c/mirror_zfs
1
0
Fork 0
mirror of https://git.proxmox.com/git/mirror_zfs.git synced 2024-11-19 19:01:00 +03:00
Code Issues Packages Projects Releases Wiki Activity
19516b69ee
mirror_zfs/module/icp/asm-x86_64/blake3/blake3_sse2.S

2324 lines
68 KiB
ArmAsm
Raw Normal View History

Introduce BLAKE3 checksums as an OpenZFS feature This commit adds BLAKE3 checksums to OpenZFS, it has similar performance to Edon-R, but without the caveats around the latter. Homepage of BLAKE3: https://github.com/BLAKE3-team/BLAKE3 Wikipedia: https://en.wikipedia.org/wiki/BLAKE_(hash_function)#BLAKE3 Short description of Wikipedia: BLAKE3 is a cryptographic hash function based on Bao and BLAKE2, created by Jack O'Connor, Jean-Philippe Aumasson, Samuel Neves, and Zooko Wilcox-O'Hearn. It was announced on January 9, 2020, at Real World Crypto. BLAKE3 is a single algorithm with many desirable features (parallelism, XOF, KDF, PRF and MAC), in contrast to BLAKE and BLAKE2, which are algorithm families with multiple variants. BLAKE3 has a binary tree structure, so it supports a practically unlimited degree of parallelism (both SIMD and multithreading) given enough input. The official Rust and C implementations are dual-licensed as public domain (CC0) and the Apache License. Along with adding the BLAKE3 hash into the OpenZFS infrastructure a new benchmarking file called chksum_bench was introduced. When read it reports the speed of the available checksum functions. On Linux: cat /proc/spl/kstat/zfs/chksum_bench On FreeBSD: sysctl kstat.zfs.misc.chksum_bench This is an example output of an i3-1005G1 test system with Debian 11: implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1196 1602 1761 1749 1762 1759 1751 skein-generic 546 591 608 615 619 612 616 sha256-generic 240 300 316 314 304 285 276 sha512-generic 353 441 467 476 472 467 426 blake3-generic 308 313 313 313 312 313 312 blake3-sse2 402 1289 1423 1446 1432 1458 1413 blake3-sse41 427 1470 1625 1704 1679 1607 1629 blake3-avx2 428 1920 3095 3343 3356 3318 3204 blake3-avx512 473 2687 4905 5836 5844 5643 5374 Output on Debian 5.10.0-10-amd64 system: (Ryzen 7 5800X) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1840 2458 2665 2719 2711 2723 2693 skein-generic 870 966 996 992 1003 1005 1009 sha256-generic 415 442 453 455 457 457 457 sha512-generic 608 690 711 718 719 720 721 blake3-generic 301 313 311 309 309 310 310 blake3-sse2 343 1865 2124 2188 2180 2181 2186 blake3-sse41 364 2091 2396 2509 2463 2482 2488 blake3-avx2 365 2590 4399 4971 4915 4802 4764 Output on Debian 5.10.0-9-powerpc64le system: (POWER 9) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1213 1703 1889 1918 1957 1902 1907 skein-generic 434 492 520 522 511 525 525 sha256-generic 167 183 187 188 188 187 188 sha512-generic 186 216 222 221 225 224 224 blake3-generic 153 152 154 153 151 153 153 blake3-sse2 391 1170 1366 1406 1428 1426 1414 blake3-sse41 352 1049 1212 1174 1262 1258 1259 Output on Debian 5.10.0-11-arm64 system: (Pi400) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 487 603 629 639 643 641 641 skein-generic 271 299 303 308 309 309 307 sha256-generic 117 127 128 130 130 129 130 sha512-generic 145 165 170 172 173 174 175 blake3-generic 81 29 71 89 89 89 89 blake3-sse2 112 323 368 379 380 371 374 blake3-sse41 101 315 357 368 369 364 360 Structurally, the new code is mainly split into these parts: - 1x cross platform generic c variant: blake3_generic.c - 4x assembly for X86-64 (SSE2, SSE4.1, AVX2, AVX512) - 2x assembly for ARMv8 (NEON converted from SSE2) - 2x assembly for PPC64-LE (POWER8 converted from SSE2) - one file for switching between the implementations Note the PPC64 assembly requires the VSX instruction set and the kfpu_begin() / kfpu_end() calls on PowerPC were updated accordingly. Reviewed-by: Felix Dörre <felix@dogcraft.de> Reviewed-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz> Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Tino Reichardt <milky-zfs@mcmilk.de> Co-authored-by: Rich Ercolani <rincebrain@gmail.com> Closes #10058 Closes #12918
2022-06-09 01:55:57 +03:00
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
Replace dead opensolaris.org license link The commit replaces all findings of the link: http://www.opensolaris.org/os/licensing with this one: https://opensource.org/licenses/CDDL-1.0 Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Tino Reichardt <milky-zfs@mcmilk.de> Closes #13619
2022-07-12 00:16:13 +03:00
* or https://opensource.org/licenses/CDDL-1.0.
Introduce BLAKE3 checksums as an OpenZFS feature This commit adds BLAKE3 checksums to OpenZFS, it has similar performance to Edon-R, but without the caveats around the latter. Homepage of BLAKE3: https://github.com/BLAKE3-team/BLAKE3 Wikipedia: https://en.wikipedia.org/wiki/BLAKE_(hash_function)#BLAKE3 Short description of Wikipedia: BLAKE3 is a cryptographic hash function based on Bao and BLAKE2, created by Jack O'Connor, Jean-Philippe Aumasson, Samuel Neves, and Zooko Wilcox-O'Hearn. It was announced on January 9, 2020, at Real World Crypto. BLAKE3 is a single algorithm with many desirable features (parallelism, XOF, KDF, PRF and MAC), in contrast to BLAKE and BLAKE2, which are algorithm families with multiple variants. BLAKE3 has a binary tree structure, so it supports a practically unlimited degree of parallelism (both SIMD and multithreading) given enough input. The official Rust and C implementations are dual-licensed as public domain (CC0) and the Apache License. Along with adding the BLAKE3 hash into the OpenZFS infrastructure a new benchmarking file called chksum_bench was introduced. When read it reports the speed of the available checksum functions. On Linux: cat /proc/spl/kstat/zfs/chksum_bench On FreeBSD: sysctl kstat.zfs.misc.chksum_bench This is an example output of an i3-1005G1 test system with Debian 11: implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1196 1602 1761 1749 1762 1759 