Public-Mirror/gbe_fork
1
0
Fork 0
mirror of https://github.com/Detanup01/gbe_fork.git synced 2024-12-27 10:54:15 +08:00
Code Issues 1 Projects Releases Wiki Activity
gbe_fork/mbedtls/bn_mul.h

1095 lines
44 KiB
C
Raw Normal View History

added a new cryptography library Mbed TLS
2023-12-23 16:05:24 +02:00
/**
* \file bn_mul.h
*
* \brief Multi-precision integer library
*/
/*
* Copyright The Mbed TLS Contributors
* SPDX-License-Identifier: Apache-2.0 OR GPL-2.0-or-later
*/
/*
* Multiply source vector [s] with b, add result
* to destination vector [d] and set carry c.
*
* Currently supports:
*
* . IA-32 (386+) . AMD64 / EM64T
* . IA-32 (SSE2) . Motorola 68000
* . PowerPC, 32-bit . MicroBlaze
* . PowerPC, 64-bit . TriCore
* . SPARC v8 . ARM v3+
* . Alpha . MIPS32
* . C, longlong . C, generic
*/
#ifndef MBEDTLS_BN_MUL_H
#define MBEDTLS_BN_MUL_H
#include "mbedtls/build_info.h"
#include "mbedtls/bignum.h"
/*
* Conversion macros for embedded constants:
* build lists of mbedtls_mpi_uint's from lists of unsigned char's grouped by 8, 4 or 2
*/
#if defined(MBEDTLS_HAVE_INT32)
#define MBEDTLS_BYTES_TO_T_UINT_4(a, b, c, d) \
((mbedtls_mpi_uint) (a) << 0) | \
((mbedtls_mpi_uint) (b) << 8) | \
((mbedtls_mpi_uint) (c) << 16) | \
((mbedtls_mpi_uint) (d) << 24)
#define MBEDTLS_BYTES_TO_T_UINT_2(a, b) \
MBEDTLS_BYTES_TO_T_UINT_4(a, b, 0, 0)
#define MBEDTLS_BYTES_TO_T_UINT_8(a, b, c, d, e, f, g, h) \
MBEDTLS_BYTES_TO_T_UINT_4(a, b, c, d), \
MBEDTLS_BYTES_TO_T_UINT_4(e, f, g, h)
#else /* 64-bits */
#define MBEDTLS_BYTES_TO_T_UINT_8(a, b, c, d, e, f, g, h) \
((mbedtls_mpi_uint) (a) << 0) | \
((mbedtls_mpi_uint) (b) << 8) | \
((mbedtls_mpi_uint) (c) << 16) | \
((mbedtls_mpi_uint) (d) << 24) | \
((mbedtls_mpi_uint) (e) << 32) | \
((mbedtls_mpi_uint) (f) << 40) | \
((mbedtls_mpi_uint) (g) << 48) | \
((mbedtls_mpi_uint) (h) << 56)
#define MBEDTLS_BYTES_TO_T_UINT_4(a, b, c, d) \
MBEDTLS_BYTES_TO_T_UINT_8(a, b, c, d, 0, 0, 0, 0)
#define MBEDTLS_BYTES_TO_T_UINT_2(a, b) \
MBEDTLS_BYTES_TO_T_UINT_8(a, b, 0, 0, 0, 0, 0, 0)
#endif /* bits in mbedtls_mpi_uint */
/* *INDENT-OFF* */
#if defined(MBEDTLS_HAVE_ASM)
/* armcc5 --gnu defines __GNUC__ but doesn't support GNU's extended asm */
#if defined(__GNUC__) && \
( !defined(__ARMCC_VERSION) || __ARMCC_VERSION >= 6000000 )
/*
* GCC < 5.0 treated the x86 ebx (which is used for the GOT) as a
* fixed reserved register when building as PIC, leading to errors
* like: bn_mul.h:46:13: error: PIC register clobbered by 'ebx' in 'asm'
*
* This is fixed by an improved register allocator in GCC 5+. From the
* release notes:
* Register allocation improvements: Reuse of the PIC hard register,
* instead of using a fixed register, was implemented on x86/x86-64
* targets. This improves generated PIC code performance as more hard
* registers can be used.
*/
#if defined(__GNUC__) && __GNUC__ < 5 && defined(__PIC__)
#define MULADDC_CANNOT_USE_EBX
#endif
/*
* Disable use of the i386 assembly code below if option -O0, to disable all
* compiler optimisations, is passed, detected with __OPTIMIZE__
* This is done as the number of registers used in the assembly code doesn't
* work with the -O0 option.
