feat: Waku v2 bridge

Issue #12610
This commit is contained in:
Michal Iskierko
2023-11-12 13:29:38 +01:00
parent 56e7bd01ca
commit 6d31343205
6716 changed files with 1982502 additions and 5891 deletions
+246
View File
@@ -0,0 +1,246 @@
// Copyright 2016 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.
// Package blake2s implements the BLAKE2s hash algorithm defined by RFC 7693
// and the extendable output function (XOF) BLAKE2Xs.
//
// BLAKE2s is optimized for 8- to 32-bit platforms and produces digests of any
// size between 1 and 32 bytes.
// For a detailed specification of BLAKE2s see https://blake2.net/blake2.pdf
// and for BLAKE2Xs see https://blake2.net/blake2x.pdf
//
// If you aren't sure which function you need, use BLAKE2s (Sum256 or New256).
// If you need a secret-key MAC (message authentication code), use the New256
// function with a non-nil key.
//
// BLAKE2X is a construction to compute hash values larger than 32 bytes. It
// can produce hash values between 0 and 65535 bytes.
package blake2s // import "golang.org/x/crypto/blake2s"
import (
"encoding/binary"
"errors"
"hash"
)
const (
// The blocksize of BLAKE2s in bytes.
BlockSize = 64
// The hash size of BLAKE2s-256 in bytes.
Size = 32
// The hash size of BLAKE2s-128 in bytes.
Size128 = 16
)
var errKeySize = errors.New("blake2s: invalid key size")
var iv = [8]uint32{
0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a,
0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19,
}
// Sum256 returns the BLAKE2s-256 checksum of the data.
func Sum256(data []byte) [Size]byte {
var sum [Size]byte
checkSum(&sum, Size, data)
return sum
}
// New256 returns a new hash.Hash computing the BLAKE2s-256 checksum. A non-nil
// key turns the hash into a MAC. The key must between zero and 32 bytes long.
// When the key is nil, the returned hash.Hash implements BinaryMarshaler
// and BinaryUnmarshaler for state (de)serialization as documented by hash.Hash.
func New256(key []byte) (hash.Hash, error) { return newDigest(Size, key) }
// New128 returns a new hash.Hash computing the BLAKE2s-128 checksum given a
// non-empty key. Note that a 128-bit digest is too small to be secure as a
// cryptographic hash and should only be used as a MAC, thus the key argument
// is not optional.
func New128(key []byte) (hash.Hash, error) {
if len(key) == 0 {
return nil, errors.New("blake2s: a key is required for a 128-bit hash")
}
return newDigest(Size128, key)
}
func newDigest(hashSize int, key []byte) (*digest, error) {
if len(key) > Size {
return nil, errKeySize
}
d := &digest{
size: hashSize,
keyLen: len(key),
}
copy(d.key[:], key)
d.Reset()
return d, nil
}
func checkSum(sum *[Size]byte, hashSize int, data []byte) {
var (
h [8]uint32
c [2]uint32
)
h = iv
h[0] ^= uint32(hashSize) | (1 << 16) | (1 << 24)
if length := len(data); length > BlockSize {
n := length &^ (BlockSize - 1)
if length == n {
n -= BlockSize
}
hashBlocks(&h, &c, 0, data[:n])
data = data[n:]
}
var block [BlockSize]byte
offset := copy(block[:], data)
remaining := uint32(BlockSize - offset)
if c[0] < remaining {
c[1]--
}
c[0] -= remaining
hashBlocks(&h, &c, 0xFFFFFFFF, block[:])
for i, v := range h {
binary.LittleEndian.PutUint32(sum[4*i:], v)
}
}
type digest struct {
h [8]uint32
c [2]uint32
size int
block [BlockSize]byte
offset int
key [BlockSize]byte
keyLen int
}
const (
magic = "b2s"
marshaledSize = len(magic) + 8*4 + 2*4 + 1 + BlockSize + 1
)
func (d *digest) MarshalBinary() ([]byte, error) {
if d.keyLen != 0 {
return nil, errors.New("crypto/blake2s: cannot marshal MACs")
}
b := make([]byte, 0, marshaledSize)
b = append(b, magic...)
