Add accelerated arm64 sha512 to libmd

As with sha256 add support for accelerated sha512 support to libmd on
arm64. This depends on clang 13+ to build as this is the first release
with the needed intrinsics. Gcc should also support them, however from
a currently unknown release.

Reviewed by:	cem
Sponsored by:	The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D33373
This commit is contained in:
Andrew Turner 2021-11-19 11:32:58 +00:00
parent 4826406b30
commit 71bf1c4cc5
4 changed files with 229 additions and 0 deletions

View File

@ -3,6 +3,7 @@
SHLIBDIR?= /lib
.include <src.opts.mk>
.include <bsd.compiler.mk>
PACKAGE= runtime
LIB= md
@ -133,6 +134,11 @@ ACFLAGS+= -DELF -Wa,--noexecstack
SRCS+= sha256c_arm64.c
CFLAGS+= -DARM64_SHA2
CFLAGS.sha256c_arm64.c+= -march=armv8-a+crypto
.if ${COMPILER_FEATURES:Maarch64-sha512}
SRCS+= sha512c_arm64.c
CFLAGS+= -DARM64_SHA512
CFLAGS.sha512c_arm64.c+= -march=armv8.2-a+sha3
.endif # aarch64-sha512
.endif
.endif # ${USE_ASM_SOURCES} != 0

View File

@ -40,6 +40,12 @@ __FBSDID("$FreeBSD$");
#include "sha512.h"
#include "sha512t.h"
#include "sha384.h"
#include "sha512c_impl.h"
#if defined(ARM64_SHA512)
#include <sys/auxv.h>
#include <machine/ifunc.h>
#endif
#if BYTE_ORDER == BIG_ENDIAN
@ -158,7 +164,11 @@ static const uint64_t K[80] = {
* the 512-bit input block to produce a new state.
*/
static void
#if defined(ARM64_SHA512)
SHA512_Transform_c(uint64_t * state, const unsigned char block[SHA512_BLOCK_LENGTH])
#else
SHA512_Transform(uint64_t * state, const unsigned char block[SHA512_BLOCK_LENGTH])
#endif
{
uint64_t W[80];
uint64_t S[8];
@ -214,6 +224,29 @@ SHA512_Transform(uint64_t * state, const unsigned char block[SHA512_BLOCK_LENGTH
state[i] += S[i];
}
#if defined(ARM64_SHA512)
static void
SHA512_Transform_arm64(uint64_t * state,
const unsigned char block[SHA512_BLOCK_LENGTH])
{
SHA512_Transform_arm64_impl(state, block, K);
}
DEFINE_UIFUNC(static, void, SHA512_Transform,
(uint64_t * state, const unsigned char block[SHA512_BLOCK_LENGTH]))
{
u_long hwcap;
if (elf_aux_info(AT_HWCAP, &hwcap, sizeof(hwcap)) == 0) {
if ((hwcap & HWCAP_SHA512) != 0) {
return (SHA512_Transform_arm64);
}
}
return (SHA512_Transform_c);
}
#endif
static unsigned char PAD[SHA512_BLOCK_LENGTH] = {
0x80, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,

