer9() -{ - if (mIndexBuffer) - { - mIndexBuffer->Release(); - mIndexBuffer = NULL; - } -} - -bool IndexBuffer9::initialize(unsigned int bufferSize, GLenum indexType, bool dynamic) -{ - if (mIndexBuffer) - { - mIndexBuffer->Release(); - mIndexBuffer = NULL; - } - - updateSerial(); - - if (bufferSize > 0) - { - D3DFORMAT format; - if (indexType == GL_UNSIGNED_SHORT || indexType == GL_UNSIGNED_BYTE) - { - format = D3DFMT_INDEX16; - } - else if (indexType == GL_UNSIGNED_INT) - { - if (mRenderer->get32BitIndexSupport()) - { - format = D3DFMT_INDEX32; - } - else - { - ERR("Attempted to create a 32-bit index buffer but renderer does not support 32-bit indices."); - return false; - } - } - else - { - ERR("Invalid index type %u.", indexType); - return false; - } - - DWORD usageFlags = D3DUSAGE_WRITEONLY; - if (dynamic) - { - usageFlags |= D3DUSAGE_DYNAMIC; - } - - HRESULT result = mRenderer->createIndexBuffer(bufferSize, usageFlags, format, &mIndexBuffer); - if (FAILED(result)) - { - ERR("Failed to create an index buffer of size %u, result: 0x%08x.", mBufferSize, result); - return false; - } - } - - mBufferSize = bufferSize; - mIndexType = indexType; - mDynamic = dynamic; - - return true; -} - -IndexBuffer9 *IndexBuffer9::makeIndexBuffer9(IndexBuffer *indexBuffer) -{ - ASSERT(HAS_DYNAMIC_TYPE(IndexBuffer9*, indexBuffer)); - return static_cast<IndexBuffer9*>(indexBuffer); -} - -bool IndexBuffer9::mapBuffer(unsigned int offset, unsigned int size, void** outMappedMemory) -{ - if (mIndexBuffer) - { - DWORD lockFlags = mDynamic ? D3DLOCK_NOOVERWRITE : 0; - - void *mapPtr = NULL; - HRESULT result = mIndexBuffer->Lock(offset, size, &mapPtr, lockFlags); - if (FAILED(result)) - { - ERR("Index buffer lock failed with error 0x%08x", result); - return false; - } - - *outMappedMemory = mapPtr; - return true; - } - else - { - ERR("Index buffer not initialized."); - return false; - } -} - -bool IndexBuffer9::unmapBuffer() -{ - if (mIndexBuffer) - { - HRESULT result = mIndexBuffer->Unlock(); - if (FAILED(result)) - { - ERR("Index buffer unlock failed with error 0x%08x", result); - return false; - } - - return true; - } - else - { - ERR("Index buffer not initialized."); - return false; - } -} - -GLenum IndexBuffer9::getIndexType() const -{ - return mIndexType; -} - -unsigned int IndexBuffer9::getBufferSize() const -{ - return mBufferSize; -} - -bool IndexBuffer9::setSize(unsigned int bufferSize, GLenum indexType) -{ - if (bufferSize > mBufferSize || indexType != mIndexType) - { - return initialize(bufferSize, indexType, mDynamic); - } - else - { - return true; - } -} - -bool IndexBuffer9::discard() -{ - if (mIndexBuffer) - { - void *dummy; - HRESULT result; - - result = mIndexBuffer->Lock(0, 1, &dummy, D3DLOCK_DISCARD); - if (FAILED(result)) - { - ERR("Discard lock failed with error 0x%08x", result); - return false; - } - - result = mIndexBuffer->Unlock(); - if (FAILED(result)) - { - ERR("Discard unlock failed with error 0x%08x", result); - return false; - } - - return true; - } - else - { - ERR("Index buffer not initialized."); - return false; - } -} - -D3DFORMAT IndexBuffer9::getIndexFormat() const -{ - switch (mIndexType) - { - case GL_UNSIGNED_BYTE: return D3DFMT_INDEX16; - case GL_UNSIGNED_SHORT: return D3DFMT_INDEX16; - case GL_UNSIGNED_INT: return D3DFMT_INDEX32; - default: UNREACHABLE(); return D3DFMT_UNKNOWN; - } -} - -IDirect3DIndexBuffer9 * IndexBuffer9::getBuffer() const -{ - return mIndexBuffer; -} - -}
\ No newline at end of file diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Query9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Query9.cpp deleted file mode 100644 index 72781cbc39..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Query9.cpp +++ /dev/null @@ -1,125 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// Query9.cpp: Defines the rx::Query9 class which implements rx::QueryImpl. - - -#include "libGLESv2/renderer/d3d9/Query9.h" -#include "libGLESv2/main.h" -#include "libGLESv2/renderer/d3d9/renderer9_utils.h" -#include "libGLESv2/renderer/d3d9/Renderer9.h" - -namespace rx -{ - -Query9::Query9(rx::Renderer9 *renderer, GLenum type) : QueryImpl(type) -{ - mRenderer = renderer; - mQuery = NULL; -} - -Query9::~Query9() -{ - if (mQuery) - { - mQuery->Release(); - mQuery = NULL; - } -} - -void Query9::begin() -{ - if (mQuery == NULL) - { - if (FAILED(mRenderer->getDevice()->CreateQuery(D3DQUERYTYPE_OCCLUSION, &mQuery))) - { - return gl::error(GL_OUT_OF_MEMORY); - } - } - - HRESULT result = mQuery->Issue(D3DISSUE_BEGIN); - ASSERT(SUCCEEDED(result)); -} - -void Query9::end() -{ - if (mQuery == NULL) - { - return gl::error(GL_INVALID_OPERATION); - } - - HRESULT result = mQuery->Issue(D3DISSUE_END); - ASSERT(SUCCEEDED(result)); - - mStatus = GL_FALSE; - mResult = GL_FALSE; -} - -GLuint Query9::getResult() -{ - if (mQuery != NULL) - { - while (!testQuery()) - { - Sleep(0); - // explicitly check for device loss - // some drivers seem to return S_FALSE even if the device is lost - // instead of D3DERR_DEVICELOST like they should - if (mRenderer->testDeviceLost(true)) - { - return gl::error(GL_OUT_OF_MEMORY, 0); - } - } - } - - return mResult; -} - -GLboolean Query9::isResultAvailable() -{ - if (mQuery != NULL) - { - testQuery(); - } - - return mStatus; -} - -GLboolean Query9::testQuery() -{ - if (mQuery != NULL && mStatus != GL_TRUE) - { - DWORD numPixels = 0; - - HRESULT hres = mQuery->GetData(&numPixels, sizeof(DWORD), D3DGETDATA_FLUSH); - if (hres == S_OK) - { - mStatus = GL_TRUE; - - switch (getType()) - { - case GL_ANY_SAMPLES_PASSED_EXT: - case GL_ANY_SAMPLES_PASSED_CONSERVATIVE_EXT: - mResult = (numPixels > 0) ? GL_TRUE : GL_FALSE; - break; - default: - ASSERT(false); - } - } - else if (d3d9::isDeviceLostError(hres)) - { - mRenderer->notifyDeviceLost(); - return gl::error(GL_OUT_OF_MEMORY, GL_TRUE); - } - - return mStatus; - } - - return GL_TRUE; // prevent blocking when query is null -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/RenderTarget9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/RenderTarget9.cpp deleted file mode 100644 index 090431db99..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/RenderTarget9.cpp +++ /dev/null @@ -1,113 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// RenderTarget9.cpp: Implements a D3D9-specific wrapper for IDirect3DSurface9 -// pointers retained by renderbuffers. - -#include "libGLESv2/renderer/d3d9/RenderTarget9.h" -#include "libGLESv2/renderer/d3d9/Renderer9.h" - -#include "libGLESv2/renderer/d3d9/renderer9_utils.h" -#include "libGLESv2/main.h" - -namespace rx -{ - -RenderTarget9::RenderTarget9(Renderer *renderer, IDirect3DSurface9 *surface) -{ - mRenderer = Renderer9::makeRenderer9(renderer); - mRenderTarget = surface; - - if (mRenderTarget) - { - D3DSURFACE_DESC description; - mRenderTarget->GetDesc(&description); - - mWidth = description.Width; - mHeight = description.Height; - - mInternalFormat = d3d9_gl::GetEquivalentFormat(description.Format); - mActualFormat = d3d9_gl::GetEquivalentFormat(description.Format); - mSamples = d3d9_gl::GetSamplesFromMultisampleType(description.MultiSampleType); - } -} - -RenderTarget9::RenderTarget9(Renderer *renderer, GLsizei width, GLsizei height, GLenum format, GLsizei samples) -{ - mRenderer = Renderer9::makeRenderer9(renderer); - mRenderTarget = NULL; - - D3DFORMAT requestedFormat = gl_d3d9::ConvertRenderbufferFormat(format); - int supportedSamples = mRenderer->getNearestSupportedSamples(requestedFormat, samples); - - if (supportedSamples == -1) - { - gl::error(GL_OUT_OF_MEMORY); - - return; - } - - HRESULT result = D3DERR_INVALIDCALL; - - if (width > 0 && height > 0) - { - if (requestedFormat == D3DFMT_D24S8) - { - result = mRenderer->getDevice()->CreateDepthStencilSurface(width, height, requestedFormat, - gl_d3d9::GetMultisampleTypeFromSamples(supportedSamples), - 0, FALSE, &mRenderTarget, NULL); - } - else - { - result = mRenderer->getDevice()->CreateRenderTarget(width, height, requestedFormat, - gl_d3d9::GetMultisampleTypeFromSamples(supportedSamples), - 0, FALSE, &mRenderTarget, NULL); - } - - if (result == D3DERR_OUTOFVIDEOMEMORY || result == E_OUTOFMEMORY) - { - gl::error(GL_OUT_OF_MEMORY); - - return; - } - - ASSERT(SUCCEEDED(result)); - } - - mWidth = width; - mHeight = height; - mInternalFormat = format; - mSamples = supportedSamples; - mActualFormat = d3d9_gl::GetEquivalentFormat(requestedFormat); -} - -RenderTarget9::~RenderTarget9() -{ - if (mRenderTarget) - { - mRenderTarget->Release(); - } -} - -RenderTarget9 *RenderTarget9::makeRenderTarget9(RenderTarget *target) -{ - ASSERT(HAS_DYNAMIC_TYPE(rx::RenderTarget9*, target)); - return static_cast<rx::RenderTarget9*>(target); -} - -IDirect3DSurface9 *RenderTarget9::getSurface() -{ - // Caller is responsible for releasing the returned surface reference. - if (mRenderTarget) - { - mRenderTarget->AddRef(); - } - - return mRenderTarget; -} - -}
\ No newline at end of file diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Renderer9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Renderer9.h deleted file mode 100644 index 24fd2bdd84..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/Renderer9.h +++ /dev/null @@ -1,356 +0,0 @@ -// -// Copyright (c) 2012-2013 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// Renderer9.h: Defines a back-end specific class for the D3D9 renderer. - -#ifndef LIBGLESV2_RENDERER_RENDERER9_H_ -#define LIBGLESV2_RENDERER_RENDERER9_H_ - -#include "common/angleutils.h" -#include "libGLESv2/mathutil.h" -#include "libGLESv2/renderer/ShaderCache.h" -#include "libGLESv2/renderer/d3d9/VertexDeclarationCache.h" -#include "libGLESv2/renderer/Renderer.h" -#include "libGLESv2/renderer/RenderTarget.h" - -namespace gl -{ -class Renderbuffer; -} - -namespace rx -{ -class VertexDataManager; -class IndexDataManager; -class StreamingIndexBufferInterface; -struct TranslatedAttribute; - -class Renderer9 : public Renderer -{ - public: - Renderer9(egl::Display *display, HDC hDc, bool softwareDevice); - virtual ~Renderer9(); - - static Renderer9 *makeRenderer9(Renderer *renderer); - - virtual EGLint initialize(); - virtual bool resetDevice(); - - virtual int generateConfigs(ConfigDesc **configDescList); - virtual void deleteConfigs(ConfigDesc *configDescList); - - void startScene(); - void endScene(); - - virtual void sync(bool block); - - virtual SwapChain *createSwapChain(HWND window, HANDLE shareHandle, GLenum backBufferFormat, GLenum depthBufferFormat); - - IDirect3DQuery9* allocateEventQuery(); - void freeEventQuery(IDirect3DQuery9* query); - - // resource creation - IDirect3DVertexShader9 *createVertexShader(const DWORD *function, size_t length); - IDirect3DPixelShader9 *createPixelShader(const DWORD *function, size_t length); - HRESULT createVertexBuffer(UINT Length, DWORD Usage, IDirect3DVertexBuffer9 **ppVertexBuffer); - HRESULT createIndexBuffer(UINT Length, DWORD Usage, D3DFORMAT Format, IDirect3DIndexBuffer9 **ppIndexBuffer); -#if 0 - void *createTexture2D(); - void *createTextureCube(); - void *createQuery(); - void *createIndexBuffer(); - void *createVertexbuffer(); - - // state setup - void applyShaders(); - void applyConstants(); -#endif - virtual void setSamplerState(gl::SamplerType type, int index, const gl::SamplerState &sampler); - virtual void setTexture(gl::SamplerType type, int index, gl::Texture *texture); - - virtual void setRasterizerState(const gl::RasterizerState &rasterState); - virtual void setBlendState(gl::Framebuffer *framebuffer, const gl::BlendState &blendState, const gl::Color &blendColor, - unsigned int sampleMask); - virtual void setDepthStencilState(const gl::DepthStencilState &depthStencilState, int stencilRef, - int stencilBackRef, bool frontFaceCCW); - - virtual void setScissorRectangle(const gl::Rectangle &scissor, bool enabled); - virtual bool setViewport(const gl::Rectangle &viewport, float zNear, float zFar, GLenum drawMode, GLenum frontFace, - bool ignoreViewport); - - virtual bool applyRenderTarget(gl::Framebuffer *frameBuffer); - virtual void applyShaders(gl::ProgramBinary *programBinary); - virtual void applyUniforms(gl::ProgramBinary *programBinary, gl::UniformArray *uniformArray); - virtual bool applyPrimitiveType(GLenum primitiveType, GLsizei elementCount); - virtual GLenum applyVertexBuffer(gl::ProgramBinary *programBinary, gl::VertexAttribute vertexAttributes[], GLint first, GLsizei count, GLsizei instances); - virtual GLenum applyIndexBuffer(const GLvoid *indices, gl::Buffer *elementArrayBuffer, GLsizei count, GLenum mode, GLenum type, TranslatedIndexData *indexInfo); - - virtual void drawArrays(GLenum mode, GLsizei count, GLsizei instances); - virtual void drawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, gl::Buffer *elementArrayBuffer, const TranslatedIndexData &indexInfo, GLsizei instances); - - virtual void clear(const gl::ClearParameters &clearParams, gl::Framebuffer *frameBuffer); - - virtual void markAllStateDirty(); - - // lost device - void notifyDeviceLost(); - virtual bool isDeviceLost(); - virtual bool testDeviceLost(bool notify); - virtual bool testDeviceResettable(); - - // Renderer capabilities - IDirect3DDevice9 *getDevice() { return mDevice; } - virtual DWORD getAdapterVendor() const; - virtual std::string getRendererDescription() const; - virtual GUID getAdapterIdentifier() const; - - virtual bool getBGRATextureSupport() const; - virtual bool getDXT1TextureSupport(); - virtual bool getDXT3TextureSupport(); - virtual bool getDXT5TextureSupport(); - virtual bool getEventQuerySupport(); - virtual bool getFloat32TextureSupport(bool *filtering, bool *renderable); - virtual bool getFloat16TextureSupport(bool *filtering, bool *renderable); - virtual bool getLuminanceTextureSupport(); - virtual bool getLuminanceAlphaTextureSupport(); - virtual unsigned int getMaxVertexTextureImageUnits() const; - virtual unsigned int getMaxCombinedTextureImageUnits() const; - virtual unsigned int getReservedVertexUniformVectors() const; - virtual unsigned int getReservedFragmentUniformVectors() const; - virtual unsigned int getMaxVertexUniformVectors() const; - virtual unsigned int getMaxFragmentUniformVectors() const; - virtual unsigned int getMaxVaryingVectors() const; - virtual bool getNonPower2TextureSupport() const; - virtual bool getDepthTextureSupport() const; - virtual bool getOcclusionQuerySupport() const; - virtual bool getInstancingSupport() const; - virtual bool getTextureFilterAnisotropySupport() const; - virtual float getTextureMaxAnisotropy() const; - virtual bool getShareHandleSupport() const; - virtual bool getDerivativeInstructionSupport() const; - virtual bool getPostSubBufferSupport() const; - - virtual int getMajorShaderModel() const; - virtual float getMaxPointSize() const; - virtual int getMaxViewportDimension() const; - virtual int getMaxTextureWidth() const; - virtual int getMaxTextureHeight() const; - virtual bool get32BitIndexSupport() const; - DWORD getCapsDeclTypes() const; - virtual int getMinSwapInterval() const; - virtual int getMaxSwapInterval() const; - - virtual GLsizei getMaxSupportedSamples() const; - int getNearestSupportedSamples(D3DFORMAT format, int requested) const; - - virtual unsigned int getMaxRenderTargets() const; - - D3DFORMAT ConvertTextureInternalFormat(GLint internalformat); - - // Pixel operations - virtual bool copyToRenderTarget(TextureStorageInterface2D *dest, TextureStorageInterface2D *source); - virtual bool copyToRenderTarget(TextureStorageInterfaceCube *dest, TextureStorageInterfaceCube *source); - - virtual bool copyImage(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, - GLint xoffset, GLint yoffset, TextureStorageInterface2D *storage, GLint level); - virtual bool copyImage(gl::Framebuffer *framebuffer, const gl::Rectangle &sourceRect, GLenum destFormat, - GLint xoffset, GLint yoffset, TextureStorageInterfaceCube *storage, GLenum target, GLint level); - - virtual bool blitRect(gl::Framebuffer *readTarget, const gl::Rectangle &readRect, gl::Framebuffer *drawTarget, const gl::Rectangle &drawRect, - bool blitRenderTarget, bool blitDepthStencil); - virtual void readPixels(gl::Framebuffer *framebuffer, GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, - GLsizei outputPitch, bool packReverseRowOrder, GLint packAlignment, void* pixels); - - // RenderTarget creation - virtual RenderTarget *createRenderTarget(SwapChain *swapChain, bool depth); - virtual RenderTarget *createRenderTarget(int width, int height, GLenum format, GLsizei samples, bool depth); - - // Shader operations - virtual ShaderExecutable *loadExecutable(const void *function, size_t length, rx::ShaderType type); - virtual ShaderExecutable *compileToExecutable(gl::InfoLog &infoLog, const char *shaderHLSL, rx::ShaderType type, D3DWorkaroundType workaround); - - // Image operations - virtual Image *createImage(); - virtual void generateMipmap(Image *dest, Image *source); - virtual TextureStorage *createTextureStorage2D(SwapChain *swapChain); - virtual TextureStorage *createTextureStorage2D(int levels, GLenum internalformat, GLenum usage, bool forceRenderable, GLsizei width, GLsizei height); - virtual TextureStorage *createTextureStorageCube(int levels, GLenum internalformat, GLenum usage, bool forceRenderable, int size); - - // Buffer creation - virtual VertexBuffer *createVertexBuffer(); - virtual IndexBuffer *createIndexBuffer(); - virtual BufferStorage *createBufferStorage(); - - // Query and Fence creation - virtual QueryImpl *createQuery(GLenum type); - virtual FenceImpl *createFence(); - - // D3D9-renderer specific methods - bool boxFilter(IDirect3DSurface9 *source, IDirect3DSurface9 *dest); - - D3DPOOL getTexturePool(DWORD usage) const; - - virtual bool getLUID(LUID *adapterLuid) const; - - private: - DISALLOW_COPY_AND_ASSIGN(Renderer9); - - void deinitialize(); - - void applyUniformnfv(gl::Uniform *targetUniform, const GLfloat *v); - void applyUniformniv(gl::Uniform *targetUniform, const GLint *v); - void applyUniformnbv(gl::Uniform *targetUniform, const GLint *v); - - void drawLineLoop(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer); - void drawIndexedPoints(GLsizei count, GLenum type, const GLvoid *indices, int minIndex, gl::Buffer *elementArrayBuffer); - - void