AcoSail/kernel_optimize_test
8
0
Fork 0
forked from luck/tmp_suning_uos_patched
Code Pull Requests Activity

crypto: sha - SHA1 transform x86_64 AVX2

Browse Source 
This git patch adds x86_64 AVX2 optimization of SHA1
transform to crypto support. The patch has been tested with 3.14.0-rc1
kernel.

On a Haswell desktop, with turbo disabled and all cpus running
at maximum frequency, tcrypt shows AVX2 performance improvement
from 3% for 256 bytes update to 16% for 1024 bytes update over
AVX implementation.

This patch adds sha1_avx2_transform(), the glue, build and
configuration changes needed for AVX2 optimization of
SHA1 transform to crypto support.

sha1-ssse3 is one module which adds the necessary optimization
support (SSSE3/AVX/AVX2) for the low-level SHA1 transform function.
With better optimization support, transform function is overridden
as the case may be. In the case of AVX2, due to performance reasons
across datablock sizes, the AVX or AVX2 transform function is used
at run-time as it suits best. The Makefile change therefore appends
the necessary objects to the linkage. Due to this, the patch merely
appends AVX2 transform to the existing build mix and Kconfig support
and leaves the configuration build support as is.

Signed-off-by: Chandramouli Narayanan <mouli@linux.intel.com>
Reviewed-by: Marek Vasut <marex@denx.de>
Acked-by: H. Peter Anvin <hpa@linux.intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
This commit is contained in:
chandramouli narayanan 2014-03-20 15:14:00 -07:00 committed by Herbert Xu
parent 130fa5bc81
commit 7c1da8d0d0
4 changed files with 759 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
arch/x86/crypto/Makefile
View File

@ -79,6 +79,9 @@ aesni-intel-y := aesni-intel_asm.o aesni-intel_glue.o fpu.o
aesni-intel-$(CONFIG_64BIT) += aesni-intel_avx-x86_64.o
ghash-clmulni-intel-y := ghash-clmulni-intel_asm.o ghash-clmulni-intel_glue.o
sha1-ssse3-y := sha1_ssse3_asm.o sha1_ssse3_glue.o
ifeq ($(avx2_supported),yes)
sha1-ssse3-y += sha1_avx2_x86_64_asm.o
endif
crc32c-intel-y := crc32c-intel_glue.o
crc32c-intel-$(CONFIG_64BIT) += crc32c-pcl-intel-asm_64.o
crc32-pclmul-y := crc32-pclmul_asm.o crc32-pclmul_glue.o