View File

@ -0,0 +1,153 @@
/*-
* Copyright (c) 2021 The FreeBSD Foundation
*
* This software was developed by Andrew Turner under sponsorship from
* the FreeBSD Foundation.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. 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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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 AUTHOR 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.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/types.h>
#include <arm_neon.h>
#include "sha512.h"
#include "sha512c_impl.h"
void __hidden
SHA512_Transform_arm64_impl(uint64_t * state,
const unsigned char block[SHA512_BLOCK_LENGTH], const uint64_t K[80])
{
uint64x2_t W[8];
uint64x2_t S[4];
uint64x2_t S_start[4];
uint64x2_t K_tmp, S_tmp;
int i;
#define A64_LOAD_W(x) \
W[x] = vld1q_u64((const uint64_t *)(&block[(x) * 16])); \
W[x] = vreinterpretq_u64_u8(vrev64q_u8(vreinterpretq_u8_u64(W[x])))
/* 1. Prepare the first part of the message schedule W. */
A64_LOAD_W(0);
A64_LOAD_W(1);
A64_LOAD_W(2);
A64_LOAD_W(3);
A64_LOAD_W(4);
A64_LOAD_W(5);
A64_LOAD_W(6);
A64_LOAD_W(7);
/* 2. Initialize working variables. */
S[0] = vld1q_u64(&state[0]);
S[1] = vld1q_u64(&state[2]);
S[2] = vld1q_u64(&state[4]);
S[3] = vld1q_u64(&state[6]);
S_start[0] = S[0];
S_start[1] = S[1];
S_start[2] = S[2];
S_start[3] = S[3];
/* 3. Mix. */
for (i = 0; i < 80; i += 16) {
/*
* The schedule array has 4 vectors:
* ab = S[( 8 - i) % 4]
* cd = S[( 9 - i) % 4]
* ef = S[(10 - i) % 4]
* gh = S[(11 - i) % 4]
*
* The following maacro:
* - Loads the round constants
* - Add them to schedule words
* - Rotates the total to switch the order of the two halves
* so they are in the correct order for gh
* - Fix the alignment
* - Extract fg from ef and gh
* - Extract de from cd and ef
* - Pass these into the first part of the sha512 calculation
* to calculate the Sigma 1 and Ch steps
* - Calculate the Sigma 0 and Maj steps and store to gh
* - Add the first part to the cd vector
*/
#define A64_RNDr(S, W, i, ii) \
K_tmp = vld1q_u64(K + (i * 2) + ii); \
K_tmp = vaddq_u64(W[i], K_tmp); \
K_tmp = vextq_u64(K_tmp, K_tmp, 1); \
K_tmp = vaddq_u64(K_tmp, S[(11 - i) % 4]); \
S_tmp = vsha512hq_u64(K_tmp, \
vextq_u64(S[(10 - i) % 4], S[(11 - i) % 4], 1), \
vextq_u64(S[(9 - i) % 4], S[(10 - i) % 4], 1)); \
S[(11 - i) % 4] = vsha512h2q_u64(S_tmp, S[(9 - i) % 4], S[(8 - i) % 4]); \
S[(9 - i) % 4] = vaddq_u64(S[(9 - i) % 4], S_tmp)
A64_RNDr(S, W, 0, i);
A64_RNDr(S, W, 1, i);
A64_RNDr(S, W, 2, i);
A64_RNDr(S, W, 3, i);
A64_RNDr(S, W, 4, i);
A64_RNDr(S, W, 5, i);
A64_RNDr(S, W, 6, i);
A64_RNDr(S, W, 7, i);
if (i == 64)
break;
/*
* Perform the Message schedule computation:
* - vsha512su0q_u64 performs the sigma 0 half and add it to
* the old value
* - vextq_u64 fixes the alignment of the vectors
* - vsha512su1q_u64 performs the sigma 1 half and adds it
* and both the above all together
*/
#define A64_MSCH(x) \
W[x] = vsha512su1q_u64( \
vsha512su0q_u64(W[x], W[(x + 1) % 8]), \
W[(x + 7) % 8], \
vextq_u64(W[(x + 4) % 8], W[(x + 5) % 8], 1))
A64_MSCH(0);
A64_MSCH(1);
A64_MSCH(2);
A64_MSCH(3);
A64_MSCH(4);
A64_MSCH(5);
A64_MSCH(6);
A64_MSCH(7);
}
/* 4. Mix local working variables into global state */
S[0] = vaddq_u64(S[0], S_start[0]);
S[1] = vaddq_u64(S[1], S_start[1]);
S[2] = vaddq_u64(S[2], S_start[2]);
S[3] = vaddq_u64(S[3], S_start[3]);
vst1q_u64(&state[0], S[0]);
vst1q_u64(&state[2], S[1]);
vst1q_u64(&state[4], S[2]);
vst1q_u64(&state[6], S[3]);
}

View File

@ -0,0 +1,37 @@
/*-
* Copyright (c) 2021 The FreeBSD Foundation
*
* This software was developed by Andrew Turner under sponsorship from
* the FreeBSD Foundation.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. 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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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 AUTHOR 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.
*/
#ifndef _SHA512_IMPL_H_
#define _SHA512_IMPL_H_
#if defined(ARM64_SHA512)
void SHA512_Transform_arm64_impl(uint64_t * state,
const unsigned char block[SHA512_BLOCK_LENGTH], const uint64_t K[80]);
#endif
#endif /* !_SHA512_IMPL_H_ */