Add a Mach-O decoder to the QPluginLoader

We already had an ELF decoder, which helped us greatly to find the
metadata and that catches most Unix systems (Solaris, QNX, HP-UXi, and
all of the free Unixes). On other Unix systems, aside from Mac OS X,
we simply scanned the entire file for the signature. On Windows, even
without a COFF-PE decoder, we use a LoadLibrary trick to load the
plugin without loading the dependent libraries. In most cases, that
works.

Unfortunately, on Mac OS X we didn't have a decoder and nor could we
do the file scan: because Mac OS X binaries could be fat binaries, we
wouldn't know which architecture's signature we had found.

No more. This adds a full Mach-O decoder to QtCore. It is also capable
of finding the boundaries of the architecture's binary, but that
functionality is disabled since all Qt 5 plugins have plugin metadata
sections.

Change-Id: I2d5c04c5ecf024864b8a43f31ab6b7e6c5eae9ce
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

5 files changed, 324 insertions, 6 deletions

diff --git a/src/corelib/global/qglobal.h b/src/corelib/global/qglobal.h
index 2132e555cd..f9629ff430 100644
--- a/src/corelib/global/qglobal.h
+++ b/src/corelib/global/qglobal.h
@@ -74,6 +74,9 @@
 #if defined (__ELF__)
 #  define Q_OF_ELF
 #endif
+#if defined (__MACH__) && defined (__APPLE__)
+#  define Q_OF_MACH_O
+#endif
 
 #ifdef __cplusplus
 
diff --git a/src/corelib/plugin/plugin.pri b/src/corelib/plugin/plugin.pri
index eb7a7f7fa8..338b3d0972 100644
--- a/src/corelib/plugin/plugin.pri
+++ b/src/corelib/plugin/plugin.pri
@@ -9,14 +9,16 @@ HEADERS += \
 	plugin/quuid.h \
 	plugin/qfactoryloader_p.h \
 	plugin/qsystemlibrary_p.h \
-	plugin/qelfparser_p.h
+        plugin/qelfparser_p.h \
+        plugin/qmachparser_p.h
 
 SOURCES += \
 	plugin/qpluginloader.cpp \
 	plugin/qfactoryloader.cpp \
 	plugin/quuid.cpp \
 	plugin/qlibrary.cpp \
-	plugin/qelfparser_p.cpp
+        plugin/qelfparser_p.cpp \
+        plugin/qmachparser.cpp
 
 win32 {
 	SOURCES += \
diff --git a/src/corelib/plugin/qlibrary.cpp b/src/corelib/plugin/qlibrary.cpp
index 3432f9619d..f015c3c236 100644
--- a/src/corelib/plugin/qlibrary.cpp
+++ b/src/corelib/plugin/qlibrary.cpp
@@ -1,7 +1,7 @@
-
 /****************************************************************************
 **
 ** Copyright (C) 2013 Digia Plc and/or its subsidiary(-ies).
+** Copyright (C) 2013 Intel Corporation
 ** Contact: http://www.qt-project.org/legal
 **
 ** This file is part of the QtCore module of the Qt Toolkit.
@@ -64,6 +64,7 @@
 #include <qjsondocument.h>
 #include <qjsonvalue.h>
 #include "qelfparser_p.h"
+#include "qmachparser_p.h"
 
 QT_BEGIN_NAMESPACE
 
@@ -180,7 +181,7 @@ QT_BEGIN_NAMESPACE
 */
 
 
-#if defined(Q_OS_UNIX) && !defined(Q_OS_MAC)
+#if defined(Q_OS_UNIX)
 
 static long qt_find_pattern(const char *s, ulong s_len,
                              const char *pattern, ulong p_len)
@@ -253,7 +254,7 @@ static bool qt_unix_query(const QString &library, QLibraryPrivate *lib)
     }
 
     /*
-       ELF binaries on GNU, have .qplugin sections.
+       ELF and Mach-O binaries with GCC have .qplugin sections.
     */
     bool hasMetaData = false;
     long pos = 0;
@@ -274,6 +275,26 @@ static bool qt_unix_query(const QString &library, QLibraryPrivate *lib)
             pos += rel;
         hasMetaData = true;
     }
+#elif defined (Q_OF_MACH_O)
+    {
+        QString errorString;
+        int r = QMachOParser::parse(filedata, fdlen, library, &errorString, &pos, &fdlen);
+        if (r == QMachOParser::NotSuitable) {
+            if (qt_debug_component())
+                qWarning("QMachOParser: %s", qPrintable(errorString));
+            if (lib)
+                lib->errorString = errorString;
+            return false;
+        }
+        // even if the metadata section was not found, the Mach-O parser will
+        // at least return the boundaries of the right architecture
+        long rel = qt_find_pattern(filedata + pos, fdlen, pattern, plen);
+        if (rel < 0)
+            pos = -1;
+        else
+            pos += rel;
+        hasMetaData = true;
+    }
 #else
     pos = qt_find_pattern(filedata, fdlen, pattern, plen);
     if (pos > 0)
@@ -690,7 +711,7 @@ void QLibraryPrivate::updatePluginState()
     }
 #endif
 