for i := 0; i < 8; i++ {
b = appendUint32(b, d.h[i])
}
b = appendUint32(b, d.c[0])
b = appendUint32(b, d.c[1])
// Maximum value for size is 32
b = append(b, byte(d.size))
b = append(b, d.block[:]...)
b = append(b, byte(d.offset))
return b, nil
}
func (d *digest) UnmarshalBinary(b []byte) error {
if len(b) < len(magic) || string(b[:len(magic)]) != magic {
return errors.New("crypto/blake2s: invalid hash state identifier")
}
if len(b) != marshaledSize {
return errors.New("crypto/blake2s: invalid hash state size")
}
b = b[len(magic):]
for i := 0; i < 8; i++ {
b, d.h[i] = consumeUint32(b)
}
b, d.c[0] = consumeUint32(b)
b, d.c[1] = consumeUint32(b)
d.size = int(b[0])
b = b[1:]
copy(d.block[:], b[:BlockSize])
b = b[BlockSize:]
d.offset = int(b[0])
return nil
}
func (d *digest) BlockSize() int { return BlockSize }
func (d *digest) Size() int { return d.size }
func (d *digest) Reset() {
d.h = iv
d.h[0] ^= uint32(d.size) | (uint32(d.keyLen) << 8) | (1 << 16) | (1 << 24)
d.offset, d.c[0], d.c[1] = 0, 0, 0
if d.keyLen > 0 {
d.block = d.key
d.offset = BlockSize
}
}
func (d *digest) Write(p []byte) (n int, err error) {
n = len(p)
if d.offset > 0 {
remaining := BlockSize - d.offset
if n <= remaining {
d.offset += copy(d.block[d.offset:], p)
return
}
copy(d.block[d.offset:], p[:remaining])
hashBlocks(&d.h, &d.c, 0, d.block[:])
d.offset = 0
p = p[remaining:]
}
if length := len(p); length > BlockSize {
nn := length &^ (BlockSize - 1)
if length == nn {
nn -= BlockSize
}
hashBlocks(&d.h, &d.c, 0, p[:nn])
p = p[nn:]
}
d.offset += copy(d.block[:], p)
return
}
func (d *digest) Sum(sum []byte) []byte {
var hash [Size]byte
d.finalize(&hash)
return append(sum, hash[:d.size]...)
}
func (d *digest) finalize(hash *[Size]byte) {
var block [BlockSize]byte
h := d.h
c := d.c
copy(block[:], d.block[:d.offset])
remaining := uint32(BlockSize - d.offset)
if c[0] < remaining {
c[1]--
}
c[0] -= remaining
hashBlocks(&h, &c, 0xFFFFFFFF, block[:])
for i, v := range h {
binary.LittleEndian.PutUint32(hash[4*i:], v)
}
}
func appendUint32(b []byte, x uint32) []byte {
var a [4]byte
binary.BigEndian.PutUint32(a[:], x)
return append(b, a[:]...)
}
func consumeUint32(b []byte) ([]byte, uint32) {
x := binary.BigEndian.Uint32(b)
return b[4:], x
}
+33
View File
@@ -0,0 +1,33 @@
// Copyright 2016 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.
//go:build 386 && gc && !purego
// +build 386,gc,!purego
package blake2s
import "golang.org/x/sys/cpu"
var (
useSSE4 = false
useSSSE3 = cpu.X86.HasSSSE3
useSSE2 = cpu.X86.HasSSE2
)
//go:noescape
func hashBlocksSSE2(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte)
//go:noescape
func hashBlocksSSSE3(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte)
func hashBlocks(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte) {
switch {
case useSSSE3:
hashBlocksSSSE3(h, c, flag, blocks)
case useSSE2:
hashBlocksSSE2(h, c, flag, blocks)
default:
hashBlocksGeneric(h, c, flag, blocks)
}
}
+430
View File
@@ -0,0 +1,430 @@
// Copyright 2016 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.