*/
#if defined(__i386__) && defined(__OPTIMIZE__) && !defined(MULADDC_CANNOT_USE_EBX)
#define MULADDC_X1_INIT \
{ mbedtls_mpi_uint t; \
asm( \
"movl %%ebx, %0 \n\t" \
"movl %5, %%esi \n\t" \
"movl %6, %%edi \n\t" \
"movl %7, %%ecx \n\t" \
"movl %8, %%ebx \n\t"
#define MULADDC_X1_CORE \
"lodsl \n\t" \
"mull %%ebx \n\t" \
"addl %%ecx, %%eax \n\t" \
"adcl $0, %%edx \n\t" \
"addl (%%edi), %%eax \n\t" \
"adcl $0, %%edx \n\t" \
"movl %%edx, %%ecx \n\t" \
"stosl \n\t"
#define MULADDC_X1_STOP \
"movl %4, %%ebx \n\t" \
"movl %%ecx, %1 \n\t" \
"movl %%edi, %2 \n\t" \
"movl %%esi, %3 \n\t" \
: "=m" (t), "=m" (c), "=m" (d), "=m" (s) \
: "m" (t), "m" (s), "m" (d), "m" (c), "m" (b) \
: "eax", "ebx", "ecx", "edx", "esi", "edi" \
); }
#if defined(MBEDTLS_HAVE_SSE2)
#define MULADDC_X8_INIT MULADDC_X1_INIT
#define MULADDC_X8_CORE \
"movd %%ecx, %%mm1 \n\t" \
"movd %%ebx, %%mm0 \n\t" \
"movd (%%edi), %%mm3 \n\t" \
"paddq %%mm3, %%mm1 \n\t" \
"movd (%%esi), %%mm2 \n\t" \
"pmuludq %%mm0, %%mm2 \n\t" \
"movd 4(%%esi), %%mm4 \n\t" \
"pmuludq %%mm0, %%mm4 \n\t" \
"movd 8(%%esi), %%mm6 \n\t" \
"pmuludq %%mm0, %%mm6 \n\t" \
"movd 12(%%esi), %%mm7 \n\t" \
"pmuludq %%mm0, %%mm7 \n\t" \
"paddq %%mm2, %%mm1 \n\t" \
"movd 4(%%edi), %%mm3 \n\t" \
"paddq %%mm4, %%mm3 \n\t" \
"movd 8(%%edi), %%mm5 \n\t" \
"paddq %%mm6, %%mm5 \n\t" \
"movd 12(%%edi), %%mm4 \n\t" \
"paddq %%mm4, %%mm7 \n\t" \
"movd %%mm1, (%%edi) \n\t" \
"movd 16(%%esi), %%mm2 \n\t" \
"pmuludq %%mm0, %%mm2 \n\t" \
"psrlq $32, %%mm1 \n\t" \
"movd 20(%%esi), %%mm4 \n\t" \
"pmuludq %%mm0, %%mm4 \n\t" \
"paddq %%mm3, %%mm1 \n\t" \
"movd 24(%%esi), %%mm6 \n\t" \
"pmuludq %%mm0, %%mm6 \n\t" \
"movd %%mm1, 4(%%edi) \n\t" \
"psrlq $32, %%mm1 \n\t" \
"movd 28(%%esi), %%mm3 \n\t" \
"pmuludq %%mm0, %%mm3 \n\t" \
"paddq %%mm5, %%mm1 \n\t" \
"movd 16(%%edi), %%mm5 \n\t" \
"paddq %%mm5, %%mm2 \n\t" \
"movd %%mm1, 8(%%edi) \n\t" \
"psrlq $32, %%mm1 \n\t" \
"paddq %%mm7, %%mm1 \n\t" \
"movd 20(%%edi), %%mm5 \n\t" \
"paddq %%mm5, %%mm4 \n\t" \
"movd %%mm1, 12(%%edi) \n\t" \
"psrlq $32, %%mm1 \n\t" \
"paddq %%mm2, %%mm1 \n\t" \
"movd 24(%%edi), %%mm5 \n\t" \
"paddq %%mm5, %%mm6 \n\t" \
"movd %%mm1, 16(%%edi) \n\t" \
"psrlq $32, %%mm1 \n\t" \
"paddq %%mm4, %%mm1 \n\t" \
"movd 28(%%edi), %%mm5 \n\t" \
"paddq %%mm5, %%mm3 \n\t" \
"movd %%mm1, 20(%%edi) \n\t" \
"psrlq $32, %%mm1 \n\t" \
"paddq %%mm6, %%mm1 \n\t" \
"movd %%mm1, 24(%%edi) \n\t" \
"psrlq $32, %%mm1 \n\t" \
"paddq %%mm3, %%mm1 \n\t" \
"movd %%mm1, 28(%%edi) \n\t" \
"addl $32, %%edi \n\t" \
"addl $32, %%esi \n\t" \
"psrlq $32, %%mm1 \n\t" \
"movd %%mm1, %%ecx \n\t"
#define MULADDC_X8_STOP \
"emms \n\t" \
"movl %4, %%ebx \n\t" \
"movl %%ecx, %1 \n\t" \
"movl %%edi, %2 \n\t" \
"movl %%esi, %3 \n\t" \
: "=m" (t), "=m" (c), "=m" (d), "=m" (s) \
: "m" (t), "m" (s), "m" (d), "m" (c), "m" (b) \
: "eax", "ebx", "ecx", "edx", "esi", "edi" \
); } \
#endif /* SSE2 */
#endif /* i386 */
#if defined(__amd64__) || defined (__x86_64__)
#define MULADDC_X1_INIT \
asm( \
"xorq %%r8, %%r8\n"
#define MULADDC_X1_CORE \
"movq (%%rsi), %%rax\n" \
"mulq %%rbx\n" \
"addq $8, %%rsi\n" \
"addq %%rcx, %%rax\n" \
"movq %%r8, %%rcx\n" \
"adcq $0, %%rdx\n" \
"nop \n" \
"addq %%rax, (%%rdi)\n" \
"adcq %%rdx, %%rcx\n" \
"addq $8, %%rdi\n"
#define MULADDC_X1_STOP \
: "+c" (c), "+D" (d), "+S" (s), "+m" (*(uint64_t (*)[16]) d) \
: "b" (b), "m" (*(const uint64_t (*)[16]) s) \
: "rax", "rdx", "r8" \
);
#endif /* AMD64 */
// The following assembly code assumes that a pointer will fit in a 64-bit register
// (including ILP32 __aarch64__ ABIs such as on watchOS, hence the 2^32 - 1)
#if defined(__aarch64__) && (UINTPTR_MAX == 0xfffffffful || UINTPTR_MAX == 0xfffffffffffffffful)
/*
* There are some issues around different compilers requiring different constraint
* syntax for updating pointers from assembly code (see notes for
* MBEDTLS_ASM_AARCH64_PTR_CONSTRAINT in common.h), especially on aarch64_32 (aka ILP32).