getMultiSampleSupport(D3DFORMAT format, bool *multiSampleArray); - bool copyToRenderTarget(IDirect3DSurface9 *dest, IDirect3DSurface9 *source, bool fromManaged); - gl::Renderbuffer *getNullColorbuffer(gl::Renderbuffer *depthbuffer); - - D3DPOOL getBufferPool(DWORD usage) const; - - HMODULE mD3d9Module; - HDC mDc; - - void initializeDevice(); - D3DPRESENT_PARAMETERS getDefaultPresentParameters(); - void releaseDeviceResources(); - - HRESULT getDeviceStatusCode(); - bool isRemovedDeviceResettable() const; - bool resetRemovedDevice(); - - UINT mAdapter; - D3DDEVTYPE mDeviceType; - bool mSoftwareDevice; // FIXME: Deprecate - IDirect3D9 *mD3d9; // Always valid after successful initialization. - IDirect3D9Ex *mD3d9Ex; // Might be null if D3D9Ex is not supported. - IDirect3DDevice9 *mDevice; - IDirect3DDevice9Ex *mDeviceEx; // Might be null if D3D9Ex is not supported. - - Blit *mBlit; - - HWND mDeviceWindow; - - bool mDeviceLost; - D3DCAPS9 mDeviceCaps; - D3DADAPTER_IDENTIFIER9 mAdapterIdentifier; - - D3DPRIMITIVETYPE mPrimitiveType; - int mPrimitiveCount; - GLsizei mRepeatDraw; - - bool mSceneStarted; - bool mSupportsNonPower2Textures; - bool mSupportsTextureFilterAnisotropy; - int mMinSwapInterval; - int mMaxSwapInterval; - - bool mOcclusionQuerySupport; - bool mEventQuerySupport; - bool mVertexTextureSupport; - - bool mDepthTextureSupport; - - bool mFloat32TextureSupport; - bool mFloat32FilterSupport; - bool mFloat32RenderSupport; - - bool mFloat16TextureSupport; - bool mFloat16FilterSupport; - bool mFloat16RenderSupport; - - bool mDXT1TextureSupport; - bool mDXT3TextureSupport; - bool mDXT5TextureSupport; - - bool mLuminanceTextureSupport; - bool mLuminanceAlphaTextureSupport; - - std::map<D3DFORMAT, bool *> mMultiSampleSupport; - GLsizei mMaxSupportedSamples; - - // current render target states - unsigned int mAppliedRenderTargetSerial; - unsigned int mAppliedDepthbufferSerial; - unsigned int mAppliedStencilbufferSerial; - bool mDepthStencilInitialized; - bool mRenderTargetDescInitialized; - rx::RenderTarget::Desc mRenderTargetDesc; - unsigned int mCurStencilSize; - unsigned int mCurDepthSize; - - IDirect3DStateBlock9 *mMaskedClearSavedState; - - // previously set render states - bool mForceSetDepthStencilState; - gl::DepthStencilState mCurDepthStencilState; - int mCurStencilRef; - int mCurStencilBackRef; - bool mCurFrontFaceCCW; - - bool mForceSetRasterState; - gl::RasterizerState mCurRasterState; - - bool mForceSetScissor; - gl::Rectangle mCurScissor; - bool mScissorEnabled; - - bool mForceSetViewport; - gl::Rectangle mCurViewport; - float mCurNear; - float mCurFar; - float mCurDepthFront; - - bool mForceSetBlendState; - gl::BlendState mCurBlendState; - gl::Color mCurBlendColor; - GLuint mCurSampleMask; - - // Currently applied sampler states - bool mForceSetVertexSamplerStates[gl::IMPLEMENTATION_MAX_VERTEX_TEXTURE_IMAGE_UNITS]; - gl::SamplerState mCurVertexSamplerStates[gl::IMPLEMENTATION_MAX_VERTEX_TEXTURE_IMAGE_UNITS]; - - bool mForceSetPixelSamplerStates[gl::MAX_TEXTURE_IMAGE_UNITS]; - gl::SamplerState mCurPixelSamplerStates[gl::MAX_TEXTURE_IMAGE_UNITS]; - - // Currently applied textures - unsigned int mCurVertexTextureSerials[gl::IMPLEMENTATION_MAX_VERTEX_TEXTURE_IMAGE_UNITS]; - unsigned int mCurPixelTextureSerials[gl::MAX_TEXTURE_IMAGE_UNITS]; - - unsigned int mAppliedIBSerial; - unsigned int mAppliedProgramBinarySerial; - - rx::dx_VertexConstants mVertexConstants; - rx::dx_PixelConstants mPixelConstants; - bool mDxUniformsDirty; - - // A pool of event queries that are currently unused. - std::vector<IDirect3DQuery9*> mEventQueryPool; - VertexShaderCache mVertexShaderCache; - PixelShaderCache mPixelShaderCache; - - VertexDataManager *mVertexDataManager; - VertexDeclarationCache mVertexDeclarationCache; - - IndexDataManager *mIndexDataManager; - StreamingIndexBufferInterface *mLineLoopIB; - - enum { NUM_NULL_COLORBUFFER_CACHE_ENTRIES = 12 }; - struct NullColorbufferCacheEntry - { - UINT lruCount; - int width; - int height; - gl::Renderbuffer *buffer; - } mNullColorbufferCache[NUM_NULL_COLORBUFFER_CACHE_ENTRIES]; - UINT mMaxNullColorbufferLRU; - -}; - -} -#endif // LIBGLESV2_RENDERER_RENDERER9_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/VertexBuffer9.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/VertexBuffer9.cpp deleted file mode 100644 index 57f5bcd256..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/VertexBuffer9.cpp +++ /dev/null @@ -1,530 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// VertexBuffer9.cpp: Defines the D3D9 VertexBuffer implementation. - -#include "libGLESv2/renderer/d3d9/VertexBuffer9.h" -#include "libGLESv2/renderer/d3d9/vertexconversion.h" -#include "libGLESv2/renderer/BufferStorage.h" -#include "libGLESv2/Context.h" -#include "libGLESv2/renderer/d3d9/Renderer9.h" - -#include "libGLESv2/Buffer.h" - -namespace rx -{ - -bool VertexBuffer9::mTranslationsInitialized = false; -VertexBuffer9::FormatConverter VertexBuffer9::mFormatConverters[NUM_GL_VERTEX_ATTRIB_TYPES][2][4]; - -VertexBuffer9::VertexBuffer9(rx::Renderer9 *const renderer) : mRenderer(renderer) -{ - mVertexBuffer = NULL; - mBufferSize = 0; - mDynamicUsage = false; - - if (!mTranslationsInitialized) - { - initializeTranslations(renderer->getCapsDeclTypes()); - mTranslationsInitialized = true; - } -} - -VertexBuffer9::~VertexBuffer9() -{ - if (mVertexBuffer) - { - mVertexBuffer->Release(); - mVertexBuffer = NULL; - } -} - -bool VertexBuffer9::initialize(unsigned int size, bool dynamicUsage) -{ - if (mVertexBuffer) - { - mVertexBuffer->Release(); - mVertexBuffer = NULL; - } - - updateSerial(); - - if (size > 0) - { - DWORD flags = D3DUSAGE_WRITEONLY; - if (dynamicUsage) - { - flags |= D3DUSAGE_DYNAMIC; - } - - HRESULT result = mRenderer->createVertexBuffer(size, flags, &mVertexBuffer); - - if (FAILED(result)) - { - ERR("Out of memory allocating a vertex buffer of size %lu.", size); - return false; - } - } - - mBufferSize = size; - mDynamicUsage = dynamicUsage; - return true; -} - -VertexBuffer9 *VertexBuffer9::makeVertexBuffer9(VertexBuffer *vertexBuffer) -{ - ASSERT(HAS_DYNAMIC_TYPE(VertexBuffer9*, vertexBuffer)); - return static_cast<VertexBuffer9*>(vertexBuffer); -} - -bool VertexBuffer9::storeVertexAttributes(const gl::VertexAttribute &attrib, GLint start, GLsizei count, - GLsizei instances, unsigned int offset) -{ - if (mVertexBuffer) - { - gl::Buffer *buffer = attrib.mBoundBuffer.get(); - - int inputStride = attrib.stride(); - int elementSize = attrib.typeSize(); - const FormatConverter &converter = formatConverter(attrib); - - DWORD lockFlags = mDynamicUsage ? D3DLOCK_NOOVERWRITE : 0; - - void *mapPtr = NULL; - - unsigned int mapSize; - if (!spaceRequired(attrib, count, instances, &mapSize)) - { - return false; - } - - HRESULT result = mVertexBuffer->Lock(offset, mapSize, &mapPtr, lockFlags); - - if (FAILED(result)) - { - ERR("Lock failed with error 0x%08x", result); - return false; - } - - const char *input = NULL; - if (buffer) - { - BufferStorage *storage = buffer->getStorage(); - input = static_cast<const char*>(storage->getData()) + static_cast<int>(attrib.mOffset); - } - else - { - input = static_cast<const char*>(attrib.mPointer); - } - - if (instances == 0 || attrib.mDivisor == 0) - { - input += inputStride * start; - } - - if (converter.identity && inputStride == elementSize) - { - memcpy(mapPtr, input, count * inputStride); - } - else - { - converter.convertArray(input, inputStride, count, mapPtr); - } - - mVertexBuffer->Unlock(); - - return true; - } - else - { - ERR("Vertex buffer not initialized."); - return false; - } -} - -bool VertexBuffer9::storeRawData(const void* data, unsigned int size, unsigned int offset) -{ - if (mVertexBuffer) - { - DWORD lockFlags = mDynamicUsage ? D3DLOCK_NOOVERWRITE : 0; - - void *mapPtr = NULL; - HRESULT result = mVertexBuffer->Lock(offset, size, &mapPtr, lockFlags); - - if (FAILED(result)) - { - ERR("Lock failed with error 0x%08x", result); - return false; - } - - memcpy(mapPtr, data, size); - - mVertexBuffer->Unlock(); - - return true; - } - else - { - ERR("Vertex buffer not initialized."); - return false; - } -} - -bool VertexBuffer9::getSpaceRequired(const gl::VertexAttribute &attrib, GLsizei count, GLsizei instances, - unsigned int *outSpaceRequired) const -{ - return spaceRequired(attrib, count, instances, outSpaceRequired); -} - -bool VertexBuffer9::requiresConversion(const gl::VertexAttribute &attrib) const -{ - return !formatConverter(attrib).identity; -} - -unsigned int VertexBuffer9::getVertexSize(const gl::VertexAttribute &attrib) const -{ - unsigned int spaceRequired; - return getSpaceRequired(attrib, 1, 0, &spaceRequired) ? spaceRequired : 0; -} - -D3DDECLTYPE VertexBuffer9::getDeclType(const gl::VertexAttribute &attrib) const -{ - return formatConverter(attrib).d3dDeclType; -} - -unsigned int VertexBuffer9::getBufferSize() const -{ - return mBufferSize; -} - -bool VertexBuffer9::setBufferSize(unsigned int size) -{ - if (size > mBufferSize) - { - return initialize(size, mDynamicUsage); - } - else - { - return true; - } -} - -bool VertexBuffer9::discard() -{ - if (mVertexBuffer) - { - void *dummy; - HRESULT result; - - result = mVertexBuffer->Lock(0, 1, &dummy, D3DLOCK_DISCARD); - if (FAILED(result)) - { - ERR("Discard lock failed with error 0x%08x", result); - return false; - } - - result = mVertexBuffer->Unlock(); - if (FAILED(result)) - { - ERR("Discard unlock failed with error 0x%08x", result); - return false; - } - - return true; - } - else - { - ERR("Vertex buffer not initialized."); - return false; - } -} - -IDirect3DVertexBuffer9 * VertexBuffer9::getBuffer() const -{ - return mVertexBuffer; -} - -// Mapping from OpenGL-ES vertex attrib type to D3D decl type: -// -// BYTE SHORT (Cast) -// BYTE-norm FLOAT (Normalize) (can't be exactly represented as SHORT-norm) -// UNSIGNED_BYTE UBYTE4 (Identity) or SHORT (Cast) -// UNSIGNED_BYTE-norm UBYTE4N (Identity) or FLOAT (Normalize) -// SHORT SHORT (Identity) -// SHORT-norm SHORT-norm (Identity) or FLOAT (Normalize) -// UNSIGNED_SHORT FLOAT (Cast) -// UNSIGNED_SHORT-norm USHORT-norm (Identity) or FLOAT (Normalize) -// FIXED (not in WebGL) FLOAT (FixedToFloat) -// FLOAT FLOAT (Identity) - -// GLToCType maps from GL type (as GLenum) to the C typedef. -template <GLenum GLType> struct GLToCType { }; - -template <> struct GLToCType<GL_BYTE> { typedef GLbyte type; }; -template <> struct GLToCType<GL_UNSIGNED_BYTE> { typedef GLubyte type; }; -template <> struct GLToCType<GL_SHORT> { typedef GLshort type; }; -template <> struct GLToCType<GL_UNSIGNED_SHORT> { typedef GLushort type; }; -template <> struct GLToCType<GL_FIXED> { typedef GLuint type; }; -template <> struct GLToCType<GL_FLOAT> { typedef GLfloat type; }; - -// This differs from D3DDECLTYPE in that it is unsized. (Size expansion is applied last.) -enum D3DVertexType -{ - D3DVT_FLOAT, - D3DVT_SHORT, - D3DVT_SHORT_NORM, - D3DVT_UBYTE, - D3DVT_UBYTE_NORM, - D3DVT_USHORT_NORM -}; - -// D3DToCType maps from D3D vertex type (as enum D3DVertexType) to the corresponding C type. -template <unsigned int D3DType> struct D3DToCType { }; - -template <> struct D3DToCType<D3DVT_FLOAT> { typedef float type; }; -template <> struct D3DToCType<D3DVT_SHORT> { typedef short type; }; -template <> struct D3DToCType<D3DVT_SHORT_NORM> { typedef short type; }; -template <> struct D3DToCType<D3DVT_UBYTE> { typedef unsigned char type; }; -template <> struct D3DToCType<D3DVT_UBYTE_NORM> { typedef unsigned char type; }; -template <> struct D3DToCType<D3DVT_USHORT_NORM> { typedef unsigned short type; }; - -// Encode the type/size combinations that D3D permits. For each type/size it expands to a widener that will provide the appropriate final size. -template <unsigned int type, int size> struct WidenRule { }; - -template <int size> struct WidenRule<D3DVT_FLOAT, size> : NoWiden<size> { }; -template <int size> struct WidenRule<D3DVT_SHORT, size> : WidenToEven<size> { }; -template <int size> struct WidenRule<D3DVT_SHORT_NORM, size> : WidenToEven<size> { }; -template <int size> struct WidenRule<D3DVT_UBYTE, size> : WidenToFour<size> { }; -template <int size> struct WidenRule<D3DVT_UBYTE_NORM, size> : WidenToFour<size> { }; -template <int size> struct WidenRule<D3DVT_USHORT_NORM, size> : WidenToEven<size> { }; - -// VertexTypeFlags encodes the D3DCAPS9::DeclType flag and vertex declaration flag for each D3D vertex type & size combination. -template <unsigned int d3dtype, int size> struct VertexTypeFlags { }; - -template <unsigned int _capflag, unsigned int _declflag> -struct VertexTypeFlagsHelper -{ - enum { capflag = _capflag }; - enum { declflag = _declflag }; -}; - -template <> struct VertexTypeFlags<D3DVT_FLOAT, 1> : VertexTypeFlagsHelper<0, D3DDECLTYPE_FLOAT1> { }; -template <> struct VertexTypeFlags<D3DVT_FLOAT, 2> : VertexTypeFlagsHelper<0, D3DDECLTYPE_FLOAT2> { }; -template <> struct VertexTypeFlags<D3DVT_FLOAT, 3> : VertexTypeFlagsHelper<0, D3DDECLTYPE_FLOAT3> { }; -template <> struct VertexTypeFlags<D3DVT_FLOAT, 4> : VertexTypeFlagsHelper<0, D3DDECLTYPE_FLOAT4> { }; -template <> struct VertexTypeFlags<D3DVT_SHORT, 2> : VertexTypeFlagsHelper<0, D3DDECLTYPE_SHORT2> { }; -template <> struct VertexTypeFlags<D3DVT_SHORT, 4> : VertexTypeFlagsHelper<0, D3DDECLTYPE_SHORT4> { }; -template <> struct VertexTypeFlags<D3DVT_SHORT_NORM, 2> : VertexTypeFlagsHelper<D3DDTCAPS_SHORT2N, D3DDECLTYPE_SHORT2N> { }; -template <> struct VertexTypeFlags<D3DVT_SHORT_NORM, 4> : VertexTypeFlagsHelper<D3DDTCAPS_SHORT4N, D3DDECLTYPE_SHORT4N> { }; -template <> struct VertexTypeFlags<D3DVT_UBYTE, 4> : VertexTypeFlagsHelper<D3DDTCAPS_UBYTE4, D3DDECLTYPE_UBYTE4> { }; -template <> struct VertexTypeFlags<D3DVT_UBYTE_NORM, 4> : VertexTypeFlagsHelper<D3DDTCAPS_UBYTE4N, D3DDECLTYPE_UBYTE4N> { }; -template <> struct VertexTypeFlags<D3DVT_USHORT_NORM, 2> : VertexTypeFlagsHelper<D3DDTCAPS_USHORT2N, D3DDECLTYPE_USHORT2N> { }; -template <> struct VertexTypeFlags<D3DVT_USHORT_NORM, 4> : VertexTypeFlagsHelper<D3DDTCAPS_USHORT4N, D3DDECLTYPE_USHORT4N> { }; - - -// VertexTypeMapping maps GL type & normalized flag to preferred and fallback D3D vertex types (as D3DVertexType enums). -template <GLenum GLtype, bool normalized> struct VertexTypeMapping { }; - -template <D3DVertexType Preferred, D3DVertexType Fallback = Preferred> -struct VertexTypeMappingBase -{ - enum { preferred = Preferred }; - enum { fallback = Fallback }; -}; - -template <> struct VertexTypeMapping<GL_BYTE, false> : VertexTypeMappingBase<D3DVT_SHORT> { }; // Cast -template <> struct VertexTypeMapping<GL_BYTE, true> : VertexTypeMappingBase<D3DVT_FLOAT> { }; // Normalize -template <> struct VertexTypeMapping<GL_UNSIGNED_BYTE, false> : VertexTypeMappingBase<D3DVT_UBYTE, D3DVT_FLOAT> { }; // Identity, Cast -template <> struct VertexTypeMapping<GL_UNSIGNED_BYTE, true> : VertexTypeMappingBase<D3DVT_UBYTE_NORM, D3DVT_FLOAT> { }; // Identity, Normalize -template <> struct VertexTypeMapping<GL_SHORT, false> : VertexTypeMappingBase<D3DVT_SHORT> { }; // Identity -template <> struct VertexTypeMapping<GL_SHORT, true> : VertexTypeMappingBase<D3DVT_SHORT_NORM, D3DVT_FLOAT> { }; // Cast, Normalize -template <> struct VertexTypeMapping<GL_UNSIGNED_SHORT, false> : VertexTypeMappingBase<D3DVT_FLOAT> { }; // Cast -template <> struct VertexTypeMapping<GL_UNSIGNED_SHORT, true> : VertexTypeMappingBase<D3DVT_USHORT_NORM, D3DVT_FLOAT> { }; // Cast, Normalize -template <bool normalized> struct VertexTypeMapping<GL_FIXED, normalized> : VertexTypeMappingBase<D3DVT_FLOAT> { }; // FixedToFloat -template <bool normalized> struct VertexTypeMapping<GL_FLOAT, normalized> : VertexTypeMappingBase<D3DVT_FLOAT> { }; // Identity - - -// Given a GL type & norm flag and a D3D type, ConversionRule provides the type conversion rule (Cast, Normalize, Identity, FixedToFloat). -// The conversion rules themselves are defined in vertexconversion.h. - -// Almost all cases are covered by Cast (including those that are actually Identity since Cast<T,T> knows it's an identity mapping). -template <GLenum fromType, bool normalized, unsigned int toType> -struct ConversionRule : Cast<typename GLToCType<fromType>::type, typename D3DToCType<toType>::type> { }; - -// All conversions from normalized types to float use the Normalize operator. -template <GLenum fromType> struct ConversionRule<fromType, true, D3DVT_FLOAT> : Normalize<typename GLToCType<fromType>::type> { }; - -// Use a full specialization for this so that it preferentially matches ahead of the generic normalize-to-float rules. -template <> struct ConversionRule<GL_FIXED, true, D3DVT_FLOAT> : FixedToFloat<GLint, 16> { }; -template <> struct ConversionRule<GL_FIXED, false, D3DVT_FLOAT> : FixedToFloat<GLint, 16> { }; - -// A 2-stage construction is used for DefaultVertexValues because float must use SimpleDefaultValues (i.e. 0/1) -// whether it is normalized or not. -template <class T, bool normalized> struct DefaultVertexValuesStage2 { }; - -template <class T> struct DefaultVertexValuesStage2<T, true> : NormalizedDefaultValues<T> { }; -template <class T> struct DefaultVertexValuesStage2<T, false> : SimpleDefaultValues<T> { }; - -// Work out the default value rule for a D3D type (expressed as the C type) and -template <class T, bool normalized> struct DefaultVertexValues : DefaultVertexValuesStage2<T, normalized> { }; -template <bool normalized> struct DefaultVertexValues<float, normalized> : SimpleDefaultValues<float> { }; - -// Policy rules for use with Converter, to choose whether to use the preferred or fallback conversion. -// The fallback conversion produces an output that all D3D9 devices must support. -template <class T> struct UsePreferred { enum { type = T::preferred }; }; -template <class T> struct UseFallback { enum { type = T::fallback }; }; - -// Converter ties it all together. Given an OpenGL type/norm/size and choice of preferred/fallback conversion, -// it provides all the members of the appropriate VertexDataConverter, the D3DCAPS9::DeclTypes flag in cap flag -// and the D3DDECLTYPE member needed for the vertex declaration in declflag. -template <GLenum fromType, bool normalized, int size, template <class T> class PreferenceRule> -struct Converter - : VertexDataConverter<typename