1751 skein-generic 546 591 608 615 619 612 616 sha256-generic 240 300 316 314 304 285 276 sha512-generic 353 441 467 476 472 467 426 blake3-generic 308 313 313 313 312 313 312 blake3-sse2 402 1289 1423 1446 1432 1458 1413 blake3-sse41 427 1470 1625 1704 1679 1607 1629 blake3-avx2 428 1920 3095 3343 3356 3318 3204 blake3-avx512 473 2687 4905 5836 5844 5643 5374 Output on Debian 5.10.0-10-amd64 system: (Ryzen 7 5800X) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1840 2458 2665 2719 2711 2723 2693 skein-generic 870 966 996 992 1003 1005 1009 sha256-generic 415 442 453 455 457 457 457 sha512-generic 608 690 711 718 719 720 721 blake3-generic 301 313 311 309 309 310 310 blake3-sse2 343 1865 2124 2188 2180 2181 2186 blake3-sse41 364 2091 2396 2509 2463 2482 2488 blake3-avx2 365 2590 4399 4971 4915 4802 4764 Output on Debian 5.10.0-9-powerpc64le system: (POWER 9) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1213 1703 1889 1918 1957 1902 1907 skein-generic 434 492 520 522 511 525 525 sha256-generic 167 183 187 188 188 187 188 sha512-generic 186 216 222 221 225 224 224 blake3-generic 153 152 154 153 151 153 153 blake3-sse2 391 1170 1366 1406 1428 1426 1414 blake3-sse41 352 1049 1212 1174 1262 1258 1259 Output on Debian 5.10.0-11-arm64 system: (Pi400) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 487 603 629 639 643 641 641 skein-generic 271 299 303 308 309 309 307 sha256-generic 117 127 128 130 130 129 130 sha512-generic 145 165 170 172 173 174 175 blake3-generic 81 29 71 89 89 89 89 blake3-sse2 112 323 368 379 380 371 374 blake3-sse41 101 315 357 368 369 364 360 Structurally, the new code is mainly split into these parts: - 1x cross platform generic c variant: blake3_generic.c - 4x assembly for X86-64 (SSE2, SSE4.1, AVX2, AVX512) - 2x assembly for ARMv8 (NEON converted from SSE2) - 2x assembly for PPC64-LE (POWER8 converted from SSE2) - one file for switching between the implementations Note the PPC64 assembly requires the VSX instruction set and the kfpu_begin() / kfpu_end() calls on PowerPC were updated accordingly. Reviewed-by: Felix Dörre <felix@dogcraft.de> Reviewed-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz> Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Tino Reichardt <milky-zfs@mcmilk.de> Co-authored-by: Rich Ercolani <rincebrain@gmail.com> Closes #10058 Closes #12918
2022-06-09 01:55:57 +03:00
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Based on BLAKE3 v1.3.1, https://github.com/BLAKE3-team/BLAKE3
* Copyright (c) 2019-2020 Samuel Neves and Matthew Krupcale
* Copyright (c) 2022 Tino Reichardt <milky-zfs@mcmilk.de>
*/
#if defined(HAVE_SSE2)
#define _ASM
#include <sys/asm_linkage.h>
#if defined(__ELF__) && defined(__CET__) && defined(__has_include)
#if __has_include(<cet.h>)
#include <cet.h>
#endif
#endif
#if !defined(_CET_ENDBR)
#define _CET_ENDBR
#endif
.intel_syntax noprefix
.global zfs_blake3_hash_many_sse2
.global zfs_blake3_compress_in_place_sse2
.global zfs_blake3_compress_xof_sse2
.text
.type zfs_blake3_hash_many_sse2,@function
.type zfs_blake3_compress_in_place_sse2,@function
.type zfs_blake3_compress_xof_sse2,@function
.p2align 6
zfs_blake3_hash_many_sse2:
_CET_ENDBR
push r15
push r14
push r13
push r12
push rbx
push rbp
mov rbp, rsp
sub rsp, 360
and rsp, 0xFFFFFFFFFFFFFFC0
neg r9d
movd xmm0, r9d
pshufd xmm0, xmm0, 0x00
movdqa xmmword ptr [rsp+0x130], xmm0
movdqa xmm1, xmm0
pand xmm1, xmmword ptr [ADD0+rip]
pand xmm0, xmmword ptr [ADD1+rip]
movdqa xmmword ptr [rsp+0x150], xmm0
movd xmm0, r8d
pshufd xmm0, xmm0, 0x00
paddd xmm0, xmm1
movdqa xmmword ptr [rsp+0x110], xmm0
pxor xmm0, xmmword ptr [CMP_MSB_MASK+rip]
pxor xmm1, xmmword ptr [CMP_MSB_MASK+rip]
pcmpgtd xmm1, xmm0
shr r8, 32
movd xmm2, r8d
pshufd xmm2, xmm2, 0x00
psubd xmm2, xmm1
movdqa xmmword ptr [rsp+0x120], xmm2
mov rbx, qword ptr [rbp+0x50]
mov r15, rdx
shl r15, 6
movzx r13d, byte ptr [rbp+0x38]
movzx r12d, byte ptr [rbp+0x48]
cmp rsi, 4
jc 3f
2:
movdqu xmm3, xmmword ptr [rcx]
pshufd xmm0, xmm3, 0x00
pshufd xmm1, xmm3, 0x55
pshufd xmm2, xmm3, 0xAA
pshufd xmm3, xmm3, 0xFF
movdqu xmm7, xmmword ptr [rcx+0x10]
pshufd xmm4, xmm7, 0x00
pshufd xmm5, xmm7, 0x55
pshufd xmm6, xmm7, 0xAA
pshufd xmm7, xmm7, 0xFF
mov r8, qword ptr [rdi]
mov r9, qword ptr [rdi+0x8]
mov r10, qword ptr [rdi+0x10]
mov r11, qword ptr [rdi+0x18]
movzx eax, byte ptr [rbp+0x40]
or eax, r13d
xor edx, edx
9:
mov r14d, eax
or eax, r12d
add rdx, 64
cmp rdx, r15
cmovne eax, r14d
movdqu xmm8, xmmword ptr [r8+rdx-0x40]
movdqu xmm9, xmmword ptr [r9+rdx-0x40]
movdqu xmm10, xmmword ptr [r10+rdx-0x40]
movdqu xmm11, xmmword ptr [r11+rdx-0x40]
movdqa xmm12, xmm8
punpckldq xmm8, xmm9
punpckhdq xmm12, xmm9
movdqa xmm14, xmm10
punpckldq xmm10, xmm11
punpckhdq xmm14, xmm11
movdqa xmm9, xmm8
punpcklqdq xmm8, xmm10
punpckhqdq xmm9, xmm10
movdqa xmm13, xmm12
punpcklqdq xmm12, xmm14
punpckhqdq xmm13, xmm14
movdqa xmmword ptr [rsp], xmm8
movdqa xmmword ptr [rsp+0x10], xmm9
movdqa xmmword ptr [rsp+0x20], xmm12
movdqa xmmword ptr [rsp+0x30], xmm13
movdqu xmm8, xmmword ptr [r8+rdx-0x30]
movdqu xmm9, xmmword ptr [r9+rdx-0x30]
movdqu xmm10, xmmword ptr [r10+rdx-0x30]
movdqu xmm11, xmmword ptr [r11+rdx-0x30]
movdqa xmm12, xmm8
punpckldq xmm8, xmm9
punpckhdq xmm12, xmm9
movdqa xmm14, xmm10
punpckldq xmm10, xmm11
punpckhdq xmm14, xmm11
movdqa xmm9, xmm8
punpcklqdq xmm8, xmm10
punpckhqdq xmm9, xmm10
movdqa xmm13, xmm12
punpcklqdq xmm12, xmm14
punpckhqdq xmm13, xmm14
movdqa xmmword ptr [rsp+0x40], xmm8
movdqa xmmword ptr [rsp+0x50], xmm9
movdqa xmmword ptr [rsp+0x60], xmm12
movdqa xmmword ptr [rsp+0x70], xmm13
movdqu xmm8, xmmword ptr [r8+rdx-0x20]
movdqu xmm9, xmmword ptr [r9+rdx-0x20]
movdqu xmm10, xmmword ptr [r10+rdx-0x20]
movdqu xmm11, xmmword ptr [r11+rdx-0x20]
movdqa xmm12, xmm8
punpckldq xmm8, xmm9
punpckhdq xmm12, xmm9
movdqa xmm14, xmm10
punpckldq xmm10, xmm11
punpckhdq xmm14, xmm11
movdqa xmm9, xmm8
punpcklqdq xmm8, xmm10
punpckhqdq xmm9, xmm10
movdqa xmm13, xmm12