*
* For this reason we cast the pointers to/from uintptr_t here.
*/
#define MULADDC_X1_INIT \
do { uintptr_t muladdc_d = (uintptr_t) d, muladdc_s = (uintptr_t) s; asm(
#define MULADDC_X1_CORE \
"ldr x4, [%x2], #8 \n\t" \
"ldr x5, [%x1] \n\t" \
"mul x6, x4, %4 \n\t" \
"umulh x7, x4, %4 \n\t" \
"adds x5, x5, x6 \n\t" \
"adc x7, x7, xzr \n\t" \
"adds x5, x5, %0 \n\t" \
"adc %0, x7, xzr \n\t" \
"str x5, [%x1], #8 \n\t"
#define MULADDC_X1_STOP \
: "+r" (c), \
"+r" (muladdc_d), \
"+r" (muladdc_s), \
"+m" (*(uint64_t (*)[16]) d) \
: "r" (b), "m" (*(const uint64_t (*)[16]) s) \
: "x4", "x5", "x6", "x7", "cc" \
); d = (mbedtls_mpi_uint *)muladdc_d; s = (mbedtls_mpi_uint *)muladdc_s; } while (0);
#endif /* Aarch64 */
#if defined(__mc68020__) || defined(__mcpu32__)
#define MULADDC_X1_INIT \
asm( \
"movl %3, %%a2 \n\t" \
"movl %4, %%a3 \n\t" \
"movl %5, %%d3 \n\t" \
"movl %6, %%d2 \n\t" \
"moveq #0, %%d0 \n\t"
#define MULADDC_X1_CORE \
"movel %%a2@+, %%d1 \n\t" \
"mulul %%d2, %%d4:%%d1 \n\t" \
"addl %%d3, %%d1 \n\t" \
"addxl %%d0, %%d4 \n\t" \
"moveq #0, %%d3 \n\t" \
"addl %%d1, %%a3@+ \n\t" \
"addxl %%d4, %%d3 \n\t"
#define MULADDC_X1_STOP \
"movl %%d3, %0 \n\t" \
"movl %%a3, %1 \n\t" \
"movl %%a2, %2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "d0", "d1", "d2", "d3", "d4", "a2", "a3" \
);
#define MULADDC_X8_INIT MULADDC_X1_INIT
#define MULADDC_X8_CORE \
"movel %%a2@+, %%d1 \n\t" \
"mulul %%d2, %%d4:%%d1 \n\t" \
"addxl %%d3, %%d1 \n\t" \
"addxl %%d0, %%d4 \n\t" \
"addl %%d1, %%a3@+ \n\t" \
"movel %%a2@+, %%d1 \n\t" \
"mulul %%d2, %%d3:%%d1 \n\t" \
"addxl %%d4, %%d1 \n\t" \
"addxl %%d0, %%d3 \n\t" \
"addl %%d1, %%a3@+ \n\t" \
"movel %%a2@+, %%d1 \n\t" \
"mulul %%d2, %%d4:%%d1 \n\t" \
"addxl %%d3, %%d1 \n\t" \
"addxl %%d0, %%d4 \n\t" \
"addl %%d1, %%a3@+ \n\t" \
"movel %%a2@+, %%d1 \n\t" \
"mulul %%d2, %%d3:%%d1 \n\t" \
"addxl %%d4, %%d1 \n\t" \
"addxl %%d0, %%d3 \n\t" \
"addl %%d1, %%a3@+ \n\t" \
"movel %%a2@+, %%d1 \n\t" \
"mulul %%d2, %%d4:%%d1 \n\t" \
"addxl %%d3, %%d1 \n\t" \
"addxl %%d0, %%d4 \n\t" \
"addl %%d1, %%a3@+ \n\t" \
"movel %%a2@+, %%d1 \n\t" \
"mulul %%d2, %%d3:%%d1 \n\t" \
"addxl %%d4, %%d1 \n\t" \
"addxl %%d0, %%d3 \n\t" \
"addl %%d1, %%a3@+ \n\t" \
"movel %%a2@+, %%d1 \n\t" \
"mulul %%d2, %%d4:%%d1 \n\t" \
"addxl %%d3, %%d1 \n\t" \
"addxl %%d0, %%d4 \n\t" \
"addl %%d1, %%a3@+ \n\t" \
"movel %%a2@+, %%d1 \n\t" \
"mulul %%d2, %%d3:%%d1 \n\t" \
"addxl %%d4, %%d1 \n\t" \
"addxl %%d0, %%d3 \n\t" \
"addl %%d1, %%a3@+ \n\t" \
"addxl %%d0, %%d3 \n\t"
#define MULADDC_X8_STOP MULADDC_X1_STOP
#endif /* MC68000 */
#if defined(__powerpc64__) || defined(__ppc64__)
#if defined(__MACH__) && defined(__APPLE__)
#define MULADDC_X1_INIT \
asm( \
"ld r3, %3 \n\t" \
"ld r4, %4 \n\t" \
"ld r5, %5 \n\t" \
"ld r6, %6 \n\t" \
"addi r3, r3, -8 \n\t" \
"addi r4, r4, -8 \n\t" \
"addic r5, r5, 0 \n\t"
#define MULADDC_X1_CORE \
"ldu r7, 8(r3) \n\t" \
