diff options
Diffstat (limited to 'vendor/github.com/minio')
52 files changed, 1391 insertions, 3370 deletions
diff --git a/vendor/github.com/minio/md5-simd/LICENSE.Golang b/vendor/github.com/minio/md5-simd/LICENSE.Golang new file mode 100644 index 000000000..6a66aea5e --- /dev/null +++ b/vendor/github.com/minio/md5-simd/LICENSE.Golang @@ -0,0 +1,27 @@ +Copyright (c) 2009 The Go Authors. All rights reserved. + +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 Google Inc. 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. diff --git a/vendor/github.com/minio/md5-simd/README.md b/vendor/github.com/minio/md5-simd/README.md index 374214d1a..fa6fce1a4 100644 --- a/vendor/github.com/minio/md5-simd/README.md +++ b/vendor/github.com/minio/md5-simd/README.md @@ -116,6 +116,8 @@ BenchmarkParallel/8MB-4 2182.48 17252.88 7.91x These measurements were performed on AWS EC2 instance of type `c5.xlarge` equipped with a Xeon Platinum 8124M CPU at 3.0 GHz. +If only one or two inputs are available the scalar calculation method will be used for the +optimal speed in these cases. ## Operation diff --git a/vendor/github.com/minio/md5-simd/block-generic.go b/vendor/github.com/minio/md5-simd/block-generic.go deleted file mode 100644 index eb333b93f..000000000 --- a/vendor/github.com/minio/md5-simd/block-generic.go +++ /dev/null @@ -1,132 +0,0 @@ -// Copyright 2013 The Go Authors. All rights reserved. -// Use of this source code is governed by a BSD-style -// license that can be found in the LICENSE file. - -// Code generated by go run gen.go -output md5block.go; DO NOT EDIT. - -package md5simd - -import ( - "encoding/binary" - "math/bits" -) - -type digest struct { - s [4]uint32 - x [BlockSize]byte - nx int - len uint64 -} - -func blockGeneric(dig *digest, p []byte) { - // load state - a, b, c, d := dig.s[0], dig.s[1], dig.s[2], dig.s[3] - - for i := 0; i <= len(p)-BlockSize; i += BlockSize { - // eliminate bounds checks on p - q := p[i:] - q = q[:BlockSize:BlockSize] - - // save current state - aa, bb, cc, dd := a, b, c, d - - // load input block - x0 := binary.LittleEndian.Uint32(q[4*0x0:]) - x1 := binary.LittleEndian.Uint32(q[4*0x1:]) - x2 := binary.LittleEndian.Uint32(q[4*0x2:]) - x3 := binary.LittleEndian.Uint32(q[4*0x3:]) - x4 := binary.LittleEndian.Uint32(q[4*0x4:]) - x5 := binary.LittleEndian.Uint32(q[4*0x5:]) - x6 := binary.LittleEndian.Uint32(q[4*0x6:]) - x7 := binary.LittleEndian.Uint32(q[4*0x7:]) - x8 := binary.LittleEndian.Uint32(q[4*0x8:]) - x9 := binary.LittleEndian.Uint32(q[4*0x9:]) - xa := binary.LittleEndian.Uint32(q[4*0xa:]) - xb := binary.LittleEndian.Uint32(q[4*0xb:]) - xc := binary.LittleEndian.Uint32(q[4*0xc:]) - xd := binary.LittleEndian.Uint32(q[4*0xd:]) - xe := binary.LittleEndian.Uint32(q[4*0xe:]) - xf := binary.LittleEndian.Uint32(q[4*0xf:]) - - // round 1 - a = b + bits.RotateLeft32((((c^d)&b)^d)+a+x0+0xd76aa478, 7) - d = a + bits.RotateLeft32((((b^c)&a)^c)+d+x1+0xe8c7b756, 12) - c = d + bits.RotateLeft32((((a^b)&d)^b)+c+x2+0x242070db, 17) - b = c + bits.RotateLeft32((((d^a)&c)^a)+b+x3+0xc1bdceee, 22) - a = b + bits.RotateLeft32((((c^d)&b)^d)+a+x4+0xf57c0faf, 7) - d = a + bits.RotateLeft32((((b^c)&a)^c)+d+x5+0x4787c62a, 12) - c = d + bits.RotateLeft32((((a^b)&d)^b)+c+x6+0xa8304613, 17) - b = c + bits.RotateLeft32((((d^a)&c)^a)+b+x7+0xfd469501, 22) - a = b + bits.RotateLeft32((((c^d)&b)^d)+a+x8+0x698098d8, 7) - d = a + bits.RotateLeft32((((b^c)&a)^c)+d+x9+0x8b44f7af, 12) - c = d + bits.RotateLeft32((((a^b)&d)^b)+c+xa+0xffff5bb1, 17) - b = c + bits.RotateLeft32((((d^a)&c)^a)+b+xb+0x895cd7be, 22) - a = b + bits.RotateLeft32((((c^d)&b)^d)+a+xc+0x6b901122, 7) - d = a + bits.RotateLeft32((((b^c)&a)^c)+d+xd+0xfd987193, 12) - c = d + bits.RotateLeft32((((a^b)&d)^b)+c+xe+0xa679438e, 17) - b = c + bits.RotateLeft32((((d^a)&c)^a)+b+xf+0x49b40821, 22) - - // round 2 - a = b + bits.RotateLeft32((((b^c)&d)^c)+a+x1+0xf61e2562, 5) - d = a + bits.RotateLeft32((((a^b)&c)^b)+d+x6+0xc040b340, 9) - c = d + bits.RotateLeft32((((d^a)&b)^a)+c+xb+0x265e5a51, 14) - b = c + bits.RotateLeft32((((c^d)&a)^d)+b+x0+0xe9b6c7aa, 20) - a = b + bits.RotateLeft32((((b^c)&d)^c)+a+x5+0xd62f105d, 5) - d = a + bits.RotateLeft32((((a^b)&c)^b)+d+xa+0x02441453, 9) - c = d + bits.RotateLeft32((((d^a)&b)^a)+c+xf+0xd8a1e681, 14) - b = c + bits.RotateLeft32((((c^d)&a)^d)+b+x4+0xe7d3fbc8, 20) - a = b + bits.RotateLeft32((((b^c)&d)^c)+a+x9+0x21e1cde6, 5) - d = a + bits.RotateLeft32((((a^b)&c)^b)+d+xe+0xc33707d6, 9) - c = d + bits.RotateLeft32((((d^a)&b)^a)+c+x3+0xf4d50d87, 14) - b = c + bits.RotateLeft32((((c^d)&a)^d)+b+x8+0x455a14ed, 20) - a = b + bits.RotateLeft32((((b^c)&d)^c)+a+xd+0xa9e3e905, 5) - d = a + bits.RotateLeft32((((a^b)&c)^b)+d+x2+0xfcefa3f8, 9) - c = d + bits.RotateLeft32((((d^a)&b)^a)+c+x7+0x676f02d9, 14) - b = c + bits.RotateLeft32((((c^d)&a)^d)+b+xc+0x8d2a4c8a, 20) - - // round 3 - a = b + bits.RotateLeft32((b^c^d)+a+x5+0xfffa3942, 4) - d = a + bits.RotateLeft32((a^b^c)+d+x8+0x8771f681, 11) - c = d + bits.RotateLeft32((d^a^b)+c+xb+0x6d9d6122, 16) - b = c + bits.RotateLeft32((c^d^a)+b+xe+0xfde5380c, 23) - a = b + bits.RotateLeft32((b^c^d)+a+x1+0xa4beea44, 4) - d = a + bits.RotateLeft32((a^b^c)+d+x4+0x4bdecfa9, 11) - c = d + bits.RotateLeft32((d^a^b)+c+x7+0xf6bb4b60, 16) - b = c + bits.RotateLeft32((c^d^a)+b+xa+0xbebfbc70, 23) - a = b + bits.RotateLeft32((b^c^d)+a+xd+0x289b7ec6, 4) - d = a + bits.RotateLeft32((a^b^c)+d+x0+0xeaa127fa, 11) - c = d + bits.RotateLeft32((d^a^b)+c+x3+0xd4ef3085, 16) - b = c + bits.RotateLeft32((c^d^a)+b+x6+0x04881d05, 23) - a = b + bits.RotateLeft32((b^c^d)+a+x9+0xd9d4d039, 4) - d = a + bits.RotateLeft32((a^b^c)+d+xc+0xe6db99e5, 11) - c = d + bits.RotateLeft32((d^a^b)+c+xf+0x1fa27cf8, 16) - b = c + bits.RotateLeft32((c^d^a)+b+x2+0xc4ac5665, 23) - - // round 4 - a = b + bits.RotateLeft32((c^(b|^d))+a+x0+0xf4292244, 6) - d = a + bits.RotateLeft32((b^(a|^c))+d+x7+0x432aff97, 10) - c = d + bits.RotateLeft32((a^(d|^b))+c+xe+0xab9423a7, 15) - b = c + bits.RotateLeft32((d^(c|^a))+b+x5+0xfc93a039, 21) - a = b + bits.RotateLeft32((c^(b|^d))+a+xc+0x655b59c3, 6) - d = a + bits.RotateLeft32((b^(a|^c))+d+x3+0x8f0ccc92, 10) - c = d + bits.RotateLeft32((a^(d|^b))+c+xa+0xffeff47d, 15) - b = c + bits.RotateLeft32((d^(c|^a))+b+x1+0x85845dd1, 21) - a = b + bits.RotateLeft32((c^(b|^d))+a+x8+0x6fa87e4f, 6) - d = a + bits.RotateLeft32((b^(a|^c))+d+xf+0xfe2ce6e0, 10) - c = d + bits.RotateLeft32((a^(d|^b))+c+x6+0xa3014314, 15) - b = c + bits.RotateLeft32((d^(c|^a))+b+xd+0x4e0811a1, 21) - a = b + bits.RotateLeft32((c^(b|^d))+a+x4+0xf7537e82, 6) - d = a + bits.RotateLeft32((b^(a|^c))+d+xb+0xbd3af235, 10) - c = d + bits.RotateLeft32((a^(d|^b))+c+x2+0x2ad7d2bb, 15) - b = c + bits.RotateLeft32((d^(c|^a))+b+x9+0xeb86d391, 21) - - // add saved state - a += aa - b += bb - c += cc - d += dd - } - - // save state - dig.s[0], dig.s[1], dig.s[2], dig.s[3] = a, b, c, d -} diff --git a/vendor/github.com/minio/md5-simd/block16_amd64.s b/vendor/github.com/minio/md5-simd/block16_amd64.s index d32c12200..be0a43a3b 100644 --- a/vendor/github.com/minio/md5-simd/block16_amd64.s +++ b/vendor/github.com/minio/md5-simd/block16_amd64.s @@ -2,70 +2,72 @@ // Use of this source code is governed by a license that can be // found in the LICENSE file. +//+build !noasm,!appengine,gc + // This is the AVX512 implementation of the MD5 block function (16-way parallel) #define prep(index) \ - KMOVQ kmask, ktmp \ + KMOVQ kmask, ktmp \ VPGATHERDD index*4(base)(ptrs*1), ktmp, mem #define ROUND1(a, b, c, d, index, const, shift) \ - VXORPS c, tmp, tmp \ - VPADDD 64*const(consts), a, a \ - VPADDD mem, a, a \ - VPTERNLOGD $0x6C, b, d, tmp \ - prep(index) \ - VPADDD tmp, a, a \ - VPROLD $shift, a, a \ - VMOVAPD c, tmp \ - VPADDD b, a, a + VPXORQ c, tmp, tmp \ + VPADDD 64*const(consts), a, a \ + VPADDD mem, a, a \ + VPTERNLOGD $0x6C, b, d, tmp \ + prep(index) \ + VPADDD tmp, a, a \ + VPROLD $shift, a, a \ + VMOVAPD c, tmp \ + VPADDD b, a, a #define ROUND1noload(a, b, c, d, const, shift) \ - VXORPS c, tmp, tmp \ - VPADDD 64*const(consts), a, a \ - VPADDD mem, a, a \ - VPTERNLOGD $0x6C, b, d, tmp \ - VPADDD tmp, a, a \ - VPROLD $shift, a, a \ - VMOVAPD c, tmp \ - VPADDD b, a, a + VPXORQ c, tmp, tmp \ + VPADDD 64*const(consts), a, a \ + VPADDD mem, a, a \ + VPTERNLOGD $0x6C, b, d, tmp \ + VPADDD tmp, a, a \ + VPROLD $shift, a, a \ + VMOVAPD c, tmp \ + VPADDD b, a, a #define ROUND2(a, b, c, d, zreg, const, shift) \ - VPADDD 64*const(consts), a, a \ - VPADDD zreg, a, a \ - VANDNPS c, tmp, tmp \ - VPTERNLOGD $0xEC, b, tmp, tmp2 \ - VMOVAPD c, tmp \ - VPADDD tmp2, a, a \ - VMOVAPD c, tmp2 \ - VPROLD $shift, a, a \ - VPADDD b, a, a + VPADDD 64*const(consts), a, a \ + VPADDD zreg, a, a \ + VANDNPD c, tmp, tmp \ + VPTERNLOGD $0xEC, b, tmp, tmp2 \ + VMOVAPD c, tmp \ + VPADDD tmp2, a, a \ + VMOVAPD c, tmp2 \ + VPROLD $shift, a, a \ + VPADDD b, a, a #define ROUND3(a, b, c, d, zreg, const, shift) \ - VPADDD 64*const(consts), a, a \ - VPADDD zreg, a, a \ - VPTERNLOGD $0x96, b, d, tmp \ - VPADDD tmp, a, a \ - VPROLD $shift, a, a \ - VMOVAPD b, tmp \ - VPADDD b, a, a + VPADDD 64*const(consts), a, a \ + VPADDD zreg, a, a \ + VPTERNLOGD $0x96, b, d, tmp \ + VPADDD tmp, a, a \ + VPROLD $shift, a, a \ + VMOVAPD b, tmp \ + VPADDD b, a, a #define ROUND4(a, b, c, d, zreg, const, shift) \ - VPADDD 64*const(consts), a, a \ - VPADDD zreg, a, a \ - VPTERNLOGD $0x36, b, c, tmp \ - VPADDD tmp, a, a \ - VPROLD $shift, a, a \ - VXORPS c, ones, tmp \ - VPADDD b, a, a - -TEXT ·block16(SB),4,$0-40 - - MOVQ state+0(FP), BX - MOVQ base+8(FP), SI - MOVQ ptrs+16(FP), AX - KMOVQ mask+24(FP), K1 - MOVQ n+32(FP), DX - MOVQ ·avx512md5consts+0(SB), DI + VPADDD 64*const(consts), a, a \ + VPADDD zreg, a, a \ + VPTERNLOGD $0x36, b, c, tmp \ + VPADDD tmp, a, a \ + VPROLD $shift, a, a \ + VPXORQ c, ones, tmp \ + VPADDD b, a, a + +TEXT ·block16(SB), 4, $0-40 + + MOVQ state+0(FP), BX + MOVQ base+8(FP), SI + MOVQ ptrs+16(FP), AX + KMOVQ mask+24(FP), K1 + MOVQ n+32(FP), DX + MOVQ ·avx512md5consts+0(SB), DI #define a Z0 #define b Z1 @@ -90,7 +92,6 @@ TEXT ·block16(SB),4,$0-40 // Registers Z16 through to Z31 are used for caching purposes // ---------------------------------------------------------- - #define dig BX #define count DX #define base SI @@ -105,7 +106,7 @@ TEXT ·block16(SB),4,$0-40 // load source pointers VMOVUPD 0x00(AX), ptrs - MOVQ $-1, AX + MOVQ $-1, AX VPBROADCASTQ AX, ones loop: @@ -190,7 +191,7 @@ loop: ROUND3(c,d,a,b, Z31,0x2e,16) ROUND3(b,c,d,a, Z18,0x2f,23) - VXORPS d, ones, tmp + VPXORQ d, ones, tmp ROUND4(a,b,c,d, Z16,0x30, 6) ROUND4(d,a,b,c, Z23,0x31,10) diff --git a/vendor/github.com/minio/md5-simd/block8_amd64.s b/vendor/github.com/minio/md5-simd/block8_amd64.s index f5f1d9cab..f57db17aa 100644 --- a/vendor/github.com/minio/md5-simd/block8_amd64.s +++ b/vendor/github.com/minio/md5-simd/block8_amd64.s @@ -1,3 +1,5 @@ +//+build !noasm,!appengine,gc + // Copyright (c) 2018 Igneous Systems // MIT License // @@ -70,7 +72,7 @@ TEXT ·block8(SB), 4, $0-40 #define consts DI #define prepmask \ - VXORPS mask, mask, mask \ + VPXOR mask, mask, mask \ VPCMPGTD mask, off, mask #define prep(index) \ @@ -86,14 +88,14 @@ TEXT ·block8(SB), 4, $0-40 #define roll(shift, a) \ VPSLLD $shift, a, rtmp1 \ VPSRLD $32-shift, a, a \ - VORPS rtmp1, a, a + VPOR rtmp1, a, a #define ROUND1(a, b, c, d, index, const, shift) \ - VXORPS c, tmp, tmp \ + VPXOR c, tmp, tmp \ VPADDD 32*const(consts), a, a \ VPADDD mem, a, a \ - VANDPS b, tmp, tmp \ - VXORPS d, tmp, tmp \ + VPAND b, tmp, tmp \ + VPXOR d, tmp, tmp \ prep(index) \ VPADDD tmp, a, a \ roll(shift,a) \ @@ -101,11 +103,11 @@ TEXT ·block8(SB), 4, $0-40 VPADDD b, a, a #define ROUND1load(a, b, c, d, index, const, shift) \ - VXORPS c, tmp, tmp \ + VXORPD c, tmp, tmp \ VPADDD 32*const(consts), a, a \ VPADDD mem, a, a \ - VANDPS b, tmp, tmp \ - VXORPS d, tmp, tmp \ + VPAND b, tmp, tmp \ + VPXOR d, tmp, tmp \ load(index) \ VPADDD tmp, a, a \ roll(shift,a) \ @@ -115,10 +117,10 @@ TEXT ·block8(SB), 4, $0-40 #define ROUND2(a, b, c, d, index, const, shift) \ VPADDD 32*const(consts), a, a \ VPADDD mem, a, a \ - VANDPS b, tmp2, tmp2 \ - VANDNPS c, tmp, tmp \ + VPAND b, tmp2, tmp2 \ + VANDNPD c, tmp, tmp \ load(index) \ - VORPS tmp, tmp2, tmp2 \ + VPOR tmp, tmp2, tmp2 \ VMOVAPD c, tmp \ VPADDD tmp2, a, a \ VMOVAPD c, tmp2 \ @@ -129,8 +131,8 @@ TEXT ·block8(SB), 4, $0-40 VPADDD 32*const(consts), a, a \ VPADDD mem, a, a \ load(index) \ - VXORPS d, tmp, tmp \ - VXORPS b, tmp, tmp \ + VPXOR d, tmp, tmp \ + VPXOR b, tmp, tmp \ VPADDD tmp, a, a \ roll(shift,a) \ VMOVAPD b, tmp \ @@ -139,12 +141,12 @@ TEXT ·block8(SB), 4, $0-40 #define ROUND4(a, b, c, d, index, const, shift) \ VPADDD 32*const(consts), a, a \ VPADDD mem, a, a \ - VORPS b, tmp, tmp \ - VXORPS c, tmp, tmp \ + VPOR b, tmp, tmp \ + VPXOR c, tmp, tmp \ VPADDD tmp, a, a \ load(index) \ roll(shift,a) \ - VXORPS c, ones, tmp \ + VPXOR c, ones, tmp \ VPADDD b, a, a // load digest into state registers @@ -242,7 +244,7 @@ loop: ROUND3(b,c,d,a, 0,0x2f,23) load(0) - VXORPS d, ones, tmp + VPXOR d, ones, tmp ROUND4(a,b,c,d, 7,0x30, 6) ROUND4(d,a,b,c,14,0x31,10) diff --git a/vendor/github.com/minio/md5-simd/block_amd64.go b/vendor/github.com/minio/md5-simd/block_amd64.go index 27d6ce00e..16edda268 100644 --- a/vendor/github.com/minio/md5-simd/block_amd64.go +++ b/vendor/github.com/minio/md5-simd/block_amd64.go @@ -9,14 +9,18 @@ package md5simd import ( "fmt" "math" - "sync" "unsafe" - "github.com/klauspost/cpuid" + "github.com/klauspost/cpuid/v2" ) var hasAVX512 bool +func init() { + // VANDNPD requires AVX512DQ. Technically it could be VPTERNLOGQ which is AVX512F. + hasAVX512 = cpuid.CPU.Supports(cpuid.AVX512F, cpuid.AVX512DQ) +} + //go:noescape func block8(state *uint32, base uintptr, bufs *int32, cache *byte, n int) @@ -82,45 +86,52 @@ var avx512md5consts = func(c []uint32) []uint32 { return inf }(md5consts[:]) -func init() { - hasAVX512 = cpuid.CPU.AVX512F() -} - // Interface function to assembly code func (s *md5Server) blockMd5_x16(d *digest16, input [16][]byte, half bool) { if hasAVX512 { blockMd5_avx512(d, input, s.allBufs, &s.maskRounds16) - } else { - d8a, d8b := digest8{}, digest8{} - for i := range d8a.v0 { - j := i + 8 - d8a.v0[i], d8a.v1[i], d8a.v2[i], d8a.v3[i] = d.v0[i], d.v1[i], d.v2[i], d.v3[i] - if !half { - d8b.v0[i], d8b.v1[i], d8b.v2[i], d8b.v3[i] = d.v0[j], d.v1[j], d.v2[j], d.v3[j] - } - } + return + } + + // Preparing data using copy is slower since copies aren't inlined. - i8 := [2][8][]byte{} - for i := range i8[0] { - i8[0][i], i8[1][i] = input[i], input[8+i] + // Calculate on this goroutine + if half { + for i := range s.i8[0][:] { + s.i8[0][i] = input[i] } - if half { - blockMd5_avx2(&d8a, i8[0], s.allBufs, &s.maskRounds8a) - } else { - wg := sync.WaitGroup{} - wg.Add(2) - go func() { blockMd5_avx2(&d8a, i8[0], s.allBufs, &s.maskRounds8a); wg.Done() }() - go func() { blockMd5_avx2(&d8b, i8[1], s.allBufs, &s.maskRounds8b); wg.Done() }() - wg.Wait() + for i := range s.d8a.v0[:] { + s.d8a.v0[i], s.d8a.v1[i], s.d8a.v2[i], s.d8a.v3[i] = d.v0[i], d.v1[i], d.v2[i], d.v3[i] } - - for i := range d8a.v0 { - j := i + 8 - d.v0[i], d.v1[i], d.v2[i], d.v3[i] = d8a.v0[i], d8a.v1[i], d8a.v2[i], d8a.v3[i] - if !half { - d.v0[j], d.v1[j], d.v2[j], d.v3[j] = d8b.v0[i], d8b.v1[i], d8b.v2[i], d8b.v3[i] - } + blockMd5_avx2(&s.d8a, s.i8[0], s.allBufs, &s.maskRounds8a) + for i := range s.d8a.v0[:] { + d.v0[i], d.v1[i], d.v2[i], d.v3[i] = s.d8a.v0[i], s.d8a.v1[i], s.d8a.v2[i], s.d8a.v3[i] } + return + } + + for i := range s.i8[0][:] { + s.i8[0][i], s.i8[1][i] = input[i], input[8+i] + } + + for i := range s.d8a.v0[:] { + j := (i + 8) & 15 + s.d8a.v0[i], s.d8a.v1[i], s.d8a.v2[i], s.d8a.v3[i] = d.v0[i], d.v1[i], d.v2[i], d.v3[i] + s.d8b.v0[i], s.d8b.v1[i], s.d8b.v2[i], s.d8b.v3[i] = d.v0[j], d.v1[j], d.v2[j], d.v3[j] + } + + // Benchmarks appears to be slightly faster when spinning up 2 goroutines instead + // of using the current for one of the blocks. + s.wg.Add(2) + go func() { blockMd5_avx2(&s.d8a, s.i8[0], s.allBufs, &s.maskRounds8a); s.wg.Done() }() + go func() { blockMd5_avx2(&s.d8b, s.i8[1], s.allBufs, &s.maskRounds8b); s.wg.Done() }() + s.wg.Wait() + for i := range s.d8a.v0[:] { + d.v0[i], d.v1[i], d.v2[i], d.v3[i] = s.d8a.v0[i], s.d8a.v1[i], s.d8a.v2[i], s.d8a.v3[i] + } + for i := range s.d8b.v0[:] { + j := (i + 8) & 15 + d.v0[j], d.v1[j], d.v2[j], d.v3[j] = s.d8b.v0[i], s.d8b.v1[i], s.d8b.v2[i], s.d8b.v3[i] } } diff --git a/vendor/github.com/minio/md5-simd/md5-digest_amd64.go b/vendor/github.com/minio/md5-simd/md5-digest_amd64.go index fe10c7531..5ea23a499 100644 --- a/vendor/github.com/minio/md5-simd/md5-digest_amd64.go +++ b/vendor/github.com/minio/md5-simd/md5-digest_amd64.go @@ -10,6 +10,7 @@ import ( "encoding/binary" "errors" "fmt" + "sync" "sync/atomic" ) @@ -121,6 +122,14 @@ func (d *md5Digest) Close() { } } +var sumChPool sync.Pool + +func init() { + sumChPool.New = func() interface{} { + return make(chan sumResult, 1) + } +} + // Sum - Return MD5 sum in bytes func (d *md5Digest) Sum(in []byte) (result []byte) { if d.blocksCh == nil { @@ -148,10 +157,11 @@ func (d *md5Digest) Sum(in []byte) (result []byte) { if len(trail)%BlockSize != 0 { panic(fmt.Errorf("internal error: sum block was not aligned. len=%d, nx=%d", len(trail), d.nx)) } - sumCh := make(chan sumResult, 1) + sumCh := sumChPool.Get().(chan sumResult) d.sendBlock(blockInput{uid: d.uid, msg: trail, sumCh: sumCh}, true) sum := <-sumCh + sumChPool.Put(sumCh) return append(in, sum.digest[:]...) } diff --git a/vendor/github.com/minio/md5-simd/md5-server_amd64.go b/vendor/github.com/minio/md5-simd/md5-server_amd64.go index 461059537..94f741c54 100644 --- a/vendor/github.com/minio/md5-simd/md5-server_amd64.go +++ b/vendor/github.com/minio/md5-simd/md5-server_amd64.go @@ -10,8 +10,9 @@ import ( "encoding/binary" "fmt" "runtime" + "sync" - "github.com/klauspost/cpuid" + "github.com/klauspost/cpuid/v2" ) // MD5 initialization constants @@ -23,6 +24,9 @@ const ( init1 = 0xefcdab89 init2 = 0x98badcfe init3 = 0x10325476 + + // Use scalar routine when below this many lanes + useScalarBelow = 3 ) // md5ServerUID - Does not start at 0 but next multiple of 16 so as to be able to @@ -56,11 +60,15 @@ type md5Server struct { maskRounds8b [8]maskRounds // Pre-allocated static array for max 8 rounds (2nd AVX2 core) allBufs []byte // Preallocated buffer. buffers chan []byte // Preallocated buffers, sliced from allBufs. + + i8 [2][8][]byte // avx2 temporary vars + d8a, d8b digest8 + wg sync.WaitGroup } // NewServer - Create new object for parallel processing handling func NewServer() Server { - if !cpuid.CPU.AVX2() { + if !cpuid.CPU.Supports(cpuid.AVX2) { return &fallbackServer{} } md5srv := &md5Server{} @@ -152,7 +160,7 @@ func (s *md5Server) process(newClients chan newClient) { sum := sumResult{} // Add end block to current digest. - blockGeneric(&dig, block.msg) + blockScalar(&dig.s, block.msg) binary.LittleEndian.PutUint32(sum.digest[0:], dig.s[0]) binary.LittleEndian.PutUint32(sum.digest[4:], dig.s[1]) @@ -262,6 +270,88 @@ func (s *md5Server) Close() { // Invoke assembly and send results back func (s *md5Server) blocks(lanes []blockInput) { + if len(lanes) < useScalarBelow { + // Use scalar routine when below this many lanes + switch len(lanes) { + case 0: + case 1: + lane := lanes[0] + var d digest + a, ok := s.digests[lane.uid] + if ok { + d.s[0] = binary.LittleEndian.Uint32(a[0:4]) + d.s[1] = binary.LittleEndian.Uint32(a[4:8]) + d.s[2] = binary.LittleEndian.Uint32(a[8:12]) + d.s[3] = binary.LittleEndian.Uint32(a[12:16]) + } else { + d.s[0] = init0 + d.s[1] = init1 + d.s[2] = init2 + d.s[3] = init3 + } + if len(lane.msg) > 0 { + // Update... + blockScalar(&d.s, lane.msg) + } + dig := [Size]byte{} + binary.LittleEndian.PutUint32(dig[0:], d.s[0]) + binary.LittleEndian.PutUint32(dig[4:], d.s[1]) + binary.LittleEndian.PutUint32(dig[8:], d.s[2]) + binary.LittleEndian.PutUint32(dig[12:], d.s[3]) + s.digests[lane.uid] = dig + + if lane.msg != nil { + s.buffers <- lane.msg + } + lanes[0] = blockInput{} + + default: + s.wg.Add(len(lanes)) + var results [useScalarBelow]digest + for i := range lanes { + lane := lanes[i] + go func(i int) { + var d digest + defer s.wg.Done() + a, ok := s.digests[lane.uid] + if ok { + d.s[0] = binary.LittleEndian.Uint32(a[0:4]) + d.s[1] = binary.LittleEndian.Uint32(a[4:8]) + d.s[2] = binary.LittleEndian.Uint32(a[8:12]) + d.s[3] = binary.LittleEndian.Uint32(a[12:16]) + } else { + d.s[0] = init0 + d.s[1] = init1 + d.s[2] = init2 + d.s[3] = init3 + } + if len(lane.msg) == 0 { + results[i] = d + return + } + // Update... + blockScalar(&d.s, lane.msg) + results[i] = d + }(i) + } + s.wg.Wait() + for i, lane := range lanes { + dig := [Size]byte{} + binary.LittleEndian.PutUint32(dig[0:], results[i].s[0]) + binary.LittleEndian.PutUint32(dig[4:], results[i].s[1]) + binary.LittleEndian.PutUint32(dig[8:], results[i].s[2]) + binary.LittleEndian.PutUint32(dig[12:], results[i].s[3]) + s.digests[lane.uid] = dig + + if lane.msg != nil { + s.buffers <- lane.msg + } + lanes[i] = blockInput{} + } + } + return + } + inputs := [16][]byte{} for i := range lanes { inputs[i] = lanes[i].msg diff --git a/vendor/github.com/minio/md5-simd/md5-util_amd64.go b/vendor/github.com/minio/md5-simd/md5-util_amd64.go index 32bbae4a0..73981b0eb 100644 --- a/vendor/github.com/minio/md5-simd/md5-util_amd64.go +++ b/vendor/github.com/minio/md5-simd/md5-util_amd64.go @@ -1,19 +1,21 @@ +//+build !noasm,!appengine,gc + // Copyright (c) 2020 MinIO Inc. All rights reserved. // Use of this source code is governed by a license that can be // found in the LICENSE file. package md5simd -import ( - "sort" -) - // Helper struct for sorting blocks based on length type lane struct { len uint pos uint } +type digest struct { + s [4]uint32 +} + // Helper struct for generating number of rounds in combination with mask for valid lanes type maskRounds struct { mask uint64 @@ -23,15 +25,22 @@ type maskRounds struct { func generateMaskAndRounds8(input [8][]byte, mr *[8]maskRounds) (rounds int) { // Sort on blocks length small to large var sorted [8]lane - for c, inpt := range input { + for c, inpt := range input[:] { sorted[c] = lane{uint(len(inpt)), uint(c)} + for i := c - 1; i >= 0; i-- { + // swap so largest is at the end... + if sorted[i].len > sorted[i+1].len { + sorted[i], sorted[i+1] = sorted[i+1], sorted[i] + continue + } + break + } } - sort.Slice(sorted[:], func(i, j int) bool { return sorted[i].len < sorted[j].len }) // Create mask array including 'rounds' (of processing blocks of 64 bytes) between masks m, round := uint64(0xff), uint64(0) - for _, s := range sorted { + for _, s := range sorted[:] { if s.len > 0 { if uint64(s.len)>>6 > round { mr[rounds] = maskRounds{m, (uint64(s.len) >> 6) - round} @@ -45,18 +54,24 @@ func generateMaskAndRounds8(input [8][]byte, mr *[8]maskRounds) (rounds int) { } func generateMaskAndRounds16(input [16][]byte, mr *[16]maskRounds) (rounds int) { - // Sort on blocks length small to large var sorted [16]lane - for c, inpt := range input { + for c, inpt := range input[:] { sorted[c] = lane{uint(len(inpt)), uint(c)} + for i := c - 1; i >= 0; i-- { + // swap so largest is at the end... + if sorted[i].len > sorted[i+1].len { + sorted[i], sorted[i+1] = sorted[i+1], sorted[i] + continue + } + break + } } - sort.Slice(sorted[:], func(i, j int) bool { return sorted[i].len < sorted[j].len }) // Create mask array including 'rounds' (of processing blocks of 64 bytes) between masks m, round := uint64(0xffff), uint64(0) - for _, s := range sorted { + for _, s := range sorted[:] { if s.len > 0 { if uint64(s.len)>>6 > round { mr[rounds] = maskRounds{m, (uint64(s.len) >> 6) - round} diff --git a/vendor/github.com/minio/md5-simd/md5.go b/vendor/github.com/minio/md5-simd/md5.go index 4f56b79d0..11b0cb962 100644 --- a/vendor/github.com/minio/md5-simd/md5.go +++ b/vendor/github.com/minio/md5-simd/md5.go @@ -27,6 +27,12 @@ type Hasher interface { Close() } +// StdlibHasher returns a Hasher that uses the stdlib for hashing. +// Used hashers are stored in a pool for fast reuse. +func StdlibHasher() Hasher { + return &md5Wrapper{Hash: md5Pool.New().