summaryrefslogtreecommitdiffstats
path: root/botan/src/math/bigint/mp_amd64
diff options
context:
space:
mode:
Diffstat (limited to 'botan/src/math/bigint/mp_amd64')
-rw-r--r--botan/src/math/bigint/mp_amd64/info.txt19
-rw-r--r--botan/src/math/bigint/mp_amd64/mp_asm.h69
-rw-r--r--botan/src/math/bigint/mp_amd64/mp_asmi.h243
3 files changed, 331 insertions, 0 deletions
diff --git a/botan/src/math/bigint/mp_amd64/info.txt b/botan/src/math/bigint/mp_amd64/info.txt
new file mode 100644
index 0000000..84a5bcf
--- /dev/null
+++ b/botan/src/math/bigint/mp_amd64/info.txt
@@ -0,0 +1,19 @@
+realname "MPI Core (x86-64)"
+
+mp_bits 64
+
+load_on dep
+
+<add>
+mp_asm.h
+mp_asmi.h
+</add>
+
+<arch>
+amd64
+</arch>
+
+<cc>
+gcc
+icc
+</cc>
diff --git a/botan/src/math/bigint/mp_amd64/mp_asm.h b/botan/src/math/bigint/mp_amd64/mp_asm.h
new file mode 100644
index 0000000..fa66d04
--- /dev/null
+++ b/botan/src/math/bigint/mp_amd64/mp_asm.h
@@ -0,0 +1,69 @@
+/*
+* Lowest Level MPI Algorithms
+* (C) 1999-2008 Jack Lloyd
+* 2006 Luca Piccarreta
+*
+* Distributed under the terms of the Botan license
+*/
+
+#ifndef BOTAN_MP_ASM_H__
+#define BOTAN_MP_ASM_H__
+
+#include <botan/mp_types.h>
+
+#if (BOTAN_MP_WORD_BITS != 64)
+ #error The mp_amd64 module requires that BOTAN_MP_WORD_BITS == 64
+#endif
+
+namespace Botan {
+
+extern "C" {
+
+/*
+* Helper Macros for amd64 Assembly
+*/
+#define ASM(x) x "\n\t"
+
+/*
+* Word Multiply
+*/
+inline word word_madd2(word a, word b, word* c)
+ {
+ asm(
+ ASM("mulq %[b]")
+ ASM("addq %[c],%[a]")
+ ASM("adcq $0,%[carry]")
+
+ : [a]"=a"(a), [b]"=rm"(b), [carry]"=&d"(*c)
+ : "0"(a), "1"(b), [c]"g"(*c) : "cc");
+
+ return a;
+ }
+
+/*
+* Word Multiply/Add
+*/
+inline word word_madd3(word a, word b, word c, word* d)
+ {
+ asm(
+ ASM("mulq %[b]")
+
+ ASM("addq %[c],%[a]")
+ ASM("adcq $0,%[carry]")
+
+ ASM("addq %[d],%[a]")
+ ASM("adcq $0,%[carry]")
+
+ : [a]"=a"(a), [b]"=rm"(b), [carry]"=&d"(*d)
+ : "0"(a), "1"(b), [c]"g"(c), [d]"g"(*d) : "cc");
+
+ return a;
+ }
+
+#undef ASM
+
+}
+
+}
+
+#endif
diff --git a/botan/src/math/bigint/mp_amd64/mp_asmi.h b/botan/src/math/bigint/mp_amd64/mp_asmi.h
new file mode 100644
index 0000000..8bccbaa
--- /dev/null
+++ b/botan/src/math/bigint/mp_amd64/mp_asmi.h
@@ -0,0 +1,243 @@
+/*
+* Lowest Level MPI Algorithms
+* (C) 1999-2007 Jack Lloyd
+* 2006 Luca Piccarreta
+*
+* Distributed under the terms of the Botan license
+*/
+
+#ifndef BOTAN_MP_ASM_INTERNAL_H__