-#if defined(Q_OS_UNIX) && !defined(Q_OS_MAC)
+#if defined(Q_OS_UNIX)
     if (!pHnd) {
         // use unix shortcut to avoid loading the library
         success = qt_unix_query(fileName, this);
diff --git a/src/corelib/plugin/qmachparser.cpp b/src/corelib/plugin/qmachparser.cpp
new file mode 100644
index 0000000000..b9a56a403c
--- /dev/null
+++ b/src/corelib/plugin/qmachparser.cpp
@@ -0,0 +1,213 @@
+/****************************************************************************
+**
+** Copyright (C) 2013 Intel Corporation
+** Contact: http://www.qt-project.org/legal
+**
+** This file is part of the QtCore module of the Qt Toolkit.
+**
+** $QT_BEGIN_LICENSE:LGPL$
+** Commercial License Usage
+** Licensees holding valid commercial Qt licenses may use this file in
+** accordance with the commercial license agreement provided with the
+** Software or, alternatively, in accordance with the terms contained in
+** a written agreement between you and Digia.  For licensing terms and
+** conditions see http://qt.digia.com/licensing.  For further information
+** use the contact form at http://qt.digia.com/contact-us.
+**
+** GNU Lesser General Public License Usage
+** Alternatively, this file may be used under the terms of the GNU Lesser
+** General Public License version 2.1 as published by the Free Software
+** Foundation and appearing in the file LICENSE.LGPL included in the
+** packaging of this file.  Please review the following information to
+** ensure the GNU Lesser General Public License version 2.1 requirements
+** will be met: http://www.gnu.org/licenses/old-licenses/lgpl-2.1.html.
+**
+** In addition, as a special exception, Digia gives you certain additional
+** rights.  These rights are described in the Digia Qt LGPL Exception
+** version 1.1, included in the file LGPL_EXCEPTION.txt in this package.
+**
+** GNU General Public License Usage
+** Alternatively, this file may be used under the terms of the GNU
+** General Public License version 3.0 as published by the Free Software
+** Foundation and appearing in the file LICENSE.GPL included in the
+** packaging of this file.  Please review the following information to
+** ensure the GNU General Public License version 3.0 requirements will be
+** met: http://www.gnu.org/copyleft/gpl.html.
+**
+**
+** $QT_END_LICENSE$
+**
+****************************************************************************/
+
+#include "qmachparser_p.h"
+
+#if defined(Q_OF_MACH_O) && !defined(QT_NO_LIBRARY)
+
+#include <qendian.h>
+#include "qlibrary_p.h"
+
+#include <mach-o/loader.h>
+#include <mach-o/fat.h>
+
+QT_BEGIN_NAMESPACE
+
+#if defined(Q_PROCESSOR_X86_64)
+#  define MACHO64
+static const cpu_type_t my_cputype = CPU_TYPE_X86_64;
+#elif defined(Q_PROCESSOR_X86_32)
+static const cpu_type_t my_cputype = CPU_TYPE_X86;
+#elif defined(Q_PROCESSOR_POWER_64)
+#  define MACHO64
+static const cpu_type_t my_cputype = CPU_TYPE_POWERPC64;
+#elif defined(Q_PROCESSOR_POWER_32)
+static const cpu_type_t my_cputype = CPU_TYPE_POWERPC;
+#elif defined(Q_PROCESSOR_ARM)
+static const cpu_type_t my_cputype = CPU_TYPE_ARM;
+#else
+#  error "Unknown CPU type"
+#endif
+
+#ifdef MACHO64
+#  undef MACHO64
+typedef mach_header_64 my_mach_header;
+typedef segment_command_64 my_segment_command;
+typedef section_64 my_section;
+static const uint32_t my_magic = MH_MAGIC_64;
+#else
+typedef mach_header my_mach_header;
+typedef segment_command my_segment_command;
+typedef section my_section;
+static const uint32_t my_magic = MH_MAGIC;
+#endif
+
+static int ns(const QString &reason, const QString &library, QString *errorString)
+{
+    if (errorString)
+        *errorString = QLibrary::tr("'%1' is not a valid Mach-O binary (%2)")
+                .arg(library, reason.isEmpty() ? QLibrary::tr("file is corrupt") : reason);
+    return QMachOParser::NotSuitable;
+}
+
+int QMachOParser::parse(const char *m_s, ulong fdlen, const QString &library, QString *errorString, long *pos, ulong *sectionlen)
+{
+    // we test all possibilities so that we report whether a file is a binary or not