(hash.Hash)} +} + // md5Wrapper is a wrapper around the builtin hasher. type md5Wrapper struct { hash.Hash diff --git a/vendor/github.com/minio/md5-simd/md5block_amd64.go b/vendor/github.com/minio/md5-simd/md5block_amd64.go new file mode 100644 index 000000000..4c2793662 --- /dev/null +++ b/vendor/github.com/minio/md5-simd/md5block_amd64.go @@ -0,0 +1,11 @@ +// Code generated by command: go run gen.go -out ../md5block_amd64.s -stubs ../md5block_amd64.go -pkg=md5simd. DO NOT EDIT. + +// +build !appengine +// +build !noasm +// +build gc + +package md5simd + +// Encode p to digest +//go:noescape +func blockScalar(dig *[4]uint32, p []byte) diff --git a/vendor/github.com/minio/md5-simd/md5block_amd64.s b/vendor/github.com/minio/md5-simd/md5block_amd64.s new file mode 100644 index 000000000..fbc4a21f2 --- /dev/null +++ b/vendor/github.com/minio/md5-simd/md5block_amd64.s @@ -0,0 +1,714 @@ +// Code generated by command: go run gen.go -out ../md5block_amd64.s -stubs ../md5block_amd64.go -pkg=md5simd. DO NOT EDIT. + +// +build !appengine +// +build !noasm +// +build gc + +// func blockScalar(dig *[4]uint32, p []byte) +TEXT ·blockScalar(SB), $0-32 + MOVQ p_len+16(FP), AX + MOVQ dig+0(FP), CX + MOVQ p_base+8(FP), DX + SHRQ $0x06, AX + SHLQ $0x06, AX + LEAQ (DX)(AX*1), AX + CMPQ DX, AX + JEQ end + MOVL (CX), BX + MOVL 4(CX), BP + MOVL 8(CX), SI + MOVL 12(CX), CX + MOVL $0xffffffff, DI + +loop: + MOVL (DX), R8 + MOVL CX, R9 + MOVL BX, R10 + MOVL BP, R11 + MOVL SI, R12 + MOVL CX, R13 + + // ROUND1 + XORL SI, R9 + ADDL $0xd76aa478, BX + ADDL R8, BX + ANDL BP, R9 + XORL CX, R9 + MOVL 4(DX), R8 + ADDL R9, BX + ROLL $0x07, BX + MOVL SI, R9 + ADDL BP, BX + XORL BP, R9 + ADDL $0xe8c7b756, CX + ADDL R8, CX + ANDL BX, R9 + XORL SI, R9 + MOVL 8(DX), R8 + ADDL R9, CX + ROLL $0x0c, CX + MOVL BP, R9 + ADDL BX, CX + XORL BX, R9 + ADDL $0x242070db, SI + ADDL R8, SI + ANDL CX, R9 + XORL BP, R9 + MOVL 12(DX), R8 + ADDL R9, SI + ROLL $0x11, SI + MOVL BX, R9 + ADDL CX, SI + XORL CX, R9 + ADDL $0xc1bdceee, BP + ADDL R8, BP + ANDL SI, R9 + XORL BX, R9 + MOVL 16(DX), R8 + ADDL R9, BP + ROLL $0x16, BP + MOVL CX, R9 + ADDL SI, BP + XORL SI, R9 + ADDL $0xf57c0faf, BX + ADDL R8, BX + ANDL BP, R9 + XORL CX, R9 + MOVL 20(DX), R8 + ADDL R9, BX + ROLL $0x07, BX + MOVL SI, R9 + ADDL BP, BX + XORL BP, R9 + ADDL $0x4787c62a, CX + ADDL R8, CX + ANDL BX, R9 + XORL SI, R9 + MOVL 24(DX), R8 + ADDL R9, CX + ROLL $0x0c, CX + MOVL BP, R9 + ADDL BX, CX + XORL BX, R9 + ADDL $0xa8304613, SI + ADDL R8, SI + ANDL CX, R9 + XORL BP, R9 + MOVL 28(DX), R8 + ADDL R9, SI + ROLL $0x11, SI + MOVL BX, R9 + ADDL CX, SI + XORL CX, R9 + ADDL $0xfd469501, BP + ADDL R8, BP + ANDL SI, R9 + XORL BX, R9 + MOVL 32(DX), R8 + ADDL R9, BP + ROLL $0x16, BP + MOVL CX, R9 + ADDL SI, BP + XORL SI, R9 + ADDL $0x698098d8, BX + ADDL R8, BX + ANDL BP, R9 + XORL CX, R9 + MOVL 36(DX), R8 + ADDL R9, BX + ROLL $0x07, BX + MOVL SI, R9 + ADDL BP, BX + XORL BP, R9 + ADDL $0x8b44f7af, CX + ADDL R8, CX + ANDL BX, R9 + XORL SI, R9 + MOVL 40(DX), R8 + ADDL R9, CX + ROLL $0x0c, CX + MOVL BP, R9 + ADDL BX, CX + XORL BX, R9 + ADDL $0xffff5bb1, SI + ADDL R8, SI + ANDL CX, R9 + XORL BP, R9 + MOVL 44(DX), R8 + ADDL R9, SI + ROLL $0x11, SI + MOVL BX, R9 + ADDL CX, SI + XORL CX, R9 + ADDL $0x895cd7be, BP + ADDL R8, BP + ANDL SI, R9 + XORL BX, R9 + MOVL 48(DX), R8 + ADDL R9, BP + ROLL $0x16, BP + MOVL CX, R9 + ADDL SI, BP + XORL SI, R9 + ADDL $0x6b901122, BX + ADDL R8, BX + ANDL BP, R9 + XORL CX, R9 + MOVL 52(DX), R8 + ADDL R9, BX + ROLL $0x07, BX + MOVL SI, R9 + ADDL BP, BX + XORL BP, R9 + ADDL $0xfd987193, CX + ADDL R8, CX + ANDL BX, R9 + XORL SI, R9 + MOVL 56(DX), R8 + ADDL R9, CX + ROLL $0x0c, CX + MOVL BP, R9 + ADDL BX, CX + XORL BX, R9 + ADDL $0xa679438e, SI + ADDL R8, SI + ANDL CX, R9 + XORL BP, R9 + MOVL 60(DX), R8 + ADDL R9, SI + ROLL $0x11, SI + MOVL BX, R9 + ADDL CX, SI + XORL CX, R9 + ADDL $0x49b40821, BP + ADDL R8, BP + ANDL SI, R9 + XORL BX, R9 + MOVL 4(DX), R8 + ADDL R9, BP + ROLL $0x16, BP + MOVL CX, R9 + ADDL SI, BP + + // ROUND2 + MOVL CX, R9 + MOVL CX, R14 + XORL DI, R9 + ADDL $0xf61e2562, BX + ADDL R8, BX + ANDL BP, R14 + ANDL SI, R9 + MOVL 24(DX), R8 + ORL R9, R14 + MOVL SI, R9 + ADDL R14, BX + MOVL SI, R14 + ROLL $0x05, BX + ADDL BP, BX + XORL DI, R9 + ADDL $0xc040b340, CX + ADDL R8, CX + ANDL BX, R14 + ANDL BP, R9 + MOVL 44(DX), R8 + ORL R9, R14 + MOVL BP, R9 + ADDL R14, CX + MOVL BP, R14 + ROLL $0x09, CX + ADDL BX, CX + XORL DI, R9 + ADDL $0x265e5a51, SI + ADDL R8, SI + ANDL CX, R14 + ANDL BX, R9 + MOVL (DX), R8 + ORL R9, R14 + MOVL BX, R9 + ADDL R14, SI + MOVL BX, R14 + ROLL $0x0e, SI + ADDL CX, SI + XORL DI, R9 + ADDL $0xe9b6c7aa, BP + ADDL R8, BP + ANDL SI, R14 + ANDL CX, R9 + MOVL 20(DX), R8 + ORL R9, R14 + MOVL CX, R9 + ADDL R14, BP + MOVL CX, R14 + ROLL $0x14, BP + ADDL SI, BP + XORL DI, R9 + ADDL $0xd62f105d, BX + ADDL R8, BX + ANDL BP, R14 + ANDL SI, R9 + MOVL 40(DX), R8 + ORL R9, R14 + MOVL SI, R9 + ADDL R14, BX + MOVL SI, R14 + ROLL $0x05, BX + ADDL BP, BX + XORL DI, R9 + ADDL $0x02441453, CX + ADDL R8, CX + ANDL BX, R14 + ANDL BP, R9 + MOVL 60(DX), R8 + ORL R9, R14 + MOVL BP, R9 + ADDL R14, CX + MOVL BP, R14 + ROLL $0x09, CX + ADDL BX, CX + XORL DI, R9 + ADDL $0xd8a1e681, SI + ADDL R8, SI + ANDL CX, R14 + ANDL BX, R9 + MOVL 16(DX), R8 + ORL R9, R14 + MOVL BX, R9 + ADDL R14, SI + MOVL BX, R14 + ROLL $0x0e, SI + ADDL CX, SI + XORL DI, R9 + ADDL $0xe7d3fbc8, BP + ADDL R8, BP + ANDL SI, R14 + ANDL CX, R9 + MOVL 36(DX), R8 + ORL R9, R14 + MOVL CX, R9 + ADDL R14, BP + MOVL CX, R14 + ROLL $0x14, BP + ADDL SI, BP + XORL DI, R9 + ADDL $0x21e1cde6, BX + ADDL R8, BX + ANDL BP, R14 + ANDL SI, R9 + MOVL 56(DX), R8 + ORL R9, R14 + MOVL SI, R9 + ADDL R14, BX + MOVL SI, R14 + ROLL $0x05, BX + ADDL BP, BX + XORL DI, R9 + ADDL $0xc33707d6, CX + ADDL R8, CX + ANDL BX, R14 + ANDL BP, R9 + MOVL 12(DX), R8 + ORL R9, R14 + MOVL BP, R9 + ADDL R14, CX + MOVL BP, R14 + ROLL $0x09, CX + ADDL BX, CX + XORL DI, R9 + ADDL $0xf4d50d87, SI + ADDL R8, SI + ANDL CX, R14 + ANDL BX, R9 + MOVL 32(DX), R8 + ORL R9, R14 + MOVL BX, R9 + ADDL R14, SI + MOVL BX, R14 + ROLL $0x0e, SI + ADDL CX, SI + XORL DI, R9 + ADDL $0x455a14ed, BP + ADDL R8, BP + ANDL SI, R14 + ANDL CX, R9 + MOVL 52(DX), R8 + ORL R9, R14 + MOVL CX, R9 + ADDL R14, BP + MOVL CX, R14 + ROLL $0x14, BP + ADDL SI, BP + XORL DI, R9 + ADDL $0xa9e3e905, BX + ADDL R8, BX + ANDL BP, R14 + ANDL SI, R9 + MOVL 8(DX), R8 + ORL R9, R14 + MOVL SI, R9 + ADDL R14, BX + MOVL SI, R14 + ROLL $0x05, BX + ADDL BP, BX + XORL DI, R9 + ADDL $0xfcefa3f8, CX + ADDL R8, CX + ANDL BX, R14 + ANDL BP, R9 + MOVL 28(DX), R8 + ORL R9, R14 + MOVL BP, R9 + ADDL R14, CX + MOVL BP, R14 + ROLL $0x09, CX + ADDL BX, CX + XORL DI, R9 + ADDL $0x676f02d9, SI + ADDL R8, SI + ANDL CX, R14 + ANDL BX, R9 + MOVL 48(DX), R8 + ORL R9, R14 + MOVL BX, R9 + ADDL R14, SI + MOVL BX, R14 + ROLL $0x0e, SI + ADDL CX, SI + XORL DI, R9 + ADDL $0x8d2a4c8a, BP + ADDL R8, BP + ANDL SI, R14 + ANDL CX, R9 + MOVL 20(DX), R8 + ORL R9, R14 + MOVL CX, R9 + ADDL R14, BP + MOVL CX, R14 + ROLL $0x14, BP + ADDL SI, BP + + // ROUND3 + MOVL SI, R9 + ADDL $0xfffa3942, BX + ADDL R8, BX + MOVL 32(DX), R8 + XORL CX, R9 + XORL BP, R9 + ADDL R9, BX + ROLL $0x04, BX + MOVL BP, R9 + ADDL BP, BX + ADDL $0x8771f681, CX + ADDL R8, CX + MOVL 44(DX), R8 + XORL SI, R9 + XORL BX, R9 + ADDL R9, CX + ROLL $0x0b, CX + MOVL BX, R9 + ADDL BX, CX + ADDL $0x6d9d6122, SI + ADDL R8, SI + MOVL 56(DX), R8 + XORL BP, R9 + XORL CX, R9 + ADDL R9, SI + ROLL $0x10, SI + MOVL CX, R9 + ADDL CX, SI + ADDL $0xfde5380c, BP + ADDL R8, BP + MOVL 4(DX), R8 + XORL BX, R9 + XORL SI, R9 + ADDL R9, BP + ROLL $0x17, BP + MOVL SI, R9 + ADDL SI, BP + ADDL $0xa4beea44, BX + ADDL R8, BX + MOVL 16(DX), R8 + XORL CX, R9 + XORL BP, R9 + ADDL R9, BX + ROLL $0x04, BX + MOVL BP, R9 + ADDL BP, BX + ADDL $0x4bdecfa9, CX + ADDL R8, CX + MOVL 28(DX), R8 + XORL SI, R9 + XORL BX, R9 + ADDL R9, CX + ROLL $0x0b, CX + MOVL BX, R9 + ADDL BX, CX + ADDL $0xf6bb4b60, SI + ADDL R8, SI + MOVL 40(DX), R8 + XORL BP, R9 + XORL CX, R9 + ADDL R9, SI + ROLL $0x10, SI + MOVL CX, R9 + ADDL CX, SI + ADDL $0xbebfbc70, BP + ADDL R8, BP + MOVL 52(DX), R8 + XORL BX, R9 + XORL SI, R9 + ADDL R9, BP + ROLL $0x17, BP + MOVL SI, R9 + ADDL SI, BP + ADDL $0x289b7ec6, BX + ADDL R8, BX + MOVL (DX), R8 + XORL CX, R9 + XORL BP, R9 + ADDL R9, BX + ROLL $0x04, BX + MOVL BP, R9 + ADDL BP, BX + ADDL $0xeaa127fa, CX + ADDL R8, CX + MOVL 12(DX), R8 + XORL SI, R9 + XORL BX, R9 + ADDL R9, CX + ROLL $0x0b, CX + MOVL BX, R9 + ADDL BX, CX + ADDL $0xd4ef3085, SI + ADDL R8, SI + MOVL 24(DX), R8 + XORL BP, R9 + XORL CX, R9 + ADDL R9, SI + ROLL $0x10, SI + MOVL CX, R9 + ADDL CX, SI + ADDL $0x04881d05, BP + ADDL R8, BP + MOVL 36(DX), R8 + XORL BX, R9 + XORL SI, R9 + ADDL R9, BP + ROLL $0x17, BP + MOVL SI, R9 + ADDL SI, BP + ADDL $0xd9d4d039, BX + ADDL R8, BX + MOVL 48(DX), R8 + XORL CX, R9 + XORL BP, R9 + ADDL R9, BX + ROLL $0x04, BX + MOVL BP, R9 + ADDL BP, BX + ADDL $0xe6db99e5, CX + ADDL R8, CX + MOVL 60(DX), R8 + XORL SI, R9 + XORL BX, R9 + ADDL R9, CX + ROLL $0x0b, CX + MOVL BX, R9 + ADDL BX, CX + ADDL $0x1fa27cf8, SI + ADDL R8, SI + MOVL 8(DX), R8 + XORL BP, R9 + XORL CX, R9 + ADDL R9, SI + ROLL $0x10, SI + MOVL CX, R9 + ADDL CX, SI + ADDL $0xc4ac5665, BP + ADDL R8, BP + MOVL (DX), R8 + XORL BX, R9 + XORL SI, R9 + ADDL R9, BP + ROLL $0x17, BP + MOVL SI, R9 + ADDL SI, BP + + // ROUND4 + MOVL DI, R9 + XORL CX, R9 + ADDL $0xf4292244, BX + ADDL R8, BX + ORL BP, R9 + XORL SI, R9 + ADDL R9, BX + MOVL 28(DX), R8 + MOVL DI, R9 + ROLL $0x06, BX + XORL SI, R9 + ADDL BP, BX + ADDL $0x432aff97, CX + ADDL R8, CX + ORL BX, R9 + XORL BP, R9 + ADDL R9, CX + MOVL 56(DX), R8 + MOVL DI, R9 + ROLL $0x0a, CX + XORL BP, R9 + ADDL BX, CX + ADDL $0xab9423a7, SI + ADDL R8, SI + ORL CX, R9 + XORL BX, R9 + ADDL R9, SI + MOVL 20(DX), R8 + MOVL DI, R9 + ROLL $0x0f, SI + XORL BX, R9 + ADDL CX, SI + ADDL $0xfc93a039, BP + ADDL R8, BP + ORL SI, R9 + XORL CX, R9 + ADDL R9, BP + MOVL 48(DX), R8 + MOVL DI, R9 + ROLL $0x15, BP + XORL CX, R9 + ADDL SI, BP + ADDL $0x655b59c3, BX + ADDL R8, BX + ORL BP, R9 + XORL SI, R9 + ADDL R9, BX + MOVL 12(DX), R8 + MOVL DI, R9 + ROLL $0x06, BX + XORL SI, R9 + ADDL BP, BX + ADDL $0x8f0ccc92, CX + ADDL R8, CX + ORL BX, R9 + XORL BP, R9 + ADDL R9, CX + MOVL 40(DX), R8 + MOVL DI, R9 + ROLL $0x0a, CX + XORL BP, R9 + ADDL BX, CX + ADDL $0xffeff47d, SI + ADDL R8, SI + ORL CX, R9 + XORL BX, R9 + ADDL R9, SI + MOVL 4(DX), R8 + MOVL DI, R9 + ROLL $0x0f, SI + XORL BX, R9 + ADDL CX, SI + ADDL $0x85845dd1, BP + ADDL R8, BP + ORL SI, R9 + XORL CX, R9 + ADDL R9, BP + MOVL 32(DX), R8 + MOVL DI, R9 + ROLL $0x15, BP + XORL CX, R9 + ADDL SI, BP + ADDL $0x6fa87e4f, BX + ADDL R8, BX + ORL BP, R9 + XORL SI, R9 + ADDL R9, BX + MOVL 60(DX), R8 + MOVL DI, R9 + ROLL $0x06, BX + XORL SI, R9 + ADDL BP, BX + ADDL $0xfe2ce6e0, CX + ADDL R8, CX + ORL BX, R9 + XORL BP, R9 + ADDL R9, CX + MOVL 24(DX), R8 + MOVL DI, R9 + ROLL $0x0a, CX + XORL BP, R9 + ADDL BX, CX + ADDL $0xa3014314, SI + ADDL R8, SI + ORL CX, R9 + XORL BX, R9 + ADDL R9, SI + MOVL 52(DX), R8 + MOVL DI, R9 + ROLL $0x0f, SI + XORL BX, R9 + ADDL CX, SI + ADDL $0x4e0811a1, BP + ADDL R8, BP + ORL SI, R9 + XORL CX, R9 + ADDL R9, BP + MOVL 16(DX), R8 + MOVL DI, R9 + ROLL $0x15, BP + XORL CX, R9 + ADDL SI, BP + ADDL $0xf7537e82, BX + ADDL R8, BX + ORL BP, R9 + XORL SI, R9 + ADDL R9, BX + MOVL 44(DX), R8 + MOVL DI, R9 + ROLL $0x06, BX + XORL SI, R9 + ADDL BP, BX + ADDL $0xbd3af235, CX + ADDL R8, CX + ORL BX, R9 + XORL BP, R9 + ADDL R9, CX + MOVL 8(DX), R8 + MOVL DI, R9 + ROLL $0x0a, CX + XORL BP, R9 + ADDL BX, CX + ADDL $0x2ad7d2bb, SI + ADDL R8, SI + ORL CX, R9 + XORL BX, R9 + ADDL R9, SI + MOVL 36(DX), R8 + MOVL DI, R9 + ROLL $0x0f, SI + XORL BX, R9 + ADDL CX, SI + ADDL $0xeb86d391, BP + ADDL R8, BP + ORL SI, R9 + XORL CX, R9 + ADDL R9, BP + ROLL $0x15, BP + ADDL SI, BP + ADDL R10, BX + ADDL R11, BP + ADDL R12, SI + ADDL R13, CX + + // Prepare next loop + ADDQ $0x40, DX + CMPQ DX, AX + JB loop + + // Write output + MOVQ dig+0(FP), AX + MOVL BX, (AX) + MOVL BP, 4(AX) + MOVL SI, 8(AX) + MOVL CX, 12(AX) + +end: + RET diff --git a/vendor/github.com/minio/minio-go/v7/api-get-object.go b/vendor/github.com/minio/minio-go/v7/api-get-object.go index b9f6ded99..b17f3146b 100644 --- a/vendor/github.com/minio/minio-go/v7/api-get-object.go +++ b/vendor/github.com/minio/minio-go/v7/api-get-object.go @@ -45,9 +45,7 @@ func (c *Client) GetObject(ctx context.Context, bucketName, objectName string, o // Detect if snowball is server location we are talking to. var snowball bool if location, ok := c.bucketLocCache.Get(bucketName); ok { - if location == "snowball" { - snowball = true - } + snowball = location == "snowball" } var ( @@ -64,161 +62,46 @@ func (c *Client) GetObject(ctx context.Context, bucketName, objectName string, o // This routine feeds partial object data as and when the caller reads. go func() { - defer close(reqCh) defer close(resCh) + defer func() { + // Close the http response body before returning. + // This ends the connection with the server. + if httpReader != nil { + httpReader.Close() + } + }() + defer cancel() // Used to verify if etag of object has changed since last read. var etag string - // Loop through the incoming control messages and read data. - for { - select { - // When context is closed exit our routine. - case <-gctx.Done(): - // Close the http response body before returning. - // This ends the connection with the server. - if httpReader != nil { - httpReader.Close() - } - return - - // Gather incoming request. - case req := <-reqCh: - // If this is the first request we may not need to do a getObject request yet. - if req.isFirstReq { - // First request is a Read/ReadAt. - if req.isReadOp { - // Differentiate between wanting the whole object and just a range. - if req.isReadAt { - // If this is a ReadAt request only get the specified range. - // Range is set with respect to the offset and length of the buffer requested. - // Do not set objectInfo from the first readAt request because it will not get - // the whole object. - opts.SetRange(req.Offset, req.Offset+int64(len(req.Buffer))-1) - } else if req.Offset > 0 { - opts.SetRange(req.Offset, 0) - } - httpReader, objectInfo, _, err = c.getObject(gctx, bucketName, objectName, opts) - if err != nil { - resCh <- getResponse{Error: err} - return - } - etag = objectInfo.ETag - // Read at least firstReq.Buffer bytes, if not we have - // reached our EOF. - size, err := readFull(httpReader, req.Buffer) - totalRead += size - if size > 0 && err == io.ErrUnexpectedEOF { - if int64(size) < objectInfo.Size { - // In situations when returned size - // is less than the expected content - // length set by the server, make sure - // we return io.ErrUnexpectedEOF - err = io.ErrUnexpectedEOF - } else { - // If an EOF happens after reading some but not - // all the bytes ReadFull returns ErrUnexpectedEOF - err = io.EOF - } - } else if size == 0 && err == io.EOF && objectInfo.Size > 0 { - // Special cases when server writes more data - // than the content-length, net/http response - // body returns an error, instead of converting - // it to io.EOF - return unexpected EOF. - err = io.ErrUnexpectedEOF - } - // Send back the first response. - resCh <- getResponse{ - objectInfo: objectInfo, - Size: size, - Error: err, - didRead: true, - } - } else { - // First request is a Stat or Seek call. - // Only need to run a StatObject until an actual Read or ReadAt request comes through. - - // Remove range header if already set, for stat Operations to get original file size. - delete(opts.headers, "Range") - objectInfo, err = c.StatObject(gctx, bucketName, objectName, StatObjectOptions(opts)) - if err != nil { - resCh <- getResponse{ - Error: err, - } - // Exit the go-routine. - return - } - etag = objectInfo.ETag - // Send back the first response. - resCh <- getResponse{ - objectInfo: objectInfo, - } + for req := range reqCh { + // If this is the first request we may not need to do a getObject request yet. + if req.isFirstReq { + // First request is a Read/ReadAt. + if req.isReadOp { + // Differentiate between wanting the whole object and just a range. + if req.isReadAt { + // If this is a ReadAt request only get the specified range. + // Range is set with respect to the offset and length of the buffer requested. + // Do not set objectInfo from the first readAt request because it will not get + // the whole object. + opts.SetRange(req.Offset, req.Offset+int64(len(req.Buffer))-1) + } else if req.Offset > 0 { + opts.SetRange(req.Offset, 0) } - } else if req.settingObjectInfo { // Request is just to get objectInfo. - // Remove range header if already set, for stat Operations to get original file size. - delete(opts.headers, "Range") - // Check whether this is snowball - // if yes do not use If-Match feature - // it doesn't work. - if etag != "" && !snowball { - opts.SetMatchETag(etag) - } - objectInfo, err := c.StatObject(gctx, bucketName, objectName, StatObjectOptions(opts)) + httpReader, objectInfo, _, err = c.getObject(gctx, bucketName, objectName, opts) if err != nil { - resCh <- getResponse{ - Error: err, - } - // Exit the goroutine. + resCh <- getResponse{Error: err} return } - // Send back the objectInfo. - resCh <- getResponse{ - objectInfo: objectInfo, - } - } else { - // Offset changes fetch the new object at an Offset. - // Because the httpReader may not be set by the first - // request if it was a stat or seek it must be checked - // if the object has been read or not to only initialize - // new ones when they haven't been already. - // All readAt requests are new requests. - if req.DidOffsetChange || !req.beenRead { - // Check whether this is snowball - // if yes do not use If-Match feature - // it doesn't work. - if etag != "" && !snowball { - opts.SetMatchETag(etag) - } - if httpReader != nil { - // Close previously opened http reader. - httpReader.Close() - } - // If this request is a readAt only get the specified range. - if req.isReadAt { - // Range is set with respect to the offset and length of the buffer requested. - opts.SetRange(req.Offset, req.Offset+int64(len(req.Buffer))-1) - } else if req.Offset > 0 { // Range is set with respect to the offset. - opts.SetRange(req.Offset, 0) - } else { - // Remove range header if already set - delete(opts.headers, "Range") - } - httpReader, objectInfo, _, err = c.getObject(gctx, bucketName, objectName, opts) - if err != nil { - resCh <- getResponse{ - Error: err, - } - return - } - totalRead = 0 - } - - // Read at least req.Buffer bytes, if not we have + etag = objectInfo.ETag + // Read at least firstReq.Buffer bytes, if not we have // reached our EOF. size, err := readFull(httpReader, req.Buffer) totalRead += size if size > 0 && err == io.ErrUnexpectedEOF { - if int64(totalRead) < objectInfo.Size { + if int64(size) < objectInfo.Size { // In situations when returned size // is less than the expected content // length set by the server, make sure @@ -236,15 +119,123 @@ func (c *Client) GetObject(ctx context.Context, bucketName, objectName string, o // it to io.EOF - return unexpected EOF. err = io.ErrUnexpectedEOF } - - // Reply back how much was read. + // Send back the first response. resCh <- getResponse{ + objectInfo: objectInfo, Size: size, Error: err, didRead: true, + } + } else { + // First request is a Stat or Seek call. + // Only need to run a StatObject until an actual Read or ReadAt request comes through. + + // Remove range header if already set, for stat Operations to get original file size. + delete(opts.headers, "Range") + objectInfo, err = c.StatObject(gctx, bucketName, objectName, StatObjectOptions(opts)) + if err != nil { + resCh <- getResponse{ + Error: err, + } + // Exit the go-routine. + return + } + etag = objectInfo.ETag + // Send back the first response. + resCh <- getResponse{ objectInfo: objectInfo, } } + } else if req.settingObjectInfo { // Request is just to get objectInfo. + // Remove range header if already set, for stat Operations to get original file size. + delete(opts.headers, "Range") + // Check whether this is snowball + // if yes do not use If-Match feature + // it doesn't work. + if etag != "" && !snowball { + opts.SetMatchETag(etag) + } + objectInfo, err := c.StatObject(gctx, bucketName, objectName, StatObjectOptions(opts)) + if err != nil { + resCh <- getResponse{ + Error: err, + } + // Exit the goroutine. + return + } + // Send back the objectInfo. + resCh <- getResponse{ + objectInfo: objectInfo, + } + } else { + // Offset changes fetch the new object at an Offset. + // Because the httpReader may not be set by the first + // request if it was a stat or seek it must be checked + // if the object has been read or not to only initialize + // new ones when they haven't been already. + // All readAt requests are new requests. + if req.DidOffsetChange || !req.beenRead { + // Check whether this is snowball + // if yes do not use If-Match feature + // it doesn't work. + if etag != "" && !snowball { + opts.SetMatchETag(etag) + } + if httpReader != nil { + // Close previously opened http reader. + httpReader.Close() + } + // If this request is a readAt only get the specified range. + if req.isReadAt { + // Range is set with respect to the offset and length of the buffer requested. + opts.SetRange(req.Offset, req.Offset+int64(len(req.Buffer))-1) + } else if req.Offset > 0 { // Range is set with respect to the offset. + opts.SetRange(req.Offset, 0) + } else { + // Remove range header if already set + delete(opts.headers, "Range") + } + httpReader, objectInfo, _, err = c.getObject(gctx, bucketName, objectName, opts) + if err != nil { + resCh <- getResponse{ + Error: err, + } + return + } + totalRead = 0 + } + + // Read at least req.Buffer bytes, if not we have + // reached our EOF. + size, err := readFull(httpReader, req.Buffer) + totalRead += size + if size > 0 && err == io.ErrUnexpectedEOF { + if int64(totalRead) < objectInfo.Size { + // In situations when returned size + // is less than the expected content + // length set by the server, make sure + // we return io.ErrUnexpectedEOF + err = io.ErrUnexpectedEOF + } else { + // If an EOF happens after reading some but not + // all the bytes ReadFull returns ErrUnexpectedEOF + err = io.EOF + } + } else if size == 0 && err == io.EOF && objectInfo.Size > 0 { + // Special cases when server writes more data + // than the content-length, net/http response + // body returns an error, instead of converting + // it to io.EOF - return unexpected EOF. + err = io.ErrUnexpectedEOF + } + + // Reply back how much was read. + resCh <- getResponse{ + Size: size, + Error: err, + didRead: true, + objectInfo: objectInfo, + } } } }() @@ -611,6 +602,7 @@ func (o *Object) Close() (err error) { if o == nil { return errInvalidArgument("Object is nil") } + // Locking. o.mutex.Lock() defer o.mutex.Unlock() @@ -623,6 +615,9 @@ func (o *Object) Close() (err error) { // Close successfully. o.cancel() + // Close the request channel to indicate the internal go-routine to exit. + close(o.reqCh) + // Save for future operations. errMsg := "Object is already closed. Bad file descriptor." o.prevErr = errors.New(errMsg) diff --git a/vendor/github.com/minio/minio-go/v7/api-put-object-multipart.go b/vendor/github.com/minio/minio-go/v7/api-put-object-multipart.go index 342a8dc2b..abcbd2981 100644 --- a/vendor/github.com/minio/minio-go/v7/api-put-object-multipart.go +++ b/vendor/github.com/minio/minio-go/v7/api-put-object-multipart.go @@ -104,7 +104,7 @@ func (c *Client) putObjectMultipartNoStream(ctx context.Context, bucketName, obj // Choose hash algorithms to be calculated by hashCopyN, // avoid sha256 with non-v4 signature request or // HTTPS