712
arch/x86/crypto/sha1_avx2_x86_64_asm.S Normal file
View File

@ -0,0 +1,712 @@
/*
* Implement fast SHA-1 with AVX2 instructions. (x86_64)
*
* This file is provided under a dual BSD/GPLv2 license. When using or
* redistributing this file, you may do so under either license.
*
* GPL LICENSE SUMMARY
*
* Copyright(c) 2014 Intel Corporation.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of version 2 of the GNU General Public License as
* published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* Contact Information:
* Ilya Albrekht <ilya.albrekht@intel.com>
* Maxim Locktyukhin <maxim.locktyukhin@intel.com>
* Ronen Zohar <ronen.zohar@intel.com>
* Chandramouli Narayanan <mouli@linux.intel.com>
*
* BSD LICENSE
*
* Copyright(c) 2014 Intel Corporation.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
* Neither the name of Intel Corporation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
/*
* SHA-1 implementation with Intel(R) AVX2 instruction set extensions.
*
*This implementation is based on the previous SSSE3 release:
*Visit http://software.intel.com/en-us/articles/
*and refer to improving-the-performance-of-the-secure-hash-algorithm-1/
*
*Updates 20-byte SHA-1 record in 'hash' for even number of
*'num_blocks' consecutive 64-byte blocks
*
*extern "C" void sha1_transform_avx2(
* int *hash, const char* input, size_t num_blocks );
*/
#include <linux/linkage.h>
#define CTX %rdi /* arg1 */
#define BUF %rsi /* arg2 */
#define CNT %rdx /* arg3 */
#define REG_A %ecx
#define REG_B %esi
#define REG_C %edi
#define REG_D %eax
#define REG_E %edx
#define REG_TB %ebx
#define REG_TA %r12d
#define REG_RA %rcx
#define REG_RB %rsi
#define REG_RC %rdi
#define REG_RD %rax
#define REG_RE %rdx
#define REG_RTA %r12
#define REG_RTB %rbx
#define REG_T1 %ebp
#define xmm_mov vmovups
#define avx2_zeroupper vzeroupper
#define RND_F1 1
#define RND_F2 2
#define RND_F3 3
.macro REGALLOC
.set A, REG_A
.set B, REG_B
.set C, REG_C
.set D, REG_D
.set E, REG_E
.set TB, REG_TB
.set TA, REG_TA
.set RA, REG_RA
.set RB, REG_RB
.set RC, REG_RC
.set RD, REG_RD
.set RE, REG_RE
.set RTA, REG_RTA
.set RTB, REG_RTB
.set T1, REG_T1
.endm
#define K_BASE %r8
#define HASH_PTR %r9
#define BUFFER_PTR %r10
#define BUFFER_PTR2 %r13
#define BUFFER_END %r11
#define PRECALC_BUF %r14
#define WK_BUF %r15
#define W_TMP %xmm0
#define WY_TMP %ymm0
#define WY_TMP2 %ymm9
# AVX2 variables
#define WY0 %ymm3
#define WY4 %ymm5
#define WY08 %ymm7
#define WY12 %ymm8
#define WY16 %ymm12
#define WY20 %ymm13
#define WY24 %ymm14
#define WY28 %ymm15
#define YMM_SHUFB_BSWAP %ymm10
/*
* Keep 2 iterations precalculated at a time:
* - 80 DWORDs per iteration * 2
*/
#define W_SIZE (80*2*2 +16)
#define WK(t) ((((t) % 80) / 4)*32 + ( (t) % 4)*4 + ((t)/80)*16 )(WK_BUF)
#define PRECALC_WK(t) ((t)*2*2)(PRECALC_BUF)
.macro UPDATE_HASH hash, val
add \hash, \val
mov \val, \hash
.endm
.macro PRECALC_RESET_WY
.set WY_00, WY0
.set WY_04, WY4
.set WY_08, WY08
.set WY_12, WY12
.set WY_16, WY16
.set WY_20, WY20