//go:build 386 && gc && !purego
// +build 386,gc,!purego
#include "textflag.h"
DATA iv0<>+0x00(SB)/4, $0x6a09e667
DATA iv0<>+0x04(SB)/4, $0xbb67ae85
DATA iv0<>+0x08(SB)/4, $0x3c6ef372
DATA iv0<>+0x0c(SB)/4, $0xa54ff53a
GLOBL iv0<>(SB), (NOPTR+RODATA), $16
DATA iv1<>+0x00(SB)/4, $0x510e527f
DATA iv1<>+0x04(SB)/4, $0x9b05688c
DATA iv1<>+0x08(SB)/4, $0x1f83d9ab
DATA iv1<>+0x0c(SB)/4, $0x5be0cd19
GLOBL iv1<>(SB), (NOPTR+RODATA), $16
DATA rol16<>+0x00(SB)/8, $0x0504070601000302
DATA rol16<>+0x08(SB)/8, $0x0D0C0F0E09080B0A
GLOBL rol16<>(SB), (NOPTR+RODATA), $16
DATA rol8<>+0x00(SB)/8, $0x0407060500030201
DATA rol8<>+0x08(SB)/8, $0x0C0F0E0D080B0A09
GLOBL rol8<>(SB), (NOPTR+RODATA), $16
DATA counter<>+0x00(SB)/8, $0x40
DATA counter<>+0x08(SB)/8, $0x0
GLOBL counter<>(SB), (NOPTR+RODATA), $16
#define ROTL_SSE2(n, t, v) \
MOVO v, t; \
PSLLL $n, t; \
PSRLL $(32-n), v; \
PXOR t, v
#define ROTL_SSSE3(c, v) \
PSHUFB c, v
#define ROUND_SSE2(v0, v1, v2, v3, m0, m1, m2, m3, t) \
PADDL m0, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSE2(16, t, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(20, t, v1); \
PADDL m1, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSE2(24, t, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(25, t, v1); \
PSHUFL $0x39, v1, v1; \
PSHUFL $0x4E, v2, v2; \
PSHUFL $0x93, v3, v3; \
PADDL m2, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSE2(16, t, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(20, t, v1); \
PADDL m3, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSE2(24, t, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(25, t, v1); \
PSHUFL $0x39, v3, v3; \
PSHUFL $0x4E, v2, v2; \
PSHUFL $0x93, v1, v1
#define ROUND_SSSE3(v0, v1, v2, v3, m0, m1, m2, m3, t, c16, c8) \
PADDL m0, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSSE3(c16, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(20, t, v1); \
PADDL m1, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSSE3(c8, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(25, t, v1); \
PSHUFL $0x39, v1, v1; \
PSHUFL $0x4E, v2, v2; \
PSHUFL $0x93, v3, v3; \
PADDL m2, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSSE3(c16, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(20, t, v1); \
PADDL m3, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSSE3(c8, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(25, t, v1); \