connection. - hashAlgos, hashSums := c.hashMaterials(opts.SendContentMd5) + hashAlgos, hashSums := c.hashMaterials(opts.SendContentMd5, !opts.DisableContentSha256) length, rErr := readFull(reader, buf) if rErr == io.EOF && partNumber > 1 { @@ -140,7 +140,9 @@ func (c *Client) putObjectMultipartNoStream(ctx context.Context, bucketName, obj // Proceed to upload the part. objPart, uerr := c.uploadPart(ctx, bucketName, objectName, uploadID, rd, partNumber, - md5Base64, sha256Hex, int64(length), opts.ServerSideEncryption) + md5Base64, sha256Hex, int64(length), + opts.ServerSideEncryption, + !opts.DisableContentSha256) if uerr != nil { return UploadInfo{}, uerr } @@ -241,7 +243,7 @@ func (c *Client) initiateMultipartUpload(ctx context.Context, bucketName, object // uploadPart - Uploads a part in a multipart upload. func (c *Client) uploadPart(ctx context.Context, bucketName, objectName, uploadID string, reader io.Reader, - partNumber int, md5Base64, sha256Hex string, size int64, sse encrypt.ServerSide, + partNumber int, md5Base64, sha256Hex string, size int64, sse encrypt.ServerSide, streamSha256 bool, ) (ObjectPart, error) { // Input validation. if err := s3utils.CheckValidBucketName(bucketName); err != nil { @@ -289,6 +291,7 @@ func (c *Client) uploadPart(ctx context.Context, bucketName, objectName, uploadI contentLength: size, contentMD5Base64: md5Base64, contentSHA256Hex: sha256Hex, + streamSha256: streamSha256, } // Execute PUT on each part. diff --git a/vendor/github.com/minio/minio-go/v7/api-put-object-streaming.go b/vendor/github.com/minio/minio-go/v7/api-put-object-streaming.go index 2497aecf3..11b3a5255 100644 --- a/vendor/github.com/minio/minio-go/v7/api-put-object-streaming.go +++ b/vendor/github.com/minio/minio-go/v7/api-put-object-streaming.go @@ -130,34 +130,44 @@ func (c *Client) putObjectMultipartStreamFromReadAt(ctx context.Context, bucketN var complMultipartUpload completeMultipartUpload // Declare a channel that sends the next part number to be uploaded. - // Buffered to 10000 because thats the maximum number of parts allowed - // by S3. - uploadPartsCh := make(chan uploadPartReq, 10000) + uploadPartsCh := make(chan uploadPartReq) // Declare a channel that sends back the response of a part upload. - // Buffered to 10000 because thats the maximum number of parts allowed - // by S3. - uploadedPartsCh := make(chan uploadedPartRes, 10000) + uploadedPartsCh := make(chan uploadedPartRes) // Used for readability, lastPartNumber is always totalPartsCount. lastPartNumber := totalPartsCount + partitionCtx, partitionCancel := context.WithCancel(ctx) + defer partitionCancel() // Send each part number to the channel to be processed. - for p := 1; p <= totalPartsCount; p++ { - uploadPartsCh <- uploadPartReq{PartNum: p} - } - close(uploadPartsCh) - - partsBuf := make([][]byte, opts.getNumThreads()) - for i := range partsBuf { - partsBuf[i] = make([]byte, 0, partSize) - } + go func() { + defer close(uploadPartsCh) + + for p := 1; p <= totalPartsCount; p++ { + select { + case <-partitionCtx.Done(): + return + case uploadPartsCh <- uploadPartReq{PartNum: p}: + } + } + }() // Receive each part number from the channel allowing three parallel uploads. for w := 1; w <= opts.getNumThreads(); w++ { - go func(w int, partSize int64) { - // Each worker will draw from the part channel and upload in parallel. - for uploadReq := range uploadPartsCh { + go func(partSize int64) { + for { + var uploadReq uploadPartReq + var ok bool + select { + case <-ctx.Done(): + return + case uploadReq, ok = <-uploadPartsCh: + if !ok { + return + } + // Each worker will draw from the part channel and upload in parallel. + } // If partNumber was not uploaded we calculate the missing // part offset and size. For all other part numbers we @@ -171,22 +181,15 @@ func (c *Client) putObjectMultipartStreamFromReadAt(ctx context.Context, bucketN partSize = lastPartSize } - n, rerr := readFull(io.NewSectionReader(reader, readOffset, partSize), partsBuf[w-1][:partSize]) - if rerr != nil && rerr != io.ErrUnexpectedEOF && rerr != io.EOF { - uploadedPartsCh <- uploadedPartRes{ - Error: rerr, - } - // Exit the goroutine. - return - } - - // Get a section reader on a particular offset. - hookReader := newHook(bytes.NewReader(partsBuf[w-1][:n]), opts.Progress) + sectionReader := newHook(io.NewSectionReader(reader, readOffset, partSize), opts.Progress) // Proceed to upload the part. objPart, err := c.uploadPart(ctx, bucketName, objectName, - uploadID, hookReader, uploadReq.PartNum, - "", "", partSize, opts.ServerSideEncryption) + uploadID, sectionReader, uploadReq.PartNum, + "", "", partSize, + opts.ServerSideEncryption, + !opts.DisableContentSha256, + ) if err != nil { uploadedPartsCh <- uploadedPartRes{ Error: err, @@ -205,23 +208,28 @@ func (c *Client) putObjectMultipartStreamFromReadAt(ctx context.Context, bucketN Part: uploadReq.Part, } } - }(w, partSize) + }(partSize) } // Gather the responses as they occur and update any // progress bar. for u := 1; u <= totalPartsCount; u++ { - uploadRes := <-uploadedPartsCh - if uploadRes.Error != nil { - return UploadInfo{}, uploadRes.Error + select { + case <-ctx.Done(): + return UploadInfo{}, ctx.Err() + case uploadRes := <-uploadedPartsCh: + if uploadRes.Error != nil { + + return UploadInfo{}, uploadRes.Error + } + + // Update the totalUploadedSize. + totalUploadedSize += uploadRes.Size + complMultipartUpload.Parts = append(complMultipartUpload.Parts, CompletePart{ + ETag: uploadRes.Part.ETag, + PartNumber: uploadRes.Part.PartNumber, + }) } - // Update the totalUploadedSize. - totalUploadedSize += uploadRes.Size - // Store the parts to be completed in order. - complMultipartUpload.Parts = append(complMultipartUpload.Parts, CompletePart{ - ETag: uploadRes.Part.ETag, - PartNumber: uploadRes.Part.PartNumber, - }) } // Verify if we uploaded all the data. @@ -322,7 +330,10 @@ func (c *Client) putObjectMultipartStreamOptionalChecksum(ctx context.Context, b objPart, uerr := c.uploadPart(ctx, bucketName, objectName, uploadID, io.LimitReader(hookReader, partSize), - partNumber, md5Base64, "", partSize, opts.ServerSideEncryption) + partNumber, md5Base64, "", partSize, + opts.ServerSideEncryption, + !opts.DisableContentSha256, + ) if uerr != nil { return UploadInfo{}, uerr } @@ -452,6 +463,7 @@ func (c *Client) putObjectDo(ctx context.Context, bucketName, objectName string, contentLength: size, contentMD5Base64: md5Base64, contentSHA256Hex: sha256Hex, + streamSha256: !opts.DisableContentSha256, } if opts.Internal.SourceVersionID != "" { if opts.Internal.SourceVersionID != nullVersionID { diff --git a/vendor/github.com/minio/minio-go/v7/api-put-object.go b/vendor/github.com/minio/minio-go/v7/api-put-object.go index 181f1a943..9328fb6c1 100644 --- a/vendor/github.com/minio/minio-go/v7/api-put-object.go +++ b/vendor/github.com/minio/minio-go/v7/api-put-object.go @@ -84,6 +84,7 @@ type PutObjectOptions struct { PartSize uint64 LegalHold LegalHoldStatus SendContentMd5 bool + DisableContentSha256 bool DisableMultipart bool Internal AdvancedPutOptions } @@ -344,7 +345,10 @@ func (c *Client) putObjectMultipartStreamNoLength(ctx context.Context, bucketNam // Proceed to upload the part. objPart, uerr := c.uploadPart(ctx, bucketName, objectName, uploadID, rd, partNumber, - md5Base64, "", int64(length), opts.ServerSideEncryption) + md5Base64, "", int64(length), + opts.ServerSideEncryption, + !opts.DisableContentSha256, + ) if uerr != nil { return UploadInfo{}, uerr } diff --git a/vendor/github.com/minio/minio-go/v7/api.go b/vendor/github.com/minio/minio-go/v7/api.go index c8a565988..2dcfc7978 100644 --- a/vendor/github.com/minio/minio-go/v7/api.go +++ b/vendor/github.com/minio/minio-go/v7/api.go @@ -111,7 +111,7 @@ type Options struct { // Global constants. const ( libraryName = "minio-go" - libraryVersion = "v7.0.29" + libraryVersion = "v7.0.36" ) // User Agent should always following the below style. @@ -315,14 +315,16 @@ func (c *Client) SetS3TransferAccelerate(accelerateEndpoint string) { // - For signature v4 request if the connection is insecure compute only sha256. // - For signature v4 request if the connection is secure compute only md5. // - For anonymous request compute md5. -func (c *Client) hashMaterials(isMd5Requested bool) (hashAlgos map[string]md5simd.Hasher, hashSums map[string][]byte) { +func (c *Client) hashMaterials(isMd5Requested, isSha256Requested bool) (hashAlgos map[string]md5simd.Hasher, hashSums map[string][]byte) { hashSums = make(map[string][]byte) hashAlgos = make(map[string]md5simd.Hasher) if c.overrideSignerType.IsV4() { if c.secure { hashAlgos["md5"] = c.md5Hasher() } else { - hashAlgos["sha256"] = c.sha256Hasher() + if isSha256Requested { + hashAlgos["sha256"] = c.sha256Hasher() + } } } else { if c.overrideSignerType.IsAnonymous() { @@ -377,21 +379,20 @@ func (c *Client) HealthCheck(hcDuration time.Duration) (context.CancelFunc, erro atomic.StoreInt32(&c.healthStatus, unknown) return case <-timer.C: - timer.Reset(duration) // Do health check the first time and ONLY if the connection is marked offline if c.IsOffline() { gctx, gcancel := context.WithTimeout(context.Background(), 3*time.Second) _, err := c.getBucketLocation(gctx, probeBucketName) gcancel() - if IsNetworkOrHostDown(err, false) { - // Still network errors do not need to do anything. - continue - } - switch ToErrorResponse(err).Code { - case "NoSuchBucket", "AccessDenied", "": - atomic.CompareAndSwapInt32(&c.healthStatus, offline, online) + if !IsNetworkOrHostDown(err, false) { + switch ToErrorResponse(err).Code { + case "NoSuchBucket", "AccessDenied", "": + atomic.CompareAndSwapInt32(&c.healthStatus, offline, online) + } } } + + timer.Reset(duration) } } }(hcDuration) @@ -417,6 +418,7 @@ type requestMetadata struct { contentLength int64 contentMD5Base64 string // carries base64 encoded md5sum contentSHA256Hex string // carries hex encoded sha256sum + streamSha256 bool } // dumpHTTP - dump HTTP request and response. @@ -593,12 +595,11 @@ func (c *Client) executeMethod(ctx context.Context, method string, metadata requ // Initiate the request. res, err = c.do(req) if err != nil { - if errors.Is(err, context.Canceled) || errors.Is(err, context.DeadlineExceeded) { - return nil, err + if isRequestErrorRetryable(err) { + // Retry the request + continue } - - // Retry the request - continue + return nil, err } // For any known successful http status, return quickly. @@ -812,7 +813,7 @@ func (c *Client) newRequest(ctx context.Context, method string, metadata request case signerType.IsV2(): // Add signature version '2' authorization header. req = signer.SignV2(*req, accessKeyID, secretAccessKey, isVirtualHost) - case metadata.objectName != "" && metadata.queryValues == nil && method == http.MethodPut && metadata.customHeader.Get("X-Amz-Copy-Source") == "" && !c.secure: + case metadata.streamSha256 && !c.secure: // Streaming signature is used by default for a PUT object request. Additionally we also // look if the initialized client is secure, if yes then we don't need to perform // streaming signature. diff --git a/vendor/github.com/minio/minio-go/v7/core.go b/vendor/github.com/minio/minio-go/v7/core.go index c2a90239f..148671eec 100644 --- a/vendor/github.com/minio/minio-go/v7/core.go +++ b/vendor/github.com/minio/minio-go/v7/core.go @@ -88,7 +88,8 @@ func (c Core) ListMultipartUploads(ctx context.Context, bucket, prefix, keyMarke // PutObjectPart - Upload an object part. func (c Core) PutObjectPart(ctx context.Context, bucket, object, uploadID string, partID int, data io.Reader, size int64, md5Base64, sha256Hex string, sse encrypt.ServerSide) (ObjectPart, error) { - return c.uploadPart(ctx, bucket, object, uploadID, data, partID, md5Base64, sha256Hex, size, sse) + streamSha256 := true + return c.uploadPart(ctx, bucket, object, uploadID, data, partID, md5Base64, sha256Hex, size, sse, streamSha256) } // ListObjectParts - List uploaded parts of an incomplete upload.x diff --git a/vendor/github.com/minio/minio-go/v7/hook-reader.go b/vendor/github.com/minio/minio-go/v7/hook-reader.go index f251c1e95..07bc7dbcf 100644 --- a/vendor/github.com/minio/minio-go/v7/hook-reader.go +++ b/vendor/github.com/minio/minio-go/v7/hook-reader.go @@ -20,6 +20,7 @@ package minio import ( "fmt" "io" + "sync" ) // hookReader hooks additional reader in the source stream. It is @@ -27,6 +28,7 @@ import ( // notified about the exact number of bytes read from the primary // source on each Read operation. type hookReader struct { + mu sync.RWMutex source io.Reader hook io.Reader } @@ -34,6 +36,9 @@ type hookReader struct { // Seek implements io.Seeker. Seeks source first, and if necessary // seeks hook if Seek method is appropriately found. func (hr *hookReader) Seek(offset int64, whence int) (n int64, err error) { + hr.mu.Lock() + defer hr.mu.Unlock() + // Verify for source has embedded Seeker, use it. sourceSeeker, ok := hr.source.(io.Seeker) if ok { @@ -43,18 +48,21 @@ func (hr *hookReader) Seek(offset int64, whence int) (n int64, err error) { } } - // Verify if hook has embedded Seeker, use it. - hookSeeker, ok := hr.hook.(io.Seeker) - if ok { - var m int64 - m, err = hookSeeker.Seek(offset, whence) - if err != nil { - return 0, err - } - if n != m { - return 0, fmt.Errorf("hook seeker seeked %d bytes, expected source %d bytes", m, n) + if hr.hook != nil { + // Verify if hook has embedded Seeker, use it. + hookSeeker, ok := hr.hook.(io.Seeker) + if ok { + var m int64 + m, err = hookSeeker.Seek(offset, whence) + if err != nil { + return 0, err + } + if n != m { + return 0, fmt.Errorf("hook seeker seeked %d bytes, expected source %d bytes", m, n) + } } } + return n, nil } @@ -62,14 +70,19 @@ func (hr *hookReader) Seek(offset int64, whence int) (n int64, err error) { // value 'n' number of bytes are reported through the hook. Returns // error for all non io.EOF conditions. func (hr *hookReader) Read(b []byte) (n int, err error) { + hr.mu.RLock() + defer hr.mu.RUnlock() + n, err = hr.source.Read(b) if err != nil && err != io.EOF { return n, err } - // Progress the hook with the total read bytes from the source. - if _, herr := hr.hook.Read(b[:n]); herr != nil { - if herr != io.EOF { - return n, herr + if hr.hook != nil { + // Progress the hook with the total read bytes from the source. + if _, herr := hr.hook.Read(b[:n]); herr != nil { + if herr != io.EOF { + return n, herr + } } } return n, err @@ -79,7 +92,10 @@ func (hr *hookReader) Read(b []byte) (n int, err error) { // reports the data read from the source to the hook. func newHook(source, hook io.Reader) io.Reader { if hook == nil { - return source + return &hookReader{source: source} + } + return &hookReader{ + source: source, + hook: hook, } - return &hookReader{source, hook} } diff --git a/vendor/github.com/minio/minio-go/v7/pkg/credentials/file_aws_credentials.go b/vendor/github.com/minio/minio-go/v7/pkg/credentials/file_aws_credentials.go index ccc8251f4..cbdcfe256 100644 --- a/vendor/github.com/minio/minio-go/v7/pkg/credentials/file_aws_credentials.go +++ b/vendor/github.com/minio/minio-go/v7/pkg/credentials/file_aws_credentials.go @@ -21,7 +21,6 @@ import ( "os" "path/filepath" - homedir "github.com/mitchellh/go-homedir" ini "gopkg.in/ini.v1" ) @@ -62,7 +61,7 @@ func (p *FileAWSCredentials) Retrieve() (Value, error) { if p.Filename == "" { p.Filename = os.Getenv("AWS_SHARED_CREDENTIALS_FILE") if p.Filename == "" { - homeDir, err := homedir.Dir() + homeDir, err := os.UserHomeDir() if err != nil { return Value{}, err } diff --git a/vendor/github.com/minio/minio-go/v7/pkg/credentials/file_minio_client.go b/vendor/github.com/minio/minio-go/v7/pkg/credentials/file_minio_client.go index dc3f3cc0b..56437edb2 100644 --- a/vendor/github.com/minio/minio-go/v7/pkg/credentials/file_minio_client.go +++ b/vendor/github.com/minio/minio-go/v7/pkg/credentials/file_minio_client.go @@ -24,7 +24,6 @@ import ( "runtime" jsoniter "github.com/json-iterator/go" - homedir "github.com/mitchellh/go-homedir" ) // A FileMinioClient retrieves credentials from the current user's home @@ -65,7 +64,7 @@ func (p *FileMinioClient) Retrieve() (Value, error) { if value, ok := os.LookupEnv("MINIO_SHARED_CREDENTIALS_FILE"); ok { p.Filename = value } else { - homeDir, err := homedir.Dir() + homeDir, err := os.UserHomeDir() if err != nil { return Value{}, err } diff --git a/vendor/github.com/minio/minio-go/v7/pkg/credentials/iam_aws.go b/vendor/github.com/minio/minio-go/v7/pkg/credentials/iam_aws.go index f7a4af4a2..14369cf10 100644 --- a/vendor/github.com/minio/minio-go/v7/pkg/credentials/iam_aws.go +++ b/vendor/github.com/minio/minio-go/v7/pkg/credentials/iam_aws.go @@ -289,7 +289,10 @@ func getCredentials(client *http.Client, endpoint string) (ec2RoleCredRespBody, } // https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/configuring-instance-metadata-service.html - token, _ := fetchIMDSToken(client, endpoint) + token, err := fetchIMDSToken(client, endpoint) + if err != nil { + return ec2RoleCredRespBody{}, err + } // http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/iam-roles-for-amazon-ec2.html u, err := getIAMRoleURL(endpoint) diff --git a/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_client_grants.go b/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_client_grants.go index 1f106ef72..34598bd8e 100644 --- a/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_client_grants.go +++ b/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_client_grants.go @@ -1,6 +1,6 @@ /* * MinIO Go Library for Amazon S3 Compatible Cloud Storage - * Copyright 2019 MinIO, Inc. + * Copyright 2019-2022 MinIO, Inc. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -25,6 +25,7 @@ import ( "io/ioutil" "net/http" "net/url" + "strings" "time" ) @@ -122,12 +123,14 @@ func getClientGrantsCredentials(clnt *http.Client, endpoint string, if err != nil { return AssumeRoleWithClientGrantsResponse{}, err } - u.RawQuery = v.Encode() - req, err := http.NewRequest(http.MethodPost, u.String(), nil) + req, err := http.NewRequest(http.MethodPost, u.String(), strings.NewReader(v.Encode())) if err != nil { return AssumeRoleWithClientGrantsResponse{}, err } + + req.Header.Set("Content-Type", "application/x-www-form-urlencoded") + resp, err := clnt.Do(req) if err != nil { return AssumeRoleWithClientGrantsResponse{}, err diff --git a/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_custom_identity.go b/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_custom_identity.go index ab588712c..e1f9ce4be 100644 --- a/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_custom_identity.go +++ b/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_custom_identity.go @@ -89,12 +89,12 @@ func (c *CustomTokenIdentity) Retrieve() (value Value, err error) { req, err := http.NewRequest(http.MethodPost, u.String(), nil) if err != nil { - return value, stripPassword(err) + return value, err } resp, err := c.Client.Do(req) if err != nil { - return value, stripPassword(err) + return value, err } defer resp.Body.Close() diff --git a/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_ldap_identity.go b/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_ldap_identity.go index 586995e86..25b45ecb0 100644 --- a/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_ldap_identity.go +++ b/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_ldap_identity.go @@ -1,6 +1,6 @@ /* * MinIO Go Library for Amazon S3 Compatible Cloud Storage - * Copyright 2019-2021 MinIO, Inc. + * Copyright 2019-2022 MinIO, Inc. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -24,6 +24,7 @@ import ( "io/ioutil" "net/http" "net/url" + "strings" "time" ) @@ -105,22 +106,6 @@ func LDAPIdentityExpiryOpt(d time.Duration) LDAPIdentityOpt { } } -func stripPassword(err error) error { - urlErr, ok := err.(*url.Error) - if ok { - u, _ := url.Parse(urlErr.URL) - if u == nil { - return urlErr - } - values := u.Query() - values.Set("LDAPPassword", "xxxxx") - u.RawQuery = values.Encode() - urlErr.URL = u.String() - return urlErr - } - return err -} - // NewLDAPIdentityWithSessionPolicy returns new credentials object that uses // LDAP Identity with a specified session policy. The `policy` parameter must be // a JSON string specifying the policy document. @@ -156,16 +141,16 @@ func (k *LDAPIdentity) Retrieve() (value Value, err error) { v.Set("DurationSeconds", fmt.Sprintf("%d", int(k.RequestedExpiry.Seconds()))) } - u.RawQuery = v.Encode() - - req, err := http.NewRequest(http.MethodPost, u.String(), nil) + req, err := http.NewRequest(http.MethodPost, u.String(), strings.NewReader(v.Encode())) if err != nil { - return value, stripPassword(err) + return value, err } + req.Header.Set("Content-Type", "application/x-www-form-urlencoded") + resp, err := k.Client.Do(req) if err != nil { - return value, stripPassword(err) + return value, err } defer resp.Body.Close() diff --git a/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_web_identity.go b/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_web_identity.go index 19bc3ddfc..50f5f1ce6 100644 --- a/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_web_identity.go +++ b/vendor/github.com/minio/minio-go/v7/pkg/credentials/sts_web_identity.go @@ -1,6 +1,6 @@ /* * MinIO Go Library for Amazon S3 Compatible Cloud Storage - * Copyright 2019 MinIO, Inc. + * Copyright 2019-2022 MinIO, Inc. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -26,6 +26,7 @@ import ( "net/http" "net/url" "strconv" + "strings" "time" ) @@ -139,13 +140,13 @@ func getWebIdentityCredentials(clnt *http.Client, endpoint, roleARN, roleSession return AssumeRoleWithWebIdentityResponse{}, err } - u.RawQuery = v.Encode() - - req, err := http.NewRequest(http.MethodPost, u.String(), nil) + req, err := http.NewRequest(http.MethodPost, u.String(), strings.NewReader(v.Encode())) if err != nil { return AssumeRoleWithWebIdentityResponse{}, err } + req.Header.Set("Content-Type", "application/x-www-form-urlencoded") + resp, err := clnt.Do(req) if err != nil { return AssumeRoleWithWebIdentityResponse{}, err diff --git a/vendor/github.com/minio/minio-go/v7/pkg/lifecycle/lifecycle.go b/vendor/github.com/minio/minio-go/v7/pkg/lifecycle/lifecycle.go index 743d8eca9..88a56b09f 100644 --- a/vendor/github.com/minio/minio-go/v7/pkg/lifecycle/lifecycle.go +++ b/vendor/github.com/minio/minio-go/v7/pkg/lifecycle/lifecycle.go @@ -329,15 +329,15 @@ type Expiration struct { XMLName xml.Name `xml:"Expiration,omitempty" json:"-"` Date ExpirationDate `xml:"Date,omitempty" json:"Date,omitempty"` Days ExpirationDays `xml:"Days,omitempty" json:"Days,omitempty"` - DeleteMarker ExpireDeleteMarker `xml:"ExpiredObjectDeleteMarker,omitempty"` + DeleteMarker ExpireDeleteMarker `xml:"ExpiredObjectDeleteMarker,omitempty" json:"ExpiredObjectDeleteMarker,omitempty"` } // MarshalJSON customizes json encoding by removing empty day/date specification. func (e Expiration) MarshalJSON() ([]byte, error) { type expiration struct { - Date *ExpirationDate `json:"Date,omitempty"` - Days *ExpirationDays `json:"Days,omitempty"` - DeleteMarker ExpireDeleteMarker + Date *ExpirationDate `json:"Date,omitempty"` + Days *ExpirationDays `json:"Days,omitempty"` + DeleteMarker ExpireDeleteMarker `json:"ExpiredObjectDeleteMarker,omitempty"` } newexp := expiration{ diff --git a/vendor/github.com/minio/minio-go/v7/post-policy.go b/vendor/github.com/minio/minio-go/v7/post-policy.go index 7aa96e0d6..7bf560304 100644 --- a/vendor/github.com/minio/minio-go/v7/post-policy.go +++ b/vendor/github.com/minio/minio-go/v7/post-policy.go @@ -197,8 +197,8 @@ func (p *PostPolicy) SetContentLengthRange(min, max int64) error { if min < 0 { return errInvalidArgument("Minimum limit cannot be negative.") } - if max < 0 { - return errInvalidArgument("Maximum limit cannot be negative.") + if max <= 0 { + return errInvalidArgument("Maximum limit cannot be non-positive.") } p.contentLengthRange.min = min p.contentLengthRange.max = max diff --git a/vendor/github.com/minio/minio-go/v7/retry.go b/vendor/github.com/minio/minio-go/v7/retry.go index f454e675c..055c14c4d 100644 --- a/vendor/github.com/minio/minio-go/v7/retry.go +++ b/vendor/github.com/minio/minio-go/v7/retry.go @@ -19,7 +19,10 @@ package minio import ( "context" + "crypto/x509" + "errors" "net/http" + "net/url" "time" ) @@ -123,3 +126,23 @@ func isHTTPStatusRetryable(httpStatusCode int) (ok bool) { _, ok = retryableHTTPStatusCodes[httpStatusCode] return ok } + +// For now, all http Do() requests are retriable except some well defined errors +func isRequestErrorRetryable(err error) bool { + if errors.Is(err, context.Canceled) || errors.Is(err, context.DeadlineExceeded) { + return false + } + if ue, ok := err.