punpcklqdq xmm12, xmm14
punpckhqdq xmm13, xmm14
movdqa xmmword ptr [rsp+0x80], xmm8
movdqa xmmword ptr [rsp+0x90], xmm9
movdqa xmmword ptr [rsp+0xA0], xmm12
movdqa xmmword ptr [rsp+0xB0], xmm13
movdqu xmm8, xmmword ptr [r8+rdx-0x10]
movdqu xmm9, xmmword ptr [r9+rdx-0x10]
movdqu xmm10, xmmword ptr [r10+rdx-0x10]
movdqu xmm11, xmmword ptr [r11+rdx-0x10]
movdqa xmm12, xmm8
punpckldq xmm8, xmm9
punpckhdq xmm12, xmm9
movdqa xmm14, xmm10
punpckldq xmm10, xmm11
punpckhdq xmm14, xmm11
movdqa xmm9, xmm8
punpcklqdq xmm8, xmm10
punpckhqdq xmm9, xmm10
movdqa xmm13, xmm12
punpcklqdq xmm12, xmm14
punpckhqdq xmm13, xmm14
movdqa xmmword ptr [rsp+0xC0], xmm8
movdqa xmmword ptr [rsp+0xD0], xmm9
movdqa xmmword ptr [rsp+0xE0], xmm12
movdqa xmmword ptr [rsp+0xF0], xmm13
movdqa xmm9, xmmword ptr [BLAKE3_IV_1+rip]
movdqa xmm10, xmmword ptr [BLAKE3_IV_2+rip]
movdqa xmm11, xmmword ptr [BLAKE3_IV_3+rip]
movdqa xmm12, xmmword ptr [rsp+0x110]
movdqa xmm13, xmmword ptr [rsp+0x120]
movdqa xmm14, xmmword ptr [BLAKE3_BLOCK_LEN+rip]
movd xmm15, eax
pshufd xmm15, xmm15, 0x00
prefetcht0 [r8+rdx+0x80]
prefetcht0 [r9+rdx+0x80]
prefetcht0 [r10+rdx+0x80]
prefetcht0 [r11+rdx+0x80]
paddd xmm0, xmmword ptr [rsp]
paddd xmm1, xmmword ptr [rsp+0x20]
paddd xmm2, xmmword ptr [rsp+0x40]
paddd xmm3, xmmword ptr [rsp+0x60]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
movdqa xmm8, xmmword ptr [BLAKE3_IV_0+rip]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0x10]
paddd xmm1, xmmword ptr [rsp+0x30]
paddd xmm2, xmmword ptr [rsp+0x50]
paddd xmm3, xmmword ptr [rsp+0x70]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0x80]
paddd xmm1, xmmword ptr [rsp+0xA0]
paddd xmm2, xmmword ptr [rsp+0xC0]
paddd xmm3, xmmword ptr [rsp+0xE0]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp+0x90]
paddd xmm1, xmmword ptr [rsp+0xB0]
paddd xmm2, xmmword ptr [rsp+0xD0]
paddd xmm3, xmmword ptr [rsp+0xF0]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp+0x20]
paddd xmm1, xmmword ptr [rsp+0x30]
paddd xmm2, xmmword ptr [rsp+0x70]
paddd xmm3, xmmword ptr [rsp+0x40]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0x60]
paddd xmm1, xmmword ptr [rsp+0xA0]
paddd xmm2, xmmword ptr [rsp]
paddd xmm3, xmmword ptr [rsp+0xD0]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0x10]
paddd xmm1, xmmword ptr [rsp+0xC0]
paddd xmm2, xmmword ptr [rsp+0x90]
paddd xmm3, xmmword ptr [rsp+0xF0]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp+0xB0]
paddd xmm1, xmmword ptr [rsp+0x50]
paddd xmm2, xmmword ptr [rsp+0xE0]
paddd xmm3, xmmword ptr [rsp+0x80]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp+0x30]
paddd xmm1, xmmword ptr [rsp+0xA0]
paddd xmm2, xmmword ptr [rsp+0xD0]
paddd xmm3, xmmword ptr [rsp+0x70]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0x40]
paddd xmm1, xmmword ptr [rsp+0xC0]
paddd xmm2, xmmword ptr [rsp+0x20]
paddd xmm3, xmmword ptr [rsp+0xE0]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0x60]
paddd xmm1, xmmword ptr [rsp+0x90]
paddd xmm2, xmmword ptr [rsp+0xB0]
paddd xmm3, xmmword ptr [rsp+0x80]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp+0x50]
paddd xmm1, xmmword ptr [rsp]
paddd xmm2, xmmword ptr [rsp+0xF0]
paddd xmm3, xmmword ptr [rsp+0x10]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp+0xA0]
paddd xmm1, xmmword ptr [rsp+0xC0]
paddd xmm2, xmmword ptr [rsp+0xE0]
paddd xmm3, xmmword ptr [rsp+0xD0]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0x70]
paddd xmm1, xmmword ptr [rsp+0x90]
paddd xmm2, xmmword ptr [rsp+0x30]
paddd xmm3, xmmword ptr [rsp+0xF0]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0x40]
paddd xmm1, xmmword ptr [rsp+0xB0]
paddd xmm2, xmmword ptr [rsp+0x50]
paddd xmm3, xmmword ptr [rsp+0x10]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp]
paddd xmm1, xmmword ptr [rsp+0x20]
paddd xmm2, xmmword ptr [rsp+0x80]
paddd xmm3, xmmword ptr [rsp+0x60]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp+0xC0]
paddd xmm1, xmmword ptr [rsp+0x90]
paddd xmm2, xmmword ptr [rsp+0xF0]
paddd xmm3, xmmword ptr [rsp+0xE0]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0xD0]
paddd xmm1, xmmword ptr [rsp+0xB0]
paddd xmm2, xmmword ptr [rsp+0xA0]
paddd xmm3, xmmword ptr [rsp+0x80]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0x70]
paddd xmm1, xmmword ptr [rsp+0x50]
paddd xmm2, xmmword ptr [rsp]
paddd xmm3, xmmword ptr [rsp+0x60]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp+0x20]
paddd xmm1, xmmword ptr [rsp+0x30]
paddd xmm2, xmmword ptr [rsp+0x10]
paddd xmm3, xmmword ptr [rsp+0x40]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp+0x90]
paddd xmm1, xmmword ptr [rsp+0xB0]
paddd xmm2, xmmword ptr [rsp+0x80]
paddd xmm3, xmmword ptr [rsp+0xF0]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0xE0]
paddd xmm1, xmmword ptr [rsp+0x50]
paddd xmm2, xmmword ptr [rsp+0xC0]
paddd xmm3, xmmword ptr [rsp+0x10]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0xD0]
paddd xmm1, xmmword ptr [rsp]
paddd xmm2, xmmword ptr [rsp+0x20]
paddd xmm3, xmmword ptr [rsp+0x40]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp+0x30]
paddd xmm1, xmmword ptr [rsp+0xA0]
paddd xmm2, xmmword ptr [rsp+0x60]
paddd xmm3, xmmword ptr [rsp+0x70]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp+0xB0]
paddd xmm1, xmmword ptr [rsp+0x50]
paddd xmm2, xmmword ptr [rsp+0x10]
paddd xmm3, xmmword ptr [rsp+0x80]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0xF0]
paddd xmm1, xmmword ptr [rsp]
paddd xmm2, xmmword ptr [rsp+0x90]
paddd xmm3, xmmword ptr [rsp+0x60]
paddd xmm0, xmm4
paddd xmm1, xmm5
paddd xmm2, xmm6
paddd xmm3, xmm7
pxor xmm12, xmm0
pxor xmm13, xmm1
pxor xmm14, xmm2
pxor xmm15, xmm3
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm12
paddd xmm9, xmm13
paddd xmm10, xmm14
paddd xmm11, xmm15
pxor xmm4, xmm8
pxor xmm5, xmm9
pxor xmm6, xmm10
pxor xmm7, xmm11
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
paddd xmm0, xmmword ptr [rsp+0xE0]
paddd xmm1, xmmword ptr [rsp+0x20]
paddd xmm2, xmmword ptr [rsp+0x30]