.set WY_24, WY24
.set WY_28, WY28
.set WY_32, WY_00
.endm
.macro PRECALC_ROTATE_WY
/* Rotate macros */
.set WY_32, WY_28
.set WY_28, WY_24
.set WY_24, WY_20
.set WY_20, WY_16
.set WY_16, WY_12
.set WY_12, WY_08
.set WY_08, WY_04
.set WY_04, WY_00
.set WY_00, WY_32
/* Define register aliases */
.set WY, WY_00
.set WY_minus_04, WY_04
.set WY_minus_08, WY_08
.set WY_minus_12, WY_12
.set WY_minus_16, WY_16
.set WY_minus_20, WY_20
.set WY_minus_24, WY_24
.set WY_minus_28, WY_28
.set WY_minus_32, WY
.endm
.macro PRECALC_00_15
.if (i == 0) # Initialize and rotate registers
PRECALC_RESET_WY
PRECALC_ROTATE_WY
.endif
/* message scheduling pre-compute for rounds 0-15 */
.if ((i & 7) == 0)
/*
* blended AVX2 and ALU instruction scheduling
* 1 vector iteration per 8 rounds
*/
vmovdqu ((i * 2) + PRECALC_OFFSET)(BUFFER_PTR), W_TMP
.elseif ((i & 7) == 1)
vinsertf128 $1, (((i-1) * 2)+PRECALC_OFFSET)(BUFFER_PTR2),\
WY_TMP, WY_TMP
.elseif ((i & 7) == 2)
vpshufb YMM_SHUFB_BSWAP, WY_TMP, WY
.elseif ((i & 7) == 4)
vpaddd K_XMM(K_BASE), WY, WY_TMP
.elseif ((i & 7) == 7)
vmovdqu WY_TMP, PRECALC_WK(i&~7)
PRECALC_ROTATE_WY
.endif
.endm
.macro PRECALC_16_31
/*
* message scheduling pre-compute for rounds 16-31
* calculating last 32 w[i] values in 8 XMM registers
* pre-calculate K+w[i] values and store to mem
* for later load by ALU add instruction
*
* "brute force" vectorization for rounds 16-31 only
* due to w[i]->w[i-3] dependency
*/
.if ((i & 7) == 0)
/*
* blended AVX2 and ALU instruction scheduling
* 1 vector iteration per 8 rounds
*/
/* w[i-14] */
vpalignr $8, WY_minus_16, WY_minus_12, WY
vpsrldq $4, WY_minus_04, WY_TMP /* w[i-3] */
.elseif ((i & 7) == 1)
vpxor WY_minus_08, WY, WY
vpxor WY_minus_16, WY_TMP, WY_TMP
.elseif ((i & 7) == 2)
vpxor WY_TMP, WY, WY
vpslldq $12, WY, WY_TMP2
.elseif ((i & 7) == 3)
vpslld $1, WY, WY_TMP
vpsrld $31, WY, WY
.elseif ((i & 7) == 4)
vpor WY, WY_TMP, WY_TMP
vpslld $2, WY_TMP2, WY
.elseif ((i & 7) == 5)
vpsrld $30, WY_TMP2, WY_TMP2
vpxor WY, WY_TMP, WY_TMP
.elseif ((i & 7) == 7)
vpxor WY_TMP2, WY_TMP, WY
vpaddd K_XMM(K_BASE), WY, WY_TMP
vmovdqu WY_TMP, PRECALC_WK(i&~7)
PRECALC_ROTATE_WY
.endif
.endm
.macro PRECALC_32_79
/*
* in SHA-1 specification:
* w[i] = (w[i-3] ^ w[i-8] ^ w[i-14] ^ w[i-16]) rol 1
* instead we do equal:
* w[i] = (w[i-6] ^ w[i-16] ^ w[i-28] ^ w[i-32]) rol 2
* allows more efficient vectorization
* since w[i]=>w[i-3] dependency is broken
*/
.if ((i & 7) == 0)
/*
* blended AVX2 and ALU instruction scheduling
* 1 vector iteration per 8 rounds
*/
vpalignr $8, WY_minus_08, WY_minus_04, WY_TMP
.elseif ((i & 7) == 1)
/* W is W_minus_32 before xor */
vpxor WY_minus_28, WY, WY
.elseif ((i & 7) == 2)
vpxor WY_minus_16, WY_TMP, WY_TMP
.elseif ((i & 7) == 3)
vpxor WY_TMP, WY, WY
.elseif ((i & 7) == 4)
vpslld $2, WY, WY_TMP
.elseif ((i & 7) == 5)
vpsrld $30, WY, WY
vpor WY, WY_TMP, WY
.elseif ((i & 7) == 7)
vpaddd K_XMM(K_BASE), WY, WY_TMP
vmovdqu WY_TMP, PRECALC_WK(i&~7)
PRECALC_ROTATE_WY
.endif
.endm
.macro PRECALC r, s
.set i, \r