(*url.Error); ok { + e := ue.Unwrap() + switch e.(type) { + // x509: certificate signed by unknown authority + case x509.UnknownAuthorityError: + return false + } + switch e.Error() { + case "http: server gave HTTP response to HTTPS client": + return false + } + } + return true +} diff --git a/vendor/github.com/minio/sha256-simd/.travis.yml b/vendor/github.com/minio/sha256-simd/.travis.yml deleted file mode 100644 index 4f85db539..000000000 --- a/vendor/github.com/minio/sha256-simd/.travis.yml +++ /dev/null @@ -1,25 +0,0 @@ -sudo: required -dist: trusty -language: go - -os: -- linux - -go: -- tip -- 1.12.x - -env: -- ARCH=x86_64 -- ARCH=i686 - -matrix: - fast_finish: true - allow_failures: - - go: tip - -script: -- diff -au <(gofmt -d .) <(printf "") -- go test -race -v ./... -- go vet -asmdecl . -- ./test-architectures.sh diff --git a/vendor/github.com/minio/sha256-simd/README.md b/vendor/github.com/minio/sha256-simd/README.md index 5282d83ad..6117488d7 100644 --- a/vendor/github.com/minio/sha256-simd/README.md +++ b/vendor/github.com/minio/sha256-simd/README.md @@ -1,14 +1,18 @@ # sha256-simd -Accelerate SHA256 computations in pure Go using AVX512, SHA Extensions and AVX2 for Intel and ARM64 for ARM. On AVX512 it provides an up to 8x improvement (over 3 GB/s per core) in comparison to AVX2. SHA Extensions give a performance boost of close to 4x over AVX2. +Accelerate SHA256 computations in pure Go using AVX512, SHA Extensions for x86 and ARM64 for ARM. +On AVX512 it provides an up to 8x improvement (over 3 GB/s per core). +SHA Extensions give a performance boost of close to 4x over native. ## Introduction -This package is designed as a replacement for `crypto/sha256`. For Intel CPUs it has two flavors for AVX512 and AVX2 (AVX/SSE are also supported). For ARM CPUs with the Cryptography Extensions, advantage is taken of the SHA2 instructions resulting in a massive performance improvement. +This package is designed as a replacement for `crypto/sha256`. +For ARM CPUs with the Cryptography Extensions, advantage is taken of the SHA2 instructions resulting in a massive performance improvement. -This package uses Golang assembly. The AVX512 version is based on the Intel's "multi-buffer crypto library for IPSec" whereas the other Intel implementations are described in "Fast SHA-256 Implementations on Intel Architecture Processors" by J. Guilford et al. +This package uses Golang assembly. +The AVX512 version is based on the Intel's "multi-buffer crypto library for IPSec" whereas the other Intel implementations are described in "Fast SHA-256 Implementations on Intel Architecture Processors" by J. Guilford et al. -## New: Support for Intel SHA Extensions +## Support for Intel SHA Extensions Support for the Intel SHA Extensions has been added by Kristofer Peterson (@svenski123), originally developed for spacemeshos [here](https://github.com/spacemeshos/POET/issues/23). On CPUs that support it (known thus far Intel Celeron J3455 and AMD Ryzen) it gives a significant boost in performance (with thanks to @AudriusButkevicius for reporting the results; full results [here](https://github.com/minio/sha256-simd/pull/37#issuecomment-451607827)). @@ -18,7 +22,9 @@ benchmark AVX2 MB/s SHA Ext MB/s speedup BenchmarkHash5M 514.40 1975.17 3.84x ``` -Thanks to Kristofer Peterson, we also added additional performance changes such as optimized padding, endian conversions which sped up all implementations i.e. Intel SHA alone while doubled performance for small sizes, the other changes increased everything roughly 50%. +Thanks to Kristofer Peterson, we also added additional performance changes such as optimized padding, +endian conversions which sped up all implementations i.e. Intel SHA alone while doubled performance for small sizes, +the other changes increased everything roughly 50%. ## Support for AVX512 @@ -58,7 +64,8 @@ More detailed information can be found in this [blog](https://blog.minio.io/acce ## Drop-In Replacement -The following code snippet shows how you can use `github.com/minio/sha256-simd`. This will automatically select the fastest method for the architecture on which it will be executed. +The following code snippet shows how you can use `github.com/minio/sha256-simd`. +This will automatically select the fastest method for the architecture on which it will be executed. ```go import "github.com/minio/sha256-simd" @@ -80,9 +87,6 @@ Below is the speed in MB/s for a single core (ranked fast to slow) for blocks la | 3.0 GHz Intel Xeon Platinum 8124M | AVX512 | 3498 | | 3.7 GHz AMD Ryzen 7 2700X | SHA Ext | 1979 | | 1.2 GHz ARM Cortex-A53 | ARM64 | 638 | -| 3.0 GHz Intel Xeon Platinum 8124M | AVX2 | 449 | -| 3.1 GHz Intel Core i7 | AVX | 362 | -| 3.1 GHz Intel Core i7 | SSE | 299 | ## asm2plan9s diff --git a/vendor/github.com/minio/sha256-simd/appveyor.yml b/vendor/github.com/minio/sha256-simd/appveyor.yml deleted file mode 100644 index a66bfa9f2..000000000 --- a/vendor/github.com/minio/sha256-simd/appveyor.yml +++ /dev/null @@ -1,32 +0,0 @@ -# version format -version: "{build}" - -# Operating system (build VM template) -os: Windows Server 2012 R2 - -# Platform. -platform: x64 - -clone_folder: c:\gopath\src\github.com\minio\sha256-simd - -# environment variables -environment: - GOPATH: c:\gopath - GO15VENDOREXPERIMENT: 1 - -# scripts that run after cloning repository -install: - - set PATH=%GOPATH%\bin;c:\go\bin;%PATH% - - go version - - go env - -# to run your custom scripts instead of automatic MSBuild -build_script: - - go test . - - go test -race . - -# to disable automatic tests -test: off - -# to disable deployment -deploy: off diff --git a/vendor/github.com/minio/sha256-simd/cpuid.go b/vendor/github.com/minio/sha256-simd/cpuid.go deleted file mode 100644 index 878ad4638..000000000 --- a/vendor/github.com/minio/sha256-simd/cpuid.go +++ /dev/null @@ -1,119 +0,0 @@ -// Minio Cloud Storage, (C) 2016 Minio, Inc. -// -// Licensed under the Apache License, Version 2.0 (the "License"); -// you may not use this file except in compliance with the License. -// You may obtain a copy of the License at -// -// http://www.apache.org/licenses/LICENSE-2.0 -// -// Unless required by applicable law or agreed to in writing, software -// distributed under the License is distributed on an "AS IS" BASIS, -// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -// See the License for the specific language governing permissions and -// limitations under the License. -// - -package sha256 - -// True when SIMD instructions are available. -var avx512 bool -var avx2 bool -var avx bool -var sse bool -var sse2 bool -var sse3 bool -var ssse3 bool -var sse41 bool -var sse42 bool -var popcnt bool -var sha bool -var armSha = haveArmSha() - -func init() { - var _xsave bool - var _osxsave bool - var _avx bool - var _avx2 bool - var _avx512f bool - var _avx512dq bool - // var _avx512pf bool - // var _avx512er bool - // var _avx512cd bool - var _avx512bw bool - var _avx512vl bool - var _sseState bool - var _avxState bool - var _opmaskState bool - var _zmmHI256State bool - var _hi16ZmmState bool - - mfi, _, _, _ := cpuid(0) - - if mfi >= 1 { - _, _, c, d := cpuid(1) - - sse = (d & (1 << 25)) != 0 - sse2 = (d & (1 << 26)) != 0 - sse3 = (c & (1 << 0)) != 0 - ssse3 = (c & (1 << 9)) != 0 - sse41 = (c & (1 << 19)) != 0 - sse42 = (c & (1 << 20)) != 0 - popcnt = (c & (1 << 23)) != 0 - _xsave = (c & (1 << 26)) != 0 - _osxsave = (c & (1 << 27)) != 0 - _avx = (c & (1 << 28)) != 0 - } - - if mfi >= 7 { - _, b, _, _ := cpuid(7) - - _avx2 = (b & (1 << 5)) != 0 - _avx512f = (b & (1 << 16)) != 0 - _avx512dq = (b & (1 << 17)) != 0 - // _avx512pf = (b & (1 << 26)) != 0 - // _avx512er = (b & (1 << 27)) != 0 - // _avx512cd = (b & (1 << 28)) != 0 - _avx512bw = (b & (1 << 30)) != 0 - _avx512vl = (b & (1 << 31)) != 0 - sha = (b & (1 << 29)) != 0 - } - - // Stop here if XSAVE unsupported or not enabled - if !_xsave || !_osxsave { - return - } - - if _xsave && _osxsave { - a, _ := xgetbv(0) - - _sseState = (a & (1 << 1)) != 0 - _avxState = (a & (1 << 2)) != 0 - _opmaskState = (a & (1 << 5)) != 0 - _zmmHI256State = (a & (1 << 6)) != 0 - _hi16ZmmState = (a & (1 << 7)) != 0 - } else { - _sseState = true - } - - // Very unlikely that OS would enable XSAVE and then disable SSE - if !_sseState { - sse = false - sse2 = false - sse3 = false - ssse3 = false - sse41 = false - sse42 = false - } - - if _avxState { - avx = _avx - avx2 = _avx2 - } - - if _opmaskState && _zmmHI256State && _hi16ZmmState { - avx512 = (_avx512f && - _avx512dq && - _avx512bw && - _avx512vl) - } -} diff --git a/vendor/github.com/minio/sha256-simd/cpuid_386.go b/vendor/github.com/minio/sha256-simd/cpuid_386.go deleted file mode 100644 index c9890be47..000000000 --- a/vendor/github.com/minio/sha256-simd/cpuid_386.go +++ /dev/null @@ -1,24 +0,0 @@ -// Minio Cloud Storage, (C) 2016 Minio, Inc. -// -// Licensed under the Apache License, Version 2.0 (the "License"); -// you may not use this file except in compliance with the License. -// You may obtain a copy of the License at -// -// http://www.apache.org/licenses/LICENSE-2.0 -// -// Unless required by applicable law or agreed to in writing, software -// distributed under the License is distributed on an "AS IS" BASIS, -// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -// See the License for the specific language governing permissions and -// limitations under the License. -// - -package sha256 - -func cpuid(op uint32) (eax, ebx, ecx, edx uint32) -func cpuidex(op, op2 uint32) (eax, ebx, ecx, edx uint32) -func xgetbv(index uint32) (eax, edx uint32) - -func haveArmSha() bool { - return false -} diff --git a/vendor/github.com/minio/sha256-simd/cpuid_386.s b/vendor/github.com/minio/sha256-simd/cpuid_386.s deleted file mode 100644 index 1511cd6f6..000000000 --- a/vendor/github.com/minio/sha256-simd/cpuid_386.s +++ /dev/null @@ -1,53 +0,0 @@ -// The MIT License (MIT) -// -// Copyright (c) 2015 Klaus Post -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE -// SOFTWARE. - -// +build 386,!gccgo - -// func cpuid(op uint32) (eax, ebx, ecx, edx uint32) -TEXT ·cpuid(SB), 7, $0 - XORL CX, CX - MOVL op+0(FP), AX - CPUID - MOVL AX, eax+4(FP) - MOVL BX, ebx+8(FP) - MOVL CX, ecx+12(FP) - MOVL DX, edx+16(FP) - RET - -// func cpuidex(op, op2 uint32) (eax, ebx, ecx, edx uint32) -TEXT ·cpuidex(SB), 7, $0 - MOVL op+0(FP), AX - MOVL op2+4(FP), CX - CPUID - MOVL AX, eax+8(FP) - MOVL BX, ebx+12(FP) - MOVL CX, ecx+16(FP) - MOVL DX, edx+20(FP) - RET - -// func xgetbv(index uint32) (eax, edx uint32) -TEXT ·xgetbv(SB), 7, $0 - MOVL index+0(FP), CX - BYTE $0x0f; BYTE $0x01; BYTE $0xd0 // XGETBV - MOVL AX, eax+4(FP) - MOVL DX, edx+8(FP) - RET diff --git a/vendor/github.com/minio/sha256-simd/cpuid_amd64.go b/vendor/github.com/minio/sha256-simd/cpuid_amd64.go deleted file mode 100644 index c9890be47..000000000 --- a/vendor/github.com/minio/sha256-simd/cpuid_amd64.go +++ /dev/null @@ -1,24 +0,0 @@ -// Minio Cloud Storage, (C) 2016 Minio, Inc. -// -// Licensed under the Apache License, Version 2.0 (the "License"); -// you may not use this file except in compliance with the License. -// You may obtain a copy of the License at -// -// http://www.apache.org/licenses/LICENSE-2.0 -// -// Unless required by applicable law or agreed to in writing, software -// distributed under the License is distributed on an "AS IS" BASIS, -// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -// See the License for the specific language governing permissions and -// limitations under the License. -// - -package sha256 - -func cpuid(op uint32) (eax, ebx, ecx, edx uint32) -func cpuidex(op, op2 uint32) (eax, ebx, ecx, edx uint32) -func xgetbv(index uint32) (eax, edx uint32) - -func haveArmSha() bool { - return false -} diff --git a/vendor/github.com/minio/sha256-simd/cpuid_amd64.s b/vendor/github.com/minio/sha256-simd/cpuid_amd64.s deleted file mode 100644 index b0f414748..000000000 --- a/vendor/github.com/minio/sha256-simd/cpuid_amd64.s +++ /dev/null @@ -1,53 +0,0 @@ -// The MIT License (MIT) -// -// Copyright (c) 2015 Klaus Post -// -// Permission is hereby granted, free of charge, to any person obtaining a copy -// of this software and associated documentation files (the "Software"), to deal -// in the Software without restriction, including without limitation the rights -// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -// copies of the Software, and to permit persons to whom the Software is -// furnished to do so, subject to the following conditions: -// -// The above copyright notice and this permission notice shall be included in all -// copies or substantial portions of the Software. -// -// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE -// SOFTWARE. - -// +build amd64,!gccgo - -// func cpuid(op uint32) (eax, ebx, ecx, edx uint32) -TEXT ·cpuid(SB), 7, $0 - XORQ CX, CX - MOVL op+0(FP), AX - CPUID - MOVL AX, eax+8(FP) - MOVL BX, ebx+12(FP) - MOVL CX, ecx+16(FP) - MOVL DX, edx+20(FP) - RET - -// func cpuidex(op, op2 uint32) (eax, ebx, ecx, edx uint32) -TEXT ·cpuidex(SB), 7, $0 - MOVL op+0(FP), AX - MOVL op2+4(FP), CX - CPUID - MOVL AX, eax+8(FP) - MOVL BX, ebx+12(FP) - MOVL CX, ecx+16(FP) - MOVL DX, edx+20(FP) - RET - -// func xgetbv(index uint32) (eax, edx uint32) -TEXT ·xgetbv(SB), 7, $0 - MOVL index+0(FP), CX - BYTE $0x0f; BYTE $0x01; BYTE $0xd0 // XGETBV - MOVL AX, eax+8(FP) - MOVL DX, edx+12(FP) - RET diff --git a/vendor/github.com/minio/sha256-simd/cpuid_arm.go b/vendor/github.com/minio/sha256-simd/cpuid_arm.go deleted file mode 100644 index 351dff4b6..000000000 --- a/vendor/github.com/minio/sha256-simd/cpuid_arm.go +++ /dev/null @@ -1,32 +0,0 @@ -// Minio Cloud Storage, (C) 2016 Minio, Inc. -// -// Licensed under the Apache License, Version 2.0 (the "License"); -// you may not use this file except in compliance with the License. -// You may obtain a copy of the License at -// -// http://www.apache.org/licenses/LICENSE-2.0 -// -// Unless required by applicable law or agreed to in writing, software -// distributed under the License is distributed on an "AS IS" BASIS, -// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -// See the License for the specific language governing permissions and -// limitations under the License. -// - -package sha256 - -func cpuid(op uint32) (eax, ebx, ecx, edx uint32) { - return 0, 0, 0, 0 -} - -func cpuidex(op, op2 uint32) (eax, ebx, ecx, edx uint32) { - return 0, 0, 0, 0 -} - -func xgetbv(index uint32) (eax, edx uint32) { - return 0, 0 -} - -func haveArmSha() bool { - return false -} diff --git a/vendor/github.com/minio/sha256-simd/cpuid_linux_arm64.go b/vendor/github.com/minio/sha256-simd/cpuid_linux_arm64.go deleted file mode 100644 index e739996d9..000000000 --- a/vendor/github.com/minio/sha256-simd/cpuid_linux_arm64.go +++ /dev/null @@ -1,49 +0,0 @@ -// +build arm64,linux - -// Minio Cloud Storage, (C) 2016 Minio, Inc. -// -// Licensed under the Apache License, Version 2.0 (the "License"); -// you may not use this file except in compliance with the License. -// You may obtain a copy of the License at -// -// http://www.apache.org/licenses/LICENSE-2.0 -// -// Unless required by applicable law or agreed to in writing, software -// distributed under the License is distributed on an "AS IS" BASIS, -// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -// See the License for the specific language governing permissions and -// limitations under the License. -// - -package sha256 - -import ( - "bytes" - "io/ioutil" -) - -func cpuid(op uint32) (eax, ebx, ecx, edx uint32) { - return 0, 0, 0, 0 -} - -func cpuidex(op, op2 uint32) (eax, ebx, ecx, edx uint32) { - return 0, 0, 0, 0 -} - -func xgetbv(index uint32) (eax, edx uint32) { - return 0, 0 -} - -// File to check for cpu capabilities. -const procCPUInfo = "/proc/cpuinfo" - -// Feature to check for. -const sha256Feature = "sha2" - -func haveArmSha() bool { - cpuInfo, err := ioutil.ReadFile(procCPUInfo) - if err != nil { - return false - } - return bytes.Contains(cpuInfo, []byte(sha256Feature)) -} diff --git a/vendor/github.com/minio/sha256-simd/cpuid_other.go b/vendor/github.com/minio/sha256-simd/cpuid_other.go index 3e4415828..cd9fbf2d9 100644 --- a/vendor/github.com/minio/sha256-simd/cpuid_other.go +++ b/vendor/github.com/minio/sha256-simd/cpuid_other.go @@ -1,4 +1,4 @@ -// Minio Cloud Storage, (C) 2016 Minio, Inc. +// Minio Cloud Storage, (C) 2021 Minio, Inc. // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. @@ -13,22 +13,34 @@ // limitations under the License. // -// +build !386,!amd64,!arm,!arm64 arm64,!linux - package sha256 -func cpuid(op uint32) (eax, ebx, ecx, edx uint32) { - return 0, 0, 0, 0 -} +import ( + "bytes" + "io/ioutil" + "runtime" -func cpuidex(op, op2 uint32) (eax, ebx, ecx, edx uint32) { - return 0, 0, 0, 0 -} + "github.com/klauspost/cpuid/v2" +) -func xgetbv(index uint32) (eax, edx uint32) { - return 0, 0 -} +func hasArmSha2() bool { + if cpuid.CPU.Has(cpuid.SHA2) { + return true + } + if runtime.GOARCH != "arm64" || runtime.GOOS != "linux" { + return false + } + + // Fall back to hacky cpuinfo parsing... + const procCPUInfo = "/proc/cpuinfo" + + // Feature to check for. + const sha256Feature = "sha2" + + cpuInfo, err := ioutil.ReadFile(procCPUInfo) + if err != nil { + return false + } + return bytes.Contains(cpuInfo, []byte(sha256Feature)) -func haveArmSha() bool { - return false } diff --git a/vendor/github.com/minio/sha256-simd/sha256.go b/vendor/github.com/minio/sha256-simd/sha256.go index 4e1f6d2f7..b137ead9f 100644 --- a/vendor/github.com/minio/sha256-simd/sha256.go +++ b/vendor/github.com/minio/sha256-simd/sha256.go @@ -21,6 +21,8 @@ import ( "encoding/binary" "hash" "runtime" + + "github.com/klauspost/cpuid/v2" ) // Size - The size of a SHA256 checksum in bytes. @@ -67,10 +69,6 @@ type blockfuncType int const ( blockfuncGeneric blockfuncType = iota - blockfuncAvx512 blockfuncType = iota - blockfuncAvx2 blockfuncType = iota - blockfuncAvx blockfuncType = iota - blockfuncSsse blockfuncType = iota blockfuncSha blockfuncType = iota blockfuncArm blockfuncType = iota ) @@ -78,26 +76,24 @@ const ( var blockfunc blockfuncType func init() { - is386bit := runtime.GOARCH == "386" - isARM := runtime.GOARCH == "arm" + blockfunc = blockfuncGeneric switch { - case is386bit || isARM: - blockfunc = blockfuncGeneric - case sha && ssse3 && sse41: + case hasSHAExtensions(): blockfunc = blockfuncSha - case avx2: - blockfunc = blockfuncAvx2 - case avx: - blockfunc = blockfuncAvx - case ssse3: - blockfunc = blockfuncSsse - case armSha: + case hasArmSha2(): blockfunc = blockfuncArm default: blockfunc = blockfuncGeneric } } +var avx512 = cpuid.CPU.Supports(cpuid.AVX512F, cpuid.AVX512DQ, cpuid.AVX512BW, cpuid.AVX512VL) + +// hasSHAExtensions return whether the cpu supports SHA extensions. +func hasSHAExtensions() bool { + return cpuid.CPU.Supports(cpuid.SHA, cpuid.SSSE3, cpuid.SSE4) && runtime.GOARCH == "amd64" +} + // New returns a new hash.Hash computing the SHA256 checksum. func New() hash.Hash { if blockfunc != blockfuncGeneric { @@ -278,12 +274,6 @@ func (d *digest) checkSum() (digest [Size]byte) { func block(dig *digest, p []byte) { if blockfunc == blockfuncSha { blockShaGo(dig, p) - } else if blockfunc == blockfuncAvx2 { - blockAvx2Go(dig, p) - } else if blockfunc == blockfuncAvx { - blockAvxGo(dig, p) - } else if blockfunc == blockfuncSsse { - blockSsseGo(dig, p) } else if blockfunc == blockfuncArm { blockArmGo(dig, p) } else if blockfunc == blockfuncGeneric { diff --git a/vendor/github.com/minio/sha256-simd/sha256blockAvx2_amd64.go b/vendor/github.com/minio/sha256-simd/sha256blockAvx2_amd64.go deleted file mode 100644 index 52fcaee6d..000000000 --- a/vendor/github.com/minio/sha256-simd/sha256blockAvx2_amd64.go +++ /dev/null @@ -1,22 +0,0 @@ -//+build !noasm,!appengine - -/* - * Minio Cloud Storage, (C) 2016 Minio, Inc. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -package sha256 - -//go:noescape -func blockAvx2(h []uint32, message []uint8) diff --git a/vendor/github.com/minio/sha256-simd/sha256blockAvx2_amd64.s b/vendor/github.com/minio/sha256-simd/sha256blockAvx2_amd64.s deleted file mode 100644 index 80b0b739b..000000000 --- a/vendor/github.com/minio/sha256-simd/sha256blockAvx2_amd64.s +++ /dev/null @@ -1,1449 +0,0 @@ -//+build !noasm,!appengine - -// SHA256 implementation for AVX2 - -// -// Minio Cloud Storage, (C) 2016 Minio, Inc. -// -// Licensed under the Apache License, Version 2.0 (the "License"); -// you may not use this file except in compliance with the License. -// You may obtain a copy of the License at -// -// http://www.apache.org/licenses/LICENSE-2.0 -// -// Unless required by applicable law or agreed to in writing, software -// distributed under the License is distributed on an "AS IS" BASIS, -// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -// See the License for the specific language governing permissions and -// limitations under the License. -// - -// -// This code is based on an Intel White-Paper: -// "Fast SHA-256 Implementations on Intel Architecture Processors" -// -// together with the reference implementation from the following authors: -// James Guilford <james.guilford@intel.com> -// Kirk Yap <kirk.s.yap@intel.com> -// Tim Chen <tim.c.chen@linux.intel.com> -// -// For Golang it has been converted to Plan 9 assembly with the help of -// github.com/minio/asm2plan9s to assemble Intel instructions to their Plan9 -// equivalents -// - -DATA K256<>+0x000(SB)/8, $0x71374491428a2f98 -DATA K256<>+0x008(SB)/8, $0xe9b5dba5b5c0fbcf -DATA K256<>+0x010(SB)/8, $0x71374491428a2f98 -DATA K256<>+0x018(SB)/8, $0xe9b5dba5b5c0fbcf -DATA K256<>+0x020(SB)/8, $0x59f111f13956c25b -DATA K256<>+0x028(SB)/8, $0xab1c5ed5923f82a4 -DATA K256<>+0x030(SB)/8, $0x59f111f13956c25b -DATA K256<>+0x038(SB)/8, $0xab1c5ed5923f82a4 -DATA K256<>+0x040(SB)/8, $0x12835b01d807aa98 -DATA K256<>+0x048(SB)/8, $0x550c7dc3243185be -DATA K256<>+0x050(SB)/8, $0x12835b01d807aa98 -DATA K256<>+0x058(SB)/8, $0x550c7dc3243185be -DATA K256<>+0x060(SB)/8, $0x80deb1fe72be5d74 -DATA K256<>+0x068(SB)/8, $0xc19bf1749bdc06a7 -DATA K256<>+0x070(SB)/8, $0x80deb1fe72be5d74 -DATA K256<>+0x078(SB)/8, $0xc19bf1749bdc06a7 -DATA K256<>+0x080(SB)/8, $0xefbe4786e49b69c1 -DATA K256<>+0x088(SB)/8, $0x240ca1cc0fc19dc6 -DATA K256<>+0x090(SB)/8, $0xefbe4786e49b69c1 -DATA K256<>+0x098(SB)/8, $0x240ca1cc0fc19dc6 -DATA K256<>+0x0a0(SB)/8, $0x4a7484aa2de92c6f -DATA K256<>+0x0a8(SB)/8, $0x76f988da5cb0a9dc -DATA K256<>+0x0b0(SB)/8, $0x4a7484aa2de92c6f -DATA K256<>+0x0b8(SB)/8, $0x76f988da5cb0a9dc -DATA K256<>+0x0c0(SB)/8, $0xa831c66d983e5152 -DATA K256<>+0x0c8(SB)/8, $0xbf597fc7b00327c8 -DATA K256<>+0x0d0(SB)/8, $0xa831c66d983e5152 -DATA K256<>+0x0d8(SB)/8, $0xbf597fc7b00327c8 -DATA K256<>+0x0e0(SB)/8, $0xd5a79147c6e00bf3 -DATA K256<>+0x0e8(SB)/8, $0x1429296706ca6351 -DATA K256<>+0x0f0(SB)/8, $0xd5a79147c6e00bf3 -DATA K256<>+0x0f8(SB)/8, $0x1429296706ca6351 -DATA K256<>+0x100(SB)/8, $0x2e1b213827b70a85 -DATA K256<>+0x108(SB)/8, $0x53380d134d2c6dfc -DATA K256<>+0x110(SB)/8, $0x2e1b213827b70a85 -DATA K256<>+0x118(SB)/8, $0x53380d134d2c6dfc -DATA K256<>+0x120(SB)/8, $0x766a0abb650a7354 -DATA K256<>+0x128(SB)/8, $0x92722c8581c2c92e -DATA K256<>+0x130(SB)/8, $0x766a0abb650a7354 -DATA K256<>+0x138(SB)/8, $0x92722c8581c2c92e -DATA K256<>+0x140(SB)/8, $0xa81a664ba2bfe8a1 -DATA K256<>+0x148(SB)/8, $0xc76c51a3c24b8b70 -DATA K256<>+0x150(SB)/8, $0xa81a664ba2bfe8a1 -DATA K256<>+0x158(SB)/8, $0xc76c51a3c24b8b70 -DATA K256<>+0x160(SB)/8, $0xd6990624d192e819 -DATA K256<>+0x168(SB)/8, $0x106aa070f40e3585 -DATA K256<>+0x170(SB)/8, $0xd6990624d192e819 -DATA K256<>+0x178(SB)/8, $0x106aa070f40e3585 -DATA K256<>+0x180(SB)/8, $0x1e376c0819a4c116 -DATA K256<>+0x188(SB)/8, $0x34b0bcb52748774c -DATA K256<>+0x190(SB)/8, $0x1e376c0819a4c116 -DATA K256<>+0x198(SB)/8, $0x34b0bcb52748774c -DATA K256<>+0x1a0(SB)/8, $0x4ed8aa4a391c0cb3 -DATA K256<>+0x1a8(SB)/8, $0x682e6ff35b9cca4f -DATA K256<>+0x1b0(SB)/8, $0x4ed8aa4a391c0cb3 -DATA K256<>+0x1b8(SB)/8, $0x682e6ff35b9cca4f -DATA K256<>+0x1c0(SB)/8, $0x78a5636f748f82ee -DATA K256<>+0x1c8(SB)/8, $0x8cc7020884c87814 -DATA K256<>+0x1d0(SB)/8, $0x78a5636f748f82ee -DATA K256<>+0x1d8(SB)/8, $0x8cc7020884c87814 -DATA K256<>+0x1e0(SB)/8, $0xa4506ceb90befffa -DATA K256<>+0x1e8(SB)/8, $0xc67178f2bef9a3f7 -DATA K256<>+0x1f0(SB)/8, $0xa4506ceb90befffa -DATA K256<>+0x1f8(SB)/8, $0xc67178f2bef9a3f7 - -DATA K256<>+0x200(SB)/8, $0x0405060700010203 -DATA K256<>+0x208(SB)/8, $0x0c0d0e0f08090a0b -DATA K256<>+0x210(SB)/8, $0x0405060700010203 -DATA K256<>+0x218(SB)/8, $0x0c0d0e0f08090a0b -DATA K256<>+0x220(SB)/8, $0x0b0a090803020100 -DATA K256<>+0x228(SB)/8, $0xffffffffffffffff -DATA K256<>+0x230(SB)/8, $0x0b0a090803020100 -DATA K256<>+0x238(SB)/8, $0xffffffffffffffff -DATA K256<>+0x240(SB)/8, $0xffffffffffffffff -DATA K256<>+0x248(SB)/8, $0x0b0a090803020100 -DATA K256<>+0x250(SB)/8, $0xffffffffffffffff -DATA K256<>+0x258(SB)/8, $0x0b0a090803020100 - -GLOBL K256<>(SB), 8, $608 - -// We need 0x220 stack space aligned on a 512 boundary, so for the -// worstcase-aligned SP we need twice this amount, being 1088 (=0x440) -// -// SP aligned end-aligned stacksize -// 100013d0 10001400 10001620 592 -// 100013d8 10001400 10001620 584 -// 100013e0 10001600 10001820 1088 -// 100013e8 10001600 10001820 1080 - -// func blockAvx2(h []uint32, message []uint8) -TEXT ·blockAvx2(SB),$1088-48 - - MOVQ h+0(FP), DI // DI: &h - MOVQ message_base+24(FP), SI // SI: &message - MOVQ message_len+32(FP), DX // len(message) - ADDQ SI, DX // end pointer of input - MOVQ SP, R11 // copy stack pointer - ADDQ $0x220, SP // sp += 0x220 - ANDQ $0xfffffffffffffe00, SP // align stack frame - ADDQ $0x1c0, SP - MOVQ DI, 0x40(SP) // save ctx - MOVQ SI, 0x48(SP) // save input - MOVQ DX, 0x50(SP) // save end pointer - MOVQ R11, 0x58(SP) // save copy of stack pointer - - WORD $0xf8c5; BYTE $0x77 // vzeroupper - ADDQ $0x40, SI // input++ - MOVL (DI), AX - MOVQ SI, R12 // borrow $T1 - MOVL 4(DI), BX - CMPQ SI, DX // $_end - MOVL 8(DI), CX - LONG $0xe4440f4c // cmove r12,rsp /* next block or random data */ - MOVL 12(DI), DX - MOVL 16(DI), R8 - MOVL 20(DI), R9 - MOVL 24(DI), R10 - MOVL 28(DI), R11 - - LEAQ K256<>(SB), BP - LONG $0x856f7dc5; LONG $0x00000220 // VMOVDQA YMM8, 0x220[rbp] /* vmovdqa ymm8,YMMWORD PTR [rip+0x220] */ - LONG $0x8d6f7dc5; LONG $0x00000240 // VMOVDQA YMM9, 0x240[rbp] /* vmovdqa ymm9,YMMWORD PTR [rip+0x240] */ - LONG $0x956f7dc5; LONG $0x00000200 // VMOVDQA YMM10, 0x200[rbp] /* vmovdqa ymm7,YMMWORD PTR [rip+0x200] */ - -loop0: - LONG $0x6f7dc1c4; BYTE $0xfa // VMOVDQA YMM7, YMM10 - - // Load first 16 dwords from two blocks - MOVOU -64(SI), X0 // vmovdqu xmm0,XMMWORD PTR [rsi-0x40] - MOVOU -48(SI), X1 // vmovdqu xmm1,XMMWORD PTR [rsi-0x30] - MOVOU -32(SI), X2 // vmovdqu xmm2,XMMWORD PTR [rsi-0x20] - MOVOU -16(SI), X3 // vmovdqu xmm3,XMMWORD PTR [rsi-0x10] - - // Byte swap data and transpose data into high/low - LONG $0x387dc3c4; WORD $0x2404; BYTE $0x01 // vinserti128 ymm0,ymm0,[r12],0x1 - LONG $0x3875c3c4; LONG $0x0110244c // vinserti128 ymm1,ymm1,0x10[r12],0x1 - LONG $0x007de2c4; BYTE $0xc7 // vpshufb ymm0,ymm0,ymm7 - LONG $0x386dc3c4; LONG $0x01202454 // vinserti128 ymm2,ymm2,0x20[r12],0x1 - LONG $0x0075e2c4; BYTE $0xcf // vpshufb ymm1,ymm1,ymm7 - LONG $0x3865c3c4; LONG $0x0130245c // vinserti128 ymm3,ymm3,0x30[r12],0x1 - - LEAQ K256<>(SB), BP - LONG $0x006de2c4; BYTE $0xd7 // vpshufb ymm2,ymm2,ymm7 - LONG $0x65fefdc5; BYTE $0x00 // vpaddd ymm4,ymm0,[rbp] - LONG $0x0065e2c4; BYTE $0xdf // vpshufb ymm3,ymm3,ymm7 - LONG $0x6dfef5c5; BYTE $0x20 // vpaddd ymm5,ymm1,0x20[rbp] - LONG $0x75feedc5; BYTE $0x40 // vpaddd ymm6,ymm2,0x40[rbp] - LONG $0x7dfee5c5; BYTE $0x60 // vpaddd ymm7,ymm3,0x60[rbp] - - LONG $0x247ffdc5; BYTE $0x24 // vmovdqa [rsp],ymm4 - XORQ R14, R14 - LONG $0x6c7ffdc5; WORD $0x2024 // vmovdqa [rsp+0x20],ymm5 - - ADDQ $-0x40, SP - MOVQ BX, DI - LONG $0x347ffdc5; BYTE $0x24 // vmovdqa [rsp],ymm6 - XORQ CX, DI // magic - LONG $0x7c7ffdc5; WORD $0x2024 // vmovdqa [rsp+0x20],ymm7 - MOVQ R9, R12 - ADDQ $0x80, BP - -loop1: - // Schedule 48 input dwords, by doing 3 rounds of 12 each - // Note: SIMD instructions are interleaved with the SHA calculations - ADDQ $-0x40, SP - LONG $0x0f75e3c4; WORD $0x04e0 // vpalignr ymm4,ymm1,ymm0,0x4 - - // ROUND(AX, BX, CX, DX, R8, R9, R10, R11, R12, R13, R14, R15, DI, SP, 0x80) - LONG $0x249c0344; LONG $0x00000080 // add r11d,[rsp+0x80] - WORD $0x2145; BYTE $0xc4 // and r12d,r8d - LONG $0xf07b43c4; WORD $0x19e8 // rorx r13d,r8d,0x19 - LONG $0x0f65e3c4; WORD $0x04fa // vpalignr ymm7,ymm3,ymm2,0x4 - LONG $0xf07b43c4; WORD $0x0bf8 // rorx r15d,r8d,0xb - LONG $0x30048d42 // lea eax,[rax+r14*1] - LONG $0x231c8d47 // lea r11d,[r11+r12*1] - LONG $0xd472cdc5; BYTE $0x07 // vpsrld ymm6,ymm4,0x7 - LONG $0xf23842c4; BYTE $0xe2 // andn r12d,r8d,r10d - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b43c4; WORD $0x06f0 // rorx r14d,r8d,0x6 - LONG $0xc7fefdc5 // vpaddd ymm0,ymm0,ymm7 - LONG $0x231c8d47 // lea r11d,[r11+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8941; BYTE $0xc7 // mov r15d,eax - LONG $0xd472c5c5; BYTE $0x03 // vpsrld ymm7,ymm4,0x3 - LONG $0xf07b63c4; WORD $0x16e0 // rorx r12d,eax,0x16 - LONG $0x2b1c8d47 // lea r11d,[r11+r13*1] - WORD $0x3141; BYTE $0xdf // xor r15d,ebx - LONG $0xf472d5c5; BYTE $0x0e // vpslld ymm5,ymm4,0xe - LONG $0xf07b63c4; WORD $0x0df0 // rorx r14d,eax,0xd - LONG $0xf07b63c4; WORD $0x02e8 // rorx r13d,eax,0x2 - LONG $0x1a148d42 // lea edx,[rdx+r11*1] - LONG $0xe6efc5c5 // vpxor ymm4,ymm7,ymm6 - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0xdf31 // xor edi,ebx - LONG $0xfb70fdc5; BYTE $0xfa // vpshufd ymm7,ymm3,0xfa - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x3b1c8d45 // lea r11d,[r11+rdi*1] - WORD $0x8945; BYTE $0xc4 // mov r12d,r8d - LONG $0xd672cdc5; BYTE $0x0b // vpsrld ymm6,ymm6,0xb - - // ROUND(R11, AX, BX, CX, DX, R8, R9, R10, R12, R13, R14, DI, R15, SP, 0x84) - LONG $0x24940344; LONG $0x00000084 // add r10d,[rsp+0x84] - WORD $0x2141; BYTE $0xd4 // and r12d,edx - LONG $0xf07b63c4; WORD $0x19ea // rorx r13d,edx,0x19 - LONG $0xe5efddc5 // vpxor ymm4,ymm4,ymm5 - LONG $0xf07be3c4; WORD $0x0bfa // rorx edi,edx,0xb - LONG $0x331c8d47 // lea r11d,[r11+r14*1] - LONG $0x22148d47 // lea r10d,[r10+r12*1] - LONG $0xf572d5c5; BYTE $0x0b // vpslld ymm5,ymm5,0xb - LONG $0xf26842c4; BYTE $0xe1 // andn r12d,edx,r9d - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b63c4; WORD $0x06f2 // rorx r14d,edx,0x6 - LONG $0xe6efddc5 // vpxor ymm4,ymm4,ymm6 - LONG $0x22148d47 // lea r10d,[r10+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8944; BYTE $0xdf // mov edi,r11d - LONG $0xd772cdc5; BYTE $0x0a // vpsrld ymm6,ymm7,0xa - LONG $0xf07b43c4; WORD $0x16e3 // rorx r12d,r11d,0x16 - LONG $0x2a148d47 // lea r10d,[r10+r13*1] - WORD $0xc731 // xor edi,eax - LONG $0xe5efddc5 // vpxor ymm4,ymm4,ymm5 - LONG $0xf07b43c4; WORD $0x0df3 // rorx r14d,r11d,0xd - LONG $0xf07b43c4; WORD $0x02eb // rorx r13d,r11d,0x2 - LONG $0x110c8d42 // lea ecx,[rcx+r10*1] - LONG $0xd773c5c5; BYTE $0x11 // vpsrlq ymm7,ymm7,0x11 - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3141; BYTE $0xc7 // xor r15d,eax - LONG $0xc4fefdc5 // vpaddd ymm0,ymm0,ymm4 - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x3a148d47 // lea r10d,[r10+r15*1] - WORD $0x8941; BYTE $0xd4 // mov r12d,edx - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - - // ROUND(R10, R11, AX, BX, CX, DX, R8, R9, R12, R13, R14, R15, DI, SP, 0x88) - LONG $0x248c0344; LONG $0x00000088 // add r9d,[rsp+0x88] - WORD $0x2141; BYTE $0xcc // and r12d,ecx - LONG $0xf07b63c4; WORD $0x19e9 // rorx r13d,ecx,0x19 - LONG $0xd773c5c5; BYTE $0x02 // vpsrlq ymm7,ymm7,0x2 - LONG $0xf07b63c4; WORD $0x0bf9 // rorx r15d,ecx,0xb - LONG $0x32148d47 // lea r10d,[r10+r14*1] - LONG $0x210c8d47 // lea r9d,[r9+r12*1] - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - LONG $0xf27042c4; BYTE $0xe0 // andn r12d,ecx,r8d - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b63c4; WORD $0x06f1 // rorx r14d,ecx,0x6 - LONG $0x004dc2c4; BYTE $0xf0 // vpshufb ymm6,ymm6,ymm8 - LONG $0x210c8d47 // lea r9d,[r9+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8945; BYTE $0xd7 // mov r15d,r10d - LONG $0xc6fefdc5 // vpaddd ymm0,ymm0,ymm6 - LONG $0xf07b43c4; WORD $0x16e2 // rorx r12d,r10d,0x16 - LONG $0x290c8d47 // lea r9d,[r9+r13*1] - WORD $0x3145; BYTE $0xdf // xor r15d,r11d - LONG $0xf870fdc5; BYTE $0x50 // vpshufd ymm7,ymm0,0x50 - LONG $0xf07b43c4; WORD $0x0df2 // rorx r14d,r10d,0xd - LONG $0xf07b43c4; WORD $0x02ea // rorx r13d,r10d,0x2 - LONG $0x0b1c8d42 // lea ebx,[rbx+r9*1] - LONG $0xd772cdc5; BYTE $0x0a // vpsrld ymm6,ymm7,0xa - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3144; BYTE $0xdf // xor edi,r11d - LONG $0xd773c5c5; BYTE $0x11 // vpsrlq ymm7,ymm7,0x11 - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x390c8d45 // lea r9d,[r9+rdi*1] - WORD $0x8941; BYTE $0xcc // mov r12d,ecx - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - - // ROUND(R9, R10, R11, AX, BX, CX, DX, R8, R12, R13, R14, DI, R15, SP, 0x8c) - LONG $0x24840344; LONG $0x0000008c // add r8d,[rsp+0x8c] - WORD $0x2141; BYTE $0xdc // and r12d,ebx - LONG $0xf07b63c4; WORD $0x19eb // rorx r13d,ebx,0x19 - LONG $0xd773c5c5; BYTE $0x02 // vpsrlq ymm7,ymm7,0x2 - LONG $0xf07be3c4; WORD $0x0bfb // rorx edi,ebx,0xb - LONG $0x310c8d47 // lea r9d,[r9+r14*1] - LONG $0x20048d47 // lea r8d,[r8+r12*1] - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - LONG $0xf26062c4; BYTE $0xe2 // andn r12d,ebx,edx - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b63c4; WORD $0x06f3 // rorx r14d,ebx,0x6 - LONG $0x004dc2c4; BYTE $0xf1 // vpshufb ymm6,ymm6,ymm9 - LONG $0x20048d47 // lea r8d,[r8+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8944; BYTE $0xcf // mov edi,r9d - LONG $0xc6fefdc5 // vpaddd ymm0,ymm0,ymm6 - LONG $0xf07b43c4; WORD $0x16e1 // rorx r12d,r9d,0x16 - LONG $0x28048d47 // lea r8d,[r8+r13*1] - WORD $0x3144; BYTE $0xd7 // xor edi,r10d - LONG $0x75fefdc5; BYTE $0x00 // vpaddd ymm6,ymm0,[rbp+0x0] - LONG $0xf07b43c4; WORD $0x0df1 // rorx r14d,r9d,0xd - LONG $0xf07b43c4; WORD $0x02e9 // rorx r13d,r9d,0x2 - LONG $0x00048d42 // lea eax,[rax+r8*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3145; BYTE $0xd7 // xor r15d,r10d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x38048d47 // lea r8d,[r8+r15*1] - WORD $0x8941; BYTE $0xdc // mov r12d,ebx - - LONG $0x347ffdc5; BYTE $0x24 // vmovdqa [rsp],ymm6 - LONG $0x0f6de3c4; WORD $0x04e1 // vpalignr ymm4,ymm2,ymm1,0x4 - - // ROUND(R8, R9, R10, R11, AX, BX, CX, DX, R12, R13, R14, R15, DI, SP, 0xa0) - LONG $0xa0249403; WORD $0x0000; BYTE $0x00 // add edx,[rsp+0xa0] - WORD $0x2141; BYTE $0xc4 // and r12d,eax - LONG $0xf07b63c4; WORD $0x19e8 // rorx r13d,eax,0x19 - LONG $0x0f7de3c4; WORD $0x04fb // vpalignr ymm7,ymm0,ymm3,0x4 - LONG $0xf07b63c4; WORD $0x0bf8 // rorx r15d,eax,0xb - LONG $0x30048d47 // lea r8d,[r8+r14*1] - LONG $0x22148d42 // lea edx,[rdx+r12*1] - LONG $0xd472cdc5; BYTE $0x07 // vpsrld ymm6,ymm4,0x7 - LONG $0xf27862c4; BYTE $0xe1 // andn r12d,eax,ecx - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b63c4; WORD $0x06f0 // rorx r14d,eax,0x6 - LONG $0xcffef5c5 // vpaddd ymm1,ymm1,ymm7 - LONG $0x22148d42 // lea edx,[rdx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8945; BYTE $0xc7 // mov r15d,r8d - LONG $0xd472c5c5; BYTE $0x03 // vpsrld ymm7,ymm4,0x3 - LONG $0xf07b43c4; WORD $0x16e0 // rorx r12d,r8d,0x16 - LONG $0x2a148d42 // lea edx,[rdx+r13*1] - WORD $0x3145; BYTE $0xcf // xor r15d,r9d - LONG $0xf472d5c5; BYTE $0x0e // vpslld ymm5,ymm4,0xe - LONG $0xf07b43c4; WORD $0x0df0 // rorx r14d,r8d,0xd - LONG $0xf07b43c4; WORD $0x02e8 // rorx r13d,r8d,0x2 - LONG $0x131c8d45 // lea r11d,[r11+rdx*1] - LONG $0xe6efc5c5 // vpxor ymm4,ymm7,ymm6 - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3144; BYTE $0xcf // xor edi,r9d - LONG $0xf870fdc5; BYTE $0xfa // vpshufd ymm7,ymm0,0xfa - WORD $0x3145; BYTE $0xee // xor r14d,r13d - WORD $0x148d; BYTE $0x3a // lea edx,[rdx+rdi*1] - WORD $0x8941; BYTE $0xc4 // mov r12d,eax - LONG $0xd672cdc5; BYTE $0x0b // vpsrld ymm6,ymm6,0xb - - // ROUND(DX, R8, R9, R10, R11, AX, BX, CX, R12, R13, R14, DI, R15, SP, 0xa4) - LONG $0xa4248c03; WORD $0x0000; BYTE $0x00 // add ecx,[rsp+0xa4] - WORD $0x2145; BYTE $0xdc // and r12d,r11d - LONG $0xf07b43c4; WORD $0x19eb // rorx r13d,r11d,0x19 - LONG $0xe5efddc5 // vpxor ymm4,ymm4,ymm5 - LONG $0xf07bc3c4; WORD $0x0bfb // rorx edi,r11d,0xb - LONG $0x32148d42 // lea edx,[rdx+r14*1] - LONG $0x210c8d42 // lea ecx,[rcx+r12*1] - LONG $0xf572d5c5; BYTE $0x0b // vpslld ymm5,ymm5,0xb - LONG $0xf22062c4; BYTE $0xe3 // andn r12d,r11d,ebx - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b43c4; WORD $0x06f3 // rorx r14d,r11d,0x6 - LONG $0xe6efddc5 // vpxor ymm4,ymm4,ymm6 - LONG $0x210c8d42 // lea ecx,[rcx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0xd789 // mov edi,edx - LONG $0xd772cdc5; BYTE $0x0a // vpsrld ymm6,ymm7,0xa - LONG $0xf07b63c4; WORD $0x16e2 // rorx r12d,edx,0x16 - LONG $0x290c8d42 // lea ecx,[rcx+r13*1] - WORD $0x3144; BYTE $0xc7 // xor edi,r8d - LONG $0xe5efddc5 // vpxor ymm4,ymm4,ymm5 - LONG $0xf07b63c4; WORD $0x0df2 // rorx r14d,edx,0xd - LONG $0xf07b63c4; WORD $0x02ea // rorx r13d,edx,0x2 - LONG $0x0a148d45 // lea r10d,[r10+rcx*1] - LONG $0xd773c5c5; BYTE $0x11 // vpsrlq ymm7,ymm7,0x11 - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3145; BYTE $0xc7 // xor r15d,r8d - LONG $0xccfef5c5 // vpaddd ymm1,ymm1,ymm4 - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x390c8d42 // lea ecx,[rcx+r15*1] - WORD $0x8945; BYTE $0xdc // mov r12d,r11d - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - - // ROUND(CX, DX, R8, R9, R10, R11, AX, BX, R12, R13, R14, R15, DI, SP, 0xa8) - LONG $0xa8249c03; WORD $0x0000; BYTE $0x00 // add ebx,[rsp+0xa8] - WORD $0x2145; BYTE $0xd4 // and r12d,r10d - LONG $0xf07b43c4; WORD $0x19ea // rorx r13d,r10d,0x19 - LONG $0xd773c5c5; BYTE $0x02 // vpsrlq ymm7,ymm7,0x2 - LONG $0xf07b43c4; WORD $0x0bfa // rorx r15d,r10d,0xb - LONG $0x310c8d42 // lea ecx,[rcx+r14*1] - LONG $0x231c8d42 // lea ebx,[rbx+r12*1] - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - LONG $0xf22862c4; BYTE $0xe0 // andn r12d,r10d,eax - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b43c4; WORD $0x06f2 // rorx r14d,r10d,0x6 - LONG $0x004dc2c4; BYTE $0xf0 // vpshufb ymm6,ymm6,ymm8 - LONG $0x231c8d42 // lea ebx,[rbx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8941; BYTE $0xcf // mov r15d,ecx - LONG $0xcefef5c5 // vpaddd ymm1,ymm1,ymm6 - LONG $0xf07b63c4; WORD $0x16e1 // rorx r12d,ecx,0x16 - LONG $0x2b1c8d42 // lea ebx,[rbx+r13*1] - WORD $0x3141; BYTE $0xd7 // xor r15d,edx - LONG $0xf970fdc5; BYTE $0x50 // vpshufd ymm7,ymm1,0x50 - LONG $0xf07b63c4; WORD $0x0df1 // rorx r14d,ecx,0xd - LONG $0xf07b63c4; WORD $0x02e9 // rorx r13d,ecx,0x2 - LONG $0x190c8d45 // lea r9d,[r9+rbx*1] - LONG $0xd772cdc5; BYTE $0x0a // vpsrld ymm6,ymm7,0xa - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0xd731 // xor edi,edx - LONG $0xd773c5c5; BYTE $0x11 // vpsrlq ymm7,ymm7,0x11 - WORD $0x3145; BYTE $0xee // xor r14d,r13d - WORD $0x1c8d; BYTE $0x3b // lea ebx,[rbx+rdi*1] - WORD $0x8945; BYTE $0xd4 // mov r12d,r10d - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - - // ROUND(BX, CX, DX, R8, R9, R10, R11, AX, R12, R13, R14, DI, R15, SP, 0xac) - LONG $0xac248403; WORD $0x0000; BYTE $0x00 // add eax,[rsp+0xac] - WORD $0x2145; BYTE $0xcc // and r12d,r9d - LONG $0xf07b43c4; WORD $0x19e9 // rorx r13d,r9d,0x19 - LONG $0xd773c5c5; BYTE $0x02 // vpsrlq ymm7,ymm7,0x2 - LONG $0xf07bc3c4; WORD $0x0bf9 // rorx edi,r9d,0xb - LONG $0x331c8d42 // lea ebx,[rbx+r14*1] - LONG $0x20048d42 // lea eax,[rax+r12*1] - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - LONG $0xf23042c4; BYTE $0xe3 // andn r12d,r9d,r11d - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b43c4; WORD $0x06f1 // rorx r14d,r9d,0x6 - LONG $0x004dc2c4; BYTE $0xf1 // vpshufb ymm6,ymm6,ymm9 - LONG $0x20048d42 // lea eax,[rax+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0xdf89 // mov edi,ebx - LONG $0xcefef5c5 // vpaddd ymm1,ymm1,ymm6 - LONG $0xf07b63c4; WORD $0x16e3 // rorx r12d,ebx,0x16 - LONG $0x28048d42 // lea eax,[rax+r13*1] - WORD $0xcf31 // xor edi,ecx - LONG $0x75fef5c5; BYTE $0x20 // vpaddd ymm6,ymm1,[rbp+0x20] - LONG $0xf07b63c4; WORD $0x0df3 // rorx r14d,ebx,0xd - LONG $0xf07b63c4; WORD $0x02eb // rorx r13d,ebx,0x2 - LONG $0x00048d45 // lea r8d,[r8+rax*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3141; BYTE $0xcf // xor r15d,ecx - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x38048d42 // lea eax,[rax+r15*1] - WORD $0x8945; BYTE $0xcc // mov r12d,r9d - - LONG $0x747ffdc5; WORD $0x2024 // vmovdqa [rsp+0x20],ymm6 - - LONG $0x24648d48; BYTE $0xc0 // lea rsp,[rsp-0x40] - LONG $0x0f65e3c4; WORD $0x04e2 // vpalignr ymm4,ymm3,ymm2,0x4 - - // ROUND(AX, BX, CX, DX, R8, R9, R10, R11, R12, R13, R14, R15, DI, SP, 0x80) - LONG $0x249c0344; LONG $0x00000080 // add r11d,[rsp+0x80] - WORD $0x2145; BYTE $0xc4 // and r12d,r8d - LONG $0xf07b43c4; WORD $0x19e8 // rorx r13d,r8d,0x19 - LONG $0x0f75e3c4; WORD $0x04f8 // vpalignr ymm7,ymm1,ymm0,0x4 - LONG $0xf07b43c4; WORD $0x0bf8 // rorx r15d,r8d,0xb - LONG $0x30048d42 // lea eax,[rax+r14*1] - LONG $0x231c8d47 // lea r11d,[r11+r12*1] - LONG $0xd472cdc5; BYTE $0x07 // vpsrld ymm6,ymm4,0x7 - LONG $0xf23842c4; BYTE $0xe2 // andn r12d,r8d,r10d - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b43c4; WORD $0x06f0 // rorx r14d,r8d,0x6 - LONG $0xd7feedc5 // vpaddd ymm2,ymm2,ymm7 - LONG $0x231c8d47 // lea r11d,[r11+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8941; BYTE $0xc7 // mov r15d,eax - LONG $0xd472c5c5; BYTE $0x03 // vpsrld ymm7,ymm4,0x3 - LONG $0xf07b63c4; WORD $0x16e0 // rorx r12d,eax,0x16 - LONG $0x2b1c8d47 // lea r11d,[r11+r13*1] - WORD $0x3141; BYTE $0xdf // xor r15d,ebx - LONG $0xf472d5c5; BYTE $0x0e // vpslld ymm5,ymm4,0xe - LONG $0xf07b63c4; WORD $0x0df0 // rorx r14d,eax,0xd - LONG $0xf07b63c4; WORD $0x02e8 // rorx r13d,eax,0x2 - LONG $0x1a148d42 // lea edx,[rdx+r11*1] - LONG $0xe6efc5c5 // vpxor ymm4,ymm7,ymm6 - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0xdf31 // xor edi,ebx - LONG $0xf970fdc5; BYTE $0xfa // vpshufd ymm7,ymm1,0xfa - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x3b1c8d45 // lea r11d,[r11+rdi*1] - WORD $0x8945; BYTE $0xc4 // mov r12d,r8d - LONG $0xd672cdc5; BYTE $0x0b // vpsrld ymm6,ymm6,0xb - - // ROUND(R11, AX, BX, CX, DX, R8, R9, R10, R12, R13, R14, DI, R15, SP, 0x84) - LONG $0x24940344; LONG $0x00000084 // add r10d,[rsp+0x84] - WORD $0x2141; BYTE $0xd4 // and r12d,edx - LONG $0xf07b63c4; WORD $0x19ea // rorx r13d,edx,0x19 - LONG $0xe5efddc5 // vpxor ymm4,ymm4,ymm5 - LONG $0xf07be3c4; WORD $0x0bfa // rorx edi,edx,0xb - LONG $0x331c8d47 // lea r11d,[r11+r14*1] - LONG $0x22148d47 // lea r10d,[r10+r12*1] - LONG $0xf572d5c5; BYTE $0x0b // vpslld ymm5,ymm5,0xb - LONG $0xf26842c4; BYTE $0xe1 // andn r12d,edx,r9d - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b63c4; WORD $0x06f2 // rorx r14d,edx,0x6 - LONG $0xe6efddc5 // vpxor ymm4,ymm4,ymm6 - LONG $0x22148d47 // lea r10d,[r10+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8944; BYTE $0xdf // mov edi,r11d - LONG $0xd772cdc5; BYTE $0x0a // vpsrld ymm6,ymm7,0xa - LONG $0xf07b43c4; WORD $0x16e3 // rorx r12d,r11d,0x16 - LONG $0x2a148d47 // lea r10d,[r10+r13*1] - WORD $0xc731 // xor edi,eax - LONG $0xe5efddc5 // vpxor ymm4,ymm4,ymm5 - LONG $0xf07b43c4; WORD $0x0df3 // rorx r14d,r11d,0xd - LONG $0xf07b43c4; WORD $0x02eb // rorx r13d,r11d,0x2 - LONG $0x110c8d42 // lea ecx,[rcx+r10*1] - LONG $0xd773c5c5; BYTE $0x11 // vpsrlq ymm7,ymm7,0x11 - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3141; BYTE $0xc7 // xor r15d,eax - LONG $0xd4feedc5 // vpaddd ymm2,ymm2,ymm4 - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x3a148d47 // lea r10d,[r10+r15*1] - WORD $0x8941; BYTE $0xd4 // mov r12d,edx - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - - // ROUND(R10, R11, AX, BX, CX, DX, R8, R9, R12, R13, R14, R15, DI, SP, 0x88) - LONG $0x248c0344; LONG $0x00000088 // add r9d,[rsp+0x88] - WORD $0x2141; BYTE $0xcc // and r12d,ecx - LONG $0xf07b63c4; WORD $0x19e9 // rorx r13d,ecx,0x19 - LONG $0xd773c5c5; BYTE $0x02 // vpsrlq ymm7,ymm7,0x2 - LONG $0xf07b63c4; WORD $0x0bf9 // rorx r15d,ecx,0xb - LONG $0x32148d47 // lea r10d,[r10+r14*1] - LONG $0x210c8d47 // lea r9d,[r9+r12*1] - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - LONG $0xf27042c4; BYTE $0xe0 // andn r12d,ecx,r8d - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b63c4; WORD $0x06f1 // rorx r14d,ecx,0x6 - LONG $0x004dc2c4; BYTE $0xf0 // vpshufb ymm6,ymm6,ymm8 - LONG $0x210c8d47 // lea r9d,[r9+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8945; BYTE $0xd7 // mov r15d,r10d - LONG $0xd6feedc5 // vpaddd ymm2,ymm2,ymm6 - LONG $0xf07b43c4; WORD $0x16e2 // rorx r12d,r10d,0x16 - LONG $0x290c8d47 // lea r9d,[r9+r13*1] - WORD $0x3145; BYTE $0xdf // xor r15d,r11d - LONG $0xfa70fdc5; BYTE $0x50 // vpshufd ymm7,ymm2,0x50 - LONG $0xf07b43c4; WORD $0x0df2 // rorx r14d,r10d,0xd - LONG $0xf07b43c4; WORD $0x02ea // rorx r13d,r10d,0x2 - LONG $0x0b1c8d42 // lea ebx,[rbx+r9*1] - LONG $0xd772cdc5; BYTE $0x0a // vpsrld ymm6,ymm7,0xa - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3144; BYTE $0xdf // xor edi,r11d - LONG $0xd773c5c5; BYTE $0x11 // vpsrlq ymm7,ymm7,0x11 - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x390c8d45 // lea r9d,[r9+rdi*1] - WORD $0x8941; BYTE $0xcc // mov r12d,ecx - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - - // ROUND(R9, R10, R11, AX, BX, CX, DX, R8, R12, R13, R14, DI, R15, SP, 0x8c) - LONG $0x24840344; LONG $0x0000008c // add r8d,[rsp+0x8c] - WORD $0x2141; BYTE $0xdc // and r12d,ebx - LONG $0xf07b63c4; WORD $0x19eb // rorx r13d,ebx,0x19 - LONG $0xd773c5c5; BYTE $0x02 // vpsrlq ymm7,ymm7,0x2 - LONG $0xf07be3c4; WORD $0x0bfb // rorx edi,ebx,0xb - LONG $0x310c8d47 // lea r9d,[r9+r14*1] - LONG $0x20048d47 // lea r8d,[r8+r12*1] - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - LONG $0xf26062c4; BYTE $0xe2 // andn r12d,ebx,edx - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b63c4; WORD $0x06f3 // rorx r14d,ebx,0x6 - LONG $0x004dc2c4; BYTE $0xf1 // vpshufb ymm6,ymm6,ymm9 - LONG $0x20048d47 // lea r8d,[r8+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8944; BYTE $0xcf // mov edi,r9d - LONG $0xd6feedc5 // vpaddd ymm2,ymm2,ymm6 - LONG $0xf07b43c4; WORD $0x16e1 // rorx r12d,r9d,0x16 - LONG $0x28048d47 // lea r8d,[r8+r13*1] - WORD $0x3144; BYTE $0xd7 // xor edi,r10d - LONG $0x75feedc5; BYTE $0x40 // vpaddd ymm6,ymm2,[rbp+0x40] - LONG $0xf07b43c4; WORD $0x0df1 // rorx r14d,r9d,0xd - LONG $0xf07b43c4; WORD $0x02e9 // rorx r13d,r9d,0x2 - LONG $0x00048d42 // lea eax,[rax+r8*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3145; BYTE $0xd7 // xor r15d,r10d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x38048d47 // lea r8d,[r8+r15*1] - WORD $0x8941; BYTE $0xdc // mov r12d,ebx - - LONG $0x347ffdc5; BYTE $0x24 // vmovdqa [rsp],ymm6 - LONG $0x0f7de3c4; WORD $0x04e3 // vpalignr