paddd xmm3, xmmword ptr [rsp+0x70]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
pshuflw xmm15, xmm15, 0xB1
pshufhw xmm15, xmm15, 0xB1
pshuflw xmm12, xmm12, 0xB1
pshufhw xmm12, xmm12, 0xB1
pshuflw xmm13, xmm13, 0xB1
pshufhw xmm13, xmm13, 0xB1
pshuflw xmm14, xmm14, 0xB1
pshufhw xmm14, xmm14, 0xB1
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
movdqa xmmword ptr [rsp+0x100], xmm8
movdqa xmm8, xmm5
psrld xmm8, 12
pslld xmm5, 20
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 12
pslld xmm6, 20
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 12
pslld xmm7, 20
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 12
pslld xmm4, 20
por xmm4, xmm8
paddd xmm0, xmmword ptr [rsp+0xA0]
paddd xmm1, xmmword ptr [rsp+0xC0]
paddd xmm2, xmmword ptr [rsp+0x40]
paddd xmm3, xmmword ptr [rsp+0xD0]
paddd xmm0, xmm5
paddd xmm1, xmm6
paddd xmm2, xmm7
paddd xmm3, xmm4
pxor xmm15, xmm0
pxor xmm12, xmm1
pxor xmm13, xmm2
pxor xmm14, xmm3
movdqa xmm8, xmm15
psrld xmm15, 8
pslld xmm8, 24
pxor xmm15, xmm8
movdqa xmm8, xmm12
psrld xmm12, 8
pslld xmm8, 24
pxor xmm12, xmm8
movdqa xmm8, xmm13
psrld xmm13, 8
pslld xmm8, 24
pxor xmm13, xmm8
movdqa xmm8, xmm14
psrld xmm14, 8
pslld xmm8, 24
pxor xmm14, xmm8
paddd xmm10, xmm15
paddd xmm11, xmm12
movdqa xmm8, xmmword ptr [rsp+0x100]
paddd xmm8, xmm13
paddd xmm9, xmm14
pxor xmm5, xmm10
pxor xmm6, xmm11
pxor xmm7, xmm8
pxor xmm4, xmm9
pxor xmm0, xmm8
pxor xmm1, xmm9
pxor xmm2, xmm10
pxor xmm3, xmm11
movdqa xmm8, xmm5
psrld xmm8, 7
pslld xmm5, 25
por xmm5, xmm8
movdqa xmm8, xmm6
psrld xmm8, 7
pslld xmm6, 25
por xmm6, xmm8
movdqa xmm8, xmm7
psrld xmm8, 7
pslld xmm7, 25
por xmm7, xmm8
movdqa xmm8, xmm4
psrld xmm8, 7
pslld xmm4, 25
por xmm4, xmm8
pxor xmm4, xmm12
pxor xmm5, xmm13
pxor xmm6, xmm14
pxor xmm7, xmm15
mov eax, r13d
jne 9b
movdqa xmm9, xmm0
punpckldq xmm0, xmm1
punpckhdq xmm9, xmm1
movdqa xmm11, xmm2
punpckldq xmm2, xmm3
punpckhdq xmm11, xmm3
movdqa xmm1, xmm0
punpcklqdq xmm0, xmm2
punpckhqdq xmm1, xmm2
movdqa xmm3, xmm9
punpcklqdq xmm9, xmm11
punpckhqdq xmm3, xmm11
movdqu xmmword ptr [rbx], xmm0
movdqu xmmword ptr [rbx+0x20], xmm1
movdqu xmmword ptr [rbx+0x40], xmm9
movdqu xmmword ptr [rbx+0x60], xmm3
movdqa xmm9, xmm4
punpckldq xmm4, xmm5
punpckhdq xmm9, xmm5
movdqa xmm11, xmm6
punpckldq xmm6, xmm7
punpckhdq xmm11, xmm7
movdqa xmm5, xmm4
punpcklqdq xmm4, xmm6
punpckhqdq xmm5, xmm6
movdqa xmm7, xmm9
punpcklqdq xmm9, xmm11
punpckhqdq xmm7, xmm11
movdqu xmmword ptr [rbx+0x10], xmm4
movdqu xmmword ptr [rbx+0x30], xmm5
movdqu xmmword ptr [rbx+0x50], xmm9
movdqu xmmword ptr [rbx+0x70], xmm7
movdqa xmm1, xmmword ptr [rsp+0x110]
movdqa xmm0, xmm1
paddd xmm1, xmmword ptr [rsp+0x150]
movdqa xmmword ptr [rsp+0x110], xmm1
pxor xmm0, xmmword ptr [CMP_MSB_MASK+rip]
pxor xmm1, xmmword ptr [CMP_MSB_MASK+rip]
pcmpgtd xmm0, xmm1
movdqa xmm1, xmmword ptr [rsp+0x120]
psubd xmm1, xmm0
movdqa xmmword ptr [rsp+0x120], xmm1
add rbx, 128
add rdi, 32
sub rsi, 4
cmp rsi, 4
jnc 2b
test rsi, rsi
jnz 3f
4:
mov rsp, rbp
pop rbp
pop rbx
pop r12
pop r13
pop r14
pop r15
Fix objtool: missing int3 after ret warning Resolve straight-line speculation warnings reported by objtool for x86_64 assembly on Linux when CONFIG_SLS is set. See the following LWN article for the complete details. https://lwn.net/Articles/877845/ Reviewed-by: Ryan Moeller <ryan@iXsystems.com> Reviewed-by: Alexander Motin <mav@FreeBSD.org> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #13528 Closes #13575
2022-06-21 02:36:21 +03:00
RET
Introduce BLAKE3 checksums as an OpenZFS feature This commit adds BLAKE3 checksums to OpenZFS, it has similar performance to Edon-R, but without the caveats around the latter. Homepage of BLAKE3: https://github.com/BLAKE3-team/BLAKE3 Wikipedia: https://en.wikipedia.org/wiki/BLAKE_(hash_function)#BLAKE3 Short description of Wikipedia: BLAKE3 is a cryptographic hash function based on Bao and BLAKE2, created by Jack O'Connor, Jean-Philippe Aumasson, Samuel Neves, and Zooko Wilcox-O'Hearn. It was announced on January 9, 2020, at Real World Crypto. BLAKE3 is a single algorithm with many desirable features (parallelism, XOF, KDF, PRF and MAC), in contrast to BLAKE and BLAKE2, which are algorithm families with multiple variants. BLAKE3 has a binary tree structure, so it supports a practically unlimited degree of parallelism (both SIMD and multithreading) given enough input. The official Rust and C implementations are dual-licensed as public domain (CC0) and the Apache License. Along with adding the BLAKE3 hash into the OpenZFS infrastructure a new benchmarking file called chksum_bench was introduced. When read it reports the speed of the available checksum functions. On Linux: cat /proc/spl/kstat/zfs/chksum_bench On FreeBSD: sysctl kstat.zfs.misc.chksum_bench This is an example output of an i3-1005G1 test system with Debian 11: implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1196 1602 1761 1749 1762 1759 1751 skein-generic 546 591 608 615 619 612 616 sha256-generic 240 300 316 314 304 285 276 sha512-generic 353 441 467 476 472 467 426 blake3-generic 308 313 313 313 312 313 312 blake3-sse2 402 1289 1423 1446 1432 1458 1413 blake3-sse41 427 1470 1625 1704 1679 1607 1629 blake3-avx2 428 1920 3095 3343 3356 3318 3204 blake3-avx512 473 2687 4905 5836 5844 5643 5374 Output on Debian 5.10.0-10-amd64 system: (Ryzen 7 5800X) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1840 2458 2665 2719 2711 2723 2693 skein-generic 870 966 996 992 1003 1005 1009 sha256-generic 415 442 453 455 457 457 457 sha512-generic 608 690 711 718 719 720 721 blake3-generic 301 313 311 309 309 310 310 blake3-sse2 343 1865 2124 2188 2180 2181 2186 blake3-sse41 364 2091 2396 2509 2463 2482 2488 blake3-avx2 365 2590 4399 4971 4915 4802 4764 Output on Debian 5.10.0-9-powerpc64le system: (POWER 9) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1213 1703 1889 1918 1957 1902 1907 skein-generic 434 492 520 522 511 525 525 sha256-generic 