+#define BOTAN_MP_ASM_INTERNAL_H__
+
+#include <botan/mp_asm.h>
+
+namespace Botan {
+
+extern "C" {
+
+/*
+* Helper Macros for amd64 Assembly
+*/
+#ifndef ASM
+ #define ASM(x) x "\n\t"
+#endif
+
+#define ADDSUB2_OP(OPERATION, INDEX) \
+ ASM("movq 8*" #INDEX "(%[y]), %[carry]") \
+ ASM(OPERATION " %[carry], 8*" #INDEX "(%[x])") \
+
+#define ADDSUB3_OP(OPERATION, INDEX) \
+ ASM("movq 8*" #INDEX "(%[x]), %[carry]") \
+ ASM(OPERATION " 8*" #INDEX "(%[y]), %[carry]") \
+ ASM("movq %[carry], 8*" #INDEX "(%[z])") \
+
+#define LINMUL_OP(WRITE_TO, INDEX) \
+ ASM("movq 8*" #INDEX "(%[x]),%%rax") \
+ ASM("mulq %[y]") \
+ ASM("addq %[carry],%%rax") \
+ ASM("adcq $0,%%rdx") \
+ ASM("movq %%rdx,%[carry]") \
+ ASM("movq %%rax, 8*" #INDEX "(%[" WRITE_TO "])")
+
+#define MULADD_OP(IGNORED, INDEX) \
+ ASM("movq 8*" #INDEX "(%[x]),%%rax") \
+ ASM("mulq %[y]") \
+ ASM("addq %[carry],%%rax") \
+ ASM("adcq $0,%%rdx") \
+ ASM("addq 8*" #INDEX "(%[z]),%%rax") \
+ ASM("adcq $0,%%rdx") \
+ ASM("movq %%rdx,%[carry]") \
+ ASM("movq %%rax, 8*" #INDEX " (%[z])")
+
+#define DO_8_TIMES(MACRO, ARG) \
+ MACRO(ARG, 0) \
+ MACRO(ARG, 1) \
+ MACRO(ARG, 2) \
+ MACRO(ARG, 3) \
+ MACRO(ARG, 4) \
+ MACRO(ARG, 5) \
+ MACRO(ARG, 6) \
+ MACRO(ARG, 7)
+
+#define ADD_OR_SUBTRACT(CORE_CODE) \
+ ASM("rorq %[carry]") \
+ CORE_CODE \
+ ASM("sbbq %[carry],%[carry]") \
+ ASM("negq %[carry]")
+
+/*
+* Word Addition
+*/
+inline word word_add(word x, word y, word* carry)
+ {
+#if 0
+ asm(
+ ADD_OR_SUBTRACT(ASM("adcq %[y],%[x]"))
+ : [x]"=r"(x), [carry]"=r"(*carry)
+ : "0"(x), [y]"rm"(y), "1"(*carry)
+ : "cc");
+ return x;
+#else
+ word z = x + y;
+ word c1 = (z < x);
+ z += *carry;
+ *carry = c1 | (z < *carry);
+ return z;
+#endif
+ }
+
+/*
+* Eight Word Block Addition, Two Argument
+*/
+inline word word8_add2(word x[8], const word y[8], word carry)
+ {
+ asm(
+ ADD_OR_SUBTRACT(DO_8_TIMES(ADDSUB2_OP, "adcq"))
+ : [carry]"=r"(carry)
+ : [x]"r"(x), [y]"r"(y), "0"(carry)
+ : "cc", "memory");
+ return carry;
+ }
+
+/*
+* Eight Word Block Addition, Three Argument
+*/
+inline word word8_add3(word z[8], const word x[8], const word y[8], word carry)
+ {
+ asm(
+ ADD_OR_SUBTRACT(DO_8_TIMES(ADDSUB3_OP, "adcq"))
+ : [carry]"=r"(carry)
+ : [x]"r"(x), [y]"r"(y), [z]"r"(z), "0"(carry)
+ : "cc", "memory");
+ return carry;
+ }
+
+/*
+* Word Subtraction
+*/
+inline word word_sub(word x, word y, word* carry)
+ {
+ asm(
+ ADD_OR_SUBTRACT(ASM("sbbq %[y],%[x]"))
+ : [x]"=r"(x), [carry]"=r"(*carry)