GLToCType<fromType>::type, - WidenRule<PreferenceRule< VertexTypeMapping<fromType, normalized> >::type, size>, - ConversionRule<fromType, - normalized, - PreferenceRule< VertexTypeMapping<fromType, normalized> >::type>, - DefaultVertexValues<typename D3DToCType<PreferenceRule< VertexTypeMapping<fromType, normalized> >::type>::type, normalized > > -{ -private: - enum { d3dtype = PreferenceRule< VertexTypeMapping<fromType, normalized> >::type }; - enum { d3dsize = WidenRule<d3dtype, size>::finalWidth }; - -public: - enum { capflag = VertexTypeFlags<d3dtype, d3dsize>::capflag }; - enum { declflag = VertexTypeFlags<d3dtype, d3dsize>::declflag }; -}; - -// Initialize a TranslationInfo -#define TRANSLATION(type, norm, size, preferred) \ - { \ - Converter<type, norm, size, preferred>::identity, \ - Converter<type, norm, size, preferred>::finalSize, \ - Converter<type, norm, size, preferred>::convertArray, \ - static_cast<D3DDECLTYPE>(Converter<type, norm, size, preferred>::declflag) \ - } - -#define TRANSLATION_FOR_TYPE_NORM_SIZE(type, norm, size) \ - { \ - Converter<type, norm, size, UsePreferred>::capflag, \ - TRANSLATION(type, norm, size, UsePreferred), \ - TRANSLATION(type, norm, size, UseFallback) \ - } - -#define TRANSLATIONS_FOR_TYPE(type) \ - { \ - { TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 1), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 2), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 3), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 4) }, \ - { TRANSLATION_FOR_TYPE_NORM_SIZE(type, true, 1), TRANSLATION_FOR_TYPE_NORM_SIZE(type, true, 2), TRANSLATION_FOR_TYPE_NORM_SIZE(type, true, 3), TRANSLATION_FOR_TYPE_NORM_SIZE(type, true, 4) }, \ - } - -#define TRANSLATIONS_FOR_TYPE_NO_NORM(type) \ - { \ - { TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 1), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 2), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 3), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 4) }, \ - { TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 1), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 2), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 3), TRANSLATION_FOR_TYPE_NORM_SIZE(type, false, 4) }, \ - } - -const VertexBuffer9::TranslationDescription VertexBuffer9::mPossibleTranslations[NUM_GL_VERTEX_ATTRIB_TYPES][2][4] = // [GL types as enumerated by typeIndex()][normalized][size-1] -{ - TRANSLATIONS_FOR_TYPE(GL_BYTE), - TRANSLATIONS_FOR_TYPE(GL_UNSIGNED_BYTE), - TRANSLATIONS_FOR_TYPE(GL_SHORT), - TRANSLATIONS_FOR_TYPE(GL_UNSIGNED_SHORT), - TRANSLATIONS_FOR_TYPE_NO_NORM(GL_FIXED), - TRANSLATIONS_FOR_TYPE_NO_NORM(GL_FLOAT) -}; - -void VertexBuffer9::initializeTranslations(DWORD declTypes) -{ - for (unsigned int i = 0; i < NUM_GL_VERTEX_ATTRIB_TYPES; i++) - { - for (unsigned int j = 0; j < 2; j++) - { - for (unsigned int k = 0; k < 4; k++) - { - if (mPossibleTranslations[i][j][k].capsFlag == 0 || (declTypes & mPossibleTranslations[i][j][k].capsFlag) != 0) - { - mFormatConverters[i][j][k] = mPossibleTranslations[i][j][k].preferredConversion; - } - else - { - mFormatConverters[i][j][k] = mPossibleTranslations[i][j][k].fallbackConversion; - } - } - } - } -} - -unsigned int VertexBuffer9::typeIndex(GLenum type) -{ - switch (type) - { - case GL_BYTE: return 0; - case GL_UNSIGNED_BYTE: return 1; - case GL_SHORT: return 2; - case GL_UNSIGNED_SHORT: return 3; - case GL_FIXED: return 4; - case GL_FLOAT: return 5; - - default: UNREACHABLE(); return 5; - } -} - -const VertexBuffer9::FormatConverter &VertexBuffer9::formatConverter(const gl::VertexAttribute &attribute) -{ - return mFormatConverters[typeIndex(attribute.mType)][attribute.mNormalized][attribute.mSize - 1]; -} - -bool VertexBuffer9::spaceRequired(const gl::VertexAttribute &attrib, std::size_t count, GLsizei instances, - unsigned int *outSpaceRequired) -{ - unsigned int elementSize = formatConverter(attrib).outputElementSize; - - if (attrib.mArrayEnabled) - { - unsigned int elementCount = 0; - if (instances == 0 || attrib.mDivisor == 0) - { - elementCount = count; - } - else - { - if (static_cast<unsigned int>(instances) < std::numeric_limits<unsigned int>::max() - (attrib.mDivisor - 1)) - { - // Round up - elementCount = (static_cast<unsigned int>(instances) + (attrib.mDivisor - 1)) / attrib.mDivisor; - } - else - { - elementCount = static_cast<unsigned int>(instances) / attrib.mDivisor; - } - } - - if (elementSize <= std::numeric_limits<unsigned int>::max() / elementCount) - { - if (outSpaceRequired) - { - *outSpaceRequired = elementSize * elementCount; - } - return true; - } - else - { - return false; - } - } - else - { - const unsigned int elementSize = 4; - if (outSpaceRequired) - { - *outSpaceRequired = elementSize * 4; - } - return true; - } -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/VertexBuffer9.h b/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/VertexBuffer9.h deleted file mode 100644 index 2f88117bda..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/VertexBuffer9.h +++ /dev/null @@ -1,91 +0,0 @@ -// -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// VertexBuffer9.h: Defines the D3D9 VertexBuffer implementation. - -#ifndef LIBGLESV2_RENDERER_VERTEXBUFFER9_H_ -#define LIBGLESV2_RENDERER_VERTEXBUFFER9_H_ - -#include "libGLESv2/renderer/VertexBuffer.h" - -namespace rx -{ -class Renderer9; - -class VertexBuffer9 : public VertexBuffer -{ - public: - explicit VertexBuffer9(rx::Renderer9 *const renderer); - virtual ~VertexBuffer9(); - - virtual bool initialize(unsigned int size, bool dynamicUsage); - - static VertexBuffer9 *makeVertexBuffer9(VertexBuffer *vertexBuffer); - - virtual bool storeVertexAttributes(const gl::VertexAttribute &attrib, GLint start, GLsizei count, GLsizei instances, - unsigned int offset); - virtual bool storeRawData(const void* data, unsigned int size, unsigned int offset); - - virtual bool getSpaceRequired(const gl::VertexAttribute &attrib, GLsizei count, GLsizei instances, unsigned int *outSpaceRequired) const; - - virtual bool requiresConversion(const gl::VertexAttribute &attrib) const; - - unsigned int getVertexSize(const gl::VertexAttribute &attrib) const; - D3DDECLTYPE getDeclType(const gl::VertexAttribute &attrib) const; - - virtual unsigned int getBufferSize() const; - virtual bool setBufferSize(unsigned int size); - virtual bool discard(); - - IDirect3DVertexBuffer9 *getBuffer() const; - - private: - DISALLOW_COPY_AND_ASSIGN(VertexBuffer9); - - rx::Renderer9 *const mRenderer; - - IDirect3DVertexBuffer9 *mVertexBuffer; - unsigned int mBufferSize; - bool mDynamicUsage; - - // Attribute format conversion - enum { NUM_GL_VERTEX_ATTRIB_TYPES = 6 }; - - struct FormatConverter - { - bool identity; - std::size_t outputElementSize; - void (*convertArray)(const void *in, std::size_t stride, std::size_t n, void *out); - D3DDECLTYPE d3dDeclType; - }; - - static bool mTranslationsInitialized; - static void initializeTranslations(DWORD declTypes); - - // [GL types as enumerated by typeIndex()][normalized][size - 1] - static FormatConverter mFormatConverters[NUM_GL_VERTEX_ATTRIB_TYPES][2][4]; - - struct TranslationDescription - { - DWORD capsFlag; - FormatConverter preferredConversion; - FormatConverter fallbackConversion; - }; - - // This table is used to generate mFormatConverters. - // [GL types as enumerated by typeIndex()][normalized][size - 1] - static const TranslationDescription mPossibleTranslations[NUM_GL_VERTEX_ATTRIB_TYPES][2][4]; - - static unsigned int typeIndex(GLenum type); - static const FormatConverter &formatConverter(const gl::VertexAttribute &attribute); - - static bool spaceRequired(const gl::VertexAttribute &attrib, std::size_t count, GLsizei instances, - unsigned int *outSpaceRequired); -}; - -} - -#endif // LIBGLESV2_RENDERER_VERTEXBUFFER9_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/renderer9_utils.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/renderer9_utils.cpp deleted file mode 100644 index b7f2ffb1d9..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/renderer9_utils.cpp +++ /dev/null @@ -1,500 +0,0 @@ -#include "precompiled.h" -// -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -// renderer9_utils.cpp: Conversion functions and other utility routines -// specific to the D3D9 renderer. - -#include "libGLESv2/renderer/d3d9/renderer9_utils.h" -#include "libGLESv2/mathutil.h" -#include "libGLESv2/Context.h" - -#include "common/debug.h" - -namespace gl_d3d9 -{ - -D3DCMPFUNC ConvertComparison(GLenum comparison) -{ - D3DCMPFUNC d3dComp = D3DCMP_ALWAYS; - switch (comparison) - { - case GL_NEVER: d3dComp = D3DCMP_NEVER; break; - case GL_ALWAYS: d3dComp = D3DCMP_ALWAYS; break; - case GL_LESS: d3dComp = D3DCMP_LESS; break; - case GL_LEQUAL: d3dComp = D3DCMP_LESSEQUAL; break; - case GL_EQUAL: d3dComp = D3DCMP_EQUAL; break; - case GL_GREATER: d3dComp = D3DCMP_GREATER; break; - case GL_GEQUAL: d3dComp = D3DCMP_GREATEREQUAL; break; - case GL_NOTEQUAL: d3dComp = D3DCMP_NOTEQUAL; break; - default: UNREACHABLE(); - } - - return d3dComp; -} - -D3DCOLOR ConvertColor(gl::Color color) -{ - return D3DCOLOR_RGBA(gl::unorm<8>(color.red), - gl::unorm<8>(color.green), - gl::unorm<8>(color.blue), - gl::unorm<8>(color.alpha)); -} - -D3DBLEND ConvertBlendFunc(GLenum blend) -{ - D3DBLEND d3dBlend = D3DBLEND_ZERO; - - switch (blend) - { - case GL_ZERO: d3dBlend = D3DBLEND_ZERO; break; - case GL_ONE: d3dBlend = D3DBLEND_ONE; break; - case GL_SRC_COLOR: d3dBlend = D3DBLEND_SRCCOLOR; break; - case GL_ONE_MINUS_SRC_COLOR: d3dBlend = D3DBLEND_INVSRCCOLOR; break; - case GL_DST_COLOR: d3dBlend = D3DBLEND_DESTCOLOR; break; - case GL_ONE_MINUS_DST_COLOR: d3dBlend = D3DBLEND_INVDESTCOLOR; break; - case GL_SRC_ALPHA: d3dBlend = D3DBLEND_SRCALPHA; break; - case GL_ONE_MINUS_SRC_ALPHA: d3dBlend = D3DBLEND_INVSRCALPHA; break; - case GL_DST_ALPHA: d3dBlend = D3DBLEND_DESTALPHA; break; - case GL_ONE_MINUS_DST_ALPHA: d3dBlend = D3DBLEND_INVDESTALPHA; break; - case GL_CONSTANT_COLOR: d3dBlend = D3DBLEND_BLENDFACTOR; break; - case GL_ONE_MINUS_CONSTANT_COLOR: d3dBlend = D3DBLEND_INVBLENDFACTOR; break; - case GL_CONSTANT_ALPHA: d3dBlend = D3DBLEND_BLENDFACTOR; break; - case GL_ONE_MINUS_CONSTANT_ALPHA: d3dBlend = D3DBLEND_INVBLENDFACTOR; break; - case GL_SRC_ALPHA_SATURATE: d3dBlend = D3DBLEND_SRCALPHASAT; break; - default: UNREACHABLE(); - } - - return d3dBlend; -} - -D3DBLENDOP ConvertBlendOp(GLenum blendOp) -{ - D3DBLENDOP d3dBlendOp = D3DBLENDOP_ADD; - - switch (blendOp) - { - case GL_FUNC_ADD: d3dBlendOp = D3DBLENDOP_ADD; break; - case GL_FUNC_SUBTRACT: d3dBlendOp = D3DBLENDOP_SUBTRACT; break; - case GL_FUNC_REVERSE_SUBTRACT: d3dBlendOp = D3DBLENDOP_REVSUBTRACT; break; - default: UNREACHABLE(); - } - - return d3dBlendOp; -} - -D3DSTENCILOP ConvertStencilOp(GLenum stencilOp) -{ - D3DSTENCILOP d3dStencilOp = D3DSTENCILOP_KEEP; - - switch (stencilOp) - { - case GL_ZERO: d3dStencilOp = D3DSTENCILOP_ZERO; break; - case GL_KEEP: d3dStencilOp = D3DSTENCILOP_KEEP; break; - case GL_REPLACE: d3dStencilOp = D3DSTENCILOP_REPLACE; break; - case GL_INCR: d3dStencilOp = D3DSTENCILOP_INCRSAT; break; - case GL_DECR: d3dStencilOp = D3DSTENCILOP_DECRSAT; break; - case GL_INVERT: d3dStencilOp = D3DSTENCILOP_INVERT; break; - case GL_INCR_WRAP: d3dStencilOp = D3DSTENCILOP_INCR; break; - case GL_DECR_WRAP: d3dStencilOp = D3DSTENCILOP_DECR; break; - default: UNREACHABLE(); - } - - return d3dStencilOp; -} - -D3DTEXTUREADDRESS ConvertTextureWrap(GLenum wrap) -{ - D3DTEXTUREADDRESS d3dWrap = D3DTADDRESS_WRAP; - - switch (wrap) - { - case GL_REPEAT: d3dWrap = D3DTADDRESS_WRAP; break; - case GL_CLAMP_TO_EDGE: d3dWrap = D3DTADDRESS_CLAMP; break; - case GL_MIRRORED_REPEAT: d3dWrap = D3DTADDRESS_MIRROR; break; - default: UNREACHABLE(); - } - - return d3dWrap; -} - -D3DCULL ConvertCullMode(GLenum cullFace, GLenum frontFace) -{ - D3DCULL cull = D3DCULL_CCW; - switch (cullFace) - { - case GL_FRONT: - cull = (frontFace == GL_CCW ? D3DCULL_CW : D3DCULL_CCW); - break; - case GL_BACK: - cull = (frontFace == GL_CCW ? D3DCULL_CCW : D3DCULL_CW); - break; - case GL_FRONT_AND_BACK: - cull = D3DCULL_NONE; // culling will be handled during draw - break; - default: UNREACHABLE(); - } - - return cull; -} - -D3DCUBEMAP_FACES ConvertCubeFace(GLenum cubeFace) -{ - D3DCUBEMAP_FACES face = D3DCUBEMAP_FACE_POSITIVE_X; - - switch (cubeFace) - { - case GL_TEXTURE_CUBE_MAP_POSITIVE_X: - face = D3DCUBEMAP_FACE_POSITIVE_X; - break; - case GL_TEXTURE_CUBE_MAP_NEGATIVE_X: - face = D3DCUBEMAP_FACE_NEGATIVE_X; - break; - case GL_TEXTURE_CUBE_MAP_POSITIVE_Y: - face = D3DCUBEMAP_FACE_POSITIVE_Y; - break; - case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y: - face = D3DCUBEMAP_FACE_NEGATIVE_Y; - break; - case GL_TEXTURE_CUBE_MAP_POSITIVE_Z: - face = D3DCUBEMAP_FACE_POSITIVE_Z; - break; - case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z: - face = D3DCUBEMAP_FACE_NEGATIVE_Z; - break; - default: UNREACHABLE(); - } - - return face; -} - -DWORD ConvertColorMask(bool red, bool green, bool blue, bool alpha) -{ - return (red ? D3DCOLORWRITEENABLE_RED : 0) | - (green ? D3DCOLORWRITEENABLE_GREEN : 0) | - (blue ? D3DCOLORWRITEENABLE_BLUE : 0) | - (alpha ? D3DCOLORWRITEENABLE_ALPHA : 0); -} - -D3DTEXTUREFILTERTYPE ConvertMagFilter(GLenum magFilter, float maxAnisotropy) -{ - if (maxAnisotropy > 1.0f) - { - return D3DTEXF_ANISOTROPIC; - } - - D3DTEXTUREFILTERTYPE d3dMagFilter = D3DTEXF_POINT; - switch (magFilter) - { - case GL_NEAREST: d3dMagFilter = D3DTEXF_POINT; break; - case GL_LINEAR: d3dMagFilter = D3DTEXF_LINEAR; break; - default: UNREACHABLE(); - } - - return d3dMagFilter; -} - -void ConvertMinFilter(GLenum minFilter, D3DTEXTUREFILTERTYPE *d3dMinFilter, D3DTEXTUREFILTERTYPE *d3dMipFilter, float maxAnisotropy) -{ - switch (minFilter) - { - case GL_NEAREST: - *d3dMinFilter = D3DTEXF_POINT; - *d3dMipFilter = D3DTEXF_NONE; - break; - case GL_LINEAR: - *d3dMinFilter = D3DTEXF_LINEAR; - *d3dMipFilter = D3DTEXF_NONE; - break; - case GL_NEAREST_MIPMAP_NEAREST: - *d3dMinFilter = D3DTEXF_POINT; - *d3dMipFilter = D3DTEXF_POINT; - break; - case GL_LINEAR_MIPMAP_NEAREST: - *d3dMinFilter = D3DTEXF_LINEAR; - *d3dMipFilter = D3DTEXF_POINT; - break; - case GL_NEAREST_MIPMAP_LINEAR: - *d3dMinFilter = D3DTEXF_POINT; - *d3dMipFilter = D3DTEXF_LINEAR; - break; - case GL_LINEAR_MIPMAP_LINEAR: - *d3dMinFilter = D3DTEXF_LINEAR; - *d3dMipFilter = D3DTEXF_LINEAR; - break; - default: - *d3dMinFilter = D3DTEXF_POINT; - *d3dMipFilter = D3DTEXF_NONE; - UNREACHABLE(); - } - - if (maxAnisotropy > 1.0f) - { - *d3dMinFilter = D3DTEXF_ANISOTROPIC; - } -} - -D3DFORMAT ConvertRenderbufferFormat(GLenum format) -{ - switch (format) - { - case GL_NONE: return D3DFMT_NULL; - case GL_RGBA4: - case GL_RGB5_A1: - case GL_RGBA8_OES: return D3DFMT_A8R8G8B8; - case GL_RGB565: return D3DFMT_R5G6B5; - case GL_RGB8_OES: return D3DFMT_X8R8G8B8; - case GL_DEPTH_COMPONENT16: - case GL_STENCIL_INDEX8: - case GL_DEPTH24_STENCIL8_OES: return D3DFMT_D24S8; - default: UNREACHABLE(); return D3DFMT_A8R8G8B8; - } -} - -D3DMULTISAMPLE_TYPE GetMultisampleTypeFromSamples(GLsizei samples) -{ - if (samples <= 1) - return D3DMULTISAMPLE_NONE; - else - return (D3DMULTISAMPLE_TYPE)samples; -} - -} - -namespace d3d9_gl -{ - -unsigned int GetStencilSize(D3DFORMAT stencilFormat) -{ - if (stencilFormat == D3DFMT_INTZ) - { - return 8; - } - switch(stencilFormat) - { - case D3DFMT_D24FS8: - case D3DFMT_D24S8: - return 8; - case D3DFMT_D24X4S4: - return 4; - case D3DFMT_D15S1: - return 1; - case D3DFMT_D16_LOCKABLE: - case D3DFMT_D32: - case D3DFMT_D24X8: - case D3DFMT_D32F_LOCKABLE: - case D3DFMT_D16: - return 0; - //case D3DFMT_D32_LOCKABLE: return 0; // DirectX 9Ex only - //case D3DFMT_S8_LOCKABLE: return 8; // DirectX 9Ex only - default: - return 0; - } -} - -unsigned int GetAlphaSize(D3DFORMAT colorFormat) -{ - switch (colorFormat) - { - case D3DFMT_A16B16G16R16F: - return 16; - case D3DFMT_A32B32G32R32F: - return 32; - case D3DFMT_A2R10G10B10: - return 2; - case D3DFMT_A8R8G8B8: - return 8; - case D3DFMT_A1R5G5B5: - return 1; - case D3DFMT_X8R8G8B8: - case D3DFMT_R5G6B5: - return 0; - default: - return 0; - } -} - -GLsizei GetSamplesFromMultisampleType(D3DMULTISAMPLE_TYPE type) -{ - if (type == D3DMULTISAMPLE_NONMASKABLE) - return 0; - else - return type; -} - -bool IsFormatChannelEquivalent(D3DFORMAT d3dformat, GLenum format) -{ - switch (d3dformat) - { - case D3DFMT_L8: - return (format == GL_LUMINANCE); - case D3DFMT_A8L8: - return (format == GL_LUMINANCE_ALPHA); - case D3DFMT_DXT1: - return (format == GL_COMPRESSED_RGBA_S3TC_DXT1_EXT || format == GL_COMPRESSED_RGB_S3TC_DXT1_EXT); - case D3DFMT_DXT3: - return (format == GL_COMPRESSED_RGBA_S3TC_DXT3_ANGLE); - case D3DFMT_DXT5: - return (format == GL_COMPRESSED_RGBA_S3TC_DXT5_ANGLE); - case D3DFMT_A8R8G8B8: - case D3DFMT_A16B16G16R16F: - case D3DFMT_A32B32G32R32F: - return (format == GL_RGBA || format == GL_BGRA_EXT); - case D3DFMT_X8R8G8B8: - return (format == GL_RGB); - default: - if (d3dformat == D3DFMT_INTZ && gl::IsDepthTexture(format)) - return true; - return false; - } -} - -GLenum ConvertBackBufferFormat(D3DFORMAT format) -{ - switch (format) - { - case D3DFMT_A4R4G4B4: return GL_RGBA4; - case D3DFMT_A8R8G8B8: return GL_RGBA8_OES; - case D3DFMT_A1R5G5B5: return GL_RGB5_A1; - case D3DFMT_R5G6B5: return GL_RGB565; - case D3DFMT_X8R8G8B8: return GL_RGB8_OES; - default: - UNREACHABLE(); - } - - return GL_RGBA4; -} - -GLenum ConvertDepthStencilFormat(D3DFORMAT format) -{ - if (format == D3DFMT_INTZ) - { - return GL_DEPTH24_STENCIL8_OES; - } - switch (format) - { - case D3DFMT_D16: - case D3DFMT_D24X8: - return GL_DEPTH_COMPONENT16; - case D3DFMT_D24S8: - return GL_DEPTH24_STENCIL8_OES; - case D3DFMT_UNKNOWN: - return GL_NONE; - default: - UNREACHABLE(); - } - - return GL_DEPTH24_STENCIL8_OES; -} - -GLenum ConvertRenderTargetFormat(D3DFORMAT format) -{ - if (format == D3DFMT_INTZ) - { - return GL_DEPTH24_STENCIL8_OES; - } - - switch (format) - { - case D3DFMT_A4R4G4B4: return GL_RGBA4; - case D3DFMT_A8R8G8B8: return GL_RGBA8_OES; - case D3DFMT_A1R5G5B5: return GL_RGB5_A1; - case D3DFMT_R5G6B5: return GL_RGB565; - case D3DFMT_X8R8G8B8: return GL_RGB8_OES; - case D3DFMT_D16: - case D3DFMT_D24X8: - return GL_DEPTH_COMPONENT16; - case