167 183 187 188 188 187 188 sha512-generic 186 216 222 221 225 224 224 blake3-generic 153 152 154 153 151 153 153 blake3-sse2 391 1170 1366 1406 1428 1426 1414 blake3-sse41 352 1049 1212 1174 1262 1258 1259 Output on Debian 5.10.0-11-arm64 system: (Pi400) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 487 603 629 639 643 641 641 skein-generic 271 299 303 308 309 309 307 sha256-generic 117 127 128 130 130 129 130 sha512-generic 145 165 170 172 173 174 175 blake3-generic 81 29 71 89 89 89 89 blake3-sse2 112 323 368 379 380 371 374 blake3-sse41 101 315 357 368 369 364 360 Structurally, the new code is mainly split into these parts: - 1x cross platform generic c variant: blake3_generic.c - 4x assembly for X86-64 (SSE2, SSE4.1, AVX2, AVX512) - 2x assembly for ARMv8 (NEON converted from SSE2) - 2x assembly for PPC64-LE (POWER8 converted from SSE2) - one file for switching between the implementations Note the PPC64 assembly requires the VSX instruction set and the kfpu_begin() / kfpu_end() calls on PowerPC were updated accordingly. Reviewed-by: Felix Dörre <felix@dogcraft.de> Reviewed-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz> Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Tino Reichardt <milky-zfs@mcmilk.de> Co-authored-by: Rich Ercolani <rincebrain@gmail.com> Closes #10058 Closes #12918
2022-06-09 01:55:57 +03:00
.p2align 5
3:
test esi, 0x2
je 3f
movups xmm0, xmmword ptr [rcx]
movups xmm1, xmmword ptr [rcx+0x10]
movaps xmm8, xmm0
movaps xmm9, xmm1
movd xmm13, dword ptr [rsp+0x110]
movd xmm14, dword ptr [rsp+0x120]
punpckldq xmm13, xmm14
movaps xmmword ptr [rsp], xmm13
movd xmm14, dword ptr [rsp+0x114]
movd xmm13, dword ptr [rsp+0x124]
punpckldq xmm14, xmm13
movaps xmmword ptr [rsp+0x10], xmm14
mov r8, qword ptr [rdi]
mov r9, qword ptr [rdi+0x8]
movzx eax, byte ptr [rbp+0x40]
or eax, r13d
xor edx, edx
2:
mov r14d, eax
or eax, r12d
add rdx, 64
cmp rdx, r15
cmovne eax, r14d
movaps xmm2, xmmword ptr [BLAKE3_IV+rip]
movaps xmm10, xmm2
movups xmm4, xmmword ptr [r8+rdx-0x40]
movups xmm5, xmmword ptr [r8+rdx-0x30]
movaps xmm3, xmm4
shufps xmm4, xmm5, 136
shufps xmm3, xmm5, 221
movaps xmm5, xmm3
movups xmm6, xmmword ptr [r8+rdx-0x20]
movups xmm7, xmmword ptr [r8+rdx-0x10]
movaps xmm3, xmm6
shufps xmm6, xmm7, 136
pshufd xmm6, xmm6, 0x93
shufps xmm3, xmm7, 221
pshufd xmm7, xmm3, 0x93
movups xmm12, xmmword ptr [r9+rdx-0x40]
movups xmm13, xmmword ptr [r9+rdx-0x30]
movaps xmm11, xmm12
shufps xmm12, xmm13, 136
shufps xmm11, xmm13, 221
movaps xmm13, xmm11
movups xmm14, xmmword ptr [r9+rdx-0x20]
movups xmm15, xmmword ptr [r9+rdx-0x10]
movaps xmm11, xmm14
shufps xmm14, xmm15, 136
pshufd xmm14, xmm14, 0x93
shufps xmm11, xmm15, 221
pshufd xmm15, xmm11, 0x93
shl rax, 0x20
or rax, 0x40
movq xmm3, rax
movdqa xmmword ptr [rsp+0x20], xmm3
movaps xmm3, xmmword ptr [rsp]
movaps xmm11, xmmword ptr [rsp+0x10]
punpcklqdq xmm3, xmmword ptr [rsp+0x20]
punpcklqdq xmm11, xmmword ptr [rsp+0x20]
mov al, 7
9:
paddd xmm0, xmm4
paddd xmm8, xmm12
movaps xmmword ptr [rsp+0x20], xmm4
movaps xmmword ptr [rsp+0x30], xmm12
paddd xmm0, xmm1
paddd xmm8, xmm9
pxor xmm3, xmm0
pxor xmm11, xmm8
pshuflw xmm3, xmm3, 0xB1
pshufhw xmm3, xmm3, 0xB1
pshuflw xmm11, xmm11, 0xB1
pshufhw xmm11, xmm11, 0xB1
paddd xmm2, xmm3
paddd xmm10, xmm11
pxor xmm1, xmm2
pxor xmm9, xmm10
movdqa xmm4, xmm1
pslld xmm1, 20
psrld xmm4, 12
por xmm1, xmm4
movdqa xmm4, xmm9
pslld xmm9, 20
psrld xmm4, 12
por xmm9, xmm4
paddd xmm0, xmm5
paddd xmm8, xmm13
movaps xmmword ptr [rsp+0x40], xmm5
movaps xmmword ptr [rsp+0x50], xmm13
paddd xmm0, xmm1
paddd xmm8, xmm9
pxor xmm3, xmm0
pxor xmm11, xmm8
movdqa xmm13, xmm3
psrld xmm3, 8
pslld xmm13, 24
pxor xmm3, xmm13
movdqa xmm13, xmm11
psrld xmm11, 8
pslld xmm13, 24
pxor xmm11, xmm13
paddd xmm2, xmm3
paddd xmm10, xmm11
pxor xmm1, xmm2
pxor xmm9, xmm10
movdqa xmm4, xmm1
pslld xmm1, 25
psrld xmm4, 7
por xmm1, xmm4
movdqa xmm4, xmm9
pslld xmm9, 25
psrld xmm4, 7
por xmm9, xmm4
pshufd xmm0, xmm0, 0x93
pshufd xmm8, xmm8, 0x93
pshufd xmm3, xmm3, 0x4E
pshufd xmm11, xmm11, 0x4E
pshufd xmm2, xmm2, 0x39
pshufd xmm10, xmm10, 0x39
paddd xmm0, xmm6
paddd xmm8, xmm14
paddd xmm0, xmm1
paddd xmm8, xmm9
pxor xmm3, xmm0
pxor xmm11, xmm8
pshuflw xmm3, xmm3, 0xB1
pshufhw xmm3, xmm3, 0xB1
pshuflw xmm11, xmm11, 0xB1
pshufhw xmm11, xmm11, 0xB1
paddd xmm2, xmm3
paddd xmm10, xmm11
pxor xmm1, xmm2
pxor xmm9, xmm10
movdqa xmm4, xmm1
pslld xmm1, 20
psrld xmm4, 12
por xmm1, xmm4
movdqa xmm4, xmm9
pslld xmm9, 20
psrld xmm4, 12
por xmm9, xmm4
paddd xmm0, xmm7
paddd xmm8, xmm15
paddd xmm0, xmm1
paddd xmm8, xmm9
pxor xmm3, xmm0
pxor xmm11, xmm8
movdqa xmm13, xmm3
psrld xmm3, 8
pslld xmm13, 24
pxor xmm3, xmm13
movdqa xmm13, xmm11
psrld xmm11, 8
pslld xmm13, 24
pxor xmm11, xmm13
paddd xmm2, xmm3
paddd xmm10, xmm11
pxor xmm1, xmm2
pxor xmm9, xmm10
movdqa xmm4, xmm1
pslld xmm1, 25
psrld xmm4, 7
por xmm1, xmm4
movdqa xmm4, xmm9
pslld xmm9, 25
psrld xmm4, 7
por xmm9, xmm4
pshufd xmm0, xmm0, 0x39
pshufd xmm8, xmm8, 0x39
pshufd xmm3, xmm3, 0x4E
pshufd xmm11, xmm11, 0x4E
pshufd xmm2, xmm2, 0x93
pshufd xmm10, xmm10, 0x93
dec al
je 9f
movdqa xmm12, xmmword ptr [rsp+0x20]
movdqa xmm5, xmmword ptr [rsp+0x40]
pshufd xmm13, xmm12, 0x0F
shufps xmm12, xmm5, 214
pshufd xmm4, xmm12, 0x39
movdqa xmm12, xmm6
shufps xmm12, xmm7, 250
pand xmm13, xmmword ptr [PBLENDW_0x33_MASK+rip]
pand xmm12, xmmword ptr [PBLENDW_0xCC_MASK+rip]
por xmm13, xmm12
movdqa xmmword ptr [rsp+0x20], xmm13
movdqa xmm12, xmm7
punpcklqdq xmm12, xmm5
movdqa xmm13, xmm6
pand xmm12, xmmword ptr [PBLENDW_0x3F_MASK+rip]
pand xmm13, xmmword ptr [PBLENDW_0xC0_MASK+rip]
por xmm12, xmm13
pshufd xmm12, xmm12, 0x78
punpckhdq xmm5, xmm7
punpckldq xmm6, xmm5
pshufd xmm7, xmm6, 0x1E
movdqa xmmword ptr [rsp+0x40], xmm12
movdqa xmm5, xmmword ptr [rsp+0x30]
movdqa xmm13, xmmword ptr [rsp+0x50]
pshufd xmm6, xmm5, 0x0F
shufps xmm5, xmm13, 214
pshufd xmm12, xmm5, 0x39
movdqa xmm5, xmm14
shufps xmm5, xmm15, 250
pand xmm6, xmmword ptr [PBLENDW_0x33_MASK+rip]