+ : "0"(x), [y]"rm"(y), "1"(*carry)
+ : "cc");
+ return x;
+ }
+
+/*
+* Eight Word Block Subtraction, Two Argument
+*/
+inline word word8_sub2(word x[8], const word y[8], word carry)
+ {
+ asm(
+ ADD_OR_SUBTRACT(DO_8_TIMES(ADDSUB2_OP, "sbbq"))
+ : [carry]"=r"(carry)
+ : [x]"r"(x), [y]"r"(y), "0"(carry)
+ : "cc", "memory");
+ return carry;
+ }
+
+/*
+* Eight Word Block Subtraction, Three Argument
+*/
+inline word word8_sub3(word z[8], const word x[8], const word y[8], word carry)
+ {
+ asm(
+ ADD_OR_SUBTRACT(DO_8_TIMES(ADDSUB3_OP, "sbbq"))
+ : [carry]"=r"(carry)
+ : [x]"r"(x), [y]"r"(y), [z]"r"(z), "0"(carry)
+ : "cc", "memory");
+ return carry;
+ }
+
+/*
+* Eight Word Block Linear Multiplication
+*/
+inline word word8_linmul2(word x[8], word y, word carry)
+ {
+ asm(
+ DO_8_TIMES(LINMUL_OP, "x")
+ : [carry]"=r"(carry)
+ : [x]"r"(x), [y]"rm"(y), "0"(carry)
+ : "cc", "%rax", "%rdx");
+ return carry;
+ }
+
+/*
+* Eight Word Block Linear Multiplication
+*/
+inline word word8_linmul3(word z[8], const word x[8], word y, word carry)
+ {
+ asm(
+ DO_8_TIMES(LINMUL_OP, "z")
+ : [carry]"=r"(carry)
+ : [z]"r"(z), [x]"r"(x), [y]"rm"(y), "0"(carry)
+ : "cc", "%rax", "%rdx");
+ return carry;
+ }
+
+/*
+* Eight Word Block Multiply/Add
+*/
+inline word word8_madd3(word z[8], const word x[8], word y, word carry)
+ {
+ asm(
+ DO_8_TIMES(MULADD_OP, "")
+ : [carry]"=r"(carry)
+ : [z]"r"(z), [x]"r"(x), [y]"rm"(y), "0"(carry)
+ : "cc", "%rax", "%rdx");
+ return carry;
+ }
+
+/*
+* Multiply-Add Accumulator
+*/
+inline void word3_muladd(word* w2, word* w1, word* w0, word x, word y)
+ {
+ asm(
+ ASM("mulq %[y]")
+
+ ASM("addq %[x],%[w0]")
+ ASM("adcq %[y],%[w1]")
+ ASM("adcq $0,%[w2]")
+
+ : [w0]"=r"(*w0), [w1]"=r"(*w1), [w2]"=r"(*w2)
+ : [x]"a"(x), [y]"d"(y), "0"(*w0), "1"(*w1), "2"(*w2)
+ : "cc");
+ }
+
+/*
+* Multiply-Add Accumulator
+*/
+inline void word3_muladd_2(word* w2, word* w1, word* w0, word x, word y)
+ {
+ asm(
+ ASM("mulq %[y]")
+
+ ASM("addq %[x],%[w0]")
+ ASM("adcq %[y],%[w1]")
+ ASM("adcq $0,%[w2]")
+
+ ASM("addq %[x],%[w0]")
+ ASM("adcq %[y],%[w1]")
+ ASM("adcq $0,%[w2]")
+
+ : [w0]"=r"(*w0), [w1]"=r"(*w1), [w2]"=r"(*w2)
+ : [x]"a"(x), [y]"d"(y), "0"(*w0), "1"(*w1), "2"(*w2)
+ : "cc");
+ }
+
+
+#undef ASM
+#undef DO_8_TIMES
+#undef ADD_OR_SUBTRACT
+#undef ADDSUB2_OP
+#undef ADDSUB3_OP
+#undef LINMUL_OP
+#undef MULADD_OP
+
+}
+
+}
+#endif
generated by cgit v1.2.3 (git 2.25.1) at 2024-06-25 04:24:34 +0000