"mulld r8, r7, r6 \n\t" \
"mulhdu r9, r7, r6 \n\t" \
"adde r8, r8, r5 \n\t" \
"ld r7, 8(r4) \n\t" \
"addze r5, r9 \n\t" \
"addc r8, r8, r7 \n\t" \
"stdu r8, 8(r4) \n\t"
#define MULADDC_X1_STOP \
"addze r5, r5 \n\t" \
"addi r4, r4, 8 \n\t" \
"addi r3, r3, 8 \n\t" \
"std r5, %0 \n\t" \
"std r4, %1 \n\t" \
"std r3, %2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "r3", "r4", "r5", "r6", "r7", "r8", "r9" \
);
#else /* __MACH__ && __APPLE__ */
#define MULADDC_X1_INIT \
asm( \
"ld %%r3, %3 \n\t" \
"ld %%r4, %4 \n\t" \
"ld %%r5, %5 \n\t" \
"ld %%r6, %6 \n\t" \
"addi %%r3, %%r3, -8 \n\t" \
"addi %%r4, %%r4, -8 \n\t" \
"addic %%r5, %%r5, 0 \n\t"
#define MULADDC_X1_CORE \
"ldu %%r7, 8(%%r3) \n\t" \
"mulld %%r8, %%r7, %%r6 \n\t" \
"mulhdu %%r9, %%r7, %%r6 \n\t" \
"adde %%r8, %%r8, %%r5 \n\t" \
"ld %%r7, 8(%%r4) \n\t" \
"addze %%r5, %%r9 \n\t" \
"addc %%r8, %%r8, %%r7 \n\t" \
"stdu %%r8, 8(%%r4) \n\t"
#define MULADDC_X1_STOP \
"addze %%r5, %%r5 \n\t" \
"addi %%r4, %%r4, 8 \n\t" \
"addi %%r3, %%r3, 8 \n\t" \
"std %%r5, %0 \n\t" \
"std %%r4, %1 \n\t" \
"std %%r3, %2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "r3", "r4", "r5", "r6", "r7", "r8", "r9" \
);
#endif /* __MACH__ && __APPLE__ */
#elif defined(__powerpc__) || defined(__ppc__) /* end PPC64/begin PPC32 */
#if defined(__MACH__) && defined(__APPLE__)
#define MULADDC_X1_INIT \
asm( \
"lwz r3, %3 \n\t" \
"lwz r4, %4 \n\t" \
"lwz r5, %5 \n\t" \
"lwz r6, %6 \n\t" \
"addi r3, r3, -4 \n\t" \
"addi r4, r4, -4 \n\t" \
"addic r5, r5, 0 \n\t"
#define MULADDC_X1_CORE \
"lwzu r7, 4(r3) \n\t" \
"mullw r8, r7, r6 \n\t" \
"mulhwu r9, r7, r6 \n\t" \
"adde r8, r8, r5 \n\t" \
"lwz r7, 4(r4) \n\t" \
"addze r5, r9 \n\t" \
"addc r8, r8, r7 \n\t" \
"stwu r8, 4(r4) \n\t"
#define MULADDC_X1_STOP \
"addze r5, r5 \n\t" \
"addi r4, r4, 4 \n\t" \
"addi r3, r3, 4 \n\t" \
"stw r5, %0 \n\t" \
"stw r4, %1 \n\t" \
"stw r3, %2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "r3", "r4", "r5", "r6", "r7", "r8", "r9" \
);
#else /* __MACH__ && __APPLE__ */
#define MULADDC_X1_INIT \
asm( \
"lwz %%r3, %3 \n\t" \
"lwz %%r4, %4 \n\t" \
"lwz %%r5, %5 \n\t" \
"lwz %%r6, %6 \n\t" \
"addi %%r3, %%r3, -4 \n\t" \
"addi %%r4, %%r4, -4 \n\t" \
"addic %%r5, %%r5, 0 \n\t"
#define MULADDC_X1_CORE \
"lwzu %%r7, 4(%%r3) \n\t" \
"mullw %%r8, %%r7, %%r6 \n\t" \
"mulhwu %%r9, %%r7, %%r6 \n\t" \
"adde %%r8, %%r8, %%r5 \n\t" \
"lwz %%r7, 4(%%r4) \n\t" \
"addze %%r5, %%r9 \n\t" \
"addc %%r8, %%r8, %%r7 \n\t" \
"stwu %%r8, 4(%%r4) \n\t"
#define MULADDC_X1_STOP \
"addze %%r5, %%r5 \n\t" \
"addi %%r4, %%r4, 4 \n\t" \
"addi %%r3, %%r3, 4 \n\t" \
"stw %%r5, %0 \n\t" \
"stw %%r4, %1 \n\t" \
"stw %%r3, %2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "r3", "r4", "r5", "r6", "r7", "r8", "r9" \
);
#endif /* __MACH__ && __APPLE__ */
#endif /* PPC32 */
/*
* The Sparc(64) assembly is reported to be broken.
* Disable it for now, until we're able to fix it.