pand xmm5, xmmword ptr [PBLENDW_0xCC_MASK+rip]
por xmm6, xmm5
movdqa xmm5, xmm15
punpcklqdq xmm5, xmm13
movdqa xmmword ptr [rsp+0x30], xmm2
movdqa xmm2, xmm14
pand xmm5, xmmword ptr [PBLENDW_0x3F_MASK+rip]
pand xmm2, xmmword ptr [PBLENDW_0xC0_MASK+rip]
por xmm5, xmm2
movdqa xmm2, xmmword ptr [rsp+0x30]
pshufd xmm5, xmm5, 0x78
punpckhdq xmm13, xmm15
punpckldq xmm14, xmm13
pshufd xmm15, xmm14, 0x1E
movdqa xmm13, xmm6
movdqa xmm14, xmm5
movdqa xmm5, xmmword ptr [rsp+0x20]
movdqa xmm6, xmmword ptr [rsp+0x40]
jmp 9b
9:
pxor xmm0, xmm2
pxor xmm1, xmm3
pxor xmm8, xmm10
pxor xmm9, xmm11
mov eax, r13d
cmp rdx, r15
jne 2b
movups xmmword ptr [rbx], xmm0
movups xmmword ptr [rbx+0x10], xmm1
movups xmmword ptr [rbx+0x20], xmm8
movups xmmword ptr [rbx+0x30], xmm9
mov eax, dword ptr [rsp+0x130]
neg eax
mov r10d, dword ptr [rsp+0x110+8*rax]
mov r11d, dword ptr [rsp+0x120+8*rax]
mov dword ptr [rsp+0x110], r10d
mov dword ptr [rsp+0x120], r11d
add rdi, 16
add rbx, 64
sub rsi, 2
3:
test esi, 0x1
je 4b
movups xmm0, xmmword ptr [rcx]
movups xmm1, xmmword ptr [rcx+0x10]
movd xmm13, dword ptr [rsp+0x110]
movd xmm14, dword ptr [rsp+0x120]
punpckldq xmm13, xmm14
mov r8, qword ptr [rdi]
movzx eax, byte ptr [rbp+0x40]
or eax, r13d
xor edx, edx
2:
mov r14d, eax
or eax, r12d
add rdx, 64
cmp rdx, r15
cmovne eax, r14d
movaps xmm2, xmmword ptr [BLAKE3_IV+rip]
shl rax, 32
or rax, 64
movq xmm12, rax
movdqa xmm3, xmm13
punpcklqdq xmm3, xmm12
movups xmm4, xmmword ptr [r8+rdx-0x40]
movups xmm5, xmmword ptr [r8+rdx-0x30]
movaps xmm8, xmm4
shufps xmm4, xmm5, 136
shufps xmm8, xmm5, 221
movaps xmm5, xmm8
movups xmm6, xmmword ptr [r8+rdx-0x20]
movups xmm7, xmmword ptr [r8+rdx-0x10]
movaps xmm8, xmm6
shufps xmm6, xmm7, 136
pshufd xmm6, xmm6, 0x93
shufps xmm8, xmm7, 221
pshufd xmm7, xmm8, 0x93
mov al, 7
9:
paddd xmm0, xmm4
paddd xmm0, xmm1
pxor xmm3, xmm0
pshuflw xmm3, xmm3, 0xB1
pshufhw xmm3, xmm3, 0xB1
paddd xmm2, xmm3
pxor xmm1, xmm2
movdqa xmm11, xmm1
pslld xmm1, 20
psrld xmm11, 12
por xmm1, xmm11
paddd xmm0, xmm5
paddd xmm0, xmm1
pxor xmm3, xmm0
movdqa xmm14, xmm3
psrld xmm3, 8
pslld xmm14, 24
pxor xmm3, xmm14
paddd xmm2, xmm3
pxor xmm1, xmm2
movdqa xmm11, xmm1
pslld xmm1, 25
psrld xmm11, 7
por xmm1, xmm11
pshufd xmm0, xmm0, 0x93
pshufd xmm3, xmm3, 0x4E
pshufd xmm2, xmm2, 0x39
paddd xmm0, xmm6
paddd xmm0, xmm1
pxor xmm3, xmm0
pshuflw xmm3, xmm3, 0xB1
pshufhw xmm3, xmm3, 0xB1
paddd xmm2, xmm3
pxor xmm1, xmm2
movdqa xmm11, xmm1
pslld xmm1, 20
psrld xmm11, 12
por xmm1, xmm11
paddd xmm0, xmm7
paddd xmm0, xmm1
pxor xmm3, xmm0
movdqa xmm14, xmm3
psrld xmm3, 8
pslld xmm14, 24
pxor xmm3, xmm14
paddd xmm2, xmm3
pxor xmm1, xmm2
movdqa xmm11, xmm1
pslld xmm1, 25
psrld xmm11, 7
por xmm1, xmm11
pshufd xmm0, xmm0, 0x39
pshufd xmm3, xmm3, 0x4E
pshufd xmm2, xmm2, 0x93
dec al
jz 9f
movdqa xmm8, xmm4
shufps xmm8, xmm5, 214
pshufd xmm9, xmm4, 0x0F
pshufd xmm4, xmm8, 0x39
movdqa xmm8, xmm6
shufps xmm8, xmm7, 250
pand xmm9, xmmword ptr [PBLENDW_0x33_MASK+rip]
pand xmm8, xmmword ptr [PBLENDW_0xCC_MASK+rip]
por xmm9, xmm8
movdqa xmm8, xmm7
punpcklqdq xmm8, xmm5
movdqa xmm10, xmm6
pand xmm8, xmmword ptr [PBLENDW_0x3F_MASK+rip]
pand xmm10, xmmword ptr [PBLENDW_0xC0_MASK+rip]
por xmm8, xmm10
pshufd xmm8, xmm8, 0x78
punpckhdq xmm5, xmm7
punpckldq xmm6, xmm5
pshufd xmm7, xmm6, 0x1E
movdqa xmm5, xmm9
movdqa xmm6, xmm8
jmp 9b
9:
pxor xmm0, xmm2
pxor xmm1, xmm3
mov eax, r13d
cmp rdx, r15
jne 2b
movups xmmword ptr [rbx], xmm0
movups xmmword ptr [rbx+0x10], xmm1
jmp 4b
.p2align 6
zfs_blake3_compress_in_place_sse2:
_CET_ENDBR
movups xmm0, xmmword ptr [rdi]
movups xmm1, xmmword ptr [rdi+0x10]
movaps xmm2, xmmword ptr [BLAKE3_IV+rip]
shl r8, 32
add rdx, r8
movq xmm3, rcx
movq xmm4, rdx
punpcklqdq xmm3, xmm4
movups xmm4, xmmword ptr [rsi]
movups xmm5, xmmword ptr [rsi+0x10]
movaps xmm8, xmm4
shufps xmm4, xmm5, 136
shufps xmm8, xmm5, 221
movaps xmm5, xmm8
movups xmm6, xmmword ptr [rsi+0x20]
movups xmm7, xmmword ptr [rsi+0x30]
movaps xmm8, xmm6
shufps xmm6, xmm7, 136
pshufd xmm6, xmm6, 0x93
shufps xmm8, xmm7, 221
pshufd xmm7, xmm8, 0x93
mov al, 7
9:
paddd xmm0, xmm4
paddd xmm0, xmm1
pxor xmm3, xmm0
pshuflw xmm3, xmm3, 0xB1
pshufhw xmm3, xmm3, 0xB1
paddd xmm2, xmm3
pxor xmm1, xmm2
movdqa xmm11, xmm1
pslld xmm1, 20
psrld xmm11, 12
por xmm1, xmm11
paddd xmm0, xmm5
paddd xmm0, xmm1
pxor xmm3, xmm0
movdqa xmm14, xmm3
psrld xmm3, 8
pslld xmm14, 24
pxor xmm3, xmm14
paddd xmm2, xmm3
pxor xmm1, xmm2
movdqa xmm11, xmm1
pslld xmm1, 25
psrld xmm11, 7
por xmm1, xmm11
pshufd xmm0, xmm0, 0x93
pshufd xmm3, xmm3, 0x4E
pshufd xmm2, xmm2, 0x39
paddd xmm0, xmm6
paddd xmm0, xmm1
pxor xmm3, xmm0
pshuflw xmm3, xmm3, 0xB1
pshufhw xmm3, xmm3, 0xB1
paddd xmm2, xmm3
pxor xmm1, xmm2
movdqa xmm11, xmm1
pslld xmm1, 20
psrld xmm11, 12
por xmm1, xmm11
paddd xmm0, xmm7
paddd xmm0, xmm1
pxor xmm3, xmm0
movdqa xmm14, xmm3
psrld xmm3, 8
pslld xmm14, 24
pxor xmm3, xmm14
paddd xmm2, xmm3
pxor xmm1, xmm2
movdqa xmm11, xmm1
pslld xmm1, 25
psrld xmm11, 7
por xmm1, xmm11
pshufd xmm0, xmm0, 0x39
pshufd xmm3, xmm3, 0x4E
pshufd xmm2, xmm2, 0x93
dec al
jz 9f
movdqa xmm8, xmm4
shufps xmm8, xmm5, 214
pshufd xmm9, xmm4, 0x0F
pshufd xmm4, xmm8, 0x39
movdqa xmm8, xmm6
shufps xmm8, xmm7, 250
pand xmm9, xmmword ptr [PBLENDW_0x33_MASK+rip]
pand xmm8, xmmword ptr [PBLENDW_0xCC_MASK+rip]
por xmm9, xmm8
movdqa xmm8, xmm7
punpcklqdq xmm8, xmm5
movdqa xmm10, xmm6
pand xmm8, xmmword ptr [PBLENDW_0x3F_MASK+rip]
pand xmm10, xmmword ptr [PBLENDW_0xC0_MASK+rip]
por xmm8, xmm10
pshufd xmm8, xmm8, 0x78
punpckhdq xmm5, xmm7
punpckldq xmm6, xmm5
pshufd xmm7, xmm6, 0x1E
movdqa xmm5, xmm9
movdqa xmm6, xmm8
jmp 9b
9:
pxor xmm0, xmm2
pxor xmm1, xmm3
movups xmmword ptr [rdi], xmm0
movups xmmword ptr [rdi+0x10], xmm1
Fix objtool: missing int3 after ret warning Resolve straight-line speculation warnings reported by objtool for x86_64 assembly on Linux when CONFIG_SLS is set. See the following LWN article for the complete details. https://lwn.net/Articles/877845/ Reviewed-by: Ryan Moeller <ryan@iXsystems.com> Reviewed-by: Alexander Motin <mav@FreeBSD.org> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #13528 Closes #13575
