aboutsummaryrefslogtreecommitdiffstats
path: root/src/asm/mp_ia32
diff options
context:
space:
mode:
Diffstat (limited to 'src/asm/mp_ia32')
-rw-r--r--src/asm/mp_ia32/modinfo.txt19
-rw-r--r--src/asm/mp_ia32/mp_asm.h65
-rw-r--r--src/asm/mp_ia32/mp_asmi.h225
3 files changed, 309 insertions, 0 deletions
diff --git a/src/asm/mp_ia32/modinfo.txt b/src/asm/mp_ia32/modinfo.txt
new file mode 100644
index 000000000..cf4959250
--- /dev/null
+++ b/src/asm/mp_ia32/modinfo.txt
@@ -0,0 +1,19 @@
+realname "x86 MPI Assembler Core"
+
+mp_bits 32
+
+load_on asm_ok
+
+<replace>
+mp_asm.h
+mp_asmi.h
+</replace>
+
+<arch>
+ia32
+</arch>
+
+<cc>
+gcc
+icc
+</cc>
diff --git a/src/asm/mp_ia32/mp_asm.h b/src/asm/mp_ia32/mp_asm.h
new file mode 100644
index 000000000..b45140321
--- /dev/null
+++ b/src/asm/mp_ia32/mp_asm.h
@@ -0,0 +1,65 @@
+/*************************************************
+* Lowest Level MPI Algorithms Header File *
+* (C) 1999-2008 Jack Lloyd *
+* 2006 Luca Piccarreta *
+*************************************************/
+
+#ifndef BOTAN_MP_ASM_H__
+#define BOTAN_MP_ASM_H__
+
+#include <botan/mp_types.h>
+
+#if (BOTAN_MP_WORD_BITS != 32)
+ #error The mp_ia32 module requires that BOTAN_MP_WORD_BITS == 32
+#endif
+
+namespace Botan {
+
+extern "C" {
+
+/*************************************************
+* Helper Macros for x86 Assembly *
+*************************************************/
+#define ASM(x) x "\n\t"
+
+/*************************************************
+* Word Multiply *
+*************************************************/
+inline word word_madd2(word a, word b, word* c)
+ {
+ asm(
+ ASM("mull %[b]")
+ ASM("addl %[c],%[a]")
+ ASM("adcl $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("mull %[b]")
+
+ ASM("addl %[c],%[a]")
+ ASM("adcl $0,%[carry]")
+
+ ASM("addl %[d],%[a]")
+ ASM("adcl $0,%[carry]")
+
+ : [a]"=a"(a), [b]"=rm"(b), [carry]"=&d"(*d)
+ : "0"(a), "1"(b), [c]"g"(c), [d]"g"(*d) : "cc");
+
+ return a;
+ }
+
+}
+
+}
+
+#endif
diff --git a/src/asm/mp_ia32/mp_asmi.h b/src/asm/mp_ia32/mp_asmi.h
new file mode 100644
index 000000000..9de0c11e3
--- /dev/null
+++ b/src/asm/mp_ia32/mp_asmi.h
@@ -0,0 +1,225 @@
+/*************************************************
+* Lowest Level MPI Algorithms Header File *
+* (C) 1999-2007 Jack Lloyd *
+* 2006 Luca Piccarreta *
+*************************************************/
+
+#ifndef BOTAN_MP_ASM_INTERNAL_H__
+#define BOTAN_MP_ASM_INTERNAL_H__
+
+#include <botan/mp_asm.h>
+
+namespace Botan {
+
+extern "C" {
+
+/*************************************************
+* Helper Macros for x86 Assembly *
+*************************************************/
+#ifndef ASM
+ #define ASM(x) x "\n\t"
+#endif
+
+#define ADDSUB2_OP(OPERATION, INDEX) \
+ ASM("movl 4*" #INDEX "(%[y]), %[carry]") \
+ ASM(OPERATION " %[carry], 4*" #INDEX "(%[x])") \
+
+#define ADDSUB3_OP(OPERATION, INDEX) \
+ ASM("movl 4*" #INDEX "(%[x]), %[carry]") \
+ ASM(OPERATION " 4*" #INDEX "(%[y]), %[carry]") \
+ ASM("movl %[carry], 4*" #INDEX "(%[z])") \
+
+#define LINMUL_OP(WRITE_TO, INDEX) \
+ ASM("movl 4*" #INDEX "(%[x]),%%eax") \
+ ASM("mull %[y]") \
+ ASM("addl %[carry],%%eax") \
+ ASM("adcl $0,%%edx") \
+ ASM("movl %%edx,%[carry]") \
+ ASM("movl %%eax, 4*" #INDEX "(%[" WRITE_TO "])")
+
+#define MULADD_OP(IGNORED, INDEX) \
+ ASM("movl 4*" #INDEX "(%[x]),%%eax") \
+ ASM("mull %[y]") \
+ ASM("addl %[carry],%%eax") \
+ ASM("adcl $0,%%edx") \
+ ASM("addl 4*" #INDEX "(%[z]),%%eax") \
+ ASM("adcl $0,%%edx") \
+ ASM("movl %%edx,%[carry]") \
+ ASM("movl %%eax, 4*" #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("rorl %[carry]") \
+ CORE_CODE \
+ ASM("sbbl %[carry],%[carry]") \
+ ASM("negl %[carry]")
+
+/*************************************************
+* Word Addition *
+*************************************************/
+inline word word_add(word x, word y, word* carry)
+ {
+ asm(
+ ADD_OR_SUBTRACT(ASM("adcl %[y],%[x]"))
+ : [x]"=r"(x), [carry]"=r"(*carry)
+ : "0"(x), [y]"rm"(y), "1"(*carry)
+ : "cc");
+ return x;
+ }
+
+/*************************************************
+* 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, "adcl"))
+ : [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, "adcl"))
+ : [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("sbbl %[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, "sbbl"))
+ : [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, "sbbl"))
+ : [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", "%eax", "%edx");
+ 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", "%eax", "%edx");
+ 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", "%eax", "%edx");
+ return carry;
+ }
+
+/*************************************************
+* Multiply-Add Accumulator *
+*************************************************/
+inline void word3_muladd(word* w2, word* w1, word* w0, word x, word y)
+ {
+ asm(
+ ASM("mull %[y]")
+
+ ASM("addl %[x],%[w0]")
+ ASM("adcl %[y],%[w1]")
+ ASM("adcl $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("mull %[y]")
+
+ ASM("addl %[x],%[w0]")
+ ASM("adcl %[y],%[w1]")
+ ASM("adcl $0,%[w2]")
+
+ ASM("addl %[x],%[w0]")
+ ASM("adcl %[y],%[w1]")
+ ASM("adcl $0,%[w2]")
+
+ : [w0]"=r"(*w0), [w1]"=r"(*w1), [w2]"=r"(*w2)
+ : [x]"a"(x), [y]"d"(y), "0"(*w0), "1"(*w1), "2"(*w2)
+ : "cc");
+ }
+
+}
+
+}
+
+#endif
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-19 14:08:05 +0000