+    const fat_header *fat = reinterpret_cast<const fat_header *>(m_s);
+    const mach_header *mh = reinterpret_cast<const mach_header *>(m_s);
+    const mach_header_64 *mh64 = reinterpret_cast<const mach_header_64 *>(m_s);
+    if (fdlen < sizeof(uint32_t) && fat->magic != qToBigEndian(FAT_MAGIC)
+            && mh->magic != MH_MAGIC && mh->magic != MH_CIGAM
+            && mh64->magic != MH_MAGIC_64 && mh64->magic != MH_CIGAM_64) {
+        if (errorString)
+            *errorString = QLibrary::tr("'%1' is not a Mach-O binary (%2)").arg(library, QLibrary::tr("invalid magic"));
+        return NotSuitable;
+    }
+
+    // find out if this is a fat Mach-O binary first
+    const my_mach_header *header = 0;
+    if (fat->magic == qToBigEndian(FAT_MAGIC)) {
+        // find our architecture in the binary
+        const fat_arch *arch = reinterpret_cast<const fat_arch *>(m_s + sizeof(*fat));
+        if (Q_UNLIKELY(fdlen < sizeof(*fat) + 2 * sizeof(*arch))) {
+            // fat binaries must contain at least two architectures
+            return ns(QLibrary::tr("file too small"), library, errorString);
+        }
+
+        int count = qFromBigEndian(fat->nfat_arch);
+        if (Q_UNLIKELY(fdlen < sizeof(*fat) + sizeof(*arch) * count))
+            return ns(QString(), library, errorString);
+
+        for (int i = 0; i < count; ++i) {
+            if (arch[i].cputype == qToBigEndian(my_cputype)) {
+                // ### should we check the CPU subtype? Maybe on ARM?
+                uint32_t size = qFromBigEndian(arch[i].size);
+                uint32_t offset = qFromBigEndian(arch[i].offset);
+                if (Q_UNLIKELY(size + offset > fdlen || size < sizeof(my_mach_header)))
+                    return ns(QString(), library, errorString);
+
+                header = reinterpret_cast<const my_mach_header *>(m_s + offset);
+                fdlen = size;
+                break;
+            }
+        }
+        if (!header)
+            return ns(QLibrary::tr("no suitable architecture in fat binary"), library, errorString);
+
+        // check the magic again
+        if (Q_UNLIKELY(header->magic != my_magic))
+            return ns(QString(), library, errorString);
+    } else {
+        header = reinterpret_cast<const my_mach_header *>(m_s);
+        fat = 0;
+
+        // check magic
+        if (header->magic != my_magic)
+            return ns(QLibrary::tr("invalid magic"), library, errorString);
+        if (Q_UNLIKELY(fdlen < sizeof(my_mach_header)))
+            return ns(QString(), library, errorString);
+    }
+
+    // from this point on, fdlen is specific to this architecture
+    // from this point on, everything is in host byte order
+    *pos = reinterpret_cast<const char *>(header) - m_s;
+
+    // (re-)check the CPU type
+    // ### should we check the CPU subtype? Maybe on ARM?
+    if (header->cputype != my_cputype) {
+        if (fat)
+            return ns(QString(), library, errorString);
+        return ns(QLibrary::tr("wrong architecture"), library, errorString);
+    }
+
+    // check the file type
+    if (Q_UNLIKELY(header->filetype != MH_BUNDLE && header->filetype != MH_DYLIB))
+        return ns(QLibrary::tr("not a dynamic library"), library, errorString);
+
+    // find the __TEXT segment, "qtmetadata" section
+    const my_segment_command *seg = reinterpret_cast<const my_segment_command *>(header + 1);
+    ulong minsize = sizeof(*header);
+
+    for (uint i = 0; i < header->ncmds; ++i,
+         seg = reinterpret_cast<const my_segment_command *>(reinterpret_cast<const char *>(seg) + seg->cmdsize)) {
+        if (Q_UNLIKELY(fdlen < minsize + sizeof(load_command)))
+            return ns(QString(), library, errorString);
+
+        minsize += seg->cmdsize;
+        if (Q_UNLIKELY(fdlen < minsize))
+            return ns(QString(), library, errorString);
+
+        const uint32_t MyLoadCommand = sizeof(void *) > 4 ? LC_SEGMENT_64 : LC_SEGMENT;
+        if (seg->cmd != MyLoadCommand)
+            continue;
+
+        // is this the __TEXT segment?
+        if (strcmp(seg->segname, "__TEXT") == 0) {