2022-06-21 02:36:21 +03:00
RET
Introduce BLAKE3 checksums as an OpenZFS feature This commit adds BLAKE3 checksums to OpenZFS, it has similar performance to Edon-R, but without the caveats around the latter. Homepage of BLAKE3: https://github.com/BLAKE3-team/BLAKE3 Wikipedia: https://en.wikipedia.org/wiki/BLAKE_(hash_function)#BLAKE3 Short description of Wikipedia: BLAKE3 is a cryptographic hash function based on Bao and BLAKE2, created by Jack O'Connor, Jean-Philippe Aumasson, Samuel Neves, and Zooko Wilcox-O'Hearn. It was announced on January 9, 2020, at Real World Crypto. BLAKE3 is a single algorithm with many desirable features (parallelism, XOF, KDF, PRF and MAC), in contrast to BLAKE and BLAKE2, which are algorithm families with multiple variants. BLAKE3 has a binary tree structure, so it supports a practically unlimited degree of parallelism (both SIMD and multithreading) given enough input. The official Rust and C implementations are dual-licensed as public domain (CC0) and the Apache License. Along with adding the BLAKE3 hash into the OpenZFS infrastructure a new benchmarking file called chksum_bench was introduced. When read it reports the speed of the available checksum functions. On Linux: cat /proc/spl/kstat/zfs/chksum_bench On FreeBSD: sysctl kstat.zfs.misc.chksum_bench This is an example output of an i3-1005G1 test system with Debian 11: implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1196 1602 1761 1749 1762 1759 1751 skein-generic 546 591 608 615 619 612 616 sha256-generic 240 300 316 314 304 285 276 sha512-generic 353 441 467 476 472 467 426 blake3-generic 308 313 313 313 312 313 312 blake3-sse2 402 1289 1423 1446 1432 1458 1413 blake3-sse41 427 1470 1625 1704 1679 1607 1629 blake3-avx2 428 1920 3095 3343 3356 3318 3204 blake3-avx512 473 2687 4905 5836 5844 5643 5374 Output on Debian 5.10.0-10-amd64 system: (Ryzen 7 5800X) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1840 2458 2665 2719 2711 2723 2693 skein-generic 870 966 996 992 1003 1005 1009 sha256-generic 415 442 453 455 457 457 457 sha512-generic 608 690 711 718 719 720 721 blake3-generic 301 313 311 309 309 310 310 blake3-sse2 343 1865 2124 2188 2180 2181 2186 blake3-sse41 364 2091 2396 2509 2463 2482 2488 blake3-avx2 365 2590 4399 4971 4915 4802 4764 Output on Debian 5.10.0-9-powerpc64le system: (POWER 9) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1213 1703 1889 1918 1957 1902 1907 skein-generic 434 492 520 522 511 525 525 sha256-generic 167 183 187 188 188 187 188 sha512-generic 186 216 222 221 225 224 224 blake3-generic 153 152 154 153 151 153 153 blake3-sse2 391 1170 1366 1406 1428 1426 1414 blake3-sse41 352 1049 1212 1174 1262 1258 1259 Output on Debian 5.10.0-11-arm64 system: (Pi400) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 487 603 629 639 643 641 641 skein-generic 271 299 303 308 309 309 307 sha256-generic 117 127 128 130 130 129 130 sha512-generic 145 165 170 172 173 174 175 blake3-generic 81 29 71 89 89 89 89 blake3-sse2 112 323 368 379 380 371 374 blake3-sse41 101 315 357 368 369 364 360 Structurally, the new code is mainly split into these parts: - 1x cross platform generic c variant: blake3_generic.c - 4x assembly for X86-64 (SSE2, SSE4.1, AVX2, AVX512) - 2x assembly for ARMv8 (NEON converted from SSE2) - 2x assembly for PPC64-LE (POWER8 converted from SSE2) - one file for switching between the implementations Note the PPC64 assembly requires the VSX instruction set and the kfpu_begin() / kfpu_end() calls on PowerPC were updated accordingly. Reviewed-by: Felix Dörre <felix@dogcraft.de> Reviewed-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz> Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Tino Reichardt <milky-zfs@mcmilk.de> Co-authored-by: Rich Ercolani <rincebrain@gmail.com> Closes #10058 Closes #12918
2022-06-09 01:55:57 +03:00
.p2align 6
zfs_blake3_compress_xof_sse2:
_CET_ENDBR
movups xmm0, xmmword ptr [rdi]
movups xmm1, xmmword ptr [rdi+0x10]
movaps xmm2, xmmword ptr [BLAKE3_IV+rip]
movzx eax, r8b
movzx edx, dl
shl rax, 32
add rdx, rax
movq xmm3, rcx
movq xmm4, rdx
punpcklqdq xmm3, xmm4
movups xmm4, xmmword ptr [rsi]
movups xmm5, xmmword ptr [rsi+0x10]
movaps xmm8, xmm4
shufps xmm4, xmm5, 136
shufps xmm8, xmm5, 221
movaps xmm5, xmm8
movups xmm6, xmmword ptr [rsi+0x20]
movups xmm7, xmmword ptr [rsi+0x30]
movaps xmm8, xmm6
shufps xmm6, xmm7, 136
pshufd xmm6, xmm6, 0x93
shufps xmm8, xmm7, 221
pshufd xmm7, xmm8, 0x93
mov al, 7
9:
paddd xmm0, xmm4
paddd xmm0, xmm1
pxor xmm3, xmm0
pshuflw xmm3, xmm3, 0xB1
pshufhw xmm3, xmm3, 0xB1
paddd xmm2, xmm3
pxor xmm1, xmm2
movdqa xmm11, xmm1
pslld xmm1, 20
psrld xmm11, 12
por xmm1, xmm11
paddd xmm0, xmm5
paddd xmm0, xmm1
pxor xmm3, xmm0
movdqa xmm14, xmm3
psrld xmm3, 8
pslld xmm14, 24
pxor xmm3, xmm14
paddd xmm2, xmm3
pxor xmm1, xmm2
movdqa xmm11, xmm1
pslld xmm1, 25
psrld xmm11, 7
por xmm1, xmm11
pshufd xmm0, xmm0, 0x93
pshufd xmm3, xmm3, 0x4E
pshufd xmm2, xmm2, 0x39
paddd xmm0, xmm6
paddd xmm0, xmm1
pxor xmm3, xmm0
pshuflw xmm3, xmm3, 0xB1
pshufhw xmm3, xmm3, 0xB1
paddd xmm2, xmm3
pxor xmm1, xmm2
movdqa xmm11, xmm1
pslld xmm1, 20
psrld xmm11, 12
por xmm1, xmm11
paddd xmm0, xmm7
paddd xmm0, xmm1
pxor xmm3, xmm0
movdqa xmm14, xmm3
psrld xmm3, 8
pslld xmm14, 24
pxor xmm3, xmm14
paddd xmm2, xmm3
pxor xmm1, xmm2
movdqa xmm11, xmm1
pslld xmm1, 25
psrld xmm11, 7
por xmm1, xmm11
pshufd xmm0, xmm0, 0x39
pshufd xmm3, xmm3, 0x4E
pshufd xmm2, xmm2, 0x93
dec al
jz 9f
movdqa xmm8, xmm4
shufps xmm8, xmm5, 214
pshufd xmm9, xmm4, 0x0F
pshufd xmm4, xmm8, 0x39
movdqa xmm8, xmm6
shufps xmm8, xmm7, 250
pand xmm9, xmmword ptr [PBLENDW_0x33_MASK+rip]
pand xmm8, xmmword ptr [PBLENDW_0xCC_MASK+rip]
por xmm9, xmm8
movdqa xmm8, xmm7
punpcklqdq xmm8, xmm5
movdqa xmm10, xmm6
pand xmm8, xmmword ptr [PBLENDW_0x3F_MASK+rip]