D3DFMT_D24S8: - return GL_DEPTH24_STENCIL8_OES; - case D3DFMT_UNKNOWN: - return GL_NONE; - default: - UNREACHABLE(); - } - - return GL_RGBA4; -} - -GLenum GetEquivalentFormat(D3DFORMAT format) -{ - if (format == D3DFMT_INTZ) - return GL_DEPTH24_STENCIL8_OES; - if (format == D3DFMT_NULL) - return GL_NONE; - - switch (format) - { - case D3DFMT_A4R4G4B4: return GL_RGBA4; - case D3DFMT_A8R8G8B8: return GL_RGBA8_OES; - case D3DFMT_A1R5G5B5: return GL_RGB5_A1; - case D3DFMT_R5G6B5: return GL_RGB565; - case D3DFMT_X8R8G8B8: return GL_RGB8_OES; - case D3DFMT_D16: return GL_DEPTH_COMPONENT16; - case D3DFMT_D24S8: return GL_DEPTH24_STENCIL8_OES; - case D3DFMT_UNKNOWN: return GL_NONE; - case D3DFMT_DXT1: return GL_COMPRESSED_RGBA_S3TC_DXT1_EXT; - case D3DFMT_DXT3: return GL_COMPRESSED_RGBA_S3TC_DXT3_ANGLE; - case D3DFMT_DXT5: return GL_COMPRESSED_RGBA_S3TC_DXT5_ANGLE; - case D3DFMT_A32B32G32R32F: return GL_RGBA32F_EXT; - case D3DFMT_A16B16G16R16F: return GL_RGBA16F_EXT; - case D3DFMT_L8: return GL_LUMINANCE8_EXT; - case D3DFMT_A8L8: return GL_LUMINANCE8_ALPHA8_EXT; - default: UNREACHABLE(); - return GL_NONE; - } -} - -} - -namespace d3d9 -{ - -bool IsCompressedFormat(D3DFORMAT surfaceFormat) -{ - switch(surfaceFormat) - { - case D3DFMT_DXT1: - case D3DFMT_DXT2: - case D3DFMT_DXT3: - case D3DFMT_DXT4: - case D3DFMT_DXT5: - return true; - default: - return false; - } -} - -size_t ComputeRowSize(D3DFORMAT format, unsigned int width) -{ - if (format == D3DFMT_INTZ) - { - return 4 * width; - } - switch (format) - { - case D3DFMT_L8: - return 1 * width; - case D3DFMT_A8L8: - return 2 * width; - case D3DFMT_X8R8G8B8: - case D3DFMT_A8R8G8B8: - return 4 * width; - case D3DFMT_A16B16G16R16F: - return 8 * width; - case D3DFMT_A32B32G32R32F: - return 16 * width; - case D3DFMT_DXT1: - return 8 * ((width + 3) / 4); - case D3DFMT_DXT3: - case D3DFMT_DXT5: - return 16 * ((width + 3) / 4); - default: - UNREACHABLE(); - return 0; - } -} - -} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/shaders/Blit.ps b/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/shaders/Blit.ps deleted file mode 100644 index dcb3bd0e76..0000000000 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/shaders/Blit.ps +++ /dev/null @@ -1,39 +0,0 @@ -// -// Copyright (c) 2012 The ANGLE Project Authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. -// - -sampler2D tex : s0; - -uniform float4 mode : c0; - -// Passthrough Pixel Shader -// Outputs texture 0 sampled at texcoord 0. -float4 passthroughps(float4 texcoord : TEXCOORD0) : COLOR -{ - return tex2D(tex, texcoord.xy); -}; - -// Luminance Conversion Pixel Shader -// Outputs sample(tex0, tc0).rrra. -// For LA output (pass A) set C0.X = 1, C0.Y = 0. -// For L output (A = 1) set C0.X = 0, C0.Y = 1. -float4 luminanceps(float4 texcoord : TEXCOORD0) : COLOR -{ - float4 tmp = tex2D(tex, texcoord.xy); - tmp.w = tmp.w * mode.x + mode.y; - return tmp.xxxw; -}; - -// RGB/A Component Mask Pixel Shader -// Outputs sample(tex0, tc0) with options to force RGB = 0 and/or A = 1. -// To force RGB = 0, set C0.X = 0, otherwise C0.X = 1. -// To force A = 1, set C0.Z = 0, C0.W = 1, otherwise C0.Z = 1, C0.W = 0. -float4 componentmaskps(float4 texcoord : TEXCOORD0) : COLOR -{ - float4 tmp = tex2D(tex, texcoord.xy); - tmp.xyz = tmp.xyz * mode.x; - tmp.w = tmp.w * mode.z + mode.w; - return tmp; -}; diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/generatemip.h b/src/3rdparty/angle/src/libGLESv2/renderer/generatemip.h index 8e1973605b..a57b00d444 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/generatemip.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/generatemip.h @@ -1,5 +1,5 @@ // -// Copyright (c) 2002-2012 The ANGLE Project Authors. All rights reserved. +// Copyright (c) 2002-2014 The ANGLE Project Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. // @@ -10,194 +10,19 @@ #ifndef LIBGLESV2_RENDERER_GENERATEMIP_H_ #define LIBGLESV2_RENDERER_GENERATEMIP_H_ -#include "libGLESv2/mathutil.h" +#include "libGLESv2/renderer/imageformats.h" +#include "libGLESv2/angletypes.h" namespace rx { -struct L8 -{ - unsigned char L; - - static void average(L8 *dst, const L8 *src1, const L8 *src2) - { - dst->L = ((src1->L ^ src2->L) >> 1) + (src1->L & src2->L); - } -}; - -typedef L8 R8; // R8 type is functionally equivalent for mip purposes -typedef L8 A8; // A8 type is functionally equivalent for mip purposes - -struct A8L8 -{ - unsigned char L; - unsigned char A; - - static void average(A8L8 *dst, const A8L8 *src1, const A8L8 *src2) - { - *(unsigned short*)dst = (((*(unsigned short*)src1 ^ *(unsigned short*)src2) & 0xFEFE) >> 1) + (*(unsigned short*)src1 & *(unsigned short*)src2); - } -}; - -typedef A8L8 R8G8; // R8G8 type is functionally equivalent for mip purposes - -struct A8R8G8B8 -{ - unsigned char B; - unsigned char G; - unsigned char R; - unsigned char A; - - static void average(A8R8G8B8 *dst, const A8R8G8B8 *src1, const A8R8G8B8 *src2) - { - *(unsigned int*)dst = (((*(unsigned int*)src1 ^ *(unsigned int*)src2) & 0xFEFEFEFE) >> 1) + (*(unsigned int*)src1 & *(unsigned int*)src2); - } -}; - -typedef A8R8G8B8 R8G8B8A8; // R8G8B8A8 type is functionally equivalent for mip purposes - -struct A16B16G16R16F -{ - unsigned short R; - unsigned short G; - unsigned short B; - unsigned short A; - - static void average(A16B16G16R16F *dst, const A16B16G16R16F *src1, const A16B16G16R16F *src2) - { - dst->R = gl::float32ToFloat16((gl::float16ToFloat32(src1->R) + gl::float16ToFloat32(src2->R)) * 0.5f); - dst->G = gl::float32ToFloat16((gl::float16ToFloat32(src1->G) + gl::float16ToFloat32(src2->G)) * 0.5f); - dst->B = gl::float32ToFloat16((gl::float16ToFloat32(src1->B) + gl::float16ToFloat32(src2->B)) * 0.5f); - dst->A = gl::float32ToFloat16((gl::float16ToFloat32(src1->A) + gl::float16ToFloat32(src2->A)) * 0.5f); - } -}; - -struct R16F -{ - unsigned short R; - - static void average(R16F *dst, const R16F *src1, const R16F *src2) - { - dst->R = gl::float32ToFloat16((gl::float16ToFloat32(src1->R) + gl::float16ToFloat32(src2->R)) * 0.5f); - } -}; - -struct R16G16F -{ - unsigned short R; - unsigned short G; - - static void average(R16G16F *dst, const R16G16F *src1, const R16G16F *src2) - { - dst->R = gl::float32ToFloat16((gl::float16ToFloat32(src1->R) + gl::float16ToFloat32(src2->R)) * 0.5f); - dst->G = gl::float32ToFloat16((gl::float16ToFloat32(src1->G) + gl::float16ToFloat32(src2->G)) * 0.5f); - } -}; - -struct A32B32G32R32F -{ - float R; - float G; - float B; - float A; - - static void average(A32B32G32R32F *dst, const A32B32G32R32F *src1, const A32B32G32R32F *src2) - { - dst->R = (src1->R + src2->R) * 0.5f; - dst->G = (src1->G + src2->G) * 0.5f; - dst->B = (src1->B + src2->B) * 0.5f; - dst->A = (src1->A + src2->A) * 0.5f; - } -}; - -struct R32F -{ - float R; - - static void average(R32F *dst, const R32F *src1, const R32F *src2) - { - dst->R = (src1->R + src2->R) * 0.5f; - } -}; - -struct R32G32F -{ - float R; - float G; - - static void average(R32G32F *dst, const R32G32F *src1, const R32G32F *src2) - { - dst->R = (src1->R + src2->R) * 0.5f; - dst->G = (src1->G + src2->G) * 0.5f; - } -}; - -struct R32G32B32F -{ - float R; - float G; - float B; - - static void average(R32G32B32F *dst, const R32G32B32F *src1, const R32G32B32F *src2) - { - dst->R = (src1->R + src2->R) * 0.5f; - dst->G = (src1->G + src2->G) * 0.5f; - dst->B = (src1->B + src2->B) * 0.5f; - } -}; template <typename T> -static void GenerateMip(unsigned int sourceWidth, unsigned int sourceHeight, - const unsigned char *sourceData, int sourcePitch, - unsigned char *destData, int destPitch) -{ - unsigned int mipWidth = std::max(1U, sourceWidth >> 1); - unsigned int mipHeight = std::max(1U, sourceHeight >> 1); - - if (sourceHeight == 1) - { - ASSERT(sourceWidth != 1); - - const T *src = (const T*)sourceData; - T *dst = (T*)destData; - - for (unsigned int x = 0; x < mipWidth; x++) - { - T::average(&dst[x], &src[x * 2], &src[x * 2 + 1]); - } - } - else if (sourceWidth == 1) - { - ASSERT(sourceHeight != 1); +inline void GenerateMip(size_t sourceWidth, size_t sourceHeight, size_t sourceDepth, + const uint8_t *sourceData, size_t sourceRowPitch, size_t sourceDepthPitch, + uint8_t *destData, size_t destRowPitch, size_t destDepthPitch); - for (unsigned int y = 0; y < mipHeight; y++) - { - const T *src0 = (const T*)(sourceData + y * 2 * sourcePitch); - const T *src1 = (const T*)(sourceData + y * 2 * sourcePitch + sourcePitch); - T *dst = (T*)(destData + y * destPitch); - - T::average(dst, src0, src1); - } - } - else - { - for (unsigned int y = 0; y < mipHeight; y++) - { - const T *src0 = (const T*)(sourceData + y * 2 * sourcePitch); - const T *src1 = (const T*)(sourceData + y * 2 * sourcePitch + sourcePitch); - T *dst = (T*)(destData + y * destPitch); - - for (unsigned int x = 0; x < mipWidth; x++) - { - T tmp0; - T tmp1; - - T::average(&tmp0, &src0[x * 2], &src0[x * 2 + 1]); - T::average(&tmp1, &src1[x * 2], &src1[x * 2 + 1]); - T::average(&dst[x], &tmp0, &tmp1); - } - } - } -} } +#include "generatemip.inl" + #endif // LIBGLESV2_RENDERER_GENERATEMIP_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/generatemip.inl b/src/3rdparty/angle/src/libGLESv2/renderer/generatemip.inl new file mode 100644 index 0000000000..6788a42f03 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/generatemip.inl @@ -0,0 +1,266 @@ +// +// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// generatemip.inl: Defines the GenerateMip function, templated on the format +// type of the image for which mip levels are being generated. + +#include "common/mathutil.h" + +namespace rx +{ + +namespace priv +{ + +template <typename T> +static inline T *GetPixel(uint8_t *data, size_t x, size_t y, size_t z, size_t rowPitch, size_t depthPitch) +{ + return reinterpret_cast<T*>(data + (x * sizeof(T)) + (y * rowPitch) + (z * depthPitch)); +} + +template <typename T> +static inline const T *GetPixel(const uint8_t *data, size_t x, size_t y, size_t z, size_t rowPitch, size_t depthPitch) +{ + return reinterpret_cast<const T*>(data + (x * sizeof(T)) + (y * rowPitch) + (z * depthPitch)); +} + +template <typename T> +static void GenerateMip_Y(size_t sourceWidth, size_t sourceHeight, size_t sourceDepth, + const uint8_t *sourceData, size_t sourceRowPitch, size_t sourceDepthPitch, + size_t destWidth, size_t destHeight, size_t destDepth, + uint8_t *destData, size_t destRowPitch, size_t destDepthPitch) +{ + ASSERT(sourceWidth == 1); + ASSERT(sourceHeight > 1); + ASSERT(sourceDepth == 1); + + for (size_t y = 0; y < destHeight; y++) + { + const T *src0 = GetPixel<T>(sourceData, 0, y * 2, 0, sourceRowPitch, sourceDepthPitch); + const T *src1 = GetPixel<T>(sourceData, 0, y * 2 + 1, 0, sourceRowPitch, sourceDepthPitch); + T *dst = GetPixel<T>(destData, 0, y, 0, destRowPitch, destDepthPitch); + + T::average(dst, src0, src1); + } +} + +template <typename T> +static void GenerateMip_X(size_t sourceWidth, size_t sourceHeight, size_t sourceDepth, + const uint8_t *sourceData, size_t sourceRowPitch, size_t sourceDepthPitch, + size_t destWidth, size_t destHeight, size_t destDepth, + uint8_t *destData, size_t destRowPitch, size_t destDepthPitch) +{ + ASSERT(sourceWidth > 1); + ASSERT(sourceHeight == 1); + ASSERT(sourceDepth == 1); + + for (size_t x = 0; x < destWidth; x++) + { + const T *src0 = GetPixel<T>(sourceData, x * 2, 0, 0, sourceRowPitch, sourceDepthPitch); + const T *src1 = GetPixel<T>(sourceData, x * 2 + 1, 0, 0, sourceRowPitch, sourceDepthPitch); + T *dst = GetPixel<T>(destData, x, 0, 0, destRowPitch, destDepthPitch); + + T::average(dst, src0, src1); + } +} + +template <typename T> +static void GenerateMip_Z(size_t sourceWidth, size_t sourceHeight, size_t sourceDepth, + const uint8_t *sourceData, size_t sourceRowPitch, size_t sourceDepthPitch, + size_t destWidth, size_t destHeight, size_t destDepth, + uint8_t *destData, size_t destRowPitch, size_t destDepthPitch) +{ + ASSERT(sourceWidth == 1); + ASSERT(sourceHeight == 1); + ASSERT(sourceDepth > 1); + + for (size_t z = 0; z < destDepth; z++) + { + const T *src0 = GetPixel<T>(sourceData, 0, 0, z * 2, sourceRowPitch, sourceDepthPitch); + const T *src1 = GetPixel<T>(sourceData, 0, 0, z * 2 + 1, sourceRowPitch, sourceDepthPitch); + T *dst = GetPixel<T>(destData, 0, 0, z, destRowPitch, destDepthPitch); + + T::average(dst, src0, src1); + } +} + +template <typename T> +static void GenerateMip_XY(size_t sourceWidth, size_t sourceHeight, size_t sourceDepth, + const uint8_t *sourceData, size_t sourceRowPitch, size_t sourceDepthPitch, + size_t destWidth, size_t destHeight, size_t destDepth, + uint8_t *destData, size_t destRowPitch, size_t destDepthPitch) +{ + ASSERT(sourceWidth > 1); + ASSERT(sourceHeight > 1); + ASSERT(sourceDepth == 1); + + for (size_t y = 0; y < destHeight; y++) + { + for (size_t x = 0; x < destWidth; x++) + { + const T *src0 = GetPixel<T>(sourceData, x * 2, y * 2, 0, sourceRowPitch, sourceDepthPitch); + const T *src1 = GetPixel<T>(sourceData, x * 2, y * 2 + 1, 0, sourceRowPitch, sourceDepthPitch); + const T *src2 = GetPixel<T>(sourceData, x * 2 + 1, y * 2, 0, sourceRowPitch, sourceDepthPitch); + const T *src3 = GetPixel<T>(sourceData, x * 2 + 1, y * 2 + 1, 0, sourceRowPitch, sourceDepthPitch); + T *dst = GetPixel<T>(destData, x, y, 0, destRowPitch, destDepthPitch); + + T tmp0, tmp1; + + T::average(&tmp0, src0, src1); + T::average(&tmp1, src2, src3); + T::average(dst, &tmp0, &tmp1); + } + } +} + +template <typename T> +static void GenerateMip_YZ(size_t sourceWidth, size_t sourceHeight, size_t sourceDepth, + const uint8_t *sourceData, size_t sourceRowPitch, size_t sourceDepthPitch, + size_t destWidth, size_t destHeight, size_t destDepth, + uint8_t *destData, size_t destRowPitch, size_t destDepthPitch) +{ + ASSERT(sourceWidth == 1); + ASSERT(sourceHeight > 1); + ASSERT(sourceDepth > 1); + + for (size_t z = 0; z < destDepth; z++) + { + for (size_t y = 0; y < destHeight; y++) + { + const T *src0 = GetPixel<T>(sourceData, 0, y * 2, z * 2, sourceRowPitch, sourceDepthPitch); + const T *src1 = GetPixel<T>(sourceData, 0, y * 2, z * 2 + 1, sourceRowPitch, sourceDepthPitch); + const T *src2 = GetPixel<T>(sourceData, 0, y * 2 + 1, z * 2, sourceRowPitch, sourceDepthPitch); + const T *src3 = GetPixel<T>(sourceData, 0, y * 2 + 1, z * 2 + 1, sourceRowPitch, sourceDepthPitch); + T *dst = GetPixel<T>(destData, 0, y, z, destRowPitch, destDepthPitch); + + T tmp0, tmp1; + + T::average(&tmp0, src0, src1); + T::average(&tmp1, src2, src3); + T::average(dst, &tmp0, &tmp1); + } + } +} + +template <typename T> +static void GenerateMip_XZ(size_t sourceWidth, size_t sourceHeight, size_t sourceDepth, + const uint8_t *sourceData, size_t sourceRowPitch, size_t sourceDepthPitch, + size_t destWidth, size_t destHeight, size_t destDepth, + uint8_t *destData, size_t destRowPitch, size_t destDepthPitch) +{ + ASSERT(sourceWidth > 1); + ASSERT(sourceHeight == 1); + ASSERT(sourceDepth > 1); + + for (size_t z = 0; z < destDepth; z++) + { + for (size_t x = 0; x < destWidth; x++) + { + const T *src0 = GetPixel<T>(sourceData, x * 2, 0, z * 2, sourceRowPitch, sourceDepthPitch); + const T *src1 = GetPixel<T>(sourceData, x * 2, 0, z * 2 + 1, sourceRowPitch, sourceDepthPitch); + const T *src2 = GetPixel<T>(sourceData, x * 2 + 1, 0, z * 2, sourceRowPitch, sourceDepthPitch); + const T *src3 = GetPixel<T>(sourceData, x * 2 + 1, 0, z * 2 + 1, sourceRowPitch, sourceDepthPitch); + T *dst = GetPixel<T>(destData, x, 0, z, destRowPitch, destDepthPitch); + + T tmp0, tmp1; + + T::average(&tmp0, src0, src1); + T::average(&tmp1, src2, src3); + T::average(dst, &tmp0, &tmp1); + } + } +} + +template <typename T> +static void GenerateMip_XYZ(size_t sourceWidth, size_t sourceHeight, size_t sourceDepth, + const uint8_t *sourceData, size_t sourceRowPitch, size_t sourceDepthPitch, + size_t destWidth, size_t destHeight, size_t destDepth, + uint8_t *destData, size_t destRowPitch, size_t destDepthPitch) +{ + ASSERT(sourceWidth > 1); + ASSERT(sourceHeight > 1); + ASSERT(sourceDepth > 1); + + for (size_t z = 0; z < destDepth; z++) + { + for (size_t y = 0; y < destHeight; y++) + { + for (size_t x = 0; x < destWidth; x++) + { + const T *src0 = GetPixel<T>(sourceData, x * 2, y * 2, z * 2, sourceRowPitch, sourceDepthPitch); + const T *src1 = GetPixel<T>(sourceData, x * 2, y * 2, z * 2 + 1, sourceRowPitch, sourceDepthPitch); + const T *src2 = GetPixel<T>(sourceData, x * 2, y * 2 + 1, z * 2, sourceRowPitch, sourceDepthPitch); + const T *src3 = GetPixel<T>(sourceData, x * 2, y * 2 + 1, z * 2 + 1, sourceRowPitch, sourceDepthPitch); + const T *src4 = GetPixel<T>(sourceData, x * 2 + 1, y * 2, z * 2, sourceRowPitch, sourceDepthPitch); + const T *src5 = GetPixel<T>(sourceData, x * 2 + 1, y * 2, z * 2 + 1, sourceRowPitch, sourceDepthPitch); + const T *src6 = GetPixel<T>(sourceData, x * 2 + 1, y * 2 + 1, z * 2, sourceRowPitch, sourceDepthPitch); + const T *src7 = GetPixel<T>(sourceData, x * 2 + 1, y * 2 + 1, z * 2 + 1, sourceRowPitch, sourceDepthPitch); + T *dst = GetPixel<T>(destData, x, y, z, destRowPitch, destDepthPitch); + + T tmp0, tmp1, tmp2, tmp3, tmp4, tmp5; + + T::average(&tmp0, src0, src1); + T::average(&tmp1, src2, src3); + T::average(&tmp2, src4, src5); + T::average(&tmp3, src6, src7); + + T::average(&tmp4, &tmp0, &tmp1); + T::average(&tmp5, &tmp2, &tmp3); + + T::average(dst, &tmp4, &tmp5); + } + } + } +} + + +typedef void (*MipGenerationFunction)(size_t sourceWidth, size_t sourceHeight, size_t sourceDepth, + const uint8_t *sourceData, size_t sourceRowPitch, size_t sourceDepthPitch, + size_t destWidth, size_t destHeight, size_t destDepth, + uint8_t *destData, size_t destRowPitch, size_t destDepthPitch); + +template <typename T> +static MipGenerationFunction GetMipGenerationFunction(size_t sourceWidth, size_t sourceHeight, size_t sourceDepth) +{ + uint8_t index = ((sourceWidth > 1) ? 1 : 0) | + ((sourceHeight > 1) ? 2 : 0) | + ((sourceDepth > 1) ? 