+            const my_section *sect = reinterpret_cast<const my_section *>(seg + 1);
+            for (uint j = 0; j < seg->nsects; ++j) {
+                // is this the "qtmetadata" section?
+                if (strcmp(sect[j].sectname, "qtmetadata") != 0)
+                    continue;
+
+                // found it!
+                if (Q_UNLIKELY(fdlen < sect[j].offset + sect[j].size))
+                    return ns(QString(), library, errorString);
+
+                *pos += sect[j].offset;
+                *sectionlen = sect[j].size;
+                return QtMetaDataSection;
+            }
+        }
+
+        // other type of segment
+        seg = reinterpret_cast<const my_segment_command *>(reinterpret_cast<const char *>(seg) + seg->cmdsize);
+    }
+
+//    // No Qt section was found, but at least we know that where the proper architecture's boundaries are
+//    return NoQtSection;
+    if (errorString)
+        *errorString = QLibrary::tr("'%1' is not a Qt plugin").arg(library);
+    return NotSuitable;
+}
+
+QT_END_NAMESPACE
+
+#endif
diff --git a/src/corelib/plugin/qmachparser_p.h b/src/corelib/plugin/qmachparser_p.h
new file mode 100644
index 0000000000..6b2774dab6
--- /dev/null
+++ b/src/corelib/plugin/qmachparser_p.h
@@ -0,0 +1,79 @@
+/****************************************************************************
+**
+** Copyright (C) 2013 Intel Corporation
+** Contact: http://www.qt-project.org/legal
+**
+** This file is part of the QtCore module of the Qt Toolkit.
+**
+** $QT_BEGIN_LICENSE:LGPL$
+** Commercial License Usage
+** Licensees holding valid commercial Qt licenses may use this file in
+** accordance with the commercial license agreement provided with the
+** Software or, alternatively, in accordance with the terms contained in
+** a written agreement between you and Digia.  For licensing terms and
+** conditions see http://qt.digia.com/licensing.  For further information
+** use the contact form at http://qt.digia.com/contact-us.
+**
+** GNU Lesser General Public License Usage
+** Alternatively, this file may be used under the terms of the GNU Lesser
+** General Public License version 2.1 as published by the Free Software
+** Foundation and appearing in the file LICENSE.LGPL included in the
+** packaging of this file.  Please review the following information to
+** ensure the GNU Lesser General Public License version 2.1 requirements
+** will be met: http://www.gnu.org/licenses/old-licenses/lgpl-2.1.html.
+**
+** In addition, as a special exception, Digia gives you certain additional
+** rights.  These rights are described in the Digia Qt LGPL Exception
+** version 1.1, included in the file LGPL_EXCEPTION.txt in this package.
+**
+** GNU General Public License Usage
+** Alternatively, this file may be used under the terms of the GNU
+** General Public License version 3.0 as published by the Free Software
+** Foundation and appearing in the file LICENSE.GPL included in the
+** packaging of this file.  Please review the following information to
+** ensure the GNU General Public License version 3.0 requirements will be
+** met: http://www.gnu.org/copyleft/gpl.html.
+**
+**
+** $QT_END_LICENSE$
+**
+****************************************************************************/
+
+#ifndef QMACHPARSER_P_H
+#define QMACHPARSER_P_H
+
+//
+//  W A R N I N G
+//  -------------
+//
+// This file is not part of the Qt API.  It exists purely as an
+// implementation detail.  This header file may change from version to
+// version without notice, or even be removed.
+//
+// We mean it.
+//
+
+#include <qendian.h>
+#include <qglobal.h>
+
+#ifndef QT_NO_LIBRARY
+#if defined(Q_OF_MACH_O)
+
+QT_BEGIN_NAMESPACE
+
+class QString;
+class QLibraryPrivate;
+
+class Q_AUTOTEST_EXPORT QMachOParser
+{
+public:
+    enum { QtMetaDataSection, NoQtSection, NotSuitable };
+    static int parse(const char *m_s, ulong fdlen, const QString &library, QString *errorString, long *pos, ulong *sectionlen);
+};
+
+QT_END_NAMESPACE
+
+#endif // defined(Q_OF_ELF) && defined(Q_CC_GNU)
+#endif // QT_NO_LIBRARY
+
+#endif // QMACHPARSER_P_H