.if (i < 40)
.set K_XMM, 32*0
.elseif (i < 80)
.set K_XMM, 32*1
.elseif (i < 120)
.set K_XMM, 32*2
.else
.set K_XMM, 32*3
.endif
.if (i<32)
PRECALC_00_15 \s
.elseif (i<64)
PRECALC_16_31 \s
.elseif (i < 160)
PRECALC_32_79 \s
.endif
.endm
.macro ROTATE_STATE
.set T_REG, E
.set E, D
.set D, C
.set C, B
.set B, TB
.set TB, A
.set A, T_REG
.set T_REG, RE
.set RE, RD
.set RD, RC
.set RC, RB
.set RB, RTB
.set RTB, RA
.set RA, T_REG
.endm
/* Macro relies on saved ROUND_Fx */
.macro RND_FUN f, r
.if (\f == RND_F1)
ROUND_F1 \r
.elseif (\f == RND_F2)
ROUND_F2 \r
.elseif (\f == RND_F3)
ROUND_F3 \r
.endif
.endm
.macro RR r
.set round_id, (\r % 80)
.if (round_id == 0) /* Precalculate F for first round */
.set ROUND_FUNC, RND_F1
mov B, TB
rorx $(32-30), B, B /* b>>>2 */
andn D, TB, T1
and C, TB
xor T1, TB
.endif
RND_FUN ROUND_FUNC, \r
ROTATE_STATE
.if (round_id == 18)
.set ROUND_FUNC, RND_F2
.elseif (round_id == 38)
.set ROUND_FUNC, RND_F3
.elseif (round_id == 58)
.set ROUND_FUNC, RND_F2
.endif
.set round_id, ( (\r+1) % 80)
RND_FUN ROUND_FUNC, (\r+1)
ROTATE_STATE
.endm
.macro ROUND_F1 r
add WK(\r), E
andn C, A, T1 /* ~b&d */
lea (RE,RTB), E /* Add F from the previous round */
rorx $(32-5), A, TA /* T2 = A >>> 5 */
rorx $(32-30),A, TB /* b>>>2 for next round */
PRECALC (\r) /* msg scheduling for next 2 blocks */
/*
* Calculate F for the next round
* (b & c) ^ andn[b, d]
*/
and B, A /* b&c */
xor T1, A /* F1 = (b&c) ^ (~b&d) */
lea (RE,RTA), E /* E += A >>> 5 */
.endm
.macro ROUND_F2 r
add WK(\r), E
lea (RE,RTB), E /* Add F from the previous round */
/* Calculate F for the next round */
rorx $(32-5), A, TA /* T2 = A >>> 5 */
.if ((round_id) < 79)
rorx $(32-30), A, TB /* b>>>2 for next round */
.endif
PRECALC (\r) /* msg scheduling for next 2 blocks */
.if ((round_id) < 79)
xor B, A
.endif
add TA, E /* E += A >>> 5 */
.if ((round_id) < 79)
xor C, A
.endif
.endm
.macro ROUND_F3 r
add WK(\r), E
PRECALC (\r) /* msg scheduling for next 2 blocks */
lea (RE,RTB), E /* Add F from the previous round */
mov B, T1
or A, T1
rorx $(32-5), A, TA /* T2 = A >>> 5 */
rorx $(32-30), A, TB /* b>>>2 for next round */
/* Calculate F for the next round
* (b and c) or (d and (b or c))
*/
and C, T1
and B, A
or T1, A
add TA, E /* E += A >>> 5 */
.endm
/*
* macro implements 80 rounds of SHA-1, for multiple blocks with s/w pipelining
*/
.macro SHA1_PIPELINED_MAIN_BODY
REGALLOC
mov (HASH_PTR), A
mov 4(HASH_PTR), B
mov 8(HASH_PTR), C
mov 12(HASH_PTR), D
mov 16(HASH_PTR), E
mov %rsp, PRECALC_BUF
lea (2*4*80+32)(%rsp), WK_BUF
# Precalc WK for first 2 blocks
PRECALC_OFFSET = 0
.set i, 0
.rept 160
PRECALC i
.set i, i + 1
.endr
PRECALC_OFFSET = 128
xchg WK_BUF, PRECALC_BUF
.align 32
_loop:
/*
* code loops through more than one block
* we use K_BASE value as a signal of a last block,
* it is set below by: cmovae BUFFER_PTR, K_BASE
*/
cmp K_BASE, BUFFER_PTR
jne _begin
.align 32
jmp _end
.align 32
_begin:
/*
* Do first block
* rounds: 0,2,4,6,8
*/
.set j, 0
.rept 5
RR j
.set j, j+2
.endr