ymm4,ymm0,ymm3,0x4 - - // ROUND(R8, R9, R10, R11, AX, BX, CX, DX, R12, R13, R14, R15, DI, SP, 0xa0) - LONG $0xa0249403; WORD $0x0000; BYTE $0x00 // add edx,[rsp+0xa0] - WORD $0x2141; BYTE $0xc4 // and r12d,eax - LONG $0xf07b63c4; WORD $0x19e8 // rorx r13d,eax,0x19 - LONG $0x0f6de3c4; WORD $0x04f9 // vpalignr ymm7,ymm2,ymm1,0x4 - LONG $0xf07b63c4; WORD $0x0bf8 // rorx r15d,eax,0xb - LONG $0x30048d47 // lea r8d,[r8+r14*1] - LONG $0x22148d42 // lea edx,[rdx+r12*1] - LONG $0xd472cdc5; BYTE $0x07 // vpsrld ymm6,ymm4,0x7 - LONG $0xf27862c4; BYTE $0xe1 // andn r12d,eax,ecx - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b63c4; WORD $0x06f0 // rorx r14d,eax,0x6 - LONG $0xdffee5c5 // vpaddd ymm3,ymm3,ymm7 - LONG $0x22148d42 // lea edx,[rdx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8945; BYTE $0xc7 // mov r15d,r8d - LONG $0xd472c5c5; BYTE $0x03 // vpsrld ymm7,ymm4,0x3 - LONG $0xf07b43c4; WORD $0x16e0 // rorx r12d,r8d,0x16 - LONG $0x2a148d42 // lea edx,[rdx+r13*1] - WORD $0x3145; BYTE $0xcf // xor r15d,r9d - LONG $0xf472d5c5; BYTE $0x0e // vpslld ymm5,ymm4,0xe - LONG $0xf07b43c4; WORD $0x0df0 // rorx r14d,r8d,0xd - LONG $0xf07b43c4; WORD $0x02e8 // rorx r13d,r8d,0x2 - LONG $0x131c8d45 // lea r11d,[r11+rdx*1] - LONG $0xe6efc5c5 // vpxor ymm4,ymm7,ymm6 - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3144; BYTE $0xcf // xor edi,r9d - LONG $0xfa70fdc5; BYTE $0xfa // vpshufd ymm7,ymm2,0xfa - WORD $0x3145; BYTE $0xee // xor r14d,r13d - WORD $0x148d; BYTE $0x3a // lea edx,[rdx+rdi*1] - WORD $0x8941; BYTE $0xc4 // mov r12d,eax - LONG $0xd672cdc5; BYTE $0x0b // vpsrld ymm6,ymm6,0xb - - // ROUND(DX, R8, R9, R10, R11, AX, BX, CX, R12, R13, R14, DI, R15, SP, 0xa4) - LONG $0xa4248c03; WORD $0x0000; BYTE $0x00 // add ecx,[rsp+0xa4] - WORD $0x2145; BYTE $0xdc // and r12d,r11d - LONG $0xf07b43c4; WORD $0x19eb // rorx r13d,r11d,0x19 - LONG $0xe5efddc5 // vpxor ymm4,ymm4,ymm5 - LONG $0xf07bc3c4; WORD $0x0bfb // rorx edi,r11d,0xb - LONG $0x32148d42 // lea edx,[rdx+r14*1] - LONG $0x210c8d42 // lea ecx,[rcx+r12*1] - LONG $0xf572d5c5; BYTE $0x0b // vpslld ymm5,ymm5,0xb - LONG $0xf22062c4; BYTE $0xe3 // andn r12d,r11d,ebx - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b43c4; WORD $0x06f3 // rorx r14d,r11d,0x6 - LONG $0xe6efddc5 // vpxor ymm4,ymm4,ymm6 - LONG $0x210c8d42 // lea ecx,[rcx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0xd789 // mov edi,edx - LONG $0xd772cdc5; BYTE $0x0a // vpsrld ymm6,ymm7,0xa - LONG $0xf07b63c4; WORD $0x16e2 // rorx r12d,edx,0x16 - LONG $0x290c8d42 // lea ecx,[rcx+r13*1] - WORD $0x3144; BYTE $0xc7 // xor edi,r8d - LONG $0xe5efddc5 // vpxor ymm4,ymm4,ymm5 - LONG $0xf07b63c4; WORD $0x0df2 // rorx r14d,edx,0xd - LONG $0xf07b63c4; WORD $0x02ea // rorx r13d,edx,0x2 - LONG $0x0a148d45 // lea r10d,[r10+rcx*1] - LONG $0xd773c5c5; BYTE $0x11 // vpsrlq ymm7,ymm7,0x11 - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3145; BYTE $0xc7 // xor r15d,r8d - LONG $0xdcfee5c5 // vpaddd ymm3,ymm3,ymm4 - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x390c8d42 // lea ecx,[rcx+r15*1] - WORD $0x8945; BYTE $0xdc // mov r12d,r11d - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - - // ROUND(CX, DX, R8, R9, R10, R11, AX, BX, R12, R13, R14, R15, DI, SP, 0xa8) - LONG $0xa8249c03; WORD $0x0000; BYTE $0x00 // add ebx,[rsp+0xa8] - WORD $0x2145; BYTE $0xd4 // and r12d,r10d - LONG $0xf07b43c4; WORD $0x19ea // rorx r13d,r10d,0x19 - LONG $0xd773c5c5; BYTE $0x02 // vpsrlq ymm7,ymm7,0x2 - LONG $0xf07b43c4; WORD $0x0bfa // rorx r15d,r10d,0xb - LONG $0x310c8d42 // lea ecx,[rcx+r14*1] - LONG $0x231c8d42 // lea ebx,[rbx+r12*1] - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - LONG $0xf22862c4; BYTE $0xe0 // andn r12d,r10d,eax - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b43c4; WORD $0x06f2 // rorx r14d,r10d,0x6 - LONG $0x004dc2c4; BYTE $0xf0 // vpshufb ymm6,ymm6,ymm8 - LONG $0x231c8d42 // lea ebx,[rbx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8941; BYTE $0xcf // mov r15d,ecx - LONG $0xdefee5c5 // vpaddd ymm3,ymm3,ymm6 - LONG $0xf07b63c4; WORD $0x16e1 // rorx r12d,ecx,0x16 - LONG $0x2b1c8d42 // lea ebx,[rbx+r13*1] - WORD $0x3141; BYTE $0xd7 // xor r15d,edx - LONG $0xfb70fdc5; BYTE $0x50 // vpshufd ymm7,ymm3,0x50 - LONG $0xf07b63c4; WORD $0x0df1 // rorx r14d,ecx,0xd - LONG $0xf07b63c4; WORD $0x02e9 // rorx r13d,ecx,0x2 - LONG $0x190c8d45 // lea r9d,[r9+rbx*1] - LONG $0xd772cdc5; BYTE $0x0a // vpsrld ymm6,ymm7,0xa - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0xd731 // xor edi,edx - LONG $0xd773c5c5; BYTE $0x11 // vpsrlq ymm7,ymm7,0x11 - WORD $0x3145; BYTE $0xee // xor r14d,r13d - WORD $0x1c8d; BYTE $0x3b // lea ebx,[rbx+rdi*1] - WORD $0x8945; BYTE $0xd4 // mov r12d,r10d - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - - // ROUND(BX, CX, DX, R8, R9, R10, R11, AX, R12, R13, R14, DI, R15, SP, 0xac) - LONG $0xac248403; WORD $0x0000; BYTE $0x00 // add eax,[rsp+0xac] - WORD $0x2145; BYTE $0xcc // and r12d,r9d - LONG $0xf07b43c4; WORD $0x19e9 // rorx r13d,r9d,0x19 - LONG $0xd773c5c5; BYTE $0x02 // vpsrlq ymm7,ymm7,0x2 - LONG $0xf07bc3c4; WORD $0x0bf9 // rorx edi,r9d,0xb - LONG $0x331c8d42 // lea ebx,[rbx+r14*1] - LONG $0x20048d42 // lea eax,[rax+r12*1] - LONG $0xf7efcdc5 // vpxor ymm6,ymm6,ymm7 - LONG $0xf23042c4; BYTE $0xe3 // andn r12d,r9d,r11d - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b43c4; WORD $0x06f1 // rorx r14d,r9d,0x6 - LONG $0x004dc2c4; BYTE $0xf1 // vpshufb ymm6,ymm6,ymm9 - LONG $0x20048d42 // lea eax,[rax+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0xdf89 // mov edi,ebx - LONG $0xdefee5c5 // vpaddd ymm3,ymm3,ymm6 - LONG $0xf07b63c4; WORD $0x16e3 // rorx r12d,ebx,0x16 - LONG $0x28048d42 // lea eax,[rax+r13*1] - WORD $0xcf31 // xor edi,ecx - LONG $0x75fee5c5; BYTE $0x60 // vpaddd ymm6,ymm3,[rbp+0x60] - LONG $0xf07b63c4; WORD $0x0df3 // rorx r14d,ebx,0xd - LONG $0xf07b63c4; WORD $0x02eb // rorx r13d,ebx,0x2 - LONG $0x00048d45 // lea r8d,[r8+rax*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3141; BYTE $0xcf // xor r15d,ecx - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x38048d42 // lea eax,[rax+r15*1] - WORD $0x8945; BYTE $0xcc // mov r12d,r9d - - LONG $0x747ffdc5; WORD $0x2024 // vmovdqa [rsp+0x20],ymm6 - ADDQ $0x80, BP - - CMPB 0x3(BP), $0x0 - JNE loop1 - - // ROUND(AX, BX, CX, DX, R8, R9, R10, R11, R12, R13, R14, R15, DI, SP, 0x40) - LONG $0x245c0344; BYTE $0x40 // add r11d,[rsp+0x40] - WORD $0x2145; BYTE $0xc4 // and r12d,r8d - LONG $0xf07b43c4; WORD $0x19e8 // rorx r13d,r8d,0x19 - LONG $0xf07b43c4; WORD $0x0bf8 // rorx r15d,r8d,0xb - LONG $0x30048d42 // lea eax,[rax+r14*1] - LONG $0x231c8d47 // lea r11d,[r11+r12*1] - LONG $0xf23842c4; BYTE $0xe2 // andn r12d,r8d,r10d - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b43c4; WORD $0x06f0 // rorx r14d,r8d,0x6 - LONG $0x231c8d47 // lea r11d,[r11+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8941; BYTE $0xc7 // mov r15d,eax - LONG $0xf07b63c4; WORD $0x16e0 // rorx r12d,eax,0x16 - LONG $0x2b1c8d47 // lea r11d,[r11+r13*1] - WORD $0x3141; BYTE $0xdf // xor r15d,ebx - LONG $0xf07b63c4; WORD $0x0df0 // rorx r14d,eax,0xd - LONG $0xf07b63c4; WORD $0x02e8 // rorx r13d,eax,0x2 - LONG $0x1a148d42 // lea edx,[rdx+r11*1] - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0xdf31 // xor edi,ebx - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x3b1c8d45 // lea r11d,[r11+rdi*1] - WORD $0x8945; BYTE $0xc4 // mov r12d,r8d - - // ROUND(R11, AX, BX, CX, DX, R8, R9, R10, R12, R13, R14, DI, R15, SP, 0x44) - LONG $0x24540344; BYTE $0x44 // add r10d,[rsp+0x44] - WORD $0x2141; BYTE $0xd4 // and r12d,edx - LONG $0xf07b63c4; WORD $0x19ea // rorx r13d,edx,0x19 - LONG $0xf07be3c4; WORD $0x0bfa // rorx edi,edx,0xb - LONG $0x331c8d47 // lea r11d,[r11+r14*1] - LONG $0x22148d47 // lea r10d,[r10+r12*1] - LONG $0xf26842c4; BYTE $0xe1 // andn r12d,edx,r9d - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b63c4; WORD $0x06f2 // rorx r14d,edx,0x6 - LONG $0x22148d47 // lea r10d,[r10+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8944; BYTE $0xdf // mov edi,r11d - LONG $0xf07b43c4; WORD $0x16e3 // rorx r12d,r11d,0x16 - LONG $0x2a148d47 // lea r10d,[r10+r13*1] - WORD $0xc731 // xor edi,eax - LONG $0xf07b43c4; WORD $0x0df3 // rorx r14d,r11d,0xd - LONG $0xf07b43c4; WORD $0x02eb // rorx r13d,r11d,0x2 - LONG $0x110c8d42 // lea ecx,[rcx+r10*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3141; BYTE $0xc7 // xor r15d,eax - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x3a148d47 // lea r10d,[r10+r15*1] - WORD $0x8941; BYTE $0xd4 // mov r12d,edx - - // ROUND(R10, R11, AX, BX, CX, DX, R8, R9, R12, R13, R14, R15, DI, SP, 0x48) - LONG $0x244c0344; BYTE $0x48 // add r9d,[rsp+0x48] - WORD $0x2141; BYTE $0xcc // and r12d,ecx - LONG $0xf07b63c4; WORD $0x19e9 // rorx r13d,ecx,0x19 - LONG $0xf07b63c4; WORD $0x0bf9 // rorx r15d,ecx,0xb - LONG $0x32148d47 // lea r10d,[r10+r14*1] - LONG $0x210c8d47 // lea r9d,[r9+r12*1] - LONG $0xf27042c4; BYTE $0xe0 // andn r12d,ecx,r8d - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b63c4; WORD $0x06f1 // rorx r14d,ecx,0x6 - LONG $0x210c8d47 // lea r9d,[r9+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8945; BYTE $0xd7 // mov r15d,r10d - LONG $0xf07b43c4; WORD $0x16e2 // rorx r12d,r10d,0x16 - LONG $0x290c8d47 // lea r9d,[r9+r13*1] - WORD $0x3145; BYTE $0xdf // xor r15d,r11d - LONG $0xf07b43c4; WORD $0x0df2 // rorx r14d,r10d,0xd - LONG $0xf07b43c4; WORD $0x02ea // rorx r13d,r10d,0x2 - LONG $0x0b1c8d42 // lea ebx,[rbx+r9*1] - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3144; BYTE $0xdf // xor edi,r11d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x390c8d45 // lea r9d,[r9+rdi*1] - WORD $0x8941; BYTE $0xcc // mov r12d,ecx - - // ROUND(R9, R10, R11, AX, BX, CX, DX, R8, R12, R13, R14, DI, R15, SP, 0x4c) - LONG $0x24440344; BYTE $0x4c // add r8d,[rsp+0x4c] - WORD $0x2141; BYTE $0xdc // and r12d,ebx - LONG $0xf07b63c4; WORD $0x19eb // rorx r13d,ebx,0x19 - LONG $0xf07be3c4; WORD $0x0bfb // rorx edi,ebx,0xb - LONG $0x310c8d47 // lea r9d,[r9+r14*1] - LONG $0x20048d47 // lea r8d,[r8+r12*1] - LONG $0xf26062c4; BYTE $0xe2 // andn r12d,ebx,edx - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b63c4; WORD $0x06f3 // rorx r14d,ebx,0x6 - LONG $0x20048d47 // lea r8d,[r8+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8944; BYTE $0xcf // mov edi,r9d - LONG $0xf07b43c4; WORD $0x16e1 // rorx r12d,r9d,0x16 - LONG $0x28048d47 // lea r8d,[r8+r13*1] - WORD $0x3144; BYTE $0xd7 // xor edi,r10d - LONG $0xf07b43c4; WORD $0x0df1 // rorx r14d,r9d,0xd - LONG $0xf07b43c4; WORD $0x02e9 // rorx r13d,r9d,0x2 - LONG $0x00048d42 // lea eax,[rax+r8*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3145; BYTE $0xd7 // xor r15d,r10d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x38048d47 // lea r8d,[r8+r15*1] - WORD $0x8941; BYTE $0xdc // mov r12d,ebx - - // ROUND(R8, R9, R10, R11, AX, BX, CX, DX, R12, R13, R14, R15, DI, SP, 0x60) - LONG $0x60245403 // add edx,[rsp+0x60] - WORD $0x2141; BYTE $0xc4 // and r12d,eax - LONG $0xf07b63c4; WORD $0x19e8 // rorx r13d,eax,0x19 - LONG $0xf07b63c4; WORD $0x0bf8 // rorx r15d,eax,0xb - LONG $0x30048d47 // lea r8d,[r8+r14*1] - LONG $0x22148d42 // lea edx,[rdx+r12*1] - LONG $0xf27862c4; BYTE $0xe1 // andn r12d,eax,ecx - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b63c4; WORD $0x06f0 // rorx r14d,eax,0x6 - LONG $0x22148d42 // lea edx,[rdx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8945; BYTE $0xc7 // mov r15d,r8d - LONG $0xf07b43c4; WORD $0x16e0 // rorx r12d,r8d,0x16 - LONG $0x2a148d42 // lea edx,[rdx+r13*1] - WORD $0x3145; BYTE $0xcf // xor r15d,r9d - LONG $0xf07b43c4; WORD $0x0df0 // rorx r14d,r8d,0xd - LONG $0xf07b43c4; WORD $0x02e8 // rorx r13d,r8d,0x2 - LONG $0x131c8d45 // lea r11d,[r11+rdx*1] - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3144; BYTE $0xcf // xor edi,r9d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - WORD $0x148d; BYTE $0x3a // lea edx,[rdx+rdi*1] - WORD $0x8941; BYTE $0xc4 // mov r12d,eax - - // ROUND(DX, R8, R9, R10, R11, AX, BX, CX, R12, R13, R14, DI, R15, SP, 0x64) - LONG $0x64244c03 // add ecx,[rsp+0x64] - WORD $0x2145; BYTE $0xdc // and r12d,r11d - LONG $0xf07b43c4; WORD $0x19eb // rorx r13d,r11d,0x19 - LONG $0xf07bc3c4; WORD $0x0bfb // rorx edi,r11d,0xb - LONG $0x32148d42 // lea edx,[rdx+r14*1] - LONG $0x210c8d42 // lea ecx,[rcx+r12*1] - LONG $0xf22062c4; BYTE $0xe3 // andn r12d,r11d,ebx - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b43c4; WORD $0x06f3 // rorx r14d,r11d,0x6 - LONG $0x210c8d42 // lea ecx,[rcx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0xd789 // mov edi,edx - LONG $0xf07b63c4; WORD $0x16e2 // rorx r12d,edx,0x16 - LONG $0x290c8d42 // lea ecx,[rcx+r13*1] - WORD $0x3144; BYTE $0xc7 // xor edi,r8d - LONG $0xf07b63c4; WORD $0x0df2 // rorx r14d,edx,0xd - LONG $0xf07b63c4; WORD $0x02ea // rorx r13d,edx,0x2 - LONG $0x0a148d45 // lea r10d,[r10+rcx*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3145; BYTE $0xc7 // xor r15d,r8d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x390c8d42 // lea ecx,[rcx+r15*1] - WORD $0x8945; BYTE $0xdc // mov r12d,r11d - - // ROUND(CX, DX, R8, R9, R10, R11, AX, BX, R12, R13, R14, R15, DI, SP, 0x68) - LONG $0x68245c03 // add ebx,[rsp+0x68] - WORD $0x2145; BYTE $0xd4 // and r12d,r10d - LONG $0xf07b43c4; WORD $0x19ea // rorx r13d,r10d,0x19 - LONG $0xf07b43c4; WORD $0x0bfa // rorx r15d,r10d,0xb - LONG $0x310c8d42 // lea ecx,[rcx+r14*1] - LONG $0x231c8d42 // lea ebx,[rbx+r12*1] - LONG $0xf22862c4; BYTE $0xe0 // andn r12d,r10d,eax - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b43c4; WORD $0x06f2 // rorx r14d,r10d,0x6 - LONG $0x231c8d42 // lea ebx,[rbx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8941; BYTE $0xcf // mov r15d,ecx - LONG $0xf07b63c4; WORD $0x16e1 // rorx r12d,ecx,0x16 - LONG $0x2b1c8d42 // lea ebx,[rbx+r13*1] - WORD $0x3141; BYTE $0xd7 // xor r15d,edx - LONG $0xf07b63c4; WORD $0x0df1 // rorx r14d,ecx,0xd - LONG $0xf07b63c4; WORD $0x02e9 // rorx r13d,ecx,0x2 - LONG $0x190c8d45 // lea r9d,[r9+rbx*1] - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0xd731 // xor edi,edx - WORD $0x3145; BYTE $0xee // xor r14d,r13d - WORD $0x1c8d; BYTE $0x3b // lea ebx,[rbx+rdi*1] - WORD $0x8945; BYTE $0xd4 // mov r12d,r10d - - // ROUND(BX, CX, DX, R8, R9, R10, R11, AX, R12, R13, R14, DI, R15, SP, 0x6c) - LONG $0x6c244403 // add eax,[rsp+0x6c] - WORD $0x2145; BYTE $0xcc // and r12d,r9d - LONG $0xf07b43c4; WORD $0x19e9 // rorx r13d,r9d,0x19 - LONG $0xf07bc3c4; WORD $0x0bf9 // rorx edi,r9d,0xb - LONG $0x331c8d42 // lea ebx,[rbx+r14*1] - LONG $0x20048d42 // lea eax,[rax+r12*1] - LONG $0xf23042c4; BYTE $0xe3 // andn r12d,r9d,r11d - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b43c4; WORD $0x06f1 // rorx r14d,r9d,0x6 - LONG $0x20048d42 // lea eax,[rax+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0xdf89 // mov edi,ebx - LONG $0xf07b63c4; WORD $0x16e3 // rorx r12d,ebx,0x16 - LONG $0x28048d42 // lea eax,[rax+r13*1] - WORD $0xcf31 // xor edi,ecx - LONG $0xf07b63c4; WORD $0x0df3 // rorx r14d,ebx,0xd - LONG $0xf07b63c4; WORD $0x02eb // rorx r13d,ebx,0x2 - LONG $0x00048d45 // lea r8d,[r8+rax*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3141; BYTE $0xcf // xor r15d,ecx - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x38048d42 // lea eax,[rax+r15*1] - WORD $0x8945; BYTE $0xcc // mov r12d,r9d - - // ROUND(AX, BX, CX, DX, R8, R9, R10, R11, R12, R13, R14, R15, DI, SP, 0x00) - LONG $0x241c0344 // add r11d,[rsp] - WORD $0x2145; BYTE $0xc4 // and r12d,r8d - LONG $0xf07b43c4; WORD $0x19e8 // rorx r13d,r8d,0x19 - LONG $0xf07b43c4; WORD $0x0bf8 // rorx r15d,r8d,0xb - LONG $0x30048d42 // lea eax,[rax+r14*1] - LONG $0x231c8d47 // lea r11d,[r11+r12*1] - LONG $0xf23842c4; BYTE $0xe2 // andn r12d,r8d,r10d - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b43c4; WORD $0x06f0 // rorx r14d,r8d,0x6 - LONG $0x231c8d47 // lea r11d,[r11+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8941; BYTE $0xc7 // mov r15d,eax - LONG $0xf07b63c4; WORD $0x16e0 // rorx r12d,eax,0x16 - LONG $0x2b1c8d47 // lea r11d,[r11+r13*1] - WORD $0x3141; BYTE $0xdf // xor r15d,ebx - LONG $0xf07b63c4; WORD $0x0df0 // rorx r14d,eax,0xd - LONG $0xf07b63c4; WORD $0x02e8 // rorx r13d,eax,0x2 - LONG $0x1a148d42 // lea edx,[rdx+r11*1] - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0xdf31 // xor edi,ebx - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x3b1c8d45 // lea r11d,[r11+rdi*1] - WORD $0x8945; BYTE $0xc4 // mov r12d,r8d - - // ROUND(R11, AX, BX, CX, DX, R8, R9, R10, R12, R13, R14, DI, R15, SP, 0x04) - LONG $0x24540344; BYTE $0x04 // add r10d,[rsp+0x4] - WORD $0x2141; BYTE $0xd4 // and r12d,edx - LONG $0xf07b63c4; WORD $0x19ea // rorx r13d,edx,0x19 - LONG $0xf07be3c4; WORD $0x0bfa // rorx edi,edx,0xb - LONG $0x331c8d47 // lea r11d,[r11+r14*1] - LONG $0x22148d47 // lea r10d,[r10+r12*1] - LONG $0xf26842c4; BYTE $0xe1 // andn r12d,edx,r9d - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b63c4; WORD $0x06f2 // rorx r14d,edx,0x6 - LONG $0x22148d47 // lea r10d,[r10+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8944; BYTE $0xdf // mov edi,r11d - LONG $0xf07b43c4; WORD $0x16e3 // rorx r12d,r11d,0x16 - LONG $0x2a148d47 // lea r10d,[r10+r13*1] - WORD $0xc731 // xor edi,eax - LONG $0xf07b43c4; WORD $0x0df3 // rorx r14d,r11d,0xd - LONG $0xf07b43c4; WORD $0x02eb // rorx r13d,r11d,0x2 - LONG $0x110c8d42 // lea ecx,[rcx+r10*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3141; BYTE $0xc7 // xor r15d,eax - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x3a148d47 // lea r10d,[r10+r15*1] - WORD $0x8941; BYTE $0xd4 // mov r12d,edx - - // ROUND(R10, R11, AX, BX, CX, DX, R8, R9, R12, R13, R14, R15, DI, SP, 0x08) - LONG $0x244c0344; BYTE $0x08 // add r9d,[rsp+0x8] - WORD $0x2141; BYTE $0xcc // and r12d,ecx - LONG $0xf07b63c4; WORD $0x19e9 // rorx r13d,ecx,0x19 - LONG $0xf07b63c4; WORD $0x0bf9 // rorx r15d,ecx,0xb - LONG $0x32148d47 // lea r10d,[r10+r14*1] - LONG $0x210c8d47 // lea r9d,[r9+r12*1] - LONG $0xf27042c4; BYTE $0xe0 // andn r12d,ecx,r8d - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b63c4; WORD $0x06f1 // rorx r14d,ecx,0x6 - LONG $0x210c8d47 // lea r9d,[r9+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8945; BYTE $0xd7 // mov r15d,r10d - LONG $0xf07b43c4; WORD $0x16e2 // rorx r12d,r10d,0x16 - LONG $0x290c8d47 // lea r9d,[r9+r13*1] - WORD $0x3145; BYTE $0xdf // xor r15d,r11d - LONG $0xf07b43c4; WORD $0x0df2 // rorx r14d,r10d,0xd - LONG $0xf07b43c4; WORD $0x02ea // rorx r13d,r10d,0x2 - LONG $0x0b1c8d42 // lea ebx,[rbx+r9*1] - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3144; BYTE $0xdf // xor edi,r11d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x390c8d45 // lea r9d,[r9+rdi*1] - WORD $0x8941; BYTE $0xcc // mov r12d,ecx - - // ROUND(R9, R10, R11, AX, BX, CX, DX, R8, R12, R13, R14, DI, R15, SP, 0x0c) - LONG $0x24440344; BYTE $0x0c // add r8d,[rsp+0xc] - WORD $0x2141; BYTE $0xdc // and r12d,ebx - LONG $0xf07b63c4; WORD $0x19eb // rorx r13d,ebx,0x19 - LONG $0xf07be3c4; WORD $0x0bfb // rorx edi,ebx,0xb - LONG $0x310c8d47 // lea r9d,[r9+r14*1] - LONG $0x20048d47 // lea r8d,[r8+r12*1] - LONG $0xf26062c4; BYTE $0xe2 // andn r12d,ebx,edx - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b63c4; WORD $0x06f3 // rorx r14d,ebx,0x6 - LONG $0x20048d47 // lea r8d,[r8+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8944; BYTE $0xcf // mov edi,r9d - LONG $0xf07b43c4; WORD $0x16e1 // rorx r12d,r9d,0x16 - LONG $0x28048d47 // lea r8d,[r8+r13*1] - WORD $0x3144; BYTE $0xd7 // xor edi,r10d - LONG $0xf07b43c4; WORD $0x0df1 // rorx r14d,r9d,0xd - LONG $0xf07b43c4; WORD $0x02e9 // rorx r13d,r9d,0x2 - LONG $0x00048d42 // lea eax,[rax+r8*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3145; BYTE $0xd7 // xor r15d,r10d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x38048d47 // lea r8d,[r8+r15*1] - WORD $0x8941; BYTE $0xdc // mov r12d,ebx - - // ROUND(R8, R9, R10, R11, AX, BX, CX, DX, R12, R13, R14, R15, DI, SP, 0x20) - LONG $0x20245403 // add edx,[rsp+0x20] - WORD $0x2141; BYTE $0xc4 // and r12d,eax - LONG $0xf07b63c4; WORD $0x19e8 // rorx r13d,eax,0x19 - LONG $0xf07b63c4; WORD $0x0bf8 // rorx r15d,eax,0xb - LONG $0x30048d47 // lea r8d,[r8+r14*1] - LONG $0x22148d42 // lea edx,[rdx+r12*1] - LONG $0xf27862c4; BYTE $0xe1 // andn r12d,eax,ecx - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b63c4; WORD $0x06f0 // rorx r14d,eax,0x6 - LONG $0x22148d42 // lea edx,[rdx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8945; BYTE $0xc7 // mov r15d,r8d - LONG $0xf07b43c4; WORD $0x16e0 // rorx r12d,r8d,0x16 - LONG $0x2a148d42 // lea edx,[rdx+r13*1] - WORD $0x3145; BYTE $0xcf // xor r15d,r9d - LONG $0xf07b43c4; WORD $0x0df0 // rorx r14d,r8d,0xd - LONG $0xf07b43c4; WORD $0x02e8 // rorx r13d,r8d,0x2 - LONG $0x131c8d45 // lea r11d,[r11+rdx*1] - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3144; BYTE $0xcf // xor edi,r9d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - WORD $0x148d; BYTE $0x3a // lea edx,[rdx+rdi*1] - WORD $0x8941; BYTE $0xc4 // mov r12d,eax - - // ROUND(DX, R8, R9, R10, R11, AX, BX, CX, R12, R13, R14, DI, R15, SP, 0x24) - LONG $0x24244c03 // add ecx,[rsp+0x24] - WORD $0x2145; BYTE $0xdc // and r12d,r11d - LONG $0xf07b43c4; WORD $0x19eb // rorx r13d,r11d,0x19 - LONG $0xf07bc3c4; WORD $0x0bfb // rorx edi,r11d,0xb - LONG $0x32148d42 // lea edx,[rdx+r14*1] - LONG $0x210c8d42 // lea ecx,[rcx+r12*1] - LONG $0xf22062c4; BYTE $0xe3 // andn r12d,r11d,ebx - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b43c4; WORD $0x06f3 // rorx r14d,r11d,0x6 - LONG $0x210c8d42 // lea ecx,[rcx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0xd789 // mov edi,edx - LONG $0xf07b63c4; WORD $0x16e2 // rorx r12d,edx,0x16 - LONG $0x290c8d42 // lea ecx,[rcx+r13*1] - WORD $0x3144; BYTE $0xc7 // xor edi,r8d - LONG $0xf07b63c4; WORD $0x0df2 // rorx r14d,edx,0xd - LONG $0xf07b63c4; WORD $0x02ea // rorx r13d,edx,0x2 - LONG $0x0a148d45 // lea r10d,[r10+rcx*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3145; BYTE $0xc7 // xor r15d,r8d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x390c8d42 // lea ecx,[rcx+r15*1] - WORD $0x8945; BYTE $0xdc // mov r12d,r11d - - // ROUND(CX, DX, R8, R9, R10, R11, AX, BX, R12, R13, R14, R15, DI, SP, 0x28) - LONG $0x28245c03 // add ebx,[rsp+0x28] - WORD $0x2145; BYTE $0xd4 // and r12d,r10d - LONG $0xf07b43c4; WORD $0x19ea // rorx r13d,r10d,0x19 - LONG $0xf07b43c4; WORD $0x0bfa // rorx r15d,r10d,0xb - LONG $0x310c8d42 // lea ecx,[rcx+r14*1] - LONG $0x231c8d42 // lea ebx,[rbx+r12*1] - LONG $0xf22862c4; BYTE $0xe0 // andn r12d,r10d,eax - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b43c4; WORD $0x06f2 // rorx r14d,r10d,0x6 - LONG $0x231c8d42 // lea ebx,[rbx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8941; BYTE $0xcf // mov r15d,ecx - LONG $0xf07b63c4; WORD $0x16e1 // rorx r12d,ecx,0x16 - LONG $0x2b1c8d42 // lea ebx,[rbx+r13*1] - WORD $0x3141; BYTE $0xd7 // xor r15d,edx - LONG $0xf07b63c4; WORD $0x0df1 // rorx r14d,ecx,0xd - LONG $0xf07b63c4; WORD $0x02e9 // rorx r13d,ecx,0x2 - LONG $0x190c8d45 // lea r9d,[r9+rbx*1] - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0xd731 // xor edi,edx - WORD $0x3145; BYTE $0xee // xor r14d,r13d - WORD $0x1c8d; BYTE $0x3b // lea ebx,[rbx+rdi*1] - WORD $0x8945; BYTE $0xd4 // mov r12d,r10d - - // ROUND(BX, CX, DX, R8, R9, R10, R11, AX, R12, R13, R14, DI, R15, SP, 0x2c) - LONG $0x2c244403 // add eax,[rsp+0x2c] - WORD $0x2145; BYTE $0xcc // and r12d,r9d - LONG $0xf07b43c4; WORD $0x19e9 // rorx r13d,r9d,0x19 - LONG $0xf07bc3c4; WORD $0x0bf9 // rorx edi,r9d,0xb - LONG $0x331c8d42 // lea ebx,[rbx+r14*1] - LONG $0x20048d42 // lea eax,[rax+r12*1] - LONG $0xf23042c4; BYTE $0xe3 // andn r12d,r9d,r11d - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b43c4; WORD $0x06f1 // rorx r14d,r9d,0x6 - LONG $0x20048d42 // lea eax,[rax+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0xdf89 // mov edi,ebx - LONG $0xf07b63c4; WORD $0x16e3 // rorx r12d,ebx,0x16 - LONG $0x28048d42 // lea eax,[rax+r13*1] - WORD $0xcf31 // xor edi,ecx - LONG $0xf07b63c4; WORD $0x0df3 // rorx r14d,ebx,0xd - LONG $0xf07b63c4; WORD $0x02eb // rorx r13d,ebx,0x2 - LONG $0x00048d45 // lea r8d,[r8+rax*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3141; BYTE $0xcf // xor r15d,ecx - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x38048d42 // lea eax,[rax+r15*1] - WORD $0x8945; BYTE $0xcc // mov r12d,r9d - - MOVQ 0x200(SP), DI // $_ctx - ADDQ R14, AX - - LEAQ 0x1c0(SP), BP - - ADDL (DI), AX - ADDL 4(DI), BX - ADDL 8(DI), CX - ADDL 12(DI), DX - ADDL 16(DI), R8 - ADDL 20(DI), R9 - ADDL 24(DI), R10 - ADDL 28(DI), R11 - - MOVL AX, (DI) - MOVL BX, 4(DI) - MOVL CX, 8(DI) - MOVL DX, 12(DI) - MOVL R8, 16(DI) - MOVL R9, 20(DI) - MOVL R10, 24(DI) - MOVL R11, 28(DI) - - CMPQ SI, 0x50(BP) // $_end - JE done - - XORQ R14, R14 - MOVQ BX, DI - XORQ CX, DI // magic - MOVQ R9, R12 - -loop2: - // ROUND(AX, BX, CX, DX, R8, R9, R10, R11, R12, R13, R14, R15, DI, BP, 0x10) - LONG $0x105d0344 // add r11d,[rbp+0x10] - WORD $0x2145; BYTE $0xc4 // and r12d,r8d - LONG $0xf07b43c4; WORD $0x19e8 // rorx r13d,r8d,0x19 - LONG $0xf07b43c4; WORD $0x0bf8 // rorx r15d,r8d,0xb - LONG $0x30048d42 // lea eax,[rax+r14*1] - LONG $0x231c8d47 // lea r11d,[r11+r12*1] - LONG $0xf23842c4; BYTE $0xe2 // andn r12d,r8d,r10d - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b43c4; WORD $0x06f0 // rorx r14d,r8d,0x6 - LONG $0x231c8d47 // lea r11d,[r11+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8941; BYTE $0xc7 // mov r15d,eax - LONG $0xf07b63c4; WORD $0x16e0 // rorx r12d,eax,0x16 - LONG $0x2b1c8d47 // lea r11d,[r11+r13*1] - WORD $0x3141; BYTE $0xdf // xor r15d,ebx - LONG $0xf07b63c4; WORD $0x0df0 // rorx r14d,eax,0xd - LONG $0xf07b63c4; WORD $0x02e8 // rorx r13d,eax,0x2 - LONG $0x1a148d42 // lea edx,[rdx+r11*1] - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0xdf31 // xor edi,ebx - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x3b1c8d45 // lea r11d,[r11+rdi*1] - WORD $0x8945; BYTE $0xc4 // mov r12d,r8d - - // ROUND(R11, AX, BX, CX, DX, R8, R9, R10, R12, R13, R14, DI, R15, BP, 0x14) - LONG $0x14550344 // add r10d,[rbp+0x14] - WORD $0x2141; BYTE $0xd4 // and r12d,edx - LONG $0xf07b63c4; WORD $0x19ea // rorx r13d,edx,0x19 - LONG $0xf07be3c4; WORD $0x0bfa // rorx edi,edx,0xb - LONG $0x331c8d47 // lea r11d,[r11+r14*1] - LONG $0x22148d47 // lea r10d,[r10+r12*1] - LONG $0xf26842c4; BYTE $0xe1 // andn r12d,edx,r9d - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b63c4; WORD $0x06f2 // rorx r14d,edx,0x6 - LONG $0x22148d47 // lea r10d,[r10+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8944; BYTE $0xdf // mov edi,r11d - LONG $0xf07b43c4; WORD $0x16e3 // rorx r12d,r11d,0x16 - LONG $0x2a148d47 // lea r10d,[r10+r13*1] - WORD $0xc731 // xor edi,eax - LONG $0xf07b43c4; WORD $0x0df3 // rorx r14d,r11d,0xd - LONG $0xf07b43c4; WORD $0x02eb // rorx r13d,r11d,0x2 - LONG $0x110c8d42 // lea ecx,[rcx+r10*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3141; BYTE $0xc7 // xor r15d,eax - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x3a148d47 // lea r10d,[r10+r15*1] - WORD $0x8941; BYTE $0xd4 // mov r12d,edx - - // ROUND(R10, R11, AX, BX, CX, DX, R8, R9, R12, R13, R14, R15, DI, BP, 0x18) - LONG $0x184d0344 // add r9d,[rbp+0x18] - WORD $0x2141; BYTE $0xcc // and r12d,ecx - LONG $0xf07b63c4; WORD $0x19e9 // rorx r13d,ecx,0x19 - LONG $0xf07b63c4; WORD $0x0bf9 // rorx r15d,ecx,0xb - LONG $0x32148d47 // lea r10d,[r10+r14*1] - LONG $0x210c8d47 // lea r9d,[r9+r12*1] - LONG $0xf27042c4; BYTE $0xe0 // andn r12d,ecx,r8d - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b63c4; WORD $0x06f1 // rorx r14d,ecx,0x6 - LONG $0x210c8d47 // lea r9d,[r9+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8945; BYTE $0xd7 // mov r15d,r10d - LONG $0xf07b43c4; WORD $0x16e2 // rorx r12d,r10d,0x16 - LONG $0x290c8d47 // lea r9d,[r9+r13*1] - WORD $0x3145; BYTE $0xdf // xor r15d,r11d - LONG $0xf07b43c4; WORD $0x0df2 // rorx r14d,r10d,0xd - LONG $0xf07b43c4; WORD $0x02ea // rorx r13d,r10d,0x2 - LONG $0x0b1c8d42 // lea ebx,[rbx+r9*1] - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3144; BYTE $0xdf // xor edi,r11d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x390c8d45 // lea r9d,[r9+rdi*1] - WORD $0x8941; BYTE $0xcc // mov r12d,ecx - - // ROUND(R9, R10, R11, AX, BX, CX, DX, R8, R12, R13, R14, DI, R15, BP, 0x1c) - LONG $0x1c450344 // add r8d,[rbp+0x1c] - WORD $0x2141; BYTE $0xdc // and r12d,ebx - LONG $0xf07b63c4; WORD $0x19eb // rorx r13d,ebx,0x19 - LONG $0xf07be3c4; WORD $0x0bfb // rorx edi,ebx,0xb - LONG $0x310c8d47 // lea r9d,[r9+r14*1] - LONG $0x20048d47 // lea r8d,[r8+r12*1] - LONG $0xf26062c4; BYTE $0xe2 // andn r12d,ebx,edx - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b63c4; WORD $0x06f3 // rorx r14d,ebx,0x6 - LONG $0x20048d47 // lea r8d,[r8+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8944; BYTE $0xcf // mov edi,r9d - LONG $0xf07b43c4; WORD $0x16e1 // rorx r12d,r9d,0x16 - LONG $0x28048d47 // lea r8d,[r8+r13*1] - WORD $0x3144; BYTE $0xd7 // xor edi,r10d - LONG $0xf07b43c4; WORD $0x0df1 // rorx r14d,r9d,0xd - LONG $0xf07b43c4; WORD $0x02e9 // rorx r13d,r9d,0x2 - LONG $0x00048d42 // lea eax,[rax+r8*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3145; BYTE $0xd7 // xor r15d,r10d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x38048d47 // lea r8d,[r8+r15*1] - WORD $0x8941; BYTE $0xdc // mov r12d,ebx - - // ROUND(R8, R9, R10, R11, AX, BX, CX, DX, R12, R13, R14, R15, DI, BP, 0x30) - WORD $0x5503; BYTE $0x30 // add edx,[rbp+0x30] - WORD $0x2141; BYTE $0xc4 // and r12d,eax - LONG $0xf07b63c4; WORD $0x19e8 // rorx r13d,eax,0x19 - LONG $0xf07b63c4; WORD $0x0bf8 // rorx r15d,eax,0xb - LONG $0x30048d47 // lea r8d,[r8+r14*1] - LONG $0x22148d42 // lea edx,[rdx+r12*1] - LONG $0xf27862c4; BYTE $0xe1 // andn r12d,eax,ecx - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b63c4; WORD $0x06f0 // rorx r14d,eax,0x6 - LONG $0x22148d42 // lea edx,[rdx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8945; BYTE $0xc7 // mov r15d,r8d - LONG $0xf07b43c4; WORD $0x16e0 // rorx r12d,r8d,0x16 - LONG $0x2a148d42 // lea edx,[rdx+r13*1] - WORD $0x3145; BYTE $0xcf // xor r15d,r9d - LONG $0xf07b43c4; WORD $0x0df0 // rorx r14d,r8d,0xd - LONG $0xf07b43c4; WORD $0x02e8 // rorx r13d,r8d,0x2 - LONG $0x131c8d45 // lea r11d,[r11+rdx*1] - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3144; BYTE $0xcf // xor edi,r9d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - WORD $0x148d; BYTE $0x3a // lea edx,[rdx+rdi*1] - WORD $0x8941; BYTE $0xc4 // mov r12d,eax - - // ROUND(DX, R8, R9, R10, R11, AX, BX, CX, R12, R13, R14, DI, R15, BP, 0x34) - WORD $0x4d03; BYTE $0x34 // add ecx,[rbp+0x34] - WORD $0x2145; BYTE $0xdc // and r12d,r11d - LONG $0xf07b43c4; WORD $0x19eb // rorx r13d,r11d,0x19 - LONG $0xf07bc3c4; WORD $0x0bfb // rorx edi,r11d,0xb - LONG $0x32148d42 // lea edx,[rdx+r14*1] - LONG $0x210c8d42 // lea ecx,[rcx+r12*1] - LONG $0xf22062c4; BYTE $0xe3 // andn r12d,r11d,ebx - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b43c4; WORD $0x06f3 // rorx r14d,r11d,0x6 - LONG $0x210c8d42 // lea ecx,[rcx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0xd789 // mov edi,edx - LONG $0xf07b63c4; WORD $0x16e2 // rorx r12d,edx,0x16 - LONG $0x290c8d42 // lea ecx,[rcx+r13*1] - WORD $0x3144; BYTE $0xc7 // xor edi,r8d - LONG $0xf07b63c4; WORD $0x0df2 // rorx r14d,edx,0xd - LONG $0xf07b63c4; WORD $0x02ea // rorx r13d,edx,0x2 - LONG $0x0a148d45 // lea r10d,[r10+rcx*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3145; BYTE $0xc7 // xor r15d,r8d - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x390c8d42 // lea ecx,[rcx+r15*1] - WORD $0x8945; BYTE $0xdc // mov r12d,r11d - - // ROUND(CX, DX, R8, R9, R10, R11, AX, BX, R12, R13, R14, R15, DI, BP, 0x38) - WORD $0x5d03; BYTE $0x38 // add ebx,[rbp+0x38] - WORD $0x2145; BYTE $0xd4 // and r12d,r10d - LONG $0xf07b43c4; WORD $0x19ea // rorx r13d,r10d,0x19 - LONG $0xf07b43c4; WORD $0x0bfa // rorx r15d,r10d,0xb - LONG $0x310c8d42 // lea ecx,[rcx+r14*1] - LONG $0x231c8d42 // lea ebx,[rbx+r12*1] - LONG $0xf22862c4; BYTE $0xe0 // andn r12d,r10d,eax - WORD $0x3145; BYTE $0xfd // xor r13d,r15d - LONG $0xf07b43c4; WORD $0x06f2 // rorx r14d,r10d,0x6 - LONG $0x231c8d42 // lea ebx,[rbx+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0x8941; BYTE $0xcf // mov r15d,ecx - LONG $0xf07b63c4; WORD $0x16e1 // rorx r12d,ecx,0x16 - LONG $0x2b1c8d42 // lea ebx,[rbx+r13*1] - WORD $0x3141; BYTE $0xd7 // xor r15d,edx - LONG $0xf07b63c4; WORD $0x0df1 // rorx r14d,ecx,0xd - LONG $0xf07b63c4; WORD $0x02e9 // rorx r13d,ecx,0x2 - LONG $0x190c8d45 // lea r9d,[r9+rbx*1] - WORD $0x2144; BYTE $0xff // and edi,r15d - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0xd731 // xor edi,edx - WORD $0x3145; BYTE $0xee // xor r14d,r13d - WORD $0x1c8d; BYTE $0x3b // lea ebx,[rbx+rdi*1] - WORD $0x8945; BYTE $0xd4 // mov r12d,r10d - - // ROUND(BX, CX, DX, R8, R9, R10, R11, AX, R12, R13, R14, DI, R15, BP, 0x3c) - WORD $0x4503; BYTE $0x3c // add eax,[rbp+0x3c] - WORD $0x2145; BYTE $0xcc // and r12d,r9d - LONG $0xf07b43c4; WORD $0x19e9 // rorx r13d,r9d,0x19 - LONG $0xf07bc3c4; WORD $0x0bf9 // rorx edi,r9d,0xb - LONG $0x331c8d42 // lea ebx,[rbx+r14*1] - LONG $0x20048d42 // lea eax,[rax+r12*1] - LONG $0xf23042c4; BYTE $0xe3 // andn r12d,r9d,r11d - WORD $0x3141; BYTE $0xfd // xor r13d,edi - LONG $0xf07b43c4; WORD $0x06f1 // rorx r14d,r9d,0x6 - LONG $0x20048d42 // lea eax,[rax+r12*1] - WORD $0x3145; BYTE $0xf5 // xor r13d,r14d - WORD $0xdf89 // mov edi,ebx - LONG $0xf07b63c4; WORD $0x16e3 // rorx r12d,ebx,0x16 - LONG $0x28048d42 // lea eax,[rax+r13*1] - WORD $0xcf31 // xor edi,ecx - LONG $0xf07b63c4; WORD $0x0df3 // rorx r14d,ebx,0xd - LONG $0xf07b63c4; WORD $0x02eb // rorx r13d,ebx,0x2 - LONG $0x00048d45 // lea r8d,[r8+rax*1] - WORD $0x2141; BYTE $0xff // and r15d,edi - WORD $0x3145; BYTE $0xe6 // xor r14d,r12d - WORD $0x3141; BYTE $0xcf // xor r15d,ecx - WORD $0x3145; BYTE $0xee // xor r14d,r13d - LONG $0x38048d42 // lea eax,[rax+r15*1] - WORD $0x8945; BYTE $0xcc // mov r12d,r9d - - ADDQ $-0x40, BP - CMPQ BP, SP - JAE loop2 - - MOVQ 0x200(SP), DI // $_ctx - ADDQ R14, AX - - ADDQ $0x1c0, SP - - ADDL (DI), AX - ADDL 4(DI), BX - ADDL 8(DI), CX - ADDL 12(DI), DX - ADDL 16(DI), R8 - ADDL 20(DI), R9 - - ADDQ $0x80, SI // input += 2 - ADDL 24(DI), R10 - MOVQ SI, R12 - ADDL 28(DI), R11 - CMPQ SI, 0x50(SP) // input == _end - - MOVL AX, (DI) - LONG $0xe4440f4c // cmove r12,rsp /* next block or stale data */ - MOVL AX, (DI) - MOVL BX, 4(DI) - MOVL CX, 8(DI) - MOVL DX, 12(DI) - MOVL R8, 16(DI) - MOVL R9, 20(DI) - MOVL R10, 24(DI) - MOVL R11, 28(DI) - - JBE loop0 - LEAQ (SP), BP - -done: - MOVQ BP, SP - MOVQ 0x58(SP), SP // restore saved stack pointer - WORD $0xf8c5; BYTE $0x77 // vzeroupper - - RET - diff --git a/vendor/github.com/minio/sha256-simd/sha256blockAvx512_amd64.go b/vendor/github.com/minio/sha256-simd/sha256blockAvx512_amd64.go index db8e48d31..b7d7c1637 100644 --- a/vendor/github.com/minio/sha256-simd/sha256blockAvx512_amd64.go +++ b/vendor/github.com/minio/sha256-simd/sha256blockAvx512_amd64.go @@ -1,4 +1,4 @@ -//+build !noasm,!appengine +//+build !noasm,!appengine,gc /* * Minio Cloud Storage, (C) 2017 Minio, Inc. diff --git a/vendor/github.com/minio/sha256-simd/sha256blockAvx_amd64.go b/vendor/github.com/minio/sha256-simd/sha256blockAvx_amd64.go deleted file mode 100644 index c2f71181f..000000000 --- a/vendor/github.com/minio/sha256-simd/sha256blockAvx_amd64.go +++ /dev/null @@ -1,22 +0,0 @@ -//+build !noasm,!appengine - -/* - * Minio Cloud Storage, (C) 2016 Minio, Inc. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -package sha256 - -//go:noescape -func blockAvx(h []uint32, message []uint8, reserved0, reserved1, reserved2, reserved3 uint64) diff --git a/vendor/github.com/minio/sha256-simd/sha256blockAvx_amd64.s b/vendor/github.com/minio/sha256-simd/sha256blockAvx_amd64.s deleted file mode 100644 index 9f444d49f..000000000 --- a/vendor/github.com/minio/sha256-simd/sha256blockAvx_amd64.s +++ /dev/null @@ -1,408 +0,0 @@ -//+build !noasm,!appengine - -// SHA256 implementation for AVX - -// -// Minio Cloud Storage, (C) 2016 Minio, Inc. -// -// Licensed under the Apache License, Version 2.0 (the "License"); -// you may not use this file except in compliance with the License. -// You may obtain a copy of the License at -// -// http://www.apache.org/licenses/LICENSE-2.0 -// -// Unless required by applicable law or agreed to in writing, software -// distributed under the License is distributed on an "AS IS" BASIS, -// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -// See the License for the specific language governing permissions and -// limitations under the License. -// - -// -// This code is based on an Intel White-Paper: -// "Fast SHA-256 Implementations on Intel Architecture Processors" -// -// together with the reference implementation from the following authors: -// James Guilford <james.guilford@intel.com> -// Kirk Yap <kirk.s.yap@intel.com> -// Tim Chen <tim.c.chen@linux.intel.com> -// -// For Golang it has been converted to Plan 9 assembly with the help of -// github.com/minio/asm2plan9s to assemble Intel instructions to their Plan9 -// equivalents -// - -#include "textflag.h" - -#define ROTATE_XS \ - MOVOU X4, X15 \ - MOVOU X5, X4 \ - MOVOU X6, X5 \ - MOVOU X7, X6 \ - MOVOU X15, X7 - -// compute s0 four at a time and s1 two at a time -// compute W[-16] + W[-7] 4 at a time -#define FOUR_ROUNDS_AND_SCHED(a, b, c, d, e, f, g, h) \ - MOVL e, R13 \ // y0 = e - ROLL $18, R13 \ // y0 = e >> (25-11) - MOVL a, R14 \ // y1 = a - LONG $0x0f41e3c4; WORD $0x04c6 \ // VPALIGNR XMM0,XMM7,XMM6,0x4 /* XTMP0 = W[-7] */ - ROLL $23, R14 \ // y1 = a >> (22-13) - XORL e, R13 \ // y0 = e ^ (e >> (25-11)) - MOVL f, R15 \ // y2 = f - ROLL $27, R13 \ // y0 = (e >> (11-6)) ^ (e >> (25-6)) - XORL a, R14 \ // y1 = a ^ (a >> (22-13) - XORL g, R15 \ // y2 = f^g - LONG $0xc4fef9c5 \ // VPADDD XMM0,XMM0,XMM4 /* XTMP0 = W[-7] + W[-16] */ - XORL e, R13 \ // y0 = e ^ (e >> (11-6)) ^ (e >> (25-6) ) - ANDL e, R15 \ // y2 = (f^g)&e - ROLL $21, R14 \ // y1 = (a >> (13-2)) ^ (a >> (22-2)) - \ - \ // compute s0 - \ - LONG $0x0f51e3c4; WORD $0x04cc \ // VPALIGNR XMM1,XMM5,XMM4,0x4 /* XTMP1 = W[-15] */ - XORL a, R14 \ // y1 = a ^ (a >> (13-2)) ^ (a >> (22-2)) - ROLL $26, R13 \ // y0 = S1 = (e>>6) & (e>>11) ^ (e>>25) - XORL g, R15 \ // y2 = CH = ((f^g)&e)^g - ROLL $30, R14 \ // y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22) - ADDL R13, R15 \ // y2 = S1 + CH - ADDL _xfer+48(FP), R15 \ // y2 = k + w + S1 + CH - MOVL a, R13 \ // y0 = a - ADDL R15, h \ // h = h + S1 + CH + k + w - \ // ROTATE_ARGS - MOVL a, R15 \ // y2 = a - LONG $0xd172e9c5; BYTE $0x07 \ // VPSRLD XMM2,XMM1,0x7 /* */ - ORL c, R13 \ // y0 = a|c - ADDL h, d \ // d = d + h + S1 + CH + k + w - ANDL c, R15 \ // y2 = a&c - LONG $0xf172e1c5; BYTE $0x19 \ // VPSLLD XMM3,XMM1,0x19 /* */ - ANDL b, R13 \ // y0 = (a|c)&b - ADDL R14, h \ // h = h + S1 + CH + k + w + S0 - LONG $0xdaebe1c5 \ // VPOR XMM3,XMM3,XMM2 /* XTMP1 = W[-15] MY_ROR 7 */ - ORL R15, R13 \ // y0 = MAJ = (a|c)&b)|(a&c) - ADDL R13, h \ // h = h + S1 + CH + k + w + S0 + MAJ - \ // ROTATE_ARGS - MOVL d, R13 \ // y0 = e - MOVL h, R14 \ // y1 = a - ROLL $18, R13 \ // y0 = e >> (25-11) - XORL d, R13 \ // y0 = e ^ (e >> (25-11)) - MOVL e, R15 \ // y2 = f - ROLL $23, R14 \ // y1 = a >> (22-13) - LONG $0xd172e9c5; BYTE $0x12 \ // VPSRLD XMM2,XMM1,0x12 /* */ - XORL h, R14 \ // y1 = a ^ (a >> (22-13) - ROLL $27, R13 \ // y0 = (e >> (11-6)) ^ (e >> (25-6)) - XORL f, R15 \ // y2 = f^g - LONG $0xd172b9c5; BYTE $0x03 \ // VPSRLD XMM8,XMM1,0x3 /* XTMP4 = W[-15] >> 3 */ - ROLL $21, R14 \ // y1 = (a >> (13-2)) ^ (a >> (22-2)) - XORL d, R13 \ // y0 = e ^ (e >> (11-6)) ^ (e >> (25-6)) - ANDL d, R15 \ // y2 = (f^g)&e - ROLL $26, R13 \ // y0 = S1 = (e>>6) & (e>>11) ^ (e>>25) - LONG $0xf172f1c5; BYTE $0x0e \ // VPSLLD XMM1,XMM1,0xe /* */ - XORL h, R14 \ // y1 = a ^ (a >> (13-2)) ^ (a >> (22-2)) - XORL f, R15 \ // y2 = CH = ((f^g)&e)^g - LONG $0xd9efe1c5 \ // VPXOR XMM3,XMM3,XMM1 /* */ - ADDL R13, R15 \ // y2 = S1 + CH - ADDL _xfer+52(FP), R15 \ // y2 = k + w + S1 + CH - ROLL $30, R14 \ // y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22) - LONG $0xdaefe1c5 \ // VPXOR XMM3,XMM3,XMM2 /* XTMP1 = W[-15] MY_ROR 7 ^ W[-15] MY_ROR */ - MOVL h, R13 \ // y0 = a - ADDL R15, g \ // h = h + S1 + CH + k + w - MOVL h, R15 \ // y2 = a - LONG $0xef61c1c4; BYTE $0xc8 \ // VPXOR XMM1,XMM3,XMM8 /* XTMP1 = s0 */ - ORL b, R13 \ // y0 = a|c - ADDL g, c \ // d = d + h + S1 + CH + k + w - ANDL b, R15 \ // y2 = a&c - \ - \ // compute low s1 - \ - LONG $0xd770f9c5; BYTE $0xfa \ // VPSHUFD XMM2,XMM7,0xfa /* XTMP2 = W[-2] {BBAA} */ - ANDL a, R13 \ // y0 = (a|c)&b - ADDL R14, g \ // h = h + S1 + CH + k + w + S0 - LONG $0xc1fef9c5 \ // VPADDD XMM0,XMM0,XMM1 /* XTMP0 = W[-16] + W[-7] + s0 */ - ORL R15, R13 \ // y0 = MAJ = (a|c)&b)|(a&c) - ADDL R13, g \ // h = h + S1 + CH + k + w + S0 + MAJ - \ // ROTATE_ARGS - MOVL c, R13 \ // y0 = e - MOVL g, R14 \ // y1 = a - ROLL $18, R13 \ // y0 = e >> (25-11) - XORL c, R13 \ // y0 = e ^ (e >> (25-11)) - ROLL $23, R14 \ // y1 = a >> (22-13) - MOVL d, R15 \ // y2 = f - XORL g, R14 \ // y1 = a ^ (a >> (22-13) - ROLL $27, R13 \ // y0 = (e >> (11-6)) ^ (e >> (25-6)) - LONG $0xd272b9c5; BYTE $0x0a \ // VPSRLD XMM8,XMM2,0xa /* XTMP4 = W[-2] >> 10 {BBAA} */ - XORL e, R15 \ // y2 = f^g - LONG $0xd273e1c5; BYTE $0x13 \ // VPSRLQ XMM3,XMM2,0x13 /* XTMP3 = W[-2] MY_ROR 19 {xBxA} */ - XORL c, R13 \ // y0 = e ^ (e >> (11-6)) ^ (e >> (25-6)) - ANDL c, R15 \ // y2 = (f^g)&e - LONG $0xd273e9c5; BYTE $0x11 \ // VPSRLQ XMM2,XMM2,0x11 /* XTMP2 = W[-2] MY_ROR 17 {xBxA} */ - ROLL $21, R14 \ // y1 = (a >> (13-2)) ^ (a >> (22-2)) - XORL g, R14 \ // y1 = a ^ (a >> (13-2)) ^ (a >> (22-2)) - XORL e, R15 \ // y2 = CH = ((f^g)&e)^g - ROLL $26, R13 \ // y0 = S1 = (e>>6) & (e>>11) ^ (e>>25) - LONG $0xd3efe9c5 \ // VPXOR XMM2,XMM2,XMM3 /* */ - ADDL R13, R15 \ // y2 = S1 + CH - ROLL $30, R14 \ // y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22) - ADDL _xfer+56(FP), R15 \ // y2 = k + w + S1 + CH - LONG $0xc2ef39c5 \ // VPXOR XMM8,XMM8,XMM2 /* XTMP4 = s1 {xBxA} */ - MOVL g, R13 \ // y0 = a - ADDL R15, f \ // h = h + S1 + CH + k + w - MOVL g, R15 \ // y2 = a - LONG $0x003942c4; BYTE $0xc2 \ // VPSHUFB XMM8,XMM8,XMM10 /* XTMP4 = s1 {00BA} */ - ORL a, R13 \ // y0 = a|c - ADDL f, b \ // d = d + h + S1 + CH + k + w - ANDL a, R15 \ // y2 = a&c - LONG $0xfe79c1c4; BYTE $0xc0 \ // VPADDD XMM0,XMM0,XMM8 /* XTMP0 = {..., ..., W[1], W[0]} */ - ANDL h, R13 \ // y0 = (a|c)&b - ADDL R14, f \ // h = h + S1 + CH + k + w + S0 - \ - \ // compute high s1 - \ - LONG $0xd070f9c5; BYTE $0x50 \ // VPSHUFD XMM2,XMM0,0x50 /* XTMP2 = W[-2] {DDCC} */ - ORL R15, R13 \ // y0 = MAJ = (a|c)&b)|(a&c) - ADDL R13, f \ // h = h + S1 + CH + k + w + S0 + MAJ - \ // ROTATE_ARGS - MOVL b, R13 \ // y0 = e - ROLL $18, R13 \ // y0 = e >> (25-11) - MOVL f, R14 \ // y1 = a - ROLL $23, R14 \ // y1 = a >> (22-13) - XORL b, R13 \ // y0 = e ^ (e >> (25-11)) - MOVL c, R15 \ // y2 = f - ROLL $27, R13 \ // y0 = (e >> (11-6)) ^ (e >> (25-6)) - LONG $0xd272a1c5; BYTE $0x0a \ // VPSRLD XMM11,XMM2,0xa /* XTMP5 = W[-2] >> 10 {DDCC} */ - XORL f, R14 \ // y1 = a ^ (a >> (22-13) - XORL d, R15 \ // y2 = f^g - LONG $0xd273e1c5; BYTE $0x13 \ // VPSRLQ XMM3,XMM2,0x13 /* XTMP3 = W[-2] MY_ROR 19 {xDxC} */ - XORL b, R13 \ // y0 = e ^ (e >> (11-6)) ^ (e >> (25-6)) - ANDL b, R15 \ // y2 = (f^g)&e - ROLL $21, R14 \ // y1 = (a >> (13-2)) ^ (a >> (22-2)) - LONG $0xd273e9c5; BYTE $0x11 \ // VPSRLQ XMM2,XMM2,0x11 /* XTMP2 = W[-2] MY_ROR 17 {xDxC} */ - XORL f, R14 \ // y1 = a ^ (a >> (13-2)) ^ (a >> (22-2)) - ROLL $26, R13 \ // y0 = S1 = (e>>6) & (e>>11) ^ (e>>25) - XORL d, R15 \ // y2 = CH = ((f^g)&e)^g - LONG $0xd3efe9c5 \ // VPXOR XMM2,XMM2,XMM3 /* */ - ROLL $30, R14 \ // y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22) - ADDL R13, R15 \ // y2 = S1 + CH - ADDL _xfer+60(FP), R15 \ // y2 = k + w + S1 + CH - LONG $0xdaef21c5 \ // VPXOR XMM11,XMM11,XMM2 /* XTMP5 = s1 {xDxC} */ - MOVL f, R13 \ // y0 = a - ADDL R15, e \ // h = h + S1 + CH + k + w - MOVL f, R15 \ // y2 = a - LONG $0x002142c4; BYTE $0xdc \ // VPSHUFB XMM11,XMM11,XMM12 /* XTMP5 = s1 {DC00} */ - ORL h, R13 \ // y0 = a|c - ADDL e, a \ // d = d + h + S1 + CH + k + w - ANDL h, R15 \ // y2 = a&c - LONG $0xe0fea1c5 \ // VPADDD XMM4,XMM11,XMM0 /* X0 = {W[3], W[2], W[1], W[0]} */ - ANDL g, R13 \ // y0 = (a|c)&b - ADDL R14, e \ // h = h + S1 + CH + k + w + S0 - ORL R15, R13 \ // y0 = MAJ = (a|c)&b)|(a&c) - ADDL R13, e \ // h = h + S1 + CH + k + w + S0 + MAJ - \ // ROTATE_ARGS - ROTATE_XS - -#define DO_ROUND(a, b, c, d, e, f, g, h, offset) \ - MOVL e, R13 \ // y0 = e - ROLL $18, R13 \ // y0 = e >> (25-11) - MOVL a, R14 \ // y1 = a - XORL e, R13 \ // y0 = e ^ (e >> (25-11)) - ROLL $23, R14 \ // y1 = a >> (22-13) - MOVL f, R15 \ // y2 = f - XORL a, R14 \ // y1 = a ^ (a >> (22-13) - ROLL $27, R13 \ // y0 = (e >> (11-6)) ^ (e >> (25-6)) - XORL g, R15 \ // y2 = f^g - XORL e, R13 \ // y0 = e ^ (e >> (11-6)) ^ (e >> (25-6)) - ROLL $21, R14 \ // y1 = (a >> (13-2)) ^ (a >> (22-2)) - ANDL e, R15 \ // y2 = (f^g)&e - XORL a, R14 \ // y1 = a ^ (a >> (13-2)) ^ (a >> (22-2)) - ROLL $26, R13 \ // y0 = S1 = (e>>6) & (e>>11) ^ (e>>25) - XORL g, R15 \ // y2 = CH = ((f^g)&e)^g - ADDL R13, R15 \ // y2 = S1 + CH - ROLL $30, R14 \ // y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22) - ADDL _xfer+offset(FP), R15 \ // y2 = k + w + S1 + CH - MOVL a, R13 \ // y0 = a - ADDL R15, h \ // h = h + S1 + CH + k + w - MOVL a, R15 \ // y2 = a - ORL c, R13 \ // y0 = a|c - ADDL h, d \ // d = d + h + S1 + CH + k + w - ANDL c, R15 \ // y2 = a&c - ANDL b, R13 \ // y0 = (a|c)&b - ADDL R14, h \ // h = h + S1 + CH + k + w + S0 - ORL R15, R13 \ // y0 = MAJ = (a|c)&b)|(a&c) - ADDL R13, h // h = h + S1 + CH + k + w + S0 + MAJ - -// func blockAvx(h []uint32, message []uint8, reserved0, reserved1, reserved2, reserved3 uint64) -TEXT ·blockAvx(SB), 7, $0-80 - - MOVQ h+0(FP), SI // SI: &h - MOVQ message_base+24(FP), R8 // &message - MOVQ message_len+32(FP), R9 // length of message - CMPQ R9, $0 - JEQ done_hash - ADDQ R8, R9 - MOVQ R9, reserved2+64(FP) // store end of message - - // Register definition - // a --> eax - // b --> ebx - // c --> ecx - // d --> r8d - // e --> edx - // f --> r9d - // g --> r10d - // h --> r11d - // - // y0 --> r13d - // y1 --> r14d - // y2 --> r15d - - MOVL (0*4)(SI), AX // a = H0 - MOVL (1*4)(SI), BX // b = H1 - MOVL (2*4)(SI), CX // c = H2 - MOVL (3*4)(SI), R8 // d = H3 - MOVL (4*4)(SI), DX // e = H4 - MOVL (5*4)(SI), R9 // f = H5 - MOVL (6*4)(SI), R10 // g = H6 - MOVL (7*4)(SI), R11 // h = H7 - - MOVOU bflipMask<>(SB), X13 - MOVOU shuf00BA<>(SB), X10 // shuffle xBxA -> 00BA - MOVOU shufDC00<>(SB), X12 // shuffle xDxC -> DC00 - - MOVQ message_base+24(FP), SI // SI: &message - -loop0: - LEAQ constants<>(SB), BP - - // byte swap first 16 dwords - MOVOU 0*16(SI), X4 - LONG $0x0059c2c4; BYTE $0xe5 // VPSHUFB XMM4, XMM4, XMM13 - MOVOU 1*16(SI), X5 - LONG $0x0051c2c4; BYTE $0xed // VPSHUFB XMM5, XMM5, XMM13 - MOVOU 2*16(SI), X6 - LONG $0x0049c2c4; BYTE $0xf5 // VPSHUFB XMM6, XMM6, XMM13 - MOVOU 3*16(SI), X7 - LONG $0x0041c2c4; BYTE $0xfd // VPSHUFB XMM7, XMM7, XMM13 - - MOVQ SI, reserved3+72(FP) - MOVD $0x3, DI - - // schedule 48 input dwords, by doing 3 rounds of 16 each -loop1: - LONG $0x4dfe59c5; BYTE $0x00 // VPADDD XMM9, XMM4, 0[RBP] /* Add 1st constant to first part of message */ - MOVOU X9, reserved0+48(FP) - FOUR_ROUNDS_AND_SCHED(AX, BX, CX, R8, DX, R9, R10, R11) - - LONG $0x4dfe59c5; BYTE $0x10 // VPADDD XMM9, XMM4, 16[RBP] /* Add 2nd constant to message */ - MOVOU X9, reserved0+48(FP) - FOUR_ROUNDS_AND_SCHED(DX, R9, R10, R11, AX, BX, CX, R8) - - LONG $0x4dfe59c5; BYTE $0x20 // VPADDD XMM9, XMM4, 32[RBP] /* Add 3rd constant to message */ - MOVOU