4 : 0); + + switch (index) + { + case 0: return NULL; + case 1: return GenerateMip_X<T>; // W x 1 x 1 + case 2: return GenerateMip_Y<T>; // 1 x H x 1 + case 3: return GenerateMip_XY<T>; // W x H x 1 + case 4: return GenerateMip_Z<T>; // 1 x 1 x D + case 5: return GenerateMip_XZ<T>; // W x 1 x D + case 6: return GenerateMip_YZ<T>; // 1 x H x D + case 7: return GenerateMip_XYZ<T>; // W x H x D + } + + UNREACHABLE(); + return NULL; +} + +} + +template <typename T> +inline void GenerateMip(size_t sourceWidth, size_t sourceHeight, size_t sourceDepth, + const uint8_t *sourceData, size_t sourceRowPitch, size_t sourceDepthPitch, + uint8_t *destData, size_t destRowPitch, size_t destDepthPitch) +{ + size_t mipWidth = std::max<size_t>(1, sourceWidth >> 1); + size_t mipHeight = std::max<size_t>(1, sourceHeight >> 1); + size_t mipDepth = std::max<size_t>(1, sourceDepth >> 1); + + priv::MipGenerationFunction generationFunction = priv::GetMipGenerationFunction<T>(sourceWidth, sourceHeight, sourceDepth); + ASSERT(generationFunction != NULL); + + generationFunction(sourceWidth, sourceHeight, sourceDepth, sourceData, sourceRowPitch, sourceDepthPitch, + mipWidth, mipHeight, mipDepth, destData, destRowPitch, destDepthPitch); +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/imageformats.h b/src/3rdparty/angle/src/libGLESv2/renderer/imageformats.h new file mode 100644 index 0000000000..2140a9ee72 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/imageformats.h @@ -0,0 +1,2029 @@ +// +// Copyright (c) 2013 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// imageformats.h: Defines image format types with functions for mip generation +// and copying. + +#ifndef LIBGLESV2_RENDERER_IMAGEFORMATS_H_ +#define LIBGLESV2_RENDERER_IMAGEFORMATS_H_ + +#include "common/mathutil.h" + +namespace rx +{ + +// Several structures share functionality for reading, writing or mipmapping but the layout +// must match the texture format which the structure represents. If collapsing or typedefing +// structs in this header, make sure the functionality and memory layout is exactly the same. + +struct L8 +{ + unsigned char L; + + static void readColor(gl::ColorF *dst, const L8 *src) + { + const float lum = gl::normalizedToFloat(src->L); + dst->red = lum; + dst->green = lum; + dst->blue = lum; + dst->alpha = 1.0f; + } + + static void writeColor(L8 *dst, const gl::ColorF *src) + { + dst->L = gl::floatToNormalized<unsigned char>((src->red + src->green + src->blue) / 3.0f); + } + + static void average(L8 *dst, const L8 *src1, const L8 *src2) + { + dst->L = gl::average(src1->L, src2->L); + } +}; + +struct R8 +{ + unsigned char R; + + static void readColor(gl::ColorF *dst, const R8 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = 0.0f; + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorUI *dst, const R8 *src) + { + dst->red = src->R; + dst->green = 0; + dst->blue = 0; + dst->alpha = 1; + } + + static void writeColor(R8 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned char>(src->red); + } + + static void writeColor(R8 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned char>(src->red); + } + + static void average(R8 *dst, const R8 *src1, const R8 *src2) + { + dst->R = gl::average(src1->R, src2->R); + } +}; + +struct A8 +{ + unsigned char A; + + static void readColor(gl::ColorF *dst, const A8 *src) + { + dst->red = 0.0f; + dst->green = 0.0f; + dst->blue = 0.0f; + dst->alpha = gl::normalizedToFloat(src->A); + } + + static void writeColor(A8 *dst, const gl::ColorF *src) + { + dst->A = gl::floatToNormalized<unsigned char>(src->alpha); + } + + static void average(A8 *dst, const A8 *src1, const A8 *src2) + { + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct L8A8 +{ + unsigned char L; + unsigned char A; + + static void readColor(gl::ColorF *dst, const L8A8 *src) + { + const float lum = gl::normalizedToFloat(src->L); + dst->red = lum; + dst->green = lum; + dst->blue = lum; + dst->alpha = gl::normalizedToFloat(src->A); + } + + static void writeColor(L8A8 *dst, const gl::ColorF *src) + { + dst->L = gl::floatToNormalized<unsigned char>((src->red + src->green + src->blue) / 3.0f); + dst->A = gl::floatToNormalized<unsigned char>(src->alpha); + } + + static void average(L8A8 *dst, const L8A8 *src1, const L8A8 *src2) + { + *(unsigned short*)dst = (((*(unsigned short*)src1 ^ *(unsigned short*)src2) & 0xFEFE) >> 1) + (*(unsigned short*)src1 & *(unsigned short*)src2); + } +}; + +struct A8L8 +{ + unsigned char A; + unsigned char L; + + static void readColor(gl::ColorF *dst, const A8L8 *src) + { + const float lum = gl::normalizedToFloat(src->L); + dst->red = lum; + dst->green = lum; + dst->blue = lum; + dst->alpha = gl::normalizedToFloat(src->A); + } + + static void writeColor(A8L8 *dst, const gl::ColorF *src) + { + dst->L = gl::floatToNormalized<unsigned char>((src->red + src->green + src->blue) / 3.0f); + dst->A = gl::floatToNormalized<unsigned char>(src->alpha); + } + + static void average(A8L8 *dst, const A8L8 *src1, const A8L8 *src2) + { + *(unsigned short*)dst = (((*(unsigned short*)src1 ^ *(unsigned short*)src2) & 0xFEFE) >> 1) + (*(unsigned short*)src1 & *(unsigned short*)src2); + } +}; + +struct R8G8 +{ + unsigned char R; + unsigned char G; + + static void readColor(gl::ColorF *dst, const R8G8 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorUI *dst, const R8G8 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = 0; + dst->alpha = 1; + } + + static void writeColor(R8G8 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned char>(src->red); + dst->G = gl::floatToNormalized<unsigned char>(src->green); + } + + static void writeColor(R8G8 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned char>(src->red); + dst->G = static_cast<unsigned char>(src->green); + } + + static void average(R8G8 *dst, const R8G8 *src1, const R8G8 *src2) + { + *(unsigned short*)dst = (((*(unsigned short*)src1 ^ *(unsigned short*)src2) & 0xFEFE) >> 1) + (*(unsigned short*)src1 & *(unsigned short*)src2); + } +}; + +struct R8G8B8 +{ + unsigned char R; + unsigned char G; + unsigned char B; + + static void readColor(gl::ColorF *dst, const R8G8B8 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorUI *dst, const R8G8B8 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->G; + dst->alpha = 1; + } + + static void writeColor(R8G8B8 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned char>(src->red); + dst->G = gl::floatToNormalized<unsigned char>(src->green); + dst->B = gl::floatToNormalized<unsigned char>(src->blue); + } + + static void writeColor(R8G8B8 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned char>(src->red); + dst->G = static_cast<unsigned char>(src->green); + dst->B = static_cast<unsigned char>(src->blue); + } + + static void average(R8G8B8 *dst, const R8G8B8 *src1, const R8G8B8 *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + } +}; + +struct B8G8R8 +{ + unsigned char B; + unsigned char G; + unsigned char R; + + static void readColor(gl::ColorF *dst, const B8G8R8 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorUI *dst, const B8G8R8 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->G; + dst->alpha = 1; + } + + static void writeColor(B8G8R8 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned char>(src->red); + dst->G = gl::floatToNormalized<unsigned char>(src->green); + dst->B = gl::floatToNormalized<unsigned char>(src->blue); + } + + static void writeColor(B8G8R8 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned char>(src->red); + dst->G = static_cast<unsigned char>(src->green); + dst->B = static_cast<unsigned char>(src->blue); + } + + static void average(B8G8R8 *dst, const B8G8R8 *src1, const B8G8R8 *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + } +}; + +struct R5G6B5 +{ + unsigned short RGB; + + static void readColor(gl::ColorF *dst, const R5G6B5 *src) + { + dst->red = gl::normalizedToFloat<5>(gl::getShiftedData<5, 11>(src->RGB)); + dst->green = gl::normalizedToFloat<6>(gl::getShiftedData<6, 5>(src->RGB)); + dst->blue = gl::normalizedToFloat<5>(gl::getShiftedData<5, 0>(src->RGB)); + dst->alpha = 1.0f; + } + + static void writeColor(R5G6B5 *dst, const gl::ColorF *src) + { + dst->RGB = gl::shiftData<5, 11>(gl::floatToNormalized<5, unsigned short>(src->red)) | + gl::shiftData<6, 5>(gl::floatToNormalized<6, unsigned short>(src->green)) | + gl::shiftData<5, 0>(gl::floatToNormalized<5, unsigned short>(src->blue)); + } + + static void average(R5G6B5 *dst, const R5G6B5 *src1, const R5G6B5 *src2) + { + dst->RGB = gl::shiftData<5, 11>(gl::average(gl::getShiftedData<5, 11>(src1->RGB), gl::getShiftedData<5, 11>(src2->RGB))) | + gl::shiftData<6, 5>(gl::average(gl::getShiftedData<6, 5>(src1->RGB), gl::getShiftedData<6, 5>(src2->RGB))) | + gl::shiftData<5, 0>(gl::average(gl::getShiftedData<5, 0>(src1->RGB), gl::getShiftedData<5, 0>(src2->RGB))); + } +}; + +struct A8R8G8B8 +{ + unsigned char A; + unsigned char R; + unsigned char G; + unsigned char B; + + static void readColor(gl::ColorF *dst, const A8R8G8B8 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = gl::normalizedToFloat(src->A); + } + + static void readColor(gl::ColorUI *dst, const A8R8G8B8 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = src->A; + } + + static void writeColor(A8R8G8B8 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned char>(src->red); + dst->G = gl::floatToNormalized<unsigned char>(src->green); + dst->B = gl::floatToNormalized<unsigned char>(src->blue); + dst->A = gl::floatToNormalized<unsigned char>(src->alpha); + } + + static void writeColor(A8R8G8B8 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned char>(src->red); + dst->G = static_cast<unsigned char>(src->green); + dst->B = static_cast<unsigned char>(src->blue); + dst->A = static_cast<unsigned char>(src->alpha); + } + + static void average(A8R8G8B8 *dst, const A8R8G8B8 *src1, const A8R8G8B8 *src2) + { + *(unsigned int*)dst = (((*(unsigned int*)src1 ^ *(unsigned int*)src2) & 0xFEFEFEFE) >> 1) + (*(unsigned int*)src1 & *(unsigned int*)src2); + } +}; + +struct R8G8B8A8 +{ + unsigned char R; + unsigned char G; + unsigned char B; + unsigned char A; + + static void readColor(gl::ColorF *dst, const R8G8B8A8 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = gl::normalizedToFloat(src->A); + } + + static void readColor(gl::ColorUI *dst, const R8G8B8A8 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = src->A; + } + + static void writeColor(R8G8B8A8 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned char>(src->red); + dst->G = gl::floatToNormalized<unsigned char>(src->green); + dst->B = gl::floatToNormalized<unsigned char>(src->blue); + dst->A = gl::floatToNormalized<unsigned char>(src->alpha); + } + + static void writeColor(R8G8B8A8 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned char>(src->red); + dst->G = static_cast<unsigned char>(src->green); + dst->B = static_cast<unsigned char>(src->blue); + dst->A = static_cast<unsigned char>(src->alpha); + } + + static void average(R8G8B8A8 *dst, const R8G8B8A8 *src1, const R8G8B8A8 *src2) + { + *(unsigned int*)dst = (((*(unsigned int*)src1 ^ *(unsigned int*)src2) & 0xFEFEFEFE) >> 1) + (*(unsigned int*)src1 & *(unsigned int*)src2); + } +}; + +struct B8G8R8A8 +{ + unsigned char B; + unsigned char G; + unsigned char R; + unsigned char A; + + static void readColor(gl::ColorF *dst, const B8G8R8A8 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = gl::normalizedToFloat(src->A); + } + + static void readColor(gl::ColorUI *dst, const B8G8R8A8 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = src->A; + } + + static void writeColor(B8G8R8A8 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned char>(src->red); + dst->G = gl::floatToNormalized<unsigned char>(src->green); + dst->B = gl::floatToNormalized<unsigned char>(src->blue); + dst->A = gl::floatToNormalized<unsigned char>(src->alpha); + } + + static void writeColor(B8G8R8A8 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned char>(src->red); + dst->G = static_cast<unsigned char>(src->green); + dst->B = static_cast<unsigned char>(src->blue); + dst->A = static_cast<unsigned char>(src->alpha); + } + + static void average(B8G8R8A8 *dst, const B8G8R8A8 *src1, const B8G8R8A8 *src2) + { + *(unsigned int*)dst = (((*(unsigned int*)src1 ^ *(unsigned int*)src2) & 0xFEFEFEFE) >> 1) + (*(unsigned int*)src1 & *(unsigned int*)src2); + } +}; + +struct B8G8R8X8 +{ + unsigned char B; + unsigned char G; + unsigned char R; + unsigned char X; + + static void readColor(gl::ColorF *dst, const B8G8R8X8 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorUI *dst, const B8G8R8X8 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = 1; + } + + static void writeColor(B8G8R8X8 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned char>(src->red); + dst->G = gl::floatToNormalized<unsigned char>(src->green); + dst->B = gl::floatToNormalized<unsigned char>(src->blue); + dst->X = 255; + } + + static void writeColor(B8G8R8X8 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned char>(src->red); + dst->G = static_cast<unsigned char>(src->green); + dst->B = static_cast<unsigned char>(src->blue); + dst->X = 255; + } + + static void average(B8G8R8X8 *dst, const B8G8R8X8 *src1, const B8G8R8X8 *src2) + { + *(unsigned int*)dst = (((*(unsigned int*)src1 ^ *(unsigned int*)src2) & 0xFEFEFEFE) >> 1) + (*(unsigned int*)src1 & *(unsigned int*)src2); + dst->X = 255; + } +}; + +struct B5G5R5A1 +{ + unsigned short BGRA; + + static void readColor(gl::ColorF *dst, const B5G5R5A1 *src) + { + dst->alpha = gl::normalizedToFloat<1>(gl::getShiftedData<1, 15>(src->BGRA)); + dst->red = gl::normalizedToFloat<5>(gl::getShiftedData<5, 10>(src->BGRA)); + dst->green = gl::normalizedToFloat<5>(gl::getShiftedData<5, 5>(src->BGRA)); + dst->blue = gl::normalizedToFloat<5>(gl::getShiftedData<5, 0>(src->BGRA)); + } + + static void writeColor(B5G5R5A1 *dst, const gl::ColorF *src) + { + dst->BGRA = gl::shiftData<1, 15>(gl::floatToNormalized<1, unsigned short>(src->alpha)) | + gl::shiftData<5, 10>(gl::floatToNormalized<5, unsigned short>(src->red)) | + gl::shiftData<5, 5>(gl::floatToNormalized<5, unsigned short>(src->green)) | + gl::shiftData<5, 0>(gl::floatToNormalized<5, unsigned short>(src->blue)); + } + + static void average(B5G5R5A1 *dst, const B5G5R5A1 *src1, const B5G5R5A1 *src2) + { + dst->BGRA = gl::shiftData<1, 15>(gl::average(gl::getShiftedData<1, 15>(src1->BGRA), gl::getShiftedData<1, 15>(src2->BGRA))) | + gl::shiftData<5, 10>(gl::average(gl::getShiftedData<5, 10>(src1->BGRA), gl::getShiftedData<5, 10>(src2->BGRA))) | + gl::shiftData<5, 5>(gl::average(gl::getShiftedData<5, 5>(src1->BGRA), gl::getShiftedData<5, 5>(src2->BGRA))) | + gl::shiftData<5, 0>(gl::average(gl::getShiftedData<5, 0>(src1->BGRA), gl::getShiftedData<5, 0>(src2->BGRA))); + } +}; + +struct R5G5B5A1 +{ + unsigned short RGBA; + + static void readColor(gl::ColorF *dst, const R5G5B5A1 *src) + { + dst->alpha = gl::normalizedToFloat<1>(gl::getShiftedData<1, 15>(src->RGBA)); + dst->blue = gl::normalizedToFloat<5>(gl::getShiftedData<5, 10>(src->RGBA)); + dst->green = gl::normalizedToFloat<5>(gl::getShiftedData<5, 5>(src->RGBA)); + dst->red = gl::normalizedToFloat<5>(gl::getShiftedData<5, 0>(src->RGBA)); + } + + static void writeColor(R5G5B5A1 *dst, const gl::ColorF *src) + { + dst->RGBA = gl::shiftData<1, 15>(gl::floatToNormalized<1, unsigned short>(src->alpha)) | + gl::shiftData<5, 10>(gl::floatToNormalized<5, unsigned short>(src->blue)) | + gl::shiftData<5, 5>(gl::floatToNormalized<5, unsigned short>(src->green)) | + gl::shiftData<5, 0>(gl::floatToNormalized<5, unsigned short>(src->red)); + } + + static void average(R5G5B5A1 *dst, const R5G5B5A1 *src1, const R5G5B5A1 *src2) + { + dst->RGBA = gl::shiftData<1, 15>(gl::average(gl::getShiftedData<1, 15>(src1->RGBA), gl::getShiftedData<1, 15>(src2->RGBA))) | + gl::shiftData<5, 10>(gl::average(gl::getShiftedData<5, 10>(src1->RGBA), gl::getShiftedData<5, 10>(src2->RGBA))) | + gl::shiftData<5, 5>(gl::average(gl::getShiftedData<5, 5>(src1->RGBA), gl::getShiftedData<5, 5>(src2->RGBA))) | + gl::shiftData<5, 0>(gl::average(gl::getShiftedData<5, 0>(src1->RGBA), gl::getShiftedData<5, 0>(src2->RGBA))); + } +}; + +struct R4G4B4A4 +{ + unsigned char R : 4; + unsigned char G : 4; + unsigned char B : 4; + unsigned char A : 4; + + static void readColor(gl::ColorF *dst, const R4G4B4A4 *src) + { + dst->red = gl::normalizedToFloat<4>(src->R); + dst->green = gl::normalizedToFloat<4>(src->G); + dst->blue = gl::normalizedToFloat<4>(src->B); + dst->alpha = gl::normalizedToFloat<4>(src->A); + } + + static void writeColor(R4G4B4A4 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<4, unsigned char>(src->red); + dst->G = gl::floatToNormalized<4, unsigned char>(src->green); + dst->B = gl::floatToNormalized<4, unsigned char>(src->blue); + dst->A = gl::floatToNormalized<4, unsigned char>(src->alpha); + } + + static void average(R4G4B4A4 *dst, const R4G4B4A4 *src1, const R4G4B4A4 *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct A4R4G4B4 +{ + unsigned char A : 4; + unsigned char R : 4; + unsigned char G : 4; + unsigned char B : 4; + + static void readColor(gl::ColorF *dst, const A4R4G4B4 *src) + { + dst->red = gl::normalizedToFloat<4>(src->R); + dst->green = gl::normalizedToFloat<4>(src->G); + dst->blue = gl::normalizedToFloat<4>(src->B); + dst->alpha = gl::normalizedToFloat<4>(src->A); + } + + static void writeColor(A4R4G4B4 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<4, unsigned char>(src->red); + dst->G = gl::floatToNormalized<4, unsigned char>(src->green); + dst->B = gl::floatToNormalized<4, unsigned char>(src->blue); + dst->A = gl::floatToNormalized<4, unsigned char>(src->alpha); + } + + static void average(A4R4G4B4 *dst, const A4R4G4B4 *src1, const A4R4G4B4 *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct B4G4R4A4 +{ + unsigned char B : 4; + unsigned char G : 4; + unsigned char R : 4; + unsigned char A : 4; + + static void readColor(gl::ColorF *dst, const B4G4R4A4 *src) + { + dst->red = gl::normalizedToFloat<4>(src->R); + dst->green = gl::normalizedToFloat<4>(src->G); + dst->blue = gl::normalizedToFloat<4>(src->B); + dst->alpha = gl::normalizedToFloat<4>(src->A); + } + + static void writeColor(B4G4R4A4 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<4, unsigned char>(src->red); + dst->G = gl::floatToNormalized<4, unsigned char>(src->green); + dst->B = gl::floatToNormalized<4, unsigned char>(src->blue); + dst->A = gl::floatToNormalized<4, unsigned char>(src->alpha); + } + + static void average(B4G4R4A4 *dst, const B4G4R4A4 *src1, const B4G4R4A4 *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct R16 +{ + unsigned short R; + + static void readColor(gl::ColorF *dst, const R16 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = 0.0f; + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorUI *dst, const R16 *src) + { + dst->red = src->R; + dst->green = 0; + dst->blue = 0; + dst->alpha = 1; + } + + static void writeColor(R16 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned short>(src->red); + } + + static void writeColor(R16 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned short>(src->red); + } + + static void average(R16 *dst, const R16 *src1, const R16 *src2) + { + dst->R = gl::average(src1->R, src2->R); + } +}; + +struct R16G16 +{ + unsigned short R; + unsigned short G; + + static void readColor(gl::ColorF *dst, const R16G16 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorUI *dst, const R16G16 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = 0; + dst->alpha = 1; + } + + static void writeColor(R16G16 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned short>(src->red); + dst->G = gl::floatToNormalized<unsigned short>(src->green); + } + + static void writeColor(R16G16 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned short>(src->red); + dst->G = static_cast<unsigned short>(src->green); + } + + static void average(R16G16 *dst, const R16G16 *src1, const R16G16 *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + } +}; + +struct R16G16B16 +{ + unsigned short R; + unsigned short G; + unsigned short B; + + static void readColor(gl::ColorF *dst, const R16G16B16 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorUI *dst, const R16G16B16 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = 1; + } + + static void writeColor(R16G16B16 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned short>(src->red); + dst->G = gl::floatToNormalized<unsigned short>(src->green); + dst->B = gl::floatToNormalized<unsigned short>(src->blue); + } + + static void writeColor(R16G16B16 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned short>(src->red); + dst->G = static_cast<unsigned short>(src->green); + dst->B = static_cast<unsigned short>(src->blue); + } + + static void average(R16G16B16 *dst, const R16G16B16 *src1, const R16G16B16 *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + } +}; + +struct R16G16B16A16 +{ + unsigned short R; + unsigned short G; + unsigned short B; + unsigned short A; + + static void readColor(gl::ColorF *dst, const R16G16B16A16 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = gl::normalizedToFloat(src->A); + } + + static void readColor(gl::ColorUI *dst, const R16G16B16A16 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = src->A; + } + + static void writeColor(R16G16B16A16 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned short>(src->red); + dst->G = gl::floatToNormalized<unsigned short>(src->green); + dst->B = gl::floatToNormalized<unsigned short>(src->blue); + dst->A = gl::floatToNormalized<unsigned short>(src->alpha); + } + + static void writeColor(R16G16B16A16 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned short>(src->red); + dst->G = static_cast<unsigned short>(src->green); + dst->B = static_cast<unsigned short>(src->blue); + dst->A = static_cast<unsigned short>(src->alpha); + } + + static void average(R16G16B16A16 *dst, const R16G16B16A16 *src1, const R16G16B16A16 *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct R32 +{ + unsigned int R; + + static void readColor(gl::ColorF *dst, const R32 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = 0.0f; + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorUI *dst, const R32 *src) + { + dst->red = src->R; + dst->green = 0; + dst->blue = 0; + dst->alpha = 1; + } + + static void writeColor(R32 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned int>(src->red); + } + + static void writeColor(R32 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned int>(src->red); + } + + static void average(R32 *dst, const R32 *src1, const R32 *src2) + { + dst->R = gl::average(src1->R, src2->R); + } +}; + +struct R32G32 +{ + unsigned int R; + unsigned int G; + + static void readColor(gl::ColorF *dst, const R32G32 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorUI *dst, const R32G32 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = 0; + dst->alpha = 1; + } + + static void writeColor(R32G32 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned int>(src->red); + dst->G = gl::floatToNormalized<unsigned int>(src->green); + } + + static void writeColor(R32G32 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned int>(src->red); + dst->G = static_cast<unsigned int>(src->green); + } + + static void average(R32G32 *dst, const R32G32 *src1, const R32G32 *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + } +}; + +struct R32G32B32 +{ + unsigned int R; + unsigned int G; + unsigned int B; + + static void readColor(gl::ColorF *dst, const R32G32B32 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorUI *dst, const R32G32B32 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = 1; + } + + static void writeColor(R32G32B32 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned int>(src->red); + dst->G = gl::floatToNormalized<unsigned int>(src->green); + dst->B = gl::floatToNormalized<unsigned int>(src->blue); + } + + static void writeColor(R32G32B32 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned int>(src->red); + dst->G = static_cast<unsigned int>(src->green); + dst->B = static_cast<unsigned int>(src->blue); + } + + static void average(R32G32B32 *dst, const R32G32B32 *src1, const R32G32B32 *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + } +}; + +struct R32G32B32A32 +{ + unsigned int R; + unsigned int G; + unsigned int B; + unsigned int A; + + static void readColor(gl::ColorF *dst, const R32G32B32A32 *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = gl::normalizedToFloat(src->A); + } + + static void readColor(gl::ColorUI *dst, const R32G32B32A32 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = src->A; + } + + static void writeColor(R32G32B32A32 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<unsigned int>(src->red); + dst->G = gl::floatToNormalized<unsigned int>(src->green); + dst->B = gl::floatToNormalized<unsigned int>(src->blue); + dst->A = gl::floatToNormalized<unsigned int>(src->alpha); + } + + static void writeColor(R32G32B32A32 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned int>(src->red); + dst->G = static_cast<unsigned int>(src->green); + dst->B = static_cast<unsigned int>(src->blue); + dst->A = static_cast<unsigned int>(src->alpha); + } + + static void average(R32G32B32A32 *dst, const R32G32B32A32 *src1, const R32G32B32A32 *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct R8S +{ + char R; + + static void readColor(gl::ColorF *dst, const R8S *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = 0.0f; + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorI *dst, const R8S *src) + { + dst->red = src->R; + dst->green = 0; + dst->blue = 0; + dst->alpha = 1; + } + + static void writeColor(R8S *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<char>(src->red); + } + + static void writeColor(R8S *dst, const gl::ColorI *src) + { + dst->R = static_cast<char>(src->red); + } + + static void average(R8S *dst, const R8S *src1, const R8S *src2) + { + dst->R = gl::average(src1->R, src2->R); + } +}; + +struct R8G8S +{ + char R; + char G; + + static void readColor(gl::ColorF *dst, const R8G8S *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorI *dst, const R8G8S *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = 0; + dst->alpha = 1; + } + + static void writeColor(R8G8S *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<char>(src->red); + dst->G = gl::floatToNormalized<char>(src->green); + } + + static void writeColor(R8G8S *dst, const gl::ColorI *src) + { + dst->R = static_cast<char>(src->red); + dst->G = static_cast<char>(src->green); + } + + static void average(R8G8S *dst, const R8G8S *src1, const R8G8S *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + } +}; + +struct R8G8B8S +{ + char R; + char G; + char B; + + static void readColor(gl::ColorF *dst, const R8G8B8S *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorI *dst, const R8G8B8S *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = 1; + } + + static void writeColor(R8G8B8S *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<char>(src->red); + dst->G = gl::floatToNormalized<char>(src->green); + dst->B = gl::floatToNormalized<char>(src->blue); + } + + static void writeColor(R8G8B8S *dst, const gl::ColorI *src) + { + dst->R = static_cast<char>(src->red); + dst->G = static_cast<char>(src->green); + dst->B = static_cast<char>(src->blue); + } + + static void average(R8G8B8S *dst, const R8G8B8S *src1, const R8G8B8S *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + } +}; + +struct R8G8B8A8S +{ + char R; + char G; + char B; + char A; + + static void readColor(gl::ColorF *dst, const R8G8B8A8S *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = gl::normalizedToFloat(src->A); + } + + static void readColor(gl::ColorI *dst, const R8G8B8A8S *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = src->A; + } + + static void writeColor(R8G8B8A8S *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<char>(src->red); + dst->G = gl::floatToNormalized<char>(src->green); + dst->B = gl::floatToNormalized<char>(src->blue); + dst->A = gl::floatToNormalized<char>(src->alpha); + } + + static void writeColor(R8G8B8A8S *dst, const gl::ColorI *src) + { + dst->R = static_cast<char>(src->red); + dst->G = static_cast<char>(src->green); + dst->B = static_cast<char>(src->blue); + dst->A = static_cast<char>(src->alpha); + } + + static void average(R8G8B8A8S *dst, const R8G8B8A8S *src1, const R8G8B8A8S *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct R16S +{ + short R; + + static void readColor(gl::ColorF *dst, const R16S *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = 0.0f; + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorI *dst, const R16S *src) + { + dst->red = src->R; + dst->green = 0; + dst->blue = 0; + dst->alpha = 1; + } + + static void writeColor(R16S *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<short>(src->red); + } + + static void writeColor(R16S *dst, const gl::ColorI *src) + { + dst->R = static_cast<short>(src->red); + } + + static void average(R16S *dst, const R16S *src1, const R16S *src2) + { + dst->R = gl::average(src1->R, src2->R); + } +}; + +struct R16G16S +{ + short R; + short G; + + static void readColor(gl::ColorF *dst, const R16G16S *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorI *dst, const R16G16S *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = 0; + dst->alpha = 1; + } + + static void writeColor(R16G16S *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<short>(src->red); + dst->G = gl::floatToNormalized<short>(src->green); + } + + static void writeColor(R16G16S *dst, const gl::ColorI *src) + { + dst->R = static_cast<short>(src->red); + dst->G = static_cast<short>(src->green); + } + + static void average(R16G16S *dst, const R16G16S *src1, const R16G16S *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + } +}; + +struct R16G16B16S +{ + short R; + short G; + short B; + + static void readColor(gl::ColorF *dst, const R16G16B16S *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorI *dst, const R16G16B16S *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = 1; + } + + static void writeColor(R16G16B16S *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<short>(src->red); + dst->G = gl::floatToNormalized<short>(src->green); + dst->B = gl::floatToNormalized<short>(src->blue); + } + + static void writeColor(R16G16B16S *dst, const gl::ColorI *src) + { + dst->R = static_cast<short>(src->red); + dst->G = static_cast<short>(src->green); + dst->B = static_cast<short>(src->blue); + } + + static void average(R16G16B16S *dst, const R16G16B16S *src1, const R16G16B16S *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + } +}; + +struct R16G16B16A16S +{ + short R; + short G; + short B; + short A; + + static void readColor(gl::ColorF *dst, const R16G16B16A16S *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = gl::normalizedToFloat(src->A); + } + + static void readColor(gl::ColorI *dst, const R16G16B16A16S *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = src->A; + } + + static void writeColor(R16G16B16A16S *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<short>(src->red); + dst->G = gl::floatToNormalized<short>(src->green); + dst->B = gl::floatToNormalized<short>(src->blue); + dst->A = gl::floatToNormalized<short>(src->alpha); + } + + static void writeColor(R16G16B16A16S *dst, const gl::ColorI *src) + { + dst->R = static_cast<short>(src->red); + dst->G = static_cast<short>(src->green); + dst->B = static_cast<short>(src->blue); + dst->A = static_cast<short>(src->alpha); + } + + static void average(R16G16B16A16S *dst, const R16G16B16A16S *src1, const R16G16B16A16S *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct R32S +{ + int R; + + static void readColor(gl::ColorF *dst, const R32S *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = 0.0f; + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorI *dst, const R32S *src) + { + dst->red = src->R; + dst->green = 0; + dst->blue = 0; + dst->alpha = 1; + } + + static void writeColor(R32S *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<int>(src->red); + } + + static void writeColor(R32S *dst, const gl::ColorI *src) + { + dst->R = static_cast<int>(src->red); + } + + static void average(R32S *dst, const R32S *src1, const R32S *src2) + { + dst->R = gl::average(src1->R, src2->R); + } +}; + +struct R32G32S +{ + int R; + int G; + + static void readColor(gl::ColorF *dst, const R32G32S *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorI *dst, const R32G32S *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = 0; + dst->alpha = 1; + } + + static void writeColor(R32G32S *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<int>(src->red); + dst->G = gl::floatToNormalized<int>(src->green); + } + + static void writeColor(R32G32S *dst, const gl::ColorI *src) + { + dst->R = static_cast<int>(src->red); + dst->G = static_cast<int>(src->green); + } + + static void average(R32G32S *dst, const R32G32S *src1, const R32G32S *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + } +}; + +struct R32G32B32S +{ + int R; + int G; + int B; + + static void readColor(gl::ColorF *dst, const R32G32B32S *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = 1.0f; + } + + static void readColor(gl::ColorI *dst, const R32G32B32S *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = 1; + } + + static void writeColor(R32G32B32S *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<int>(src->red); + dst->G = gl::floatToNormalized<int>(src->green); + dst->B = gl::floatToNormalized<int>(src->blue); + } + + static void writeColor(R32G32B32S *dst, const gl::ColorI *src) + { + dst->R = static_cast<int>(src->red); + dst->G = static_cast<int>(src->green); + dst->B = static_cast<int>(src->blue); + } + + static void average(R32G32B32S *dst, const R32G32B32S *src1, const R32G32B32S *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + } +}; + +struct R32G32B32A32S +{ + int R; + int G; + int B; + int A; + + static void readColor(gl::ColorF *dst, const R32G32B32A32S *src) + { + dst->red = gl::normalizedToFloat(src->R); + dst->green = gl::normalizedToFloat(src->G); + dst->blue = gl::normalizedToFloat(src->B); + dst->alpha = gl::normalizedToFloat(src->A); + } + + static void readColor(gl::ColorI *dst, const R32G32B32A32S *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = src->A; + } + + static void writeColor(R32G32B32A32S *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<int>(src->red); + dst->G = gl::floatToNormalized<int>(src->green); + dst->B = gl::floatToNormalized<int>(src->blue); + dst->A = gl::floatToNormalized<int>(src->alpha); + } + + static void writeColor(R32G32B32A32S *dst, const gl::ColorI *src) + { + dst->R = static_cast<int>(src->red); + dst->G = static_cast<int>(src->green); + dst->B = static_cast<int>(src->blue); + dst->A = static_cast<int>(src->alpha); + } + + static void average(R32G32B32A32S *dst, const R32G32B32A32S *src1, const R32G32B32A32S *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct A16B16G16R16F +{ + unsigned short A; + unsigned short R; + unsigned short G; + unsigned short B; + + static void readColor(gl::ColorF *dst, const A16B16G16R16F *src) + { + dst->red = gl::float16ToFloat32(src->R); + dst->green = gl::float16ToFloat32(src->G); + dst->blue = gl::float16ToFloat32(src->B); + dst->alpha = gl::float16ToFloat32(src->A); + } + + static void writeColor(A16B16G16R16F *dst, const gl::ColorF *src) + { + dst->R = gl::float32ToFloat16(src->red); + dst->G = gl::float32ToFloat16(src->green); + dst->B = gl::float32ToFloat16(src->blue); + dst->A = gl::float32ToFloat16(src->alpha); + } + + static void average(A16B16G16R16F *dst, const A16B16G16R16F *src1, const A16B16G16R16F *src2) + { + dst->R = gl::averageHalfFloat(src1->R, src2->R); + dst->G = gl::averageHalfFloat(src1->G, src2->G); + dst->B = gl::averageHalfFloat(src1->B, src2->B); + dst->A = gl::averageHalfFloat(src1->A, src2->A); + } +}; + +struct R16G16B16A16F +{ + unsigned short R; + unsigned short G; + unsigned short B; + unsigned short A; + + static void readColor(gl::ColorF *dst, const R16G16B16A16F *src) + { + dst->red = gl::float16ToFloat32(src->R); + dst->green = gl::float16ToFloat32(src->G); + dst->blue = gl::float16ToFloat32(src->B); + dst->alpha = gl::float16ToFloat32(src->A); + } + + static void writeColor(R16G16B16A16F *dst, const gl::ColorF *src) + { + dst->R = gl::float32ToFloat16(src->red); + dst->G = gl::float32ToFloat16(src->green); + dst->B = gl::float32ToFloat16(src->blue); + dst->A = gl::float32ToFloat16(src->alpha); + } + + static void average(R16G16B16A16F *dst, const R16G16B16A16F *src1, const R16G16B16A16F *src2) + { + dst->R = gl::averageHalfFloat(src1->R, src2->R); + dst->G = gl::averageHalfFloat(src1->G, src2->G); + dst->B = gl::averageHalfFloat(src1->B, src2->B); + dst->A = gl::averageHalfFloat(src1->A, src2->A); + } +}; + +struct