jmp _loop0
_loop0:
/*
* rounds:
* 10,12,14,16,18
* 20,22,24,26,28
* 30,32,34,36,38
* 40,42,44,46,48
* 50,52,54,56,58
*/
.rept 25
RR j
.set j, j+2
.endr
add $(2*64), BUFFER_PTR /* move to next odd-64-byte block */
cmp BUFFER_END, BUFFER_PTR /* is current block the last one? */
cmovae K_BASE, BUFFER_PTR /* signal the last iteration smartly */
/*
* rounds
* 60,62,64,66,68
* 70,72,74,76,78
*/
.rept 10
RR j
.set j, j+2
.endr
UPDATE_HASH (HASH_PTR), A
UPDATE_HASH 4(HASH_PTR), TB
UPDATE_HASH 8(HASH_PTR), C
UPDATE_HASH 12(HASH_PTR), D
UPDATE_HASH 16(HASH_PTR), E
cmp K_BASE, BUFFER_PTR /* is current block the last one? */
je _loop
mov TB, B
/* Process second block */
/*
* rounds
* 0+80, 2+80, 4+80, 6+80, 8+80
* 10+80,12+80,14+80,16+80,18+80
*/
.set j, 0
.rept 10
RR j+80
.set j, j+2
.endr
jmp _loop1
_loop1:
/*
* rounds
* 20+80,22+80,24+80,26+80,28+80
* 30+80,32+80,34+80,36+80,38+80
*/
.rept 10
RR j+80
.set j, j+2
.endr
jmp _loop2
_loop2:
/*
* rounds
* 40+80,42+80,44+80,46+80,48+80
* 50+80,52+80,54+80,56+80,58+80
*/
.rept 10
RR j+80
.set j, j+2
.endr
add $(2*64), BUFFER_PTR2 /* move to next even-64-byte block */
cmp BUFFER_END, BUFFER_PTR2 /* is current block the last one */
cmovae K_BASE, BUFFER_PTR /* signal the last iteration smartly */
jmp _loop3
_loop3:
/*
* rounds
* 60+80,62+80,64+80,66+80,68+80
* 70+80,72+80,74+80,76+80,78+80
*/
.rept 10
RR j+80
.set j, j+2
.endr
UPDATE_HASH (HASH_PTR), A
UPDATE_HASH 4(HASH_PTR), TB
UPDATE_HASH 8(HASH_PTR), C
UPDATE_HASH 12(HASH_PTR), D
UPDATE_HASH 16(HASH_PTR), E
/* Reset state for AVX2 reg permutation */
mov A, TA
mov TB, A
mov C, TB
mov E, C
mov D, B
mov TA, D
REGALLOC
xchg WK_BUF, PRECALC_BUF
jmp _loop
.align 32
_end:
.endm
/*
* macro implements SHA-1 function's body for several 64-byte blocks
* param: function's name
*/
.macro SHA1_VECTOR_ASM name
ENTRY(\name)
.align 4096
push %rbx
push %rbp
push %r12
push %r13
push %r14
push %r15
RESERVE_STACK = (W_SIZE*4 + 8+24)
/* Align stack */
mov %rsp, %rbx
and $(0x1000-1), %rbx
sub $(8+32), %rbx
sub %rbx, %rsp
push %rbx
sub $RESERVE_STACK, %rsp
avx2_zeroupper
lea K_XMM_AR(%rip), K_BASE
mov CTX, HASH_PTR
mov BUF, BUFFER_PTR
lea 64(BUF), BUFFER_PTR2
shl $6, CNT /* mul by 64 */
add BUF, CNT
add $64, CNT
mov CNT, BUFFER_END
cmp BUFFER_END, BUFFER_PTR2
cmovae K_BASE, BUFFER_PTR2
xmm_mov BSWAP_SHUFB_CTL(%rip), YMM_SHUFB_BSWAP
SHA1_PIPELINED_MAIN_BODY
avx2_zeroupper
add $RESERVE_STACK, %rsp
pop %rbx
add %rbx, %rsp
pop %r15
pop %r14
pop %r13
pop %r12
pop %rbp
pop %rbx
ret
ENDPROC(\name)
.endm
.section .rodata
#define K1 0x5a827999
#define K2 0x6ed9eba1
#define K3 0x8f1bbcdc
#define K4 0xca62c1d6
.align 128
K_XMM_AR:
.long K1, K1, K1, K1
.long K1, K1, K1, K1
.long K2, K2, K2, K2
.long K2, K2, K2, K2
.long K3, K3, K3, K3
.long K3, K3, K3, K3
.long K4, K4, K4, K4
.long K4, K4, K4, K4
BSWAP_SHUFB_CTL:
.long 0x00010203
.long 0x04050607
.long 0x08090a0b
.long 0x0c0d0e0f
.long 0x00010203
.long 0x04050607
.long 0x08090a0b
.long 0x0c0d0e0f
.text
SHA1_VECTOR_ASM sha1_transform_avx2