X9, reserved0+48(FP) - FOUR_ROUNDS_AND_SCHED(AX, BX, CX, R8, DX, R9, R10, R11) - - LONG $0x4dfe59c5; BYTE $0x30 // VPADDD XMM9, XMM4, 48[RBP] /* Add 4th constant to message */ - MOVOU X9, reserved0+48(FP) - ADDQ $64, BP - FOUR_ROUNDS_AND_SCHED(DX, R9, R10, R11, AX, BX, CX, R8) - - SUBQ $1, DI - JNE loop1 - - MOVD $0x2, DI - -loop2: - LONG $0x4dfe59c5; BYTE $0x00 // VPADDD XMM9, XMM4, 0[RBP] /* Add 1st constant to first part of message */ - MOVOU X9, reserved0+48(FP) - DO_ROUND( AX, BX, CX, R8, DX, R9, R10, R11, 48) - DO_ROUND(R11, AX, BX, CX, R8, DX, R9, R10, 52) - DO_ROUND(R10, R11, AX, BX, CX, R8, DX, R9, 56) - DO_ROUND( R9, R10, R11, AX, BX, CX, R8, DX, 60) - - LONG $0x4dfe51c5; BYTE $0x10 // VPADDD XMM9, XMM5, 16[RBP] /* Add 2nd constant to message */ - MOVOU X9, reserved0+48(FP) - ADDQ $32, BP - DO_ROUND( DX, R9, R10, R11, AX, BX, CX, R8, 48) - DO_ROUND( R8, DX, R9, R10, R11, AX, BX, CX, 52) - DO_ROUND( CX, R8, DX, R9, R10, R11, AX, BX, 56) - DO_ROUND( BX, CX, R8, DX, R9, R10, R11, AX, 60) - - MOVOU X6, X4 - MOVOU X7, X5 - - SUBQ $1, DI - JNE loop2 - - MOVQ h+0(FP), SI // SI: &h - ADDL (0*4)(SI), AX // H0 = a + H0 - MOVL AX, (0*4)(SI) - ADDL (1*4)(SI), BX // H1 = b + H1 - MOVL BX, (1*4)(SI) - ADDL (2*4)(SI), CX // H2 = c + H2 - MOVL CX, (2*4)(SI) - ADDL (3*4)(SI), R8 // H3 = d + H3 - MOVL R8, (3*4)(SI) - ADDL (4*4)(SI), DX // H4 = e + H4 - MOVL DX, (4*4)(SI) - ADDL (5*4)(SI), R9 // H5 = f + H5 - MOVL R9, (5*4)(SI) - ADDL (6*4)(SI), R10 // H6 = g + H6 - MOVL R10, (6*4)(SI) - ADDL (7*4)(SI), R11 // H7 = h + H7 - MOVL R11, (7*4)(SI) - - MOVQ reserved3+72(FP), SI - ADDQ $64, SI - CMPQ reserved2+64(FP), SI - JNE loop0 - -done_hash: - RET - -// Constants table -DATA constants<>+0x0(SB)/8, $0x71374491428a2f98 -DATA constants<>+0x8(SB)/8, $0xe9b5dba5b5c0fbcf -DATA constants<>+0x10(SB)/8, $0x59f111f13956c25b -DATA constants<>+0x18(SB)/8, $0xab1c5ed5923f82a4 -DATA constants<>+0x20(SB)/8, $0x12835b01d807aa98 -DATA constants<>+0x28(SB)/8, $0x550c7dc3243185be -DATA constants<>+0x30(SB)/8, $0x80deb1fe72be5d74 -DATA constants<>+0x38(SB)/8, $0xc19bf1749bdc06a7 -DATA constants<>+0x40(SB)/8, $0xefbe4786e49b69c1 -DATA constants<>+0x48(SB)/8, $0x240ca1cc0fc19dc6 -DATA constants<>+0x50(SB)/8, $0x4a7484aa2de92c6f -DATA constants<>+0x58(SB)/8, $0x76f988da5cb0a9dc -DATA constants<>+0x60(SB)/8, $0xa831c66d983e5152 -DATA constants<>+0x68(SB)/8, $0xbf597fc7b00327c8 -DATA constants<>+0x70(SB)/8, $0xd5a79147c6e00bf3 -DATA constants<>+0x78(SB)/8, $0x1429296706ca6351 -DATA constants<>+0x80(SB)/8, $0x2e1b213827b70a85 -DATA constants<>+0x88(SB)/8, $0x53380d134d2c6dfc -DATA constants<>+0x90(SB)/8, $0x766a0abb650a7354 -DATA constants<>+0x98(SB)/8, $0x92722c8581c2c92e -DATA constants<>+0xa0(SB)/8, $0xa81a664ba2bfe8a1 -DATA constants<>+0xa8(SB)/8, $0xc76c51a3c24b8b70 -DATA constants<>+0xb0(SB)/8, $0xd6990624d192e819 -DATA constants<>+0xb8(SB)/8, $0x106aa070f40e3585 -DATA constants<>+0xc0(SB)/8, $0x1e376c0819a4c116 -DATA constants<>+0xc8(SB)/8, $0x34b0bcb52748774c -DATA constants<>+0xd0(SB)/8, $0x4ed8aa4a391c0cb3 -DATA constants<>+0xd8(SB)/8, $0x682e6ff35b9cca4f -DATA constants<>+0xe0(SB)/8, $0x78a5636f748f82ee -DATA constants<>+0xe8(SB)/8, $0x8cc7020884c87814 -DATA constants<>+0xf0(SB)/8, $0xa4506ceb90befffa -DATA constants<>+0xf8(SB)/8, $0xc67178f2bef9a3f7 - -DATA bflipMask<>+0x00(SB)/8, $0x0405060700010203 -DATA bflipMask<>+0x08(SB)/8, $0x0c0d0e0f08090a0b - -DATA shuf00BA<>+0x00(SB)/8, $0x0b0a090803020100 -DATA shuf00BA<>+0x08(SB)/8, $0xFFFFFFFFFFFFFFFF - -DATA shufDC00<>+0x00(SB)/8, $0xFFFFFFFFFFFFFFFF -DATA shufDC00<>+0x08(SB)/8, $0x0b0a090803020100 - -GLOBL constants<>(SB), 8, $256 -GLOBL bflipMask<>(SB), (NOPTR+RODATA), $16 -GLOBL shuf00BA<>(SB), (NOPTR+RODATA), $16 -GLOBL shufDC00<>(SB), (NOPTR+RODATA), $16 diff --git a/vendor/github.com/minio/sha256-simd/sha256blockSha_amd64.go b/vendor/github.com/minio/sha256-simd/sha256blockSha_amd64.go index 483689ef0..bef949419 100644 --- a/vendor/github.com/minio/sha256-simd/sha256blockSha_amd64.go +++ b/vendor/github.com/minio/sha256-simd/sha256blockSha_amd64.go @@ -1,4 +1,4 @@ -//+build !noasm,!appengine +//+build !noasm,!appengine,gc package sha256 diff --git a/vendor/github.com/minio/sha256-simd/sha256blockSsse_amd64.go b/vendor/github.com/minio/sha256-simd/sha256blockSsse_amd64.go deleted file mode 100644 index 1ae2320bd..000000000 --- a/vendor/github.com/minio/sha256-simd/sha256blockSsse_amd64.go +++ /dev/null @@ -1,22 +0,0 @@ -//+build !noasm,!appengine - -/* - * Minio Cloud Storage, (C) 2016 Minio, Inc. - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -package sha256 - -//go:noescape -func blockSsse(h []uint32, message []uint8, reserved0, reserved1, reserved2, reserved3 uint64) diff --git a/vendor/github.com/minio/sha256-simd/sha256blockSsse_amd64.s b/vendor/github.com/minio/sha256-simd/sha256blockSsse_amd64.s deleted file mode 100644 index 7afb45c87..000000000 --- a/vendor/github.com/minio/sha256-simd/sha256blockSsse_amd64.s +++ /dev/null @@ -1,429 +0,0 @@ -//+build !noasm,!appengine - -// SHA256 implementation for SSSE3 - -// -// Minio Cloud Storage, (C) 2016 Minio, Inc. -// -// Licensed under the Apache License, Version 2.0 (the "License"); -// you may not use this file except in compliance with the License. -// You may obtain a copy of the License at -// -// http://www.apache.org/licenses/LICENSE-2.0 -// -// Unless required by applicable law or agreed to in writing, software -// distributed under the License is distributed on an "AS IS" BASIS, -// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -// See the License for the specific language governing permissions and -// limitations under the License. -// - -// -// This code is based on an Intel White-Paper: -// "Fast SHA-256 Implementations on Intel Architecture Processors" -// -// together with the reference implementation from the following authors: -// James Guilford <james.guilford@intel.com> -// Kirk Yap <kirk.s.yap@intel.com> -// Tim Chen <tim.c.chen@linux.intel.com> -// -// For Golang it has been converted to Plan 9 assembly with the help of -// github.com/minio/asm2plan9s to assemble Intel instructions to their Plan9 -// equivalents -// - -#include "textflag.h" - -#define ROTATE_XS \ - MOVOU X4, X15 \ - MOVOU X5, X4 \ - MOVOU X6, X5 \ - MOVOU X7, X6 \ - MOVOU X15, X7 - -// compute s0 four at a time and s1 two at a time -// compute W[-16] + W[-7] 4 at a time -#define FOUR_ROUNDS_AND_SCHED(a, b, c, d, e, f, g, h) \ - MOVL e, R13 \ // y0 = e - ROLL $18, R13 \ // y0 = e >> (25-11) - MOVL a, R14 \ // y1 = a - MOVOU X7, X0 \ - LONG $0x0f3a0f66; WORD $0x04c6 \ // PALIGNR XMM0,XMM6,0x4 /* XTMP0 = W[-7] */ - ROLL $23, R14 \ // y1 = a >> (22-13) - XORL e, R13 \ // y0 = e ^ (e >> (25-11)) - MOVL f, R15 \ // y2 = f - ROLL $27, R13 \ // y0 = (e >> (11-6)) ^ (e >> (25-6)) - XORL a, R14 \ // y1 = a ^ (a >> (22-13) - XORL g, R15 \ // y2 = f^g - LONG $0xc4fe0f66 \ // PADDD XMM0,XMM4 /* XTMP0 = W[-7] + W[-16] */ - XORL e, R13 \ // y0 = e ^ (e >> (11-6)) ^ (e >> (25-6) ) - ANDL e, R15 \ // y2 = (f^g)&e - ROLL $21, R14 \ // y1 = (a >> (13-2)) ^ (a >> (22-2)) - \ - \ // compute s0 - \ - MOVOU X5, X1 \ - LONG $0x0f3a0f66; WORD $0x04cc \ // PALIGNR XMM1,XMM4,0x4 /* XTMP1 = W[-15] */ - XORL a, R14 \ // y1 = a ^ (a >> (13-2)) ^ (a >> (22-2)) - ROLL $26, R13 \ // y0 = S1 = (e>>6) & (e>>11) ^ (e>>25) - XORL g, R15 \ // y2 = CH = ((f^g)&e)^g - ROLL $30, R14 \ // y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22) - ADDL R13, R15 \ // y2 = S1 + CH - ADDL _xfer+48(FP), R15 \ // y2 = k + w + S1 + CH - MOVL a, R13 \ // y0 = a - ADDL R15, h \ // h = h + S1 + CH + k + w - \ // ROTATE_ARGS - MOVL a, R15 \ // y2 = a - MOVOU X1, X2 \ - LONG $0xd2720f66; BYTE $0x07 \ // PSRLD XMM2,0x7 /* */ - ORL c, R13 \ // y0 = a|c - ADDL h, d \ // d = d + h + S1 + CH + k + w - ANDL c, R15 \ // y2 = a&c - MOVOU X1, X3 \ - LONG $0xf3720f66; BYTE $0x19 \ // PSLLD XMM3,0x19 /* */ - ANDL b, R13 \ // y0 = (a|c)&b - ADDL R14, h \ // h = h + S1 + CH + k + w + S0 - LONG $0xdaeb0f66 \ // POR XMM3,XMM2 /* XTMP1 = W[-15] MY_ROR 7 */ - ORL R15, R13 \ // y0 = MAJ = (a|c)&b)|(a&c) - ADDL R13, h \ // h = h + S1 + CH + k + w + S0 + MAJ - \ // ROTATE_ARGS - MOVL d, R13 \ // y0 = e - MOVL h, R14 \ // y1 = a - ROLL $18, R13 \ // y0 = e >> (25-11) - XORL d, R13 \ // y0 = e ^ (e >> (25-11)) - MOVL e, R15 \ // y2 = f - ROLL $23, R14 \ // y1 = a >> (22-13) - MOVOU X1, X2 \ - LONG $0xd2720f66; BYTE $0x12 \ // PSRLD XMM2,0x12 /* */ - XORL h, R14 \ // y1 = a ^ (a >> (22-13) - ROLL $27, R13 \ // y0 = (e >> (11-6)) ^ (e >> (25-6)) - XORL f, R15 \ // y2 = f^g - MOVOU X1, X8 \ - LONG $0x720f4166; WORD $0x03d0 \ // PSRLD XMM8,0x3 /* XTMP4 = W[-15] >> 3 */ - ROLL $21, R14 \ // y1 = (a >> (13-2)) ^ (a >> (22-2)) - XORL d, R13 \ // y0 = e ^ (e >> (11-6)) ^ (e >> (25-6)) - ANDL d, R15 \ // y2 = (f^g)&e - ROLL $26, R13 \ // y0 = S1 = (e>>6) & (e>>11) ^ (e>>25) - LONG $0xf1720f66; BYTE $0x0e \ // PSLLD XMM1,0xe /* */ - XORL h, R14 \ // y1 = a ^ (a >> (13-2)) ^ (a >> (22-2)) - XORL f, R15 \ // y2 = CH = ((f^g)&e)^g - LONG $0xd9ef0f66 \ // PXOR XMM3,XMM1 /* */ - ADDL R13, R15 \ // y2 = S1 + CH - ADDL _xfer+52(FP), R15 \ // y2 = k + w + S1 + CH - ROLL $30, R14 \ // y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22) - LONG $0xdaef0f66 \ // PXOR XMM3,XMM2 /* XTMP1 = W[-15] MY_ROR 7 ^ W[-15] MY_ROR */ - MOVL h, R13 \ // y0 = a - ADDL R15, g \ // h = h + S1 + CH + k + w - MOVL h, R15 \ // y2 = a - MOVOU X3, X1 \ - LONG $0xef0f4166; BYTE $0xc8 \ // PXOR XMM1,XMM8 /* XTMP1 = s0 */ - ORL b, R13 \ // y0 = a|c - ADDL g, c \ // d = d + h + S1 + CH + k + w - ANDL b, R15 \ // y2 = a&c - \ - \ // compute low s1 - \ - LONG $0xd7700f66; BYTE $0xfa \ // PSHUFD XMM2,XMM7,0xfa /* XTMP2 = W[-2] {BBAA} */ - ANDL a, R13 \ // y0 = (a|c)&b - ADDL R14, g \ // h = h + S1 + CH + k + w + S0 - LONG $0xc1fe0f66 \ // PADDD XMM0,XMM1 /* XTMP0 = W[-16] + W[-7] + s0 */ - ORL R15, R13 \ // y0 = MAJ = (a|c)&b)|(a&c) - ADDL R13, g \ // h = h + S1 + CH + k + w + S0 + MAJ - \ // ROTATE_ARGS - MOVL c, R13 \ // y0 = e - MOVL g, R14 \ // y1 = a - ROLL $18, R13 \ // y0 = e >> (25-11) - XORL c, R13 \ // y0 = e ^ (e >> (25-11)) - ROLL $23, R14 \ // y1 = a >> (22-13) - MOVL d, R15 \ // y2 = f - XORL g, R14 \ // y1 = a ^ (a >> (22-13) - ROLL $27, R13 \ // y0 = (e >> (11-6)) ^ (e >> (25-6)) - MOVOU X2, X8 \ - LONG $0x720f4166; WORD $0x0ad0 \ // PSRLD XMM8,0xa /* XTMP4 = W[-2] >> 10 {BBAA} */ - XORL e, R15 \ // y2 = f^g - MOVOU X2, X3 \ - LONG $0xd3730f66; BYTE $0x13 \ // PSRLQ XMM3,0x13 /* XTMP3 = W[-2] MY_ROR 19 {xBxA} */ - XORL c, R13 \ // y0 = e ^ (e >> (11-6)) ^ (e >> (25-6)) - ANDL c, R15 \ // y2 = (f^g)&e - LONG $0xd2730f66; BYTE $0x11 \ // PSRLQ XMM2,0x11 /* XTMP2 = W[-2] MY_ROR 17 {xBxA} */ - ROLL $21, R14 \ // y1 = (a >> (13-2)) ^ (a >> (22-2)) - XORL g, R14 \ // y1 = a ^ (a >> (13-2)) ^ (a >> (22-2)) - XORL e, R15 \ // y2 = CH = ((f^g)&e)^g - ROLL $26, R13 \ // y0 = S1 = (e>>6) & (e>>11) ^ (e>>25) - LONG $0xd3ef0f66 \ // PXOR XMM2,XMM3 /* */ - ADDL R13, R15 \ // y2 = S1 + CH - ROLL $30, R14 \ // y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22) - ADDL _xfer+56(FP), R15 \ // y2 = k + w + S1 + CH - LONG $0xef0f4466; BYTE $0xc2 \ // PXOR XMM8,XMM2 /* XTMP4 = s1 {xBxA} */ - MOVL g, R13 \ // y0 = a - ADDL R15, f \ // h = h + S1 + CH + k + w - MOVL g, R15 \ // y2 = a - LONG $0x380f4566; WORD $0xc200 \ // PSHUFB XMM8,XMM10 /* XTMP4 = s1 {00BA} */ - ORL a, R13 \ // y0 = a|c - ADDL f, b \ // d = d + h + S1 + CH + k + w - ANDL a, R15 \ // y2 = a&c - LONG $0xfe0f4166; BYTE $0xc0 \ // PADDD XMM0,XMM8 /* XTMP0 = {..., ..., W[1], W[0]} */ - ANDL h, R13 \ // y0 = (a|c)&b - ADDL R14, f \ // h = h + S1 + CH + k + w + S0 - \ - \ // compute high s1 - \ - LONG $0xd0700f66; BYTE $0x50 \ // PSHUFD XMM2,XMM0,0x50 /* XTMP2 = W[-2] {DDCC} */ - ORL R15, R13 \ // y0 = MAJ = (a|c)&b)|(a&c) - ADDL R13, f \ // h = h + S1 + CH + k + w + S0 + MAJ - \ // ROTATE_ARGS - MOVL b, R13 \ // y0 = e - ROLL $18, R13 \ // y0 = e >> (25-11) - MOVL f, R14 \ // y1 = a - ROLL $23, R14 \ // y1 = a >> (22-13) - XORL b, R13 \ // y0 = e ^ (e >> (25-11)) - MOVL c, R15 \ // y2 = f - ROLL $27, R13 \ // y0 = (e >> (11-6)) ^ (e >> (25-6)) - MOVOU X2, X11 \ - LONG $0x720f4166; WORD $0x0ad3 \ // PSRLD XMM11,0xa /* XTMP5 = W[-2] >> 10 {DDCC} */ - XORL f, R14 \ // y1 = a ^ (a >> (22-13) - XORL d, R15 \ // y2 = f^g - MOVOU X2, X3 \ - LONG $0xd3730f66; BYTE $0x13 \ // PSRLQ XMM3,0x13 /* XTMP3 = W[-2] MY_ROR 19 {xDxC} */ - XORL b, R13 \ // y0 = e ^ (e >> (11-6)) ^ (e >> (25-6)) - ANDL b, R15 \ // y2 = (f^g)&e - ROLL $21, R14 \ // y1 = (a >> (13-2)) ^ (a >> (22-2)) - LONG $0xd2730f66; BYTE $0x11 \ // PSRLQ XMM2,0x11 /* XTMP2 = W[-2] MY_ROR 17 {xDxC} */ - XORL f, R14 \ // y1 = a ^ (a >> (13-2)) ^ (a >> (22-2)) - ROLL $26, R13 \ // y0 = S1 = (e>>6) & (e>>11) ^ (e>>25) - XORL d, R15 \ // y2 = CH = ((f^g)&e)^g - LONG $0xd3ef0f66 \ // PXOR XMM2,XMM3 /* */ - ROLL $30, R14 \ // y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22) - ADDL R13, R15 \ // y2 = S1 + CH - ADDL _xfer+60(FP), R15 \ // y2 = k + w + S1 + CH - LONG $0xef0f4466; BYTE $0xda \ // PXOR XMM11,XMM2 /* XTMP5 = s1 {xDxC} */ - MOVL f, R13 \ // y0 = a - ADDL R15, e \ // h = h + S1 + CH + k + w - MOVL f, R15 \ // y2 = a - LONG $0x380f4566; WORD $0xdc00 \ // PSHUFB XMM11,XMM12 /* XTMP5 = s1 {DC00} */ - ORL h, R13 \ // y0 = a|c - ADDL e, a \ // d = d + h + S1 + CH + k + w - ANDL h, R15 \ // y2 = a&c - MOVOU X11, X4 \ - LONG $0xe0fe0f66 \ // PADDD XMM4,XMM0 /* X0 = {W[3], W[2], W[1], W[0]} */ - ANDL g, R13 \ // y0 = (a|c)&b - ADDL R14, e \ // h = h + S1 + CH + k + w + S0 - ORL R15, R13 \ // y0 = MAJ = (a|c)&b)|(a&c) - ADDL R13, e \ // h = h + S1 + CH + k + w + S0 + MAJ - \ // ROTATE_ARGS - ROTATE_XS - -#define DO_ROUND(a, b, c, d, e, f, g, h, offset) \ - MOVL e, R13 \ // y0 = e - ROLL $18, R13 \ // y0 = e >> (25-11) - MOVL a, R14 \ // y1 = a - XORL e, R13 \ // y0 = e ^ (e >> (25-11)) - ROLL $23, R14 \ // y1 = a >> (22-13) - MOVL f, R15 \ // y2 = f - XORL a, R14 \ // y1 = a ^ (a >> (22-13) - ROLL $27, R13 \ // y0 = (e >> (11-6)) ^ (e >> (25-6)) - XORL g, R15 \ // y2 = f^g - XORL e, R13 \ // y0 = e ^ (e >> (11-6)) ^ (e >> (25-6)) - ROLL $21, R14 \ // y1 = (a >> (13-2)) ^ (a >> (22-2)) - ANDL e, R15 \ // y2 = (f^g)&e - XORL a, R14 \ // y1 = a ^ (a >> (13-2)) ^ (a >> (22-2)) - ROLL $26, R13 \ // y0 = S1 = (e>>6) & (e>>11) ^ (e>>25) - XORL g, R15 \ // y2 = CH = ((f^g)&e)^g - ADDL R13, R15 \ // y2 = S1 + CH - ROLL $30, R14 \ // y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22) - ADDL _xfer+offset(FP), R15 \ // y2 = k + w + S1 + CH - MOVL a, R13 \ // y0 = a - ADDL R15, h \ // h = h + S1 + CH + k + w - MOVL a, R15 \ // y2 = a - ORL c, R13 \ // y0 = a|c - ADDL h, d \ // d = d + h + S1 + CH + k + w - ANDL c, R15 \ // y2 = a&c - ANDL b, R13 \ // y0 = (a|c)&b - ADDL R14, h \ // h = h + S1 + CH + k + w + S0 - ORL R15, R13 \ // y0 = MAJ = (a|c)&b)|(a&c) - ADDL R13, h // h = h + S1 + CH + k + w + S0 + MAJ - -// func blockSsse(h []uint32, message []uint8, reserved0, reserved1, reserved2, reserved3 uint64) -TEXT ·blockSsse(SB), 7, $0-80 - - MOVQ h+0(FP), SI // SI: &h - MOVQ message_base+24(FP), R8 // &message - MOVQ message_len+32(FP), R9 // length of message - CMPQ R9, $0 - JEQ done_hash - ADDQ R8, R9 - MOVQ R9, reserved2+64(FP) // store end of message - - // Register definition - // a --> eax - // b --> ebx - // c --> ecx - // d --> r8d - // e --> edx - // f --> r9d - // g --> r10d - // h --> r11d - // - // y0 --> r13d - // y1 --> r14d - // y2 --> r15d - - MOVL (0*4)(SI), AX // a = H0 - MOVL (1*4)(SI), BX // b = H1 - MOVL (2*4)(SI), CX // c = H2 - MOVL (3*4)(SI), R8 // d = H3 - MOVL (4*4)(SI), DX // e = H4 - MOVL (5*4)(SI), R9 // f = H5 - MOVL (6*4)(SI), R10 // g = H6 - MOVL (7*4)(SI), R11 // h = H7 - - MOVOU bflipMask<>(SB), X13 - MOVOU shuf00BA<>(SB), X10 // shuffle xBxA -> 00BA - MOVOU shufDC00<>(SB), X12 // shuffle xDxC -> DC00 - - MOVQ message_base+24(FP), SI // SI: &message - -loop0: - LEAQ constants<>(SB), BP - - // byte swap first 16 dwords - MOVOU 0*16(SI), X4 - LONG $0x380f4166; WORD $0xe500 // PSHUFB XMM4, XMM13 - MOVOU 1*16(SI), X5 - LONG $0x380f4166; WORD $0xed00 // PSHUFB XMM5, XMM13 - MOVOU 2*16(SI), X6 - LONG $0x380f4166; WORD $0xf500 // PSHUFB XMM6, XMM13 - MOVOU 3*16(SI), X7 - LONG $0x380f4166; WORD $0xfd00 // PSHUFB XMM7, XMM13 - - MOVQ SI, reserved3+72(FP) - MOVD $0x3, DI - - // Align - // nop WORD PTR [rax+rax*1+0x0] - - // schedule 48 input dwords, by doing 3 rounds of 16 each -loop1: - MOVOU X4, X9 - LONG $0xfe0f4466; WORD $0x004d // PADDD XMM9, 0[RBP] /* Add 1st constant to first part of message */ - MOVOU X9, reserved0+48(FP) - FOUR_ROUNDS_AND_SCHED(AX, BX, CX, R8, DX, R9, R10, R11) - - MOVOU X4, X9 - LONG $0xfe0f4466; WORD $0x104d // PADDD XMM9, 16[RBP] /* Add 2nd constant to message */ - MOVOU X9, reserved0+48(FP) - FOUR_ROUNDS_AND_SCHED(DX, R9, R10, R11, AX, BX, CX, R8) - - MOVOU X4, X9 - LONG $0xfe0f4466; WORD $0x204d // PADDD XMM9, 32[RBP] /* Add 3rd constant to message */ - MOVOU X9, reserved0+48(FP) - FOUR_ROUNDS_AND_SCHED(AX, BX, CX, R8, DX, R9, R10, R11) - - MOVOU X4, X9 - LONG $0xfe0f4466; WORD $0x304d // PADDD XMM9, 48[RBP] /* Add 4th constant to message */ - MOVOU X9, reserved0+48(FP) - ADDQ $64, BP - FOUR_ROUNDS_AND_SCHED(DX, R9, R10, R11, AX, BX, CX, R8) - - SUBQ $1, DI - JNE loop1 - - MOVD $0x2, DI - -loop2: - MOVOU X4, X9 - LONG $0xfe0f4466; WORD $0x004d // PADDD XMM9, 0[RBP] /* Add 1st constant to first part of message */ - MOVOU X9, reserved0+48(FP) - DO_ROUND( AX, BX, CX, R8, DX, R9, R10, R11, 48) - DO_ROUND(R11, AX, BX, CX, R8, DX, R9, R10, 52) - DO_ROUND(R10, R11, AX, BX, CX, R8, DX, R9, 56) - DO_ROUND( R9, R10, R11, AX, BX, CX, R8, DX, 60) - - MOVOU X5, X9 - LONG $0xfe0f4466; WORD $0x104d // PADDD XMM9, 16[RBP] /* Add 2nd constant to message */ - MOVOU X9, reserved0+48(FP) - ADDQ $32, BP - DO_ROUND( DX, R9, R10, R11, AX, BX, CX, R8, 48) - DO_ROUND( R8, DX, R9, R10, R11, AX, BX, CX, 52) - DO_ROUND( CX, R8, DX, R9, R10, R11, AX, BX, 56) - DO_ROUND( BX, CX, R8, DX, R9, R10, R11, AX, 60) - - MOVOU X6, X4 - MOVOU X7, X5 - - SUBQ $1, DI - JNE loop2 - - MOVQ h+0(FP), SI // SI: &h - ADDL (0*4)(SI), AX // H0 = a + H0 - MOVL AX, (0*4)(SI) - ADDL (1*4)(SI), BX // H1 = b + H1 - MOVL BX, (1*4)(SI) - ADDL (2*4)(SI), CX // H2 = c + H2 - MOVL CX, (2*4)(SI) - ADDL (3*4)(SI), R8 // H3 = d + H3 - MOVL R8, (3*4)(SI) - ADDL (4*4)(SI), DX // H4 = e + H4 - MOVL DX, (4*4)(SI) - ADDL (5*4)(SI), R9 // H5 = f + H5 - MOVL R9, (5*4)(SI) - ADDL (6*4)(SI), R10 // H6 = g + H6 - MOVL R10, (6*4)(SI) - ADDL (7*4)(SI), R11 // H7 = h + H7 - MOVL R11, (7*4)(SI) - - MOVQ reserved3+72(FP), SI - ADDQ $64, SI - CMPQ reserved2+64(FP), SI - JNE loop0 - -done_hash: - RET - -// Constants table -DATA constants<>+0x0(SB)/8, $0x71374491428a2f98 -DATA constants<>+0x8(SB)/8, $0xe9b5dba5b5c0fbcf -DATA constants<>+0x10(SB)/8, $0x59f111f13956c25b -DATA constants<>+0x18(SB)/8, $0xab1c5ed5923f82a4 -DATA constants<>+0x20(SB)/8, $0x12835b01d807aa98 -DATA constants<>+0x28(SB)/8, $0x550c7dc3243185be -DATA constants<>+0x30(SB)/8, $0x80deb1fe72be5d74 -DATA constants<>+0x38(SB)/8, $0xc19bf1749bdc06a7 -DATA constants<>+0x40(SB)/8, $0xefbe4786e49b69c1 -DATA constants<>+0x48(SB)/8, $0x240ca1cc0fc19dc6 -DATA constants<>+0x50(SB)/8, $0x4a7484aa2de92c6f -DATA constants<>+0x58(SB)/8, $0x76f988da5cb0a9dc -DATA constants<>+0x60(SB)/8, $0xa831c66d983e5152 -DATA constants<>+0x68(SB)/8, $0xbf597fc7b00327c8 -DATA constants<>+0x70(SB)/8, $0xd5a79147c6e00bf3 -DATA constants<>+0x78(SB)/8, $0x1429296706ca6351 -DATA constants<>+0x80(SB)/8, $0x2e1b213827b70a85 -DATA constants<>+0x88(SB)/8, $0x53380d134d2c6dfc -DATA constants<>+0x90(SB)/8, $0x766a0abb650a7354 -DATA constants<>+0x98(SB)/8, $0x92722c8581c2c92e -DATA constants<>+0xa0(SB)/8, $0xa81a664ba2bfe8a1 -DATA constants<>+0xa8(SB)/8, $0xc76c51a3c24b8b70 -DATA constants<>+0xb0(SB)/8, $0xd6990624d192e819 -DATA constants<>+0xb8(SB)/8, $0x106aa070f40e3585 -DATA constants<>+0xc0(SB)/8, $0x1e376c0819a4c116 -DATA constants<>+0xc8(SB)/8, $0x34b0bcb52748774c -DATA constants<>+0xd0(SB)/8, $0x4ed8aa4a391c0cb3 -DATA constants<>+0xd8(SB)/8, $0x682e6ff35b9cca4f -DATA constants<>+0xe0(SB)/8, $0x78a5636f748f82ee -DATA constants<>+0xe8(SB)/8, $0x8cc7020884c87814 -DATA constants<>+0xf0(SB)/8, $0xa4506ceb90befffa -DATA constants<>+0xf8(SB)/8, $0xc67178f2bef9a3f7 - -DATA bflipMask<>+0x00(SB)/8, $0x0405060700010203 -DATA bflipMask<>+0x08(SB)/8, $0x0c0d0e0f08090a0b - -DATA shuf00BA<>+0x00(SB)/8, $0x0b0a090803020100 -DATA shuf00BA<>+0x08(SB)/8, $0xFFFFFFFFFFFFFFFF - -DATA shufDC00<>+0x00(SB)/8, $0xFFFFFFFFFFFFFFFF -DATA shufDC00<>+0x08(SB)/8, $0x0b0a090803020100 - -GLOBL constants<>(SB), 8, $256 -GLOBL bflipMask<>(SB), (NOPTR+RODATA), $16 -GLOBL shuf00BA<>(SB), (NOPTR+RODATA), $16 -GLOBL shufDC00<>(SB), (NOPTR+RODATA), $16 diff --git a/vendor/github.com/minio/sha256-simd/sha256block_amd64.go b/vendor/github.com/minio/sha256-simd/sha256block_amd64.go index 1c4d97f0c..0c48d45f8 100644 --- a/vendor/github.com/minio/sha256-simd/sha256block_amd64.go +++ b/vendor/github.com/minio/sha256-simd/sha256block_amd64.go @@ -1,4 +1,4 @@ -//+build !noasm,!appengine +//+build !noasm,!appengine,gc /* * Minio Cloud Storage, (C) 2016 Minio, Inc. @@ -18,36 +18,10 @@ package sha256 -func blockArmGo(dig *digest, p []byte) {} - -func blockAvxGo(dig *digest, p []byte) { - - h := []uint32{dig.h[0], dig.h[1], dig.h[2], dig.h[3], dig.h[4], dig.h[5], dig.h[6], dig.h[7]} - - blockAvx(h[:], p[:], 0, 0, 0, 0) - - dig.h[0], dig.h[1], dig.h[2], dig.h[3], dig.h[4], dig.h[5], dig.h[6], dig.h[7] = h[0], h[1], h[2], h[3], h[4], h[5], h[6], h[7] -} - -func blockAvx2Go(dig *digest, p []byte) { - - h := []uint32{dig.h[0], dig.h[1], dig.h[2], dig.h[3], dig.h[4], dig.h[5], dig.h[6], dig.h[7]} - - blockAvx2(h[:], p[:]) - - dig.h[0], dig.h[1], dig.h[2], dig.h[3], dig.h[4], dig.h[5], dig.h[6], dig.h[7] = h[0], h[1], h[2], h[3], h[4], h[5], h[6], h[7] -} - -func blockSsseGo(dig *digest, p []byte) { - - h := []uint32{dig.h[0], dig.h[1], dig.h[2], dig.h[3], dig.h[4], dig.h[5], dig.h[6], dig.h[7]} - - blockSsse(h[:], p[:], 0, 0, 0, 0) - - dig.h[0], dig.h[1], dig.h[2], dig.h[3], dig.h[4], dig.h[5], dig.h[6], dig.h[7] = h[0], h[1], h[2], h[3], h[4], h[5], h[6], h[7] +func blockArmGo(dig *digest, p []byte) { + panic("blockArmGo called unexpectedly") } func blockShaGo(dig *digest, p []byte) { - blockSha(&dig.h, p) } diff --git a/vendor/github.com/minio/sha256-simd/sha256block_arm64.go b/vendor/github.com/minio/sha256-simd/sha256block_arm64.go index 0979c20ae..58ccf6eb5 100644 --- a/vendor/github.com/minio/sha256-simd/sha256block_arm64.go +++ b/vendor/github.com/minio/sha256-simd/sha256block_arm64.go @@ -1,4 +1,4 @@ -//+build !noasm,!appengine +//+build !noasm,!appengine,gc /* * Minio Cloud Storage, (C) 2016 Minio, Inc. @@ -18,10 +18,9 @@ package sha256 -func blockAvx2Go(dig *digest, p []byte) {} -func blockAvxGo(dig *digest, p []byte) {} -func blockSsseGo(dig *digest, p []byte) {} -func blockShaGo(dig *digest, p []byte) {} +func blockShaGo(dig *digest, p []byte) { + panic("blockShaGoc called unexpectedly") +} //go:noescape func blockArm(h []uint32, message []uint8) diff --git a/vendor/github.com/minio/sha256-simd/sha256block_other.go b/vendor/github.com/minio/sha256-simd/sha256block_other.go index 0187c950a..ec586c060 100644 --- a/vendor/github.com/minio/sha256-simd/sha256block_other.go +++ b/vendor/github.com/minio/sha256-simd/sha256block_other.go @@ -1,4 +1,4 @@ -//+build appengine noasm !amd64,!arm64 +//+build appengine noasm !amd64,!arm64 !gc /* * Minio Cloud Storage, (C) 2019 Minio, Inc. @@ -18,8 +18,11 @@ package sha256 -func blockAvx2Go(dig *digest, p []byte) {} -func blockAvxGo(dig *digest, p []byte) {} -func blockSsseGo(dig *digest, p []byte) {} -func blockShaGo(dig *digest, p []byte) {} -func blockArmGo(dig *digest, p []byte) {} +func blockShaGo(dig *digest, p []byte) { + panic("blockShaGo called unexpectedly") + +} + +func blockArmGo(dig *digest, p []byte) { + panic("blockArmGo called unexpectedly") +} |