R16F +{ + unsigned short R; + + static void readColor(gl::ColorF *dst, const R16F *src) + { + dst->red = gl::float16ToFloat32(src->R); + dst->green = 0.0f; + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void writeColor(R16F *dst, const gl::ColorF *src) + { + dst->R = gl::float32ToFloat16(src->red); + } + + static void average(R16F *dst, const R16F *src1, const R16F *src2) + { + dst->R = gl::averageHalfFloat(src1->R, src2->R); + } +}; + +struct A16F +{ + unsigned short A; + + static void readColor(gl::ColorF *dst, const A16F *src) + { + dst->red = 0.0f; + dst->green = 0.0f; + dst->blue = 0.0f; + dst->alpha = gl::float16ToFloat32(src->A); + } + + static void writeColor(A16F *dst, const gl::ColorF *src) + { + dst->A = gl::float32ToFloat16(src->alpha); + } + + static void average(A16F *dst, const A16F *src1, const A16F *src2) + { + dst->A = gl::averageHalfFloat(src1->A, src2->A); + } +}; + +struct L16F +{ + unsigned short L; + + static void readColor(gl::ColorF *dst, const L16F *src) + { + float lum = gl::float16ToFloat32(src->L); + dst->red = lum; + dst->green = lum; + dst->blue = lum; + dst->alpha = 1.0f; + } + + static void writeColor(L16F *dst, const gl::ColorF *src) + { + dst->L = gl::float32ToFloat16((src->red + src->green + src->blue) / 3.0f); + } + + static void average(L16F *dst, const L16F *src1, const L16F *src2) + { + dst->L = gl::averageHalfFloat(src1->L, src2->L); + } +}; + +struct L16A16F +{ + unsigned short L; + unsigned short A; + + static void readColor(gl::ColorF *dst, const L16A16F *src) + { + float lum = gl::float16ToFloat32(src->L); + dst->red = lum; + dst->green = lum; + dst->blue = lum; + dst->alpha = gl::float16ToFloat32(src->A); + } + + static void writeColor(L16A16F *dst, const gl::ColorF *src) + { + dst->L = gl::float32ToFloat16((src->red + src->green + src->blue) / 3.0f); + dst->A = gl::float32ToFloat16(src->alpha); + } + + static void average(L16A16F *dst, const L16A16F *src1, const L16A16F *src2) + { + dst->L = gl::averageHalfFloat(src1->L, src2->L); + dst->A = gl::averageHalfFloat(src1->A, src2->A); + } +}; + +struct R16G16F +{ + unsigned short R; + unsigned short G; + + static void readColor(gl::ColorF *dst, const R16G16F *src) + { + dst->red = gl::float16ToFloat32(src->R); + dst->green = gl::float16ToFloat32(src->G); + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void writeColor(R16G16F *dst, const gl::ColorF *src) + { + dst->R = gl::float32ToFloat16(src->red); + dst->G = gl::float32ToFloat16(src->green); + } + + static void average(R16G16F *dst, const R16G16F *src1, const R16G16F *src2) + { + dst->R = gl::averageHalfFloat(src1->R, src2->R); + dst->G = gl::averageHalfFloat(src1->G, src2->G); + } +}; + +struct R16G16B16F +{ + unsigned short R; + unsigned short G; + unsigned short B; + + static void readColor(gl::ColorF *dst, const R16G16B16F *src) + { + dst->red = gl::float16ToFloat32(src->R); + dst->green = gl::float16ToFloat32(src->G); + dst->blue = gl::float16ToFloat32(src->B); + dst->alpha = 1.0f; + } + + static void writeColor(R16G16B16F *dst, const gl::ColorF *src) + { + dst->R = gl::float32ToFloat16(src->red); + dst->G = gl::float32ToFloat16(src->green); + dst->B = gl::float32ToFloat16(src->blue); + } + + static void average(R16G16B16F *dst, const R16G16B16F *src1, const R16G16B16F *src2) + { + dst->R = gl::averageHalfFloat(src1->R, src2->R); + dst->G = gl::averageHalfFloat(src1->G, src2->G); + dst->B = gl::averageHalfFloat(src1->B, src2->B); + } +}; + +struct A32B32G32R32F +{ + float A; + float R; + float G; + float B; + + static void readColor(gl::ColorF *dst, const A32B32G32R32F *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = src->A; + } + + static void writeColor(A32B32G32R32F *dst, const gl::ColorF *src) + { + dst->R = src->red; + dst->G = src->green; + dst->B = src->blue; + dst->A = src->alpha; + } + + static void average(A32B32G32R32F *dst, const A32B32G32R32F *src1, const A32B32G32R32F *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct R32G32B32A32F +{ + float R; + float G; + float B; + float A; + + static void readColor(gl::ColorF *dst, const R32G32B32A32F *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = src->A; + } + + static void writeColor(R32G32B32A32F *dst, const gl::ColorF *src) + { + dst->R = src->red; + dst->G = src->green; + dst->B = src->blue; + dst->A = src->alpha; + } + + static void average(R32G32B32A32F *dst, const R32G32B32A32F *src1, const R32G32B32A32F *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct R32F +{ + float R; + + static void readColor(gl::ColorF *dst, const R32F *src) + { + dst->red = src->R; + dst->green = 0.0f; + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void writeColor(R32F *dst, const gl::ColorF *src) + { + dst->R = src->red; + } + + static void average(R32F *dst, const R32F *src1, const R32F *src2) + { + dst->R = gl::average(src1->R, src2->R); + } +}; + +struct A32F +{ + float A; + + static void readColor(gl::ColorF *dst, const A32F *src) + { + dst->red = 0.0f; + dst->green = 0.0f; + dst->blue = 0.0f; + dst->alpha = src->A; + } + + static void writeColor(A32F *dst, const gl::ColorF *src) + { + dst->A = src->alpha; + } + + static void average(A32F *dst, const A32F *src1, const A32F *src2) + { + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct L32F +{ + float L; + + static void readColor(gl::ColorF *dst, const L32F *src) + { + dst->red = src->L; + dst->green = src->L; + dst->blue = src->L; + dst->alpha = 1.0f; + } + + static void writeColor(L32F *dst, const gl::ColorF *src) + { + dst->L = (src->red + src->green + src->blue) / 3.0f; + } + + static void average(L32F *dst, const L32F *src1, const L32F *src2) + { + dst->L = gl::average(src1->L, src2->L); + } +}; + +struct L32A32F +{ + float L; + float A; + + static void readColor(gl::ColorF *dst, const L32A32F *src) + { + dst->red = src->L; + dst->green = src->L; + dst->blue = src->L; + dst->alpha = src->A; + } + + static void writeColor(L32A32F *dst, const gl::ColorF *src) + { + dst->L = (src->red + src->green + src->blue) / 3.0f; + dst->A = src->alpha; + } + + static void average(L32A32F *dst, const L32A32F *src1, const L32A32F *src2) + { + dst->L = gl::average(src1->L, src2->L); + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct R32G32F +{ + float R; + float G; + + static void readColor(gl::ColorF *dst, const R32G32F *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = 0.0f; + dst->alpha = 1.0f; + } + + static void writeColor(R32G32F *dst, const gl::ColorF *src) + { + dst->R = src->red; + dst->G = src->green; + } + + static void average(R32G32F *dst, const R32G32F *src1, const R32G32F *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + } +}; + +struct R32G32B32F +{ + float R; + float G; + float B; + + static void readColor(gl::ColorF *dst, const R32G32B32F *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = 1.0f; + } + + static void writeColor(R32G32B32F *dst, const gl::ColorF *src) + { + dst->R = src->red; + dst->G = src->green; + dst->B = src->blue; + } + + static void average(R32G32B32F *dst, const R32G32B32F *src1, const R32G32B32F *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + } +}; + +struct R10G10B10A2 +{ + unsigned int R : 10; + unsigned int G : 10; + unsigned int B : 10; + unsigned int A : 2; + + static void readColor(gl::ColorF *dst, const R10G10B10A2 *src) + { + dst->red = gl::normalizedToFloat<10>(src->R); + dst->green = gl::normalizedToFloat<10>(src->G); + dst->blue = gl::normalizedToFloat<10>(src->B); + dst->alpha = gl::normalizedToFloat< 2>(src->A); + } + + static void readColor(gl::ColorUI *dst, const R10G10B10A2 *src) + { + dst->red = src->R; + dst->green = src->G; + dst->blue = src->B; + dst->alpha = src->A; + } + + static void writeColor(R10G10B10A2 *dst, const gl::ColorF *src) + { + dst->R = gl::floatToNormalized<10, unsigned int>(src->red); + dst->G = gl::floatToNormalized<10, unsigned int>(src->green); + dst->B = gl::floatToNormalized<10, unsigned int>(src->blue); + dst->A = gl::floatToNormalized< 2, unsigned int>(src->alpha); + } + + static void writeColor(R10G10B10A2 *dst, const gl::ColorUI *src) + { + dst->R = static_cast<unsigned int>(src->red); + dst->G = static_cast<unsigned int>(src->green); + dst->B = static_cast<unsigned int>(src->blue); + dst->A = static_cast<unsigned int>(src->alpha); + } + + static void average(R10G10B10A2 *dst, const R10G10B10A2 *src1, const R10G10B10A2 *src2) + { + dst->R = gl::average(src1->R, src2->R); + dst->G = gl::average(src1->G, src2->G); + dst->B = gl::average(src1->B, src2->B); + dst->A = gl::average(src1->A, src2->A); + } +}; + +struct R9G9B9E5 +{ + unsigned int R : 9; + unsigned int G : 9; + unsigned int B : 9; + unsigned int E : 5; + + static void readColor(gl::ColorF *dst, const R9G9B9E5 *src) + { + gl::convert999E5toRGBFloats(gl::bitCast<unsigned int>(*src), &dst->red, &dst->green, &dst->blue); + dst->alpha = 1.0f; + } + + static void writeColor(R9G9B9E5 *dst, const gl::ColorF *src) + { + *reinterpret_cast<unsigned int*>(dst) = gl::convertRGBFloatsTo999E5(src->red, + src->green, + src->blue); + } + + static void average(R9G9B9E5 *dst, const R9G9B9E5 *src1, const R9G9B9E5 *src2) + { + float r1, g1, b1; + gl::convert999E5toRGBFloats(*reinterpret_cast<const unsigned int*>(src1), &r1, &g1, &b1); + + float r2, g2, b2; + gl::convert999E5toRGBFloats(*reinterpret_cast<const unsigned int*>(src2), &r2, &g2, &b2); + + *reinterpret_cast<unsigned int*>(dst) = gl::convertRGBFloatsTo999E5(gl::average(r1, r2), + gl::average(g1, g2), + gl::average(b1, b2)); + } +}; + +struct R11G11B10F +{ + unsigned int R : 11; + unsigned int G : 11; + unsigned int B : 10; + + static void readColor(gl::ColorF *dst, const R11G11B10F *src) + { + dst->red = gl::float11ToFloat32(src->R); + dst->green = gl::float11ToFloat32(src->G); + dst->blue = gl::float10ToFloat32(src->B); + dst->alpha = 1.0f; + } + + static void writeColor(R11G11B10F *dst, const gl::ColorF *src) + { + dst->R = gl::float32ToFloat11(src->red); + dst->G = gl::float32ToFloat11(src->green); + dst->B = gl::float32ToFloat10(src->blue); + } + + static void average(R11G11B10F *dst, const R11G11B10F *src1, const R11G11B10F *src2) + { + dst->R = gl::averageFloat11(src1->R, src2->R); + dst->G = gl::averageFloat11(src1->G, src2->G); + dst->B = gl::averageFloat10(src1->B, src2->B); + } +}; + +} + +#endif // LIBGLESV2_RENDERER_IMAGEFORMATS_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/loadimage.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/loadimage.cpp new file mode 100644 index 0000000000..1986191a75 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/loadimage.cpp @@ -0,0 +1,661 @@ +// +// Copyright (c) 2013-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// loadimage.cpp: Defines image loading functions. + +#include "libGLESv2/renderer/loadimage.h" + +namespace rx +{ + +void LoadA8ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint8_t *source = OffsetDataPointer<uint8_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint32_t *dest = OffsetDataPointer<uint32_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[x] = static_cast<uint32_t>(source[x]) << 24; + } + } + } +} + +void LoadA8ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + // Same as loading to RGBA + LoadA8ToRGBA8(width, height, depth, input, inputRowPitch, inputDepthPitch, output, outputRowPitch, outputDepthPitch); +} + +void LoadA32FToRGBA32F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const float *source = OffsetDataPointer<float>(input, y, z, inputRowPitch, inputDepthPitch); + float *dest = OffsetDataPointer<float>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[4 * x + 0] = 0.0f; + dest[4 * x + 1] = 0.0f; + dest[4 * x + 2] = 0.0f; + dest[4 * x + 3] = source[x]; + } + } + } +} + +void LoadA16FToRGBA16F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint16_t *dest = OffsetDataPointer<uint16_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[4 * x + 0] = 0; + dest[4 * x + 1] = 0; + dest[4 * x + 2] = 0; + dest[4 * x + 3] = source[x]; + } + } + } +} + +void LoadL8ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint8_t *source = OffsetDataPointer<uint8_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[4 * x + 0] = source[x]; + dest[4 * x + 1] = source[x]; + dest[4 * x + 2] = source[x]; + dest[4 * x + 3] = 0xFF; + } + } + } +} + +void LoadL8ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + // Same as loading to RGBA + LoadL8ToRGBA8(width, height, depth, input, inputRowPitch, inputDepthPitch, output, outputRowPitch, outputDepthPitch); +} + +void LoadL32FToRGBA32F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const float *source = OffsetDataPointer<float>(input, y, z, inputRowPitch, inputDepthPitch); + float *dest = OffsetDataPointer<float>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[4 * x + 0] = source[x]; + dest[4 * x + 1] = source[x]; + dest[4 * x + 2] = source[x]; + dest[4 * x + 3] = 1.0f; + } + } + } +} + +void LoadL16FToRGBA16F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint16_t *dest = OffsetDataPointer<uint16_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[4 * x + 0] = source[x]; + dest[4 * x + 1] = source[x]; + dest[4 * x + 2] = source[x]; + dest[4 * x + 3] = gl::Float16One; + } + } + } +} + +void LoadLA8ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint8_t *source = OffsetDataPointer<uint8_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[4 * x + 0] = source[2 * x + 0]; + dest[4 * x + 1] = source[2 * x + 0]; + dest[4 * x + 2] = source[2 * x + 0]; + dest[4 * x + 3] = source[2 * x + 1]; + } + } + } +} + +void LoadLA8ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + // Same as loading to RGBA + LoadLA8ToRGBA8(width, height, depth, input, inputRowPitch, inputDepthPitch, output, outputRowPitch, outputDepthPitch); +} + +void LoadLA32FToRGBA32F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const float *source = OffsetDataPointer<float>(input, y, z, inputRowPitch, inputDepthPitch); + float *dest = OffsetDataPointer<float>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[4 * x + 0] = source[2 * x + 0]; + dest[4 * x + 1] = source[2 * x + 0]; + dest[4 * x + 2] = source[2 * x + 0]; + dest[4 * x + 3] = source[2 * x + 1]; + } + } + } +} + +void LoadLA16FToRGBA16F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint16_t *dest = OffsetDataPointer<uint16_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[4 * x + 0] = source[2 * x + 0]; + dest[4 * x + 1] = source[2 * x + 0]; + dest[4 * x + 2] = source[2 * x + 0]; + dest[4 * x + 3] = source[2 * x + 1]; + } + } + } +} + +void LoadRGB8ToBGRX8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint8_t *source = OffsetDataPointer<uint8_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[4 * x + 0] = source[x * 3 + 2]; + dest[4 * x + 1] = source[x * 3 + 1]; + dest[4 * x + 2] = source[x * 3 + 0]; + dest[4 * x + 3] = 0xFF; + } + } + } +} + +void LoadRG8ToBGRX8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint8_t *source = OffsetDataPointer<uint8_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[4 * x + 0] = 0x00; + dest[4 * x + 1] = source[x * 2 + 1]; + dest[4 * x + 2] = source[x * 2 + 0]; + dest[4 * x + 3] = 0xFF; + } + } + } +} + +void LoadR8ToBGRX8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint8_t *source = OffsetDataPointer<uint8_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[4 * x + 0] = 0x00; + dest[4 * x + 1] = 0x00; + dest[4 * x + 2] = source[x]; + dest[4 * x + 3] = 0xFF; + } + } + } +} + +void LoadR5G6B5ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + uint16_t rgb = source[x]; + dest[4 * x + 0] = ((rgb & 0x001F) << 3) | ((rgb & 0x001F) >> 2); + dest[4 * x + 1] = ((rgb & 0x07E0) >> 3) | ((rgb & 0x07E0) >> 9); + dest[4 * x + 2] = ((rgb & 0xF800) >> 8) | ((rgb & 0xF800) >> 13); + dest[4 * x + 3] = 0xFF; + } + } + } +} + +void LoadR5G6B5ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + uint16_t rgb = source[x]; + dest[4 * x + 0] = ((rgb & 0xF800) >> 8) | ((rgb & 0xF800) >> 13); + dest[4 * x + 1] = ((rgb & 0x07E0) >> 3) | ((rgb & 0x07E0) >> 9); + dest[4 * x + 2] = ((rgb & 0x001F) << 3) | ((rgb & 0x001F) >> 2); + dest[4 * x + 3] = 0xFF; + } + } + } +} + +void LoadRGBA8ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint32_t *source = OffsetDataPointer<uint32_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint32_t *dest = OffsetDataPointer<uint32_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + uint32_t rgba = source[x]; + dest[x] = (_rotl(rgba, 16) & 0x00ff00ff) | (rgba & 0xff00ff00); + } + } + } +} + +void LoadRGBA4ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + uint16_t rgba = source[x]; + dest[4 * x + 0] = ((rgba & 0x00F0) << 0) | ((rgba & 0x00F0) >> 4); + dest[4 * x + 1] = ((rgba & 0x0F00) >> 4) | ((rgba & 0x0F00) >> 8); + dest[4 * x + 2] = ((rgba & 0xF000) >> 8) | ((rgba & 0xF000) >> 12); + dest[4 * x + 3] = ((rgba & 0x000F) << 4) | ((rgba & 0x000F) >> 0); + } + } + } +} + +void LoadRGBA4ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + uint16_t rgba = source[x]; + dest[4 * x + 0] = ((rgba & 0xF000) >> 8) | ((rgba & 0xF000) >> 12); + dest[4 * x + 1] = ((rgba & 0x0F00) >> 4) | ((rgba & 0x0F00) >> 8); + dest[4 * x + 2] = ((rgba & 0x00F0) << 0) | ((rgba & 0x00F0) >> 4); + dest[4 * x + 3] = ((rgba & 0x000F) << 4) | ((rgba & 0x000F) >> 0); + } + } + } +} + +void LoadBGRA4ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + uint16_t bgra = source[x]; + dest[4 * x + 0] = ((bgra & 0xF000) >> 8) | ((bgra & 0xF000) >> 12); + dest[4 * x + 1] = ((bgra & 0x0F00) >> 4) | ((bgra & 0x0F00) >> 8); + dest[4 * x + 2] = ((bgra & 0x00F0) << 0) | ((bgra & 0x00F0) >> 4); + dest[4 * x + 3] = ((bgra & 0x000F) << 4) | ((bgra & 0x000F) >> 0); + } + } + } +} + +void LoadRGB5A1ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + uint16_t rgba = source[x]; + dest[4 * x + 0] = ((rgba & 0x003E) << 2) | ((rgba & 0x003E) >> 3); + dest[4 * x + 1] = ((rgba & 0x07C0) >> 3) | ((rgba & 0x07C0) >> 8); + dest[4 * x + 2] = ((rgba & 0xF800) >> 8) | ((rgba & 0xF800) >> 13); + dest[4 * x + 3] = (rgba & 0x0001) ? 