*/
#if 0 && defined(__sparc__)
#if defined(__sparc64__)
#define MULADDC_X1_INIT \
asm( \
"ldx %3, %%o0 \n\t" \
"ldx %4, %%o1 \n\t" \
"ld %5, %%o2 \n\t" \
"ld %6, %%o3 \n\t"
#define MULADDC_X1_CORE \
"ld [%%o0], %%o4 \n\t" \
"inc 4, %%o0 \n\t" \
"ld [%%o1], %%o5 \n\t" \
"umul %%o3, %%o4, %%o4 \n\t" \
"addcc %%o4, %%o2, %%o4 \n\t" \
"rd %%y, %%g1 \n\t" \
"addx %%g1, 0, %%g1 \n\t" \
"addcc %%o4, %%o5, %%o4 \n\t" \
"st %%o4, [%%o1] \n\t" \
"addx %%g1, 0, %%o2 \n\t" \
"inc 4, %%o1 \n\t"
#define MULADDC_X1_STOP \
"st %%o2, %0 \n\t" \
"stx %%o1, %1 \n\t" \
"stx %%o0, %2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "g1", "o0", "o1", "o2", "o3", "o4", \
"o5" \
);
#else /* __sparc64__ */
#define MULADDC_X1_INIT \
asm( \
"ld %3, %%o0 \n\t" \
"ld %4, %%o1 \n\t" \
"ld %5, %%o2 \n\t" \
"ld %6, %%o3 \n\t"
#define MULADDC_X1_CORE \
"ld [%%o0], %%o4 \n\t" \
"inc 4, %%o0 \n\t" \
"ld [%%o1], %%o5 \n\t" \
"umul %%o3, %%o4, %%o4 \n\t" \
"addcc %%o4, %%o2, %%o4 \n\t" \
"rd %%y, %%g1 \n\t" \
"addx %%g1, 0, %%g1 \n\t" \
"addcc %%o4, %%o5, %%o4 \n\t" \
"st %%o4, [%%o1] \n\t" \
"addx %%g1, 0, %%o2 \n\t" \
"inc 4, %%o1 \n\t"
#define MULADDC_X1_STOP \
"st %%o2, %0 \n\t" \
"st %%o1, %1 \n\t" \
"st %%o0, %2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "g1", "o0", "o1", "o2", "o3", "o4", \
"o5" \
);
#endif /* __sparc64__ */
#endif /* __sparc__ */
#if defined(__microblaze__) || defined(microblaze)
#define MULADDC_X1_INIT \
asm( \
"lwi r3, %3 \n\t" \
"lwi r4, %4 \n\t" \
"lwi r5, %5 \n\t" \
"lwi r6, %6 \n\t" \
"andi r7, r6, 0xffff \n\t" \
"bsrli r6, r6, 16 \n\t"
#if(__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)
#define MULADDC_LHUI \
"lhui r9, r3, 0 \n\t" \
"addi r3, r3, 2 \n\t" \
"lhui r8, r3, 0 \n\t"
#else
#define MULADDC_LHUI \
"lhui r8, r3, 0 \n\t" \
"addi r3, r3, 2 \n\t" \
"lhui r9, r3, 0 \n\t"
#endif
#define MULADDC_X1_CORE \
MULADDC_LHUI \
"addi r3, r3, 2 \n\t" \
"mul r10, r9, r6 \n\t" \
"mul r11, r8, r7 \n\t" \
"mul r12, r9, r7 \n\t" \
"mul r13, r8, r6 \n\t" \
"bsrli r8, r10, 16 \n\t" \
"bsrli r9, r11, 16 \n\t" \
"add r13, r13, r8 \n\t" \
"add r13, r13, r9 \n\t" \
"bslli r10, r10, 16 \n\t" \
"bslli r11, r11, 16 \n\t" \
"add r12, r12, r10 \n\t" \
"addc r13, r13, r0 \n\t" \
"add r12, r12, r11 \n\t" \
"addc r13, r13, r0 \n\t" \
"lwi r10, r4, 0 \n\t" \
"add r12, r12, r10 \n\t" \
"addc r13, r13, r0 \n\t" \
"add r12, r12, r5 \n\t" \
"addc r5, r13, r0 \n\t" \
"swi r12, r4, 0 \n\t" \
"addi r4, r4, 4 \n\t"
#define MULADDC_X1_STOP \
"swi r5, %0 \n\t" \
"swi r4, %1 \n\t" \
"swi r3, %2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "r3", "r4", "r5", "r6", "r7", "r8", \
"r9", "r10", "r11", "r12", "r13" \
);
#endif /* MicroBlaze */
#if defined(__tricore__)
#define MULADDC_X1_INIT \
asm( \
"ld.a %%a2, %3 \n\t" \
"ld.a %%a3, %4 \n\t" \
"ld.w %%d4, %5 \n\t" \
"ld.w %%d1, %6 \n\t" \
"xor %%d5, %%d5 \n\t"
#define MULADDC_X1_CORE \
"ld.w %%d0, [%%a2+] \n\t" \
"madd.u %%e2, %%e4, %%d0, %%d1 \n\t" \
"ld.w %%d0, [%%a3] \n\t" \
"addx %%d2, %%d2, %%d0 \n\t" \
"addc %%d3, %%d3, 0 \n\t" \
"mov %%d4, %%d3 \n\t" \
"st.w [%%a3+], %%d2 \n\t"
#define MULADDC_X1_STOP \
"st.w %0, %%d4 \n\t" \
"st.a %1, %%a3 \n\t" \
"st.a %2, %%a2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "d0", "d1", "e2", "d4", "a2", "a3" \
);
#endif /* TriCore */
#if defined(__arm__)
#if defined(__thumb__) && !defined(__thumb2__)
#if defined(MBEDTLS_COMPILER_IS_GCC)
/*
* Thumb 1 ISA. This code path has only been tested successfully on gcc;
* it does not compile on clang or armclang.
*/
#if !defined(__OPTIMIZE__) && defined(__GNUC__)
/*
* Note, gcc -O0 by default uses r7 for the frame pointer, so it complains about
* our use of r7 below, unless -fomit-frame-pointer is passed.
*
* On the other hand, -fomit-frame-pointer is implied by any -Ox options with
* x !=0, which we can detect using __OPTIMIZE__ (which is also defined by
* clang and armcc5 under the same conditions).