pand xmm10, xmmword ptr [PBLENDW_0xC0_MASK+rip]
por xmm8, xmm10
pshufd xmm8, xmm8, 0x78
punpckhdq xmm5, xmm7
punpckldq xmm6, xmm5
pshufd xmm7, xmm6, 0x1E
movdqa xmm5, xmm9
movdqa xmm6, xmm8
jmp 9b
9:
movdqu xmm4, xmmword ptr [rdi]
movdqu xmm5, xmmword ptr [rdi+0x10]
pxor xmm0, xmm2
pxor xmm1, xmm3
pxor xmm2, xmm4
pxor xmm3, xmm5
movups xmmword ptr [r9], xmm0
movups xmmword ptr [r9+0x10], xmm1
movups xmmword ptr [r9+0x20], xmm2
movups xmmword ptr [r9+0x30], xmm3
Fix objtool: missing int3 after ret warning Resolve straight-line speculation warnings reported by objtool for x86_64 assembly on Linux when CONFIG_SLS is set. See the following LWN article for the complete details. https://lwn.net/Articles/877845/ Reviewed-by: Ryan Moeller <ryan@iXsystems.com> Reviewed-by: Alexander Motin <mav@FreeBSD.org> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #13528 Closes #13575
2022-06-21 02:36:21 +03:00
RET
Introduce BLAKE3 checksums as an OpenZFS feature This commit adds BLAKE3 checksums to OpenZFS, it has similar performance to Edon-R, but without the caveats around the latter. Homepage of BLAKE3: https://github.com/BLAKE3-team/BLAKE3 Wikipedia: https://en.wikipedia.org/wiki/BLAKE_(hash_function)#BLAKE3 Short description of Wikipedia: BLAKE3 is a cryptographic hash function based on Bao and BLAKE2, created by Jack O'Connor, Jean-Philippe Aumasson, Samuel Neves, and Zooko Wilcox-O'Hearn. It was announced on January 9, 2020, at Real World Crypto. BLAKE3 is a single algorithm with many desirable features (parallelism, XOF, KDF, PRF and MAC), in contrast to BLAKE and BLAKE2, which are algorithm families with multiple variants. BLAKE3 has a binary tree structure, so it supports a practically unlimited degree of parallelism (both SIMD and multithreading) given enough input. The official Rust and C implementations are dual-licensed as public domain (CC0) and the Apache License. Along with adding the BLAKE3 hash into the OpenZFS infrastructure a new benchmarking file called chksum_bench was introduced. When read it reports the speed of the available checksum functions. On Linux: cat /proc/spl/kstat/zfs/chksum_bench On FreeBSD: sysctl kstat.zfs.misc.chksum_bench This is an example output of an i3-1005G1 test system with Debian 11: implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1196 1602 1761 1749 1762 1759 1751 skein-generic 546 591 608 615 619 612 616 sha256-generic 240 300 316 314 304 285 276 sha512-generic 353 441 467 476 472 467 426 blake3-generic 308 313 313 313 312 313 312 blake3-sse2 402 1289 1423 1446 1432 1458 1413 blake3-sse41 427 1470 1625 1704 1679 1607 1629 blake3-avx2 428 1920 3095 3343 3356 3318 3204 blake3-avx512 473 2687 4905 5836 5844 5643 5374 Output on Debian 5.10.0-10-amd64 system: (Ryzen 7 5800X) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1840 2458 2665 2719 2711 2723 2693 skein-generic 870 966 996 992 1003 1005 1009 sha256-generic 415 442 453 455 457 457 457 sha512-generic 608 690 711 718 719 720 721 blake3-generic 301 313 311 309 309 310 310 blake3-sse2 343 1865 2124 2188 2180 2181 2186 blake3-sse41 364 2091 2396 2509 2463 2482 2488 blake3-avx2 365 2590 4399 4971 4915 4802 4764 Output on Debian 5.10.0-9-powerpc64le system: (POWER 9) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 1213 1703 1889 1918 1957 1902 1907 skein-generic 434 492 520 522 511 525 525 sha256-generic 167 183 187 188 188 187 188 sha512-generic 186 216 222 221 225 224 224 blake3-generic 153 152 154 153 151 153 153 blake3-sse2 391 1170 1366 1406 1428 1426 1414 blake3-sse41 352 1049 1212 1174 1262 1258 1259 Output on Debian 5.10.0-11-arm64 system: (Pi400) implementation 1k 4k 16k 64k 256k 1m 4m edonr-generic 487 603 629 639 643 641 641 skein-generic 271 299 303 308 309 309 307 sha256-generic 117 127 128 130 130 129 130 sha512-generic 145 165 170 172 173 174 175 blake3-generic 81 29 71 89 89 89 89 blake3-sse2 112 323 368 379 380 371 374 blake3-sse41 101 315 357 368 369 364 360 Structurally, the new code is mainly split into these parts: - 1x cross platform generic c variant: blake3_generic.c - 4x assembly for X86-64 (SSE2, SSE4.1, AVX2, AVX512) - 2x assembly for ARMv8 (NEON converted from SSE2) - 2x assembly for PPC64-LE (POWER8 converted from SSE2) - one file for switching between the implementations Note the PPC64 assembly requires the VSX instruction set and the kfpu_begin() / kfpu_end() calls on PowerPC were updated accordingly. Reviewed-by: Felix Dörre <felix@dogcraft.de> Reviewed-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz> Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Tino Reichardt <milky-zfs@mcmilk.de> Co-authored-by: Rich Ercolani <rincebrain@gmail.com> Closes #10058 Closes #12918
2022-06-09 01:55:57 +03:00
.size zfs_blake3_hash_many_sse2, . - zfs_blake3_hash_many_sse2
.size zfs_blake3_compress_in_place_sse2, . - zfs_blake3_compress_in_place_sse2
.size zfs_blake3_compress_xof_sse2, . - zfs_blake3_compress_xof_sse2
#ifdef __APPLE__
.static_data
#else
.section .rodata
#endif
.p2align 6
BLAKE3_IV:
.long 0x6A09E667, 0xBB67AE85
.long 0x3C6EF372, 0xA54FF53A
ADD0:
.long 0, 1, 2, 3
ADD1:
.long 4, 4, 4, 4
BLAKE3_IV_0:
.long 0x6A09E667, 0x6A09E667, 0x6A09E667, 0x6A09E667
BLAKE3_IV_1:
.long 0xBB67AE85, 0xBB67AE85, 0xBB67AE85, 0xBB67AE85
BLAKE3_IV_2:
.long 0x3C6EF372, 0x3C6EF372, 0x3C6EF372, 0x3C6EF372
BLAKE3_IV_3:
.long 0xA54FF53A, 0xA54FF53A, 0xA54FF53A, 0xA54FF53A
BLAKE3_BLOCK_LEN:
.long 64, 64, 64, 64
CMP_MSB_MASK:
.long 0x80000000, 0x80000000, 0x80000000, 0x80000000
PBLENDW_0x33_MASK:
.long 0xFFFFFFFF, 0x00000000, 0xFFFFFFFF, 0x00000000
PBLENDW_0xCC_MASK:
.long 0x00000000, 0xFFFFFFFF, 0x00000000, 0xFFFFFFFF
PBLENDW_0x3F_MASK:
.long 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0x00000000
PBLENDW_0xC0_MASK:
.long 0x00000000, 0x00000000, 0x00000000, 0xFFFFFFFF
#endif /* HAVE_SSE2 */
#ifdef __ELF__
.section .note.GNU-stack,"",%progbits
#endif
Reference in New Issue Copy Permalink