PSHUFL $0x39, v3, v3; \
PSHUFL $0x4E, v2, v2; \
PSHUFL $0x93, v1, v1
#define PRECOMPUTE(dst, off, src, t) \
MOVL 0*4(src), t; \
MOVL t, 0*4+off+0(dst); \
MOVL t, 9*4+off+64(dst); \
MOVL t, 5*4+off+128(dst); \
MOVL t, 14*4+off+192(dst); \
MOVL t, 4*4+off+256(dst); \
MOVL t, 2*4+off+320(dst); \
MOVL t, 8*4+off+384(dst); \
MOVL t, 12*4+off+448(dst); \
MOVL t, 3*4+off+512(dst); \
MOVL t, 15*4+off+576(dst); \
MOVL 1*4(src), t; \
MOVL t, 4*4+off+0(dst); \
MOVL t, 8*4+off+64(dst); \
MOVL t, 14*4+off+128(dst); \
MOVL t, 5*4+off+192(dst); \
MOVL t, 12*4+off+256(dst); \
MOVL t, 11*4+off+320(dst); \
MOVL t, 1*4+off+384(dst); \
MOVL t, 6*4+off+448(dst); \
MOVL t, 10*4+off+512(dst); \
MOVL t, 3*4+off+576(dst); \
MOVL 2*4(src), t; \
MOVL t, 1*4+off+0(dst); \
MOVL t, 13*4+off+64(dst); \
MOVL t, 6*4+off+128(dst); \
MOVL t, 8*4+off+192(dst); \
MOVL t, 2*4+off+256(dst); \
MOVL t, 0*4+off+320(dst); \
MOVL t, 14*4+off+384(dst); \
MOVL t, 11*4+off+448(dst); \
MOVL t, 12*4+off+512(dst); \
MOVL t, 4*4+off+576(dst); \
MOVL 3*4(src), t; \
MOVL t, 5*4+off+0(dst); \
MOVL t, 15*4+off+64(dst); \
MOVL t, 9*4+off+128(dst); \
MOVL t, 1*4+off+192(dst); \
MOVL t, 11*4+off+256(dst); \
MOVL t, 7*4+off+320(dst); \
MOVL t, 13*4+off+384(dst); \
MOVL t, 3*4+off+448(dst); \
MOVL t, 6*4+off+512(dst); \
MOVL t, 10*4+off+576(dst); \
MOVL 4*4(src), t; \
MOVL t, 2*4+off+0(dst); \
MOVL t, 1*4+off+64(dst); \
MOVL t, 15*4+off+128(dst); \
MOVL t, 10*4+off+192(dst); \
MOVL t, 6*4+off+256(dst); \
MOVL t, 8*4+off+320(dst); \
MOVL t, 3*4+off+384(dst); \
MOVL t, 13*4+off+448(dst); \
MOVL t, 14*4+off+512(dst); \
MOVL t, 5*4+off+576(dst); \
MOVL 5*4(src), t; \
MOVL t, 6*4+off+0(dst); \
MOVL t, 11*4+off+64(dst); \
MOVL t, 2*4+off+128(dst); \
MOVL t, 9*4+off+192(dst); \
MOVL t, 1*4+off+256(dst); \
MOVL t, 13*4+off+320(dst); \
MOVL t, 4*4+off+384(dst); \
MOVL t, 8*4+off+448(dst); \
MOVL t, 15*4+off+512(dst); \
MOVL t, 7*4+off+576(dst); \
MOVL 6*4(src), t; \
MOVL t, 3*4+off+0(dst); \
MOVL t, 7*4+off+64(dst); \
MOVL t, 13*4+off+128(dst); \
MOVL t, 12*4+off+192(dst); \
MOVL t, 10*4+off+256(dst); \
MOVL t, 1*4+off+320(dst); \
MOVL t, 9*4+off+384(dst); \
MOVL t, 14*4+off+448(dst); \
MOVL t, 0*4+off+512(dst); \
MOVL t, 6*4+off+576(dst); \
MOVL 7*4(src), t; \
MOVL t, 7*4+off+0(dst); \