*
* If gcc needs to use r7, we use r1 as a scratch register and have a few extra
* instructions to preserve/restore it; otherwise, we can use r7 and avoid
* the preserve/restore overhead.
*/
#define MULADDC_SCRATCH "RS .req r1 \n\t"
#define MULADDC_PRESERVE_SCRATCH "mov r10, r1 \n\t"
#define MULADDC_RESTORE_SCRATCH "mov r1, r10 \n\t"
#define MULADDC_SCRATCH_CLOBBER "r10"
#else /* !defined(__OPTIMIZE__) && defined(__GNUC__) */
#define MULADDC_SCRATCH "RS .req r7 \n\t"
#define MULADDC_PRESERVE_SCRATCH ""
#define MULADDC_RESTORE_SCRATCH ""
#define MULADDC_SCRATCH_CLOBBER "r7"
#endif /* !defined(__OPTIMIZE__) && defined(__GNUC__) */
#define MULADDC_X1_INIT \
asm( \
MULADDC_SCRATCH \
"ldr r0, %3 \n\t" \
"ldr r1, %4 \n\t" \
"ldr r2, %5 \n\t" \
"ldr r3, %6 \n\t" \
"lsr r4, r3, #16 \n\t" \
"mov r9, r4 \n\t" \
"lsl r4, r3, #16 \n\t" \
"lsr r4, r4, #16 \n\t" \
"mov r8, r4 \n\t" \
#define MULADDC_X1_CORE \
MULADDC_PRESERVE_SCRATCH \
"ldmia r0!, {r6} \n\t" \
"lsr RS, r6, #16 \n\t" \
"lsl r6, r6, #16 \n\t" \
"lsr r6, r6, #16 \n\t" \
"mov r4, r8 \n\t" \
"mul r4, r6 \n\t" \
"mov r3, r9 \n\t" \
"mul r6, r3 \n\t" \
"mov r5, r9 \n\t" \
"mul r5, RS \n\t" \
"mov r3, r8 \n\t" \
"mul RS, r3 \n\t" \
"lsr r3, r6, #16 \n\t" \
"add r5, r5, r3 \n\t" \
"lsr r3, RS, #16 \n\t" \
"add r5, r5, r3 \n\t" \
"add r4, r4, r2 \n\t" \
"mov r2, #0 \n\t" \
"adc r5, r2 \n\t" \
"lsl r3, r6, #16 \n\t" \
"add r4, r4, r3 \n\t" \
"adc r5, r2 \n\t" \
"lsl r3, RS, #16 \n\t" \
"add r4, r4, r3 \n\t" \
"adc r5, r2 \n\t" \
MULADDC_RESTORE_SCRATCH \
"ldr r3, [r1] \n\t" \
"add r4, r4, r3 \n\t" \
"adc r2, r5 \n\t" \
"stmia r1!, {r4} \n\t"
#define MULADDC_X1_STOP \
"str r2, %0 \n\t" \
"str r1, %1 \n\t" \
"str r0, %2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "r0", "r1", "r2", "r3", "r4", "r5", \
"r6", MULADDC_SCRATCH_CLOBBER, "r8", "r9", "cc" \
);
#endif /* !defined(__ARMCC_VERSION) && !defined(__clang__) */
#elif (__ARM_ARCH >= 6) && \
defined (__ARM_FEATURE_DSP) && (__ARM_FEATURE_DSP == 1)
/* Armv6-M (or later) with DSP Instruction Set Extensions.
* Requires support for either Thumb 2 or Arm ISA.
*/
#define MULADDC_X1_INIT \
{ \
mbedtls_mpi_uint tmp_a, tmp_b; \
asm volatile (
#define MULADDC_X1_CORE \
".p2align 2 \n\t" \
"ldr %[a], [%[in]], #4 \n\t" \
"ldr %[b], [%[acc]] \n\t" \
"umaal %[b], %[carry], %[scalar], %[a] \n\t" \
"str %[b], [%[acc]], #4 \n\t"
#define MULADDC_X1_STOP \
: [a] "=&r" (tmp_a), \
[b] "=&r" (tmp_b), \
[in] "+r" (s), \
[acc] "+r" (d), \
[carry] "+l" (c) \
: [scalar] "r" (b) \
: "memory" \
); \
}
#define MULADDC_X2_INIT \
{ \
mbedtls_mpi_uint tmp_a0, tmp_b0; \
mbedtls_mpi_uint tmp_a1, tmp_b1; \
asm volatile (
/* - Make sure loop is 4-byte aligned to avoid stalls
* upon repeated non-word aligned instructions in
* some microarchitectures.
* - Don't use ldm with post-increment or back-to-back
* loads with post-increment and same address register
* to avoid stalls on some microarchitectures.