49
arch/x86/crypto/sha1_ssse3_glue.c
View File

@ -10,6 +10,7 @@
* Copyright (c) Andrew McDonald <andrew@mcdonald.org.uk>
* Copyright (c) Jean-Francois Dive <jef@linuxbe.org>
* Copyright (c) Mathias Krause <minipli@googlemail.com>
* Copyright (c) Chandramouli Narayanan <mouli@linux.intel.com>
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License as published by the Free
@ -39,6 +40,12 @@ asmlinkage void sha1_transform_ssse3(u32 *digest, const char *data,
asmlinkage void sha1_transform_avx(u32 *digest, const char *data,
unsigned int rounds);
#endif
#ifdef CONFIG_AS_AVX2
#define SHA1_AVX2_BLOCK_OPTSIZE 4 /* optimal 4*64 bytes of SHA1 blocks */
asmlinkage void sha1_transform_avx2(u32 *digest, const char *data,
unsigned int rounds);
#endif
static asmlinkage void (*sha1_transform_asm)(u32 *, const char *, unsigned int);
@ -165,6 +172,18 @@ static int sha1_ssse3_import(struct shash_desc *desc, const void *in)
return 0;
}
#ifdef CONFIG_AS_AVX2
static void sha1_apply_transform_avx2(u32 *digest, const char *data,
unsigned int rounds)
{
/* Select the optimal transform based on data block size */
if (rounds >= SHA1_AVX2_BLOCK_OPTSIZE)
sha1_transform_avx2(digest, data, rounds);
else
sha1_transform_avx(digest, data, rounds);
}
#endif
static struct shash_alg alg = {
.digestsize = SHA1_DIGEST_SIZE,
.init = sha1_ssse3_init,
@ -189,7 +208,11 @@ static bool __init avx_usable(void)
{
u64 xcr0;
#if defined(CONFIG_AS_AVX2)
if (!cpu_has_avx || !cpu_has_avx2 || !cpu_has_osxsave)
#else
if (!cpu_has_avx || !cpu_has_osxsave)
#endif
return false;
xcr0 = xgetbv(XCR_XFEATURE_ENABLED_MASK);
@ -205,23 +228,35 @@ static bool __init avx_usable(void)
static int __init sha1_ssse3_mod_init(void)
{
char *algo_name;
/* test for SSSE3 first */
if (cpu_has_ssse3)
if (cpu_has_ssse3) {
sha1_transform_asm = sha1_transform_ssse3;
algo_name = "SSSE3";
}
#ifdef CONFIG_AS_AVX
/* allow AVX to override SSSE3, it's a little faster */
if (avx_usable())
sha1_transform_asm = sha1_transform_avx;
if (avx_usable()) {
if (cpu_has_avx) {
sha1_transform_asm = sha1_transform_avx;
algo_name = "AVX";
}
#ifdef CONFIG_AS_AVX2
if (cpu_has_avx2 && boot_cpu_has(X86_FEATURE_BMI2)) {
/* allow AVX2 to override AVX, it's a little faster */
sha1_transform_asm = sha1_apply_transform_avx2;
algo_name = "AVX2";
}
#endif
}
#endif
if (sha1_transform_asm) {
pr_info("Using %s optimized SHA-1 implementation\n",
sha1_transform_asm == sha1_transform_ssse3 ? "SSSE3"
: "AVX");
pr_info("Using %s optimized SHA-1 implementation\n", algo_name);
return crypto_register_shash(&alg);
}
pr_info("Neither AVX nor SSSE3 is available/usable.\n");
pr_info("Neither AVX nor AVX2 nor SSSE3 is available/usable.\n");
return -ENODEV;
}

4
crypto/Kconfig
View File

@ -491,14 +491,14 @@ config CRYPTO_SHA1
SHA-1 secure hash standard (FIPS 180-1/DFIPS 180-2).
config CRYPTO_SHA1_SSSE3
tristate "SHA1 digest algorithm (SSSE3/AVX)"
tristate "SHA1 digest algorithm (SSSE3/AVX/AVX2)"
depends on X86 && 64BIT
select CRYPTO_SHA1
select CRYPTO_HASH
help
SHA-1 secure hash standard (FIPS 180-1/DFIPS 180-2) implemented
using Supplemental SSE3 (SSSE3) instructions or Advanced Vector
Extensions (AVX), when available.
Extensions (AVX/AVX2), when available.
config CRYPTO_SHA256_SSSE3
tristate "SHA256 digest algorithm (SSSE3/AVX/AVX2)"