MOVL t, 14*4+off+64(dst); \
MOVL t, 10*4+off+128(dst); \
MOVL t, 0*4+off+192(dst); \
MOVL t, 5*4+off+256(dst); \
MOVL t, 9*4+off+320(dst); \
MOVL t, 12*4+off+384(dst); \
MOVL t, 1*4+off+448(dst); \
MOVL t, 13*4+off+512(dst); \
MOVL t, 2*4+off+576(dst); \
MOVL 8*4(src), t; \
MOVL t, 8*4+off+0(dst); \
MOVL t, 5*4+off+64(dst); \
MOVL t, 4*4+off+128(dst); \
MOVL t, 15*4+off+192(dst); \
MOVL t, 14*4+off+256(dst); \
MOVL t, 3*4+off+320(dst); \
MOVL t, 11*4+off+384(dst); \
MOVL t, 10*4+off+448(dst); \
MOVL t, 7*4+off+512(dst); \
MOVL t, 1*4+off+576(dst); \
MOVL 9*4(src), t; \
MOVL t, 12*4+off+0(dst); \
MOVL t, 2*4+off+64(dst); \
MOVL t, 11*4+off+128(dst); \
MOVL t, 4*4+off+192(dst); \
MOVL t, 0*4+off+256(dst); \
MOVL t, 15*4+off+320(dst); \
MOVL t, 10*4+off+384(dst); \
MOVL t, 7*4+off+448(dst); \
MOVL t, 5*4+off+512(dst); \
MOVL t, 9*4+off+576(dst); \
MOVL 10*4(src), t; \
MOVL t, 9*4+off+0(dst); \
MOVL t, 4*4+off+64(dst); \
MOVL t, 8*4+off+128(dst); \
MOVL t, 13*4+off+192(dst); \
MOVL t, 3*4+off+256(dst); \
MOVL t, 5*4+off+320(dst); \
MOVL t, 7*4+off+384(dst); \
MOVL t, 15*4+off+448(dst); \
MOVL t, 11*4+off+512(dst); \
MOVL t, 0*4+off+576(dst); \
MOVL 11*4(src), t; \
MOVL t, 13*4+off+0(dst); \
MOVL t, 10*4+off+64(dst); \
MOVL t, 0*4+off+128(dst); \
MOVL t, 3*4+off+192(dst); \
MOVL t, 9*4+off+256(dst); \
MOVL t, 6*4+off+320(dst); \
MOVL t, 15*4+off+384(dst); \
MOVL t, 4*4+off+448(dst); \
MOVL t, 2*4+off+512(dst); \
MOVL t, 12*4+off+576(dst); \
MOVL 12*4(src), t; \
MOVL t, 10*4+off+0(dst); \
MOVL t, 12*4+off+64(dst); \
MOVL t, 1*4+off+128(dst); \
MOVL t, 6*4+off+192(dst); \
MOVL t, 13*4+off+256(dst); \
MOVL t, 4*4+off+320(dst); \
MOVL t, 0*4+off+384(dst); \
MOVL t, 2*4+off+448(dst); \
MOVL t, 8*4+off+512(dst); \
MOVL t, 14*4+off+576(dst); \
MOVL 13*4(src), t; \
MOVL t, 14*4+off+0(dst); \
MOVL t, 3*4+off+64(dst); \
MOVL t, 7*4+off+128(dst); \
MOVL t, 2*4+off+192(dst); \
MOVL t, 15*4+off+256(dst); \
MOVL t, 12*4+off+320(dst); \
MOVL t, 6*4+off+384(dst); \
MOVL t, 0*4+off+448(dst); \
MOVL t, 9*4+off+512(dst); \
MOVL t, 11*4+off+576(dst); \
MOVL 14*4(src), t; \
MOVL t, 11*4+off+0(dst); \
MOVL t, 0*4+off+64(dst); \
MOVL t, 12*4+off+128(dst); \
MOVL t, 7*4+off+192(dst); \
MOVL t, 8*4+off+256(dst); \
MOVL t, 14*4+off+320(dst); \
MOVL t, 2*4+off+384(dst); \
MOVL t, 5*4+off+448(dst); \
MOVL t, 1*4+off+512(dst); \