0xFF : 0; + } + } + } +} + +void LoadRGB5A1ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + uint16_t rgba = source[x]; + dest[4 * x + 0] = ((rgba & 0xF800) >> 8) | ((rgba & 0xF800) >> 13); + dest[4 * x + 1] = ((rgba & 0x07C0) >> 3) | ((rgba & 0x07C0) >> 8); + dest[4 * x + 2] = ((rgba & 0x003E) << 2) | ((rgba & 0x003E) >> 3); + dest[4 * x + 3] = (rgba & 0x0001) ? 0xFF : 0; + } + } + } +} + + +void LoadBGR5A1ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + uint16_t bgra = source[x]; + dest[4 * x + 0] = ((bgra & 0xF800) >> 8) | ((bgra & 0xF800) >> 13); + dest[4 * x + 1] = ((bgra & 0x07C0) >> 3) | ((bgra & 0x07C0) >> 8); + dest[4 * x + 2] = ((bgra & 0x003E) << 2) | ((bgra & 0x003E) >> 3); + dest[4 * x + 3] = (bgra & 0x0001) ? 0xFF : 0; + } + } + } +} + +void LoadRGB10A2ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint32_t *source = OffsetDataPointer<uint32_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + uint32_t rgba = source[x]; + dest[4 * x + 0] = (rgba & 0x000003FF) >> 2; + dest[4 * x + 1] = (rgba & 0x000FFC00) >> 12; + dest[4 * x + 2] = (rgba & 0x3FF00000) >> 22; + dest[4 * x + 3] = ((rgba & 0xC0000000) >> 30) * 0x55; + } + } + } +} + +void LoadRGB16FToRGB9E5(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint32_t *dest = OffsetDataPointer<uint32_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[x] = gl::convertRGBFloatsTo999E5(gl::float16ToFloat32(source[x * 3 + 0]), + gl::float16ToFloat32(source[x * 3 + 1]), + gl::float16ToFloat32(source[x * 3 + 2])); + } + } + } +} + +void LoadRGB32FToRGB9E5(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const float *source = OffsetDataPointer<float>(input, y, z, inputRowPitch, inputDepthPitch); + uint32_t *dest = OffsetDataPointer<uint32_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[x] = gl::convertRGBFloatsTo999E5(source[x * 3 + 0], source[x * 3 + 1], source[x * 3 + 2]); + } + } + } +} + +void LoadRGB16FToRG11B10F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint16_t *source = OffsetDataPointer<uint16_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint32_t *dest = OffsetDataPointer<uint32_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[x] = (gl::float32ToFloat11(gl::float16ToFloat32(source[x * 3 + 0])) << 0) | + (gl::float32ToFloat11(gl::float16ToFloat32(source[x * 3 + 1])) << 11) | + (gl::float32ToFloat10(gl::float16ToFloat32(source[x * 3 + 2])) << 22); + } + } + } +} + +void LoadRGB32FToRG11B10F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const float *source = OffsetDataPointer<float>(input, y, z, inputRowPitch, inputDepthPitch); + uint32_t *dest = OffsetDataPointer<uint32_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[x] = (gl::float32ToFloat11(source[x * 3 + 0]) << 0) | + (gl::float32ToFloat11(source[x * 3 + 1]) << 11) | + (gl::float32ToFloat10(source[x * 3 + 2]) << 22); + } + } + } +} + +void LoadG8R24ToR24G8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint32_t *source = OffsetDataPointer<uint32_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint32_t *dest = OffsetDataPointer<uint32_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + uint32_t d = source[x] >> 8; + uint8_t s = source[x] & 0xFF; + dest[x] = d | (s << 24); + } + } + } +} + +void LoadRGB32FToRGBA16F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const float *source = OffsetDataPointer<float>(input, y, z, inputRowPitch, inputDepthPitch); + uint16_t *dest = OffsetDataPointer<uint16_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[x * 4 + 0] = gl::float32ToFloat16(source[x * 3 + 0]); + dest[x * 4 + 1] = gl::float32ToFloat16(source[x * 3 + 1]); + dest[x * 4 + 2] = gl::float32ToFloat16(source[x * 3 + 2]); + dest[x * 4 + 3] = gl::Float16One; + } + } + } +} + +void LoadR32ToR16(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint32_t *source = OffsetDataPointer<uint32_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint16_t *dest = OffsetDataPointer<uint16_t>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[x] = source[x] >> 16; + } + } + } +} + +void LoadR32ToR24G8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint32_t *source = OffsetDataPointer<uint32_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint32_t *dest = OffsetDataPointer<uint32_t>(output, y, z, outputRowPitch, outputDepthPitch); + + for (size_t x = 0; x < width; x++) + { + dest[x] = source[x] >> 8; + } + } + } +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/loadimage.h b/src/3rdparty/angle/src/libGLESv2/renderer/loadimage.h new file mode 100644 index 0000000000..bcdff24a66 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/loadimage.h @@ -0,0 +1,193 @@ +// +// Copyright (c) 2013-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// loadimage.h: Defines image loading functions + +#ifndef LIBGLESV2_RENDERER_LOADIMAGE_H_ +#define LIBGLESV2_RENDERER_LOADIMAGE_H_ + +#include "libGLESv2/angletypes.h" + +#include <cstdint> + +namespace rx +{ + +void LoadA8ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadA8ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadA8ToBGRA8_SSE2(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadA32FToRGBA32F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadA16FToRGBA16F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadL8ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadL8ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadL32FToRGBA32F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadL16FToRGBA16F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadLA8ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadLA8ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadLA32FToRGBA32F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadLA16FToRGBA16F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGB8ToBGRX8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRG8ToBGRX8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadR8ToBGRX8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadR5G6B5ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadR5G6B5ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGBA8ToBGRA8_SSE2(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGBA8ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGBA4ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGBA4ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadBGRA4ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGB5A1ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGB5A1ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadBGR5A1ToBGRA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGB10A2ToRGBA8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGB16FToRGB9E5(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGB32FToRGB9E5(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGB16FToRG11B10F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGB32FToRG11B10F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadG8R24ToR24G8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +template <typename type, size_t componentCount> +inline void LoadToNative(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +template <typename type, uint32_t fourthComponentBits> +inline void LoadToNative3To4(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +template <size_t componentCount> +inline void Load32FTo16F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadRGB32FToRGBA16F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +template <size_t blockWidth, size_t blockHeight, size_t blockSize> +inline void LoadCompressedToNative(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadR32ToR16(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +template <typename type, uint32_t firstBits, uint32_t secondBits, uint32_t thirdBits, uint32_t fourthBits> +inline void Initialize4ComponentData(size_t width, size_t height, size_t depth, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +void LoadR32ToR24G8(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch); + +template <typename T> +inline T *OffsetDataPointer(uint8_t *data, size_t y, size_t z, size_t rowPitch, size_t depthPitch); + +template <typename T> +inline const T *OffsetDataPointer(const uint8_t *data, size_t y, size_t z, size_t rowPitch, size_t depthPitch); + +} + +#include "loadimage.inl" + +#endif // LIBGLESV2_RENDERER_LOADIMAGE_H_ diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/loadimage.inl b/src/3rdparty/angle/src/libGLESv2/renderer/loadimage.inl new file mode 100644 index 0000000000..abd0a3673c --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/loadimage.inl @@ -0,0 +1,156 @@ +// +// Copyright (c) 2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +#include "common/mathutil.h" + +namespace rx +{ + +template <typename T> +inline T *OffsetDataPointer(uint8_t *data, size_t y, size_t z, size_t rowPitch, size_t depthPitch) +{ + return reinterpret_cast<T*>(data + (y * rowPitch) + (z * depthPitch)); +} + +template <typename T> +inline const T *OffsetDataPointer(const uint8_t *data, size_t y, size_t z, size_t rowPitch, size_t depthPitch) +{ + return reinterpret_cast<const T*>(data + (y * rowPitch) + (z * depthPitch)); +} + +template <typename type, size_t componentCount> +inline void LoadToNative(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + const size_t rowSize = width * sizeof(type) * componentCount; + const size_t layerSize = rowSize * height; + const size_t imageSize = layerSize * depth; + + if (layerSize == inputDepthPitch && layerSize == outputDepthPitch) + { + ASSERT(rowSize == inputRowPitch && rowSize == outputRowPitch); + memcpy(output, input, imageSize); + } + else if (rowSize == inputRowPitch && rowSize == outputRowPitch) + { + for (size_t z = 0; z < depth; z++) + { + const type *source = OffsetDataPointer<type>(input, 0, z, inputRowPitch, inputDepthPitch); + type *dest = OffsetDataPointer<type>(output, 0, z, outputRowPitch, outputDepthPitch); + + memcpy(dest, source, layerSize); + } + } + else + { + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const type *source = OffsetDataPointer<type>(input, y, z, inputRowPitch, inputDepthPitch); + type *dest = OffsetDataPointer<type>(output, y, z, outputRowPitch, outputDepthPitch); + memcpy(dest, source, width * sizeof(type) * componentCount); + } + } + } +} + +template <typename type, uint32_t fourthComponentBits> +inline void LoadToNative3To4(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + const type fourthValue = gl::bitCast<type>(fourthComponentBits); + + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const type *source = OffsetDataPointer<type>(input, y, z, inputRowPitch, inputDepthPitch); + type *dest = OffsetDataPointer<type>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + dest[x * 4 + 0] = source[x * 3 + 0]; + dest[x * 4 + 1] = source[x * 3 + 1]; + dest[x * 4 + 2] = source[x * 3 + 2]; + dest[x * 4 + 3] = fourthValue; + } + } + } +} + +template <size_t componentCount> +inline void Load32FTo16F(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + const size_t elementWidth = componentCount * width; + + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const float *source = OffsetDataPointer<float>(input, y, z, inputRowPitch, inputDepthPitch); + uint16_t *dest = OffsetDataPointer<uint16_t>(output, y, z, outputRowPitch, outputDepthPitch); + + for (size_t x = 0; x < elementWidth; x++) + { + dest[x] = gl::float32ToFloat16(source[x]); + } + } + } +} + +template <size_t blockWidth, size_t blockHeight, size_t blockSize> +inline void LoadCompressedToNative(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + const size_t columns = (width + (blockWidth - 1)) / blockWidth; + const size_t rows = (height + (blockHeight - 1)) / blockHeight; + + for (size_t z = 0; z < depth; ++z) + { + for (size_t y = 0; y < rows; ++y) + { + const uint8_t *source = OffsetDataPointer<uint8_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint8_t *dest = OffsetDataPointer<uint8_t>(output, y, z, outputRowPitch, outputDepthPitch); + memcpy(dest, source, columns * blockSize); + } + } +} + +template <typename type, uint32_t firstBits, uint32_t secondBits, uint32_t thirdBits, uint32_t fourthBits> +inline void Initialize4ComponentData(size_t width, size_t height, size_t depth, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ + type writeValues[4] = + { + gl::bitCast<type>(firstBits), + gl::bitCast<type>(secondBits), + gl::bitCast<type>(thirdBits), + gl::bitCast<type>(fourthBits), + }; + + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + type *destRow = OffsetDataPointer<type>(output, y, z, outputRowPitch, outputDepthPitch); + for (size_t x = 0; x < width; x++) + { + type* destPixel = destRow + x * 4; + + // This could potentially be optimized by generating an entire row of initialization + // data and copying row by row instead of pixel by pixel. + memcpy(destPixel, writeValues, sizeof(type) * 4); + } + } + } +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/loadimageSSE2.cpp b/src/3rdparty/angle/src/libGLESv2/renderer/loadimageSSE2.cpp new file mode 100644 index 0000000000..f777b30be6 --- /dev/null +++ b/src/3rdparty/angle/src/libGLESv2/renderer/loadimageSSE2.cpp @@ -0,0 +1,112 @@ +// +// Copyright (c) 2002-2014 The ANGLE Project Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +// + +// loadimageSSE2.cpp: Defines image loading functions. It's +// in a separated file for GCC, which can enable SSE usage only per-file, +// not for code blocks that use SSE2 explicitly. + +#include "libGLESv2/renderer/loadimage.h" + +#if !defined(__SSE2__) && (defined(_M_X64) || _M_IX86_FP == 2) +#define __SSE2__ +#endif + +namespace rx +{ + +void LoadA8ToBGRA8_SSE2(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ +#ifdef __SSE2__ + __m128i zeroWide = _mm_setzero_si128(); + + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint8_t *source = OffsetDataPointer<uint8_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint32_t *dest = OffsetDataPointer<uint32_t>(output, y, z, outputRowPitch, outputDepthPitch); + + size_t x = 0; + + // Make output writes aligned + for (; ((reinterpret_cast<intptr_t>(&dest[x]) & 0xF) != 0 && x < width); x++) + { + dest[x] = static_cast<uint32_t>(source[x]) << 24; + } + + for (; x + 7 < width; x += 8) + { + __m128i sourceData = _mm_loadl_epi64(reinterpret_cast<const __m128i*>(&source[x])); + // Interleave each byte to 16bit, make the lower byte to zero + sourceData = _mm_unpacklo_epi8(zeroWide, sourceData); + // Interleave each 16bit to 32bit, make the lower 16bit to zero + __m128i lo = _mm_unpacklo_epi16(zeroWide, sourceData); + __m128i hi = _mm_unpackhi_epi16(zeroWide, sourceData); + + _mm_store_si128(reinterpret_cast<__m128i*>(&dest[x]), lo); + _mm_store_si128(reinterpret_cast<__m128i*>(&dest[x + 4]), hi); + } + + // Handle the remainder + for (; x < width; x++) + { + dest[x] = static_cast<uint32_t>(source[x]) << 24; + } + } + } +#endif +} + +void LoadRGBA8ToBGRA8_SSE2(size_t width, size_t height, size_t depth, + const uint8_t *input, size_t inputRowPitch, size_t inputDepthPitch, + uint8_t *output, size_t outputRowPitch, size_t outputDepthPitch) +{ +#ifdef __SSE2__ + __m128i brMask = _mm_set1_epi32(0x00ff00ff); + + for (size_t z = 0; z < depth; z++) + { + for (size_t y = 0; y < height; y++) + { + const uint32_t *source = OffsetDataPointer<uint32_t>(input, y, z, inputRowPitch, inputDepthPitch); + uint32_t *dest = OffsetDataPointer<uint32_t>(output, y, z, outputRowPitch, outputDepthPitch); + + size_t x = 0; + + // Make output writes aligned + for (; ((reinterpret_cast<intptr_t>(&dest[x]) & 15) != 0) && x < width; x++) + { + uint32_t rgba = source[x]; + dest[x] = (_rotl(rgba, 16) & 0x00ff00ff) | (rgba & 0xff00ff00); + } + + for (; x + 3 < width; x += 4) + { + __m128i sourceData = _mm_loadu_si128(reinterpret_cast<const __m128i*>(&source[x])); + // Mask out g and a, which don't change + __m128i gaComponents = _mm_andnot_si128(brMask, sourceData); + // Mask out b and r + __m128i brComponents = _mm_and_si128(sourceData, brMask); + // Swap b and r + __m128i brSwapped = _mm_shufflehi_epi16(_mm_shufflelo_epi16(brComponents, _MM_SHUFFLE(2, 3, 0, 1)), _MM_SHUFFLE(2, 3, 0, 1)); + __m128i result = _mm_or_si128(gaComponents, brSwapped); + _mm_store_si128(reinterpret_cast<__m128i*>(&dest[x]), result); + } + + // Perform leftover writes + for (; x < width; x++) + { + uint32_t rgba = source[x]; + dest[x] = (_rotl(rgba, 16) & 0x00ff00ff) | (rgba & 0xff00ff00); + } + } + } +#endif +} + +} diff --git a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/vertexconversion.h b/src/3rdparty/angle/src/libGLESv2/renderer/vertexconversion.h index 590b9d48a3..81ba8a0767 100644 --- a/src/3rdparty/angle/src/libGLESv2/renderer/d3d9/vertexconversion.h +++ b/src/3rdparty/angle/src/libGLESv2/renderer/vertexconversion.h @@ -10,6 +10,10 @@ #ifndef LIBGLESV2_VERTEXCONVERSION_H_ #define LIBGLESV2_VERTEXCONVERSION_H_ +#include <limits> +#include <cstdint> +#include <cstddef> + namespace rx { @@ -154,11 +158,13 @@ struct VertexDataConverter static const bool identity = (WidenRule::initialWidth == WidenRule::finalWidth) && Converter::identity; static const std::size_t finalSize = WidenRule::finalWidth * sizeof(OutputType); - static void convertArray(const InputType *in, std::size_t stride, std::size_t n, OutputType *out) + static void convertArray(const uint8_t *input, size_t stride, size_t n, uint8_t *output) { + OutputType *out = reinterpret_cast<OutputType*>(output); + for (std::size_t i = 0; i < n; i++) { - const InputType *ein = pointerAddBytes(in, i * stride); + const InputType *ein = reinterpret_cast<const InputType*>(input + i * stride); copyComponent(out, ein, 0, static_cast<OutputType>(DefaultValueRule::zero())); copyComponent(out, ein, 1, static_cast<OutputType>(DefaultValueRule::zero())); @@ -169,19 +175,7 @@ struct VertexDataConverter } } - static void convertArray(const void *in, std::size_t stride, std::size_t n, void *out) - { - return convertArray(static_cast<const InputType*>(in), stride, n, static_cast<OutputType*>(out)); - } - private: - // Advance the given pointer by a number of bytes (not pointed-to elements). - template <class T> - static T *pointerAddBytes(T *basePtr, std::size_t numBytes) - { - return reinterpret_cast<T *>(reinterpret_cast<uintptr_t>(basePtr) + numBytes); - } - static void copyComponent(OutputType *out, const InputType *in, std::size_t elementindex, OutputType defaultvalue) { if (WidenRule::finalWidth > elementindex) |