* - Bunch loads and stores to reduce latency on some
* microarchitectures. E.g., on Cortex-M4, the first
* in a series of load/store operations has latency
* 2 cycles, while subsequent loads/stores are single-cycle. */
#define MULADDC_X2_CORE \
".p2align 2 \n\t" \
"ldr %[a0], [%[in]], #+8 \n\t" \
"ldr %[b0], [%[acc]], #+8 \n\t" \
"ldr %[a1], [%[in], #-4] \n\t" \
"ldr %[b1], [%[acc], #-4] \n\t" \
"umaal %[b0], %[carry], %[scalar], %[a0] \n\t" \
"umaal %[b1], %[carry], %[scalar], %[a1] \n\t" \
"str %[b0], [%[acc], #-8] \n\t" \
"str %[b1], [%[acc], #-4] \n\t"
#define MULADDC_X2_STOP \
: [a0] "=&r" (tmp_a0), \
[b0] "=&r" (tmp_b0), \
[a1] "=&r" (tmp_a1), \
[b1] "=&r" (tmp_b1), \
[in] "+r" (s), \
[acc] "+r" (d), \
[carry] "+l" (c) \
: [scalar] "r" (b) \
: "memory" \
); \
}
#else /* Thumb 2 or Arm ISA, without DSP extensions */
#define MULADDC_X1_INIT \
asm( \
"ldr r0, %3 \n\t" \
"ldr r1, %4 \n\t" \
"ldr r2, %5 \n\t" \
"ldr r3, %6 \n\t"
#define MULADDC_X1_CORE \
"ldr r4, [r0], #4 \n\t" \
"mov r5, #0 \n\t" \
"ldr r6, [r1] \n\t" \
"umlal r2, r5, r3, r4 \n\t" \
"adds r4, r6, r2 \n\t" \
"adc r2, r5, #0 \n\t" \
"str r4, [r1], #4 \n\t"
#define MULADDC_X1_STOP \
"str r2, %0 \n\t" \
"str r1, %1 \n\t" \
"str r0, %2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "r0", "r1", "r2", "r3", "r4", "r5", \
"r6", "cc" \
);
#endif /* ISA codepath selection */
#endif /* defined(__arm__) */
#if defined(__alpha__)
#define MULADDC_X1_INIT \
asm( \
"ldq $1, %3 \n\t" \
"ldq $2, %4 \n\t" \
"ldq $3, %5 \n\t" \
"ldq $4, %6 \n\t"
#define MULADDC_X1_CORE \
"ldq $6, 0($1) \n\t" \
"addq $1, 8, $1 \n\t" \
"mulq $6, $4, $7 \n\t" \
"umulh $6, $4, $6 \n\t" \
"addq $7, $3, $7 \n\t" \
"cmpult $7, $3, $3 \n\t" \
"ldq $5, 0($2) \n\t" \
"addq $7, $5, $7 \n\t" \
"cmpult $7, $5, $5 \n\t" \
"stq $7, 0($2) \n\t" \
"addq $2, 8, $2 \n\t" \
"addq $6, $3, $3 \n\t" \
"addq $5, $3, $3 \n\t"
#define MULADDC_X1_STOP \
"stq $3, %0 \n\t" \
"stq $2, %1 \n\t" \
"stq $1, %2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "$1", "$2", "$3", "$4", "$5", "$6", "$7" \
);
#endif /* Alpha */
#if defined(__mips__) && !defined(__mips64)
#define MULADDC_X1_INIT \
asm( \
"lw $10, %3 \n\t" \
"lw $11, %4 \n\t" \
"lw $12, %5 \n\t" \
"lw $13, %6 \n\t"
#define MULADDC_X1_CORE \
"lw $14, 0($10) \n\t" \
"multu $13, $14 \n\t" \
"addi $10, $10, 4 \n\t" \
"mflo $14 \n\t" \
"mfhi $9 \n\t" \
"addu $14, $12, $14 \n\t" \
"lw $15, 0($11) \n\t" \
"sltu $12, $14, $12 \n\t" \
"addu $15, $14, $15 \n\t" \
"sltu $14, $15, $14 \n\t" \
"addu $12, $12, $9 \n\t" \
"sw $15, 0($11) \n\t" \
"addu $12, $12, $14 \n\t" \
"addi $11, $11, 4 \n\t"
#define MULADDC_X1_STOP \
"sw $12, %0 \n\t" \
"sw $11, %1 \n\t" \
"sw $10, %2 \n\t" \
: "=m" (c), "=m" (d), "=m" (s) \
: "m" (s), "m" (d), "m" (c), "m" (b) \
: "$9", "$10", "$11", "$12", "$13", "$14", "$15", "lo", "hi" \
);
#endif /* MIPS */
#endif /* GNUC */
#if (defined(_MSC_VER) && defined(_M_IX86)) || defined(__WATCOMC__)
#define MULADDC_X1_INIT \
__asm mov esi, s \
__asm mov edi, d \
__asm mov ecx, c \
__asm mov ebx, b
#define MULADDC_X1_CORE \
__asm lodsd \
__asm mul ebx \
__asm add eax, ecx \
__asm adc edx, 0 \
__asm add eax, [edi] \
__asm adc edx, 0 \
__asm mov ecx, edx \
__asm stosd
#define MULADDC_X1_STOP \
__asm mov c, ecx \
__asm mov d, edi \
__asm mov s, esi
#if defined(MBEDTLS_HAVE_SSE2)
#define EMIT __asm _emit
#define MULADDC_X8_INIT MULADDC_X1_INIT
#define MULADDC_X8_CORE \
EMIT 0x0F EMIT 0x6E EMIT 0xC9 \
EMIT 0x0F EMIT 0x6E EMIT 0xC3 \
EMIT 0x0F EMIT 0x6E EMIT 0x1F \
EMIT 0x0F EMIT 0xD4 EMIT 0xCB \
EMIT 0x0F EMIT 0x6E EMIT 0x16 \
EMIT 0x0F EMIT 0xF4 EMIT 0xD0 \
EMIT 0x0F EMIT 0x6E EMIT 0x66 EMIT 0x04 \
EMIT 0x0F EMIT 0xF4 EMIT 0xE0 \
EMIT 0x0F EMIT 0x6E EMIT 0x76 EMIT 0x08 \
EMIT 0x0F EMIT 0xF4 EMIT 0xF0 \
EMIT 0x0F EMIT 0x6E EMIT 0x7E EMIT 0x0C \
EMIT 0x0F EMIT 0xF4 EMIT 0xF8 \
EMIT 0x0F EMIT 0xD4 EMIT 0xCA \
EMIT 0x0F EMIT 0x6E EMIT 0x5F EMIT 0x04 \