MOVL t, 13*4+off+576(dst); \
MOVL 15*4(src), t; \
MOVL t, 15*4+off+0(dst); \
MOVL t, 6*4+off+64(dst); \
MOVL t, 3*4+off+128(dst); \
MOVL t, 11*4+off+192(dst); \
MOVL t, 7*4+off+256(dst); \
MOVL t, 10*4+off+320(dst); \
MOVL t, 5*4+off+384(dst); \
MOVL t, 9*4+off+448(dst); \
MOVL t, 4*4+off+512(dst); \
MOVL t, 8*4+off+576(dst)
// func hashBlocksSSE2(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte)
TEXT ·hashBlocksSSE2(SB), 0, $672-24 // frame = 656 + 16 byte alignment
MOVL h+0(FP), AX
MOVL c+4(FP), BX
MOVL flag+8(FP), CX
MOVL blocks_base+12(FP), SI
MOVL blocks_len+16(FP), DX
MOVL SP, DI
ADDL $15, DI
ANDL $~15, DI
MOVL CX, 8(DI)
MOVL 0(BX), CX
MOVL CX, 0(DI)
MOVL 4(BX), CX
MOVL CX, 4(DI)
XORL CX, CX
MOVL CX, 12(DI)
MOVOU 0(AX), X0
MOVOU 16(AX), X1
MOVOU counter<>(SB), X2
loop:
MOVO X0, X4
MOVO X1, X5
MOVOU iv0<>(SB), X6
MOVOU iv1<>(SB), X7
MOVO 0(DI), X3
PADDQ X2, X3
PXOR X3, X7
MOVO X3, 0(DI)
PRECOMPUTE(DI, 16, SI, CX)
ROUND_SSE2(X4, X5, X6, X7, 16(DI), 32(DI), 48(DI), 64(DI), X3)
ROUND_SSE2(X4, X5, X6, X7, 16+64(DI), 32+64(DI), 48+64(DI), 64+64(DI), X3)
ROUND_SSE2(X4, X5, X6, X7, 16+128(DI), 32+128(DI), 48+128(DI), 64+128(DI), X3)
ROUND_SSE2(X4, X5, X6, X7, 16+192(DI), 32+192(DI), 48+192(DI), 64+192(DI), X3)
ROUND_SSE2(X4, X5, X6, X7, 16+256(DI), 32+256(DI), 48+256(DI), 64+256(DI), X3)
ROUND_SSE2(X4, X5, X6, X7, 16+320(DI), 32+320(DI), 48+320(DI), 64+320(DI), X3)
ROUND_SSE2(X4, X5, X6, X7, 16+384(DI), 32+384(DI), 48+384(DI), 64+384(DI), X3)
ROUND_SSE2(X4, X5, X6, X7, 16+448(DI), 32+448(DI), 48+448(DI), 64+448(DI), X3)
ROUND_SSE2(X4, X5, X6, X7, 16+512(DI), 32+512(DI), 48+512(DI), 64+512(DI), X3)
ROUND_SSE2(X4, X5, X6, X7, 16+576(DI), 32+576(DI), 48+576(DI), 64+576(DI), X3)
PXOR X4, X0
PXOR X5, X1
PXOR X6, X0
PXOR X7, X1
LEAL 64(SI), SI
SUBL $64, DX
JNE loop
MOVL 0(DI), CX
MOVL CX, 0(BX)
MOVL 4(DI), CX
MOVL CX, 4(BX)
MOVOU X0, 0(AX)
MOVOU X1, 16(AX)
RET
// func hashBlocksSSSE3(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte)
TEXT ·hashBlocksSSSE3(SB), 0, $704-24 // frame = 688 + 16 byte alignment
MOVL h+0(FP), AX
MOVL c+4(FP), BX
MOVL flag+8(FP), CX
MOVL blocks_base+12(FP), SI
MOVL blocks_len+16(FP), DX
MOVL SP, DI
ADDL $15, DI
ANDL $~15, DI
MOVL CX, 8(DI)
MOVL 0(BX), CX
MOVL CX, 0(DI)
MOVL 4(BX), CX
MOVL CX, 4(DI)
XORL CX, CX
MOVL CX, 12(DI)
MOVOU 0(AX), X0
MOVOU 16(AX), X1
MOVOU counter<>(SB), X2
loop:
MOVO X0, 656(DI)
MOVO X1, 672(DI)
MOVO X0, X4
MOVO X1, X5
MOVOU iv0<>(SB), X6
MOVOU iv1<>(SB), X7
MOVO 0(DI), X3
PADDQ X2, X3
PXOR X3, X7
MOVO X3, 0(DI)
MOVOU rol16<>(SB), X0
MOVOU rol8<>(SB), X1
PRECOMPUTE(DI, 16, SI, CX)
ROUND_SSSE3(X4, X5, X6, X7, 16(DI), 32(DI), 48(DI), 64(DI), X3, X0, X1)
ROUND_SSSE3(X4, X5, X6, X7, 16+64(DI), 32+64(DI), 48+64(DI), 64+64(DI), X3, X0, X1)
ROUND_SSSE3(X4, X5, X6, X7, 16+128(DI), 32+128(DI), 48+128(DI), 64+128(DI), X3, X0, X1)
ROUND_SSSE3(X4, X5, X6, X7, 16+192(DI), 32+192(DI), 48+192(DI), 64+192(DI), X3, X0, X1)
ROUND_SSSE3(X4, X5, X6, X7, 16+256(DI), 32+256(DI), 48+256(DI), 64+256(DI), X3, X0, X1)
ROUND_SSSE3(X4, X5, X6, X7, 16+320(DI), 32+320(DI), 48+320(DI), 64+320(DI), X3, X0, X1)
ROUND_SSSE3(X4, X5, X6, X7, 16+384(DI), 32+384(DI), 48+384(DI), 64+384(DI), X3, X0, X1)
ROUND_SSSE3(X4, X5, X6, X7, 16+448(DI), 32+448(DI), 48+448(DI), 64+448(DI), X3, X0, X1)
ROUND_SSSE3(X4, X5, X6, X7, 16+512(DI), 32+512(DI), 48+512(DI), 64+512(DI), X3, X0, X1)
ROUND_SSSE3(X4, X5, X6, X7, 16+576(DI), 32+576(DI), 48+576(DI), 64+576(DI), X3, X0, X1)
MOVO 656(DI), X0
MOVO 672(DI), X1
PXOR X4, X0
PXOR X5, X1
PXOR X6, X0
PXOR X7, X1
LEAL 64(SI), SI
SUBL $64, DX
JNE loop
MOVL 0(DI), CX
MOVL CX, 0(BX)
MOVL 4(DI), CX
MOVL CX, 4(BX)
MOVOU X0, 0(AX)
MOVOU X1, 16(AX)
RET
+38
View File
@@ -0,0 +1,38 @@
// Copyright 2016 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.
//go:build amd64 && gc && !purego
// +build amd64,gc,!purego
package blake2s
import "golang.org/x/sys/cpu"
var (
useSSE4 = cpu.X86.HasSSE41
useSSSE3 = cpu.X86.HasSSSE3
useSSE2 = cpu.X86.HasSSE2
)
//go:noescape
func hashBlocksSSE2(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte)
//go:noescape
func hashBlocksSSSE3(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte)
//go:noescape
func hashBlocksSSE4(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte)
func hashBlocks(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte) {
switch {
case useSSE4:
hashBlocksSSE4(h, c, flag, blocks)
case useSSSE3:
hashBlocksSSSE3(h, c, flag, blocks)
case useSSE2:
hashBlocksSSE2(h, c, flag, blocks)
default:
hashBlocksGeneric(h, c, flag, blocks)
}
}
+433
View File
@@ -0,0 +1,433 @@
// Copyright 2016 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.