EMIT 0x0F EMIT 0xD4 EMIT 0xDC \
EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x08 \
EMIT 0x0F EMIT 0xD4 EMIT 0xEE \
EMIT 0x0F EMIT 0x6E EMIT 0x67 EMIT 0x0C \
EMIT 0x0F EMIT 0xD4 EMIT 0xFC \
EMIT 0x0F EMIT 0x7E EMIT 0x0F \
EMIT 0x0F EMIT 0x6E EMIT 0x56 EMIT 0x10 \
EMIT 0x0F EMIT 0xF4 EMIT 0xD0 \
EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
EMIT 0x0F EMIT 0x6E EMIT 0x66 EMIT 0x14 \
EMIT 0x0F EMIT 0xF4 EMIT 0xE0 \
EMIT 0x0F EMIT 0xD4 EMIT 0xCB \
EMIT 0x0F EMIT 0x6E EMIT 0x76 EMIT 0x18 \
EMIT 0x0F EMIT 0xF4 EMIT 0xF0 \
EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x04 \
EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
EMIT 0x0F EMIT 0x6E EMIT 0x5E EMIT 0x1C \
EMIT 0x0F EMIT 0xF4 EMIT 0xD8 \
EMIT 0x0F EMIT 0xD4 EMIT 0xCD \
EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x10 \
EMIT 0x0F EMIT 0xD4 EMIT 0xD5 \
EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x08 \
EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
EMIT 0x0F EMIT 0xD4 EMIT 0xCF \
EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x14 \
EMIT 0x0F EMIT 0xD4 EMIT 0xE5 \
EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x0C \
EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
EMIT 0x0F EMIT 0xD4 EMIT 0xCA \
EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x18 \
EMIT 0x0F EMIT 0xD4 EMIT 0xF5 \
EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x10 \
EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
EMIT 0x0F EMIT 0xD4 EMIT 0xCC \
EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x1C \
EMIT 0x0F EMIT 0xD4 EMIT 0xDD \
EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x14 \
EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
EMIT 0x0F EMIT 0xD4 EMIT 0xCE \
EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x18 \
EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
EMIT 0x0F EMIT 0xD4 EMIT 0xCB \
EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x1C \
EMIT 0x83 EMIT 0xC7 EMIT 0x20 \
EMIT 0x83 EMIT 0xC6 EMIT 0x20 \
EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
EMIT 0x0F EMIT 0x7E EMIT 0xC9
#define MULADDC_X8_STOP \
EMIT 0x0F EMIT 0x77 \
__asm mov c, ecx \
__asm mov d, edi \
__asm mov s, esi
#endif /* SSE2 */
#endif /* MSVC */
#endif /* MBEDTLS_HAVE_ASM */
#if !defined(MULADDC_X1_CORE)
#if defined(MBEDTLS_HAVE_UDBL)
#define MULADDC_X1_INIT \
{ \
mbedtls_t_udbl r; \
mbedtls_mpi_uint r0, r1;
#define MULADDC_X1_CORE \
r = *(s++) * (mbedtls_t_udbl) b; \
r0 = (mbedtls_mpi_uint) r; \
r1 = (mbedtls_mpi_uint)( r >> biL ); \
r0 += c; r1 += (r0 < c); \
r0 += *d; r1 += (r0 < *d); \
c = r1; *(d++) = r0;
#define MULADDC_X1_STOP \
}
#else /* MBEDTLS_HAVE_UDBL */
#define MULADDC_X1_INIT \
{ \
mbedtls_mpi_uint s0, s1, b0, b1; \
mbedtls_mpi_uint r0, r1, rx, ry; \
b0 = ( b << biH ) >> biH; \
b1 = ( b >> biH );
#define MULADDC_X1_CORE \
s0 = ( *s << biH ) >> biH; \
s1 = ( *s >> biH ); s++; \
rx = s0 * b1; r0 = s0 * b0; \
ry = s1 * b0; r1 = s1 * b1; \
r1 += ( rx >> biH ); \
r1 += ( ry >> biH ); \
rx <<= biH; ry <<= biH; \
r0 += rx; r1 += (r0 < rx); \
r0 += ry; r1 += (r0 < ry); \
r0 += c; r1 += (r0 < c); \
r0 += *d; r1 += (r0 < *d); \
c = r1; *(d++) = r0;
#define MULADDC_X1_STOP \
}
#endif /* C (longlong) */
#endif /* C (generic) */
#if !defined(MULADDC_X2_CORE)
#define MULADDC_X2_INIT MULADDC_X1_INIT
#define MULADDC_X2_STOP MULADDC_X1_STOP
#define MULADDC_X2_CORE MULADDC_X1_CORE MULADDC_X1_CORE
#endif /* MULADDC_X2_CORE */
#if !defined(MULADDC_X4_CORE)
#define MULADDC_X4_INIT MULADDC_X2_INIT
#define MULADDC_X4_STOP MULADDC_X2_STOP
#define MULADDC_X4_CORE MULADDC_X2_CORE MULADDC_X2_CORE
#endif /* MULADDC_X4_CORE */
#if !defined(MULADDC_X8_CORE)
#define MULADDC_X8_INIT MULADDC_X4_INIT
#define MULADDC_X8_STOP MULADDC_X4_STOP
#define MULADDC_X8_CORE MULADDC_X4_CORE MULADDC_X4_CORE
#endif /* MULADDC_X8_CORE */
/* *INDENT-ON* */
#endif /* bn_mul.h */
Reference in New Issue Copy Permalink