//go:build amd64 && gc && !purego
// +build amd64,gc,!purego
#include "textflag.h"
DATA iv0<>+0x00(SB)/4, $0x6a09e667
DATA iv0<>+0x04(SB)/4, $0xbb67ae85
DATA iv0<>+0x08(SB)/4, $0x3c6ef372
DATA iv0<>+0x0c(SB)/4, $0xa54ff53a
GLOBL iv0<>(SB), (NOPTR+RODATA), $16
DATA iv1<>+0x00(SB)/4, $0x510e527f
DATA iv1<>+0x04(SB)/4, $0x9b05688c
DATA iv1<>+0x08(SB)/4, $0x1f83d9ab
DATA iv1<>+0x0c(SB)/4, $0x5be0cd19
GLOBL iv1<>(SB), (NOPTR+RODATA), $16
DATA rol16<>+0x00(SB)/8, $0x0504070601000302
DATA rol16<>+0x08(SB)/8, $0x0D0C0F0E09080B0A
GLOBL rol16<>(SB), (NOPTR+RODATA), $16
DATA rol8<>+0x00(SB)/8, $0x0407060500030201
DATA rol8<>+0x08(SB)/8, $0x0C0F0E0D080B0A09
GLOBL rol8<>(SB), (NOPTR+RODATA), $16
DATA counter<>+0x00(SB)/8, $0x40
DATA counter<>+0x08(SB)/8, $0x0
GLOBL counter<>(SB), (NOPTR+RODATA), $16
#define ROTL_SSE2(n, t, v) \
MOVO v, t; \
PSLLL $n, t; \
PSRLL $(32-n), v; \
PXOR t, v
#define ROTL_SSSE3(c, v) \
PSHUFB c, v
#define ROUND_SSE2(v0, v1, v2, v3, m0, m1, m2, m3, t) \
PADDL m0, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSE2(16, t, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(20, t, v1); \
PADDL m1, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSE2(24, t, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(25, t, v1); \
PSHUFL $0x39, v1, v1; \
PSHUFL $0x4E, v2, v2; \
PSHUFL $0x93, v3, v3; \
PADDL m2, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSE2(16, t, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(20, t, v1); \
PADDL m3, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSE2(24, t, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(25, t, v1); \
PSHUFL $0x39, v3, v3; \
PSHUFL $0x4E, v2, v2; \
PSHUFL $0x93, v1, v1
#define ROUND_SSSE3(v0, v1, v2, v3, m0, m1, m2, m3, t, c16, c8) \
PADDL m0, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSSE3(c16, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(20, t, v1); \
PADDL m1, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSSE3(c8, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(25, t, v1); \
PSHUFL $0x39, v1, v1; \
PSHUFL $0x4E, v2, v2; \
PSHUFL $0x93, v3, v3; \
PADDL m2, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSSE3(c16, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(20, t, v1); \
PADDL m3, v0; \
PADDL v1, v0; \
PXOR v0, v3; \
ROTL_SSSE3(c8, v3); \
PADDL v3, v2; \
PXOR v2, v1; \
ROTL_SSE2(25, t, v1); \
PSHUFL $0x39, v3, v3; \
PSHUFL $0x4E, v2, v2; \
PSHUFL $0x93, v1, v1
#define LOAD_MSG_SSE4(m0, m1, m2, m3, src, i0, i1, i2, i3, i4, i5, i6, i7, i8, i9, i10, i11, i12, i13, i14, i15) \
MOVL i0*4(src), m0; \
PINSRD $1, i1*4(src), m0; \
PINSRD $2, i2*4(src), m0; \
PINSRD $3, i3*4(src), m0; \
MOVL i4*4(src), m1; \
PINSRD $1, i5*4(src), m1; \
PINSRD $2, i6*4(src), m1; \
PINSRD $3, i7*4(src), m1; \
MOVL i8*4(src), m2; \
PINSRD $1, i9*4(src), m2; \
PINSRD $2, i10*4(src), m2; \
PINSRD $3, i11*4(src), m2; \
MOVL i12*4(src), m3; \
PINSRD $1, i13*4(src), m3; \
PINSRD $2, i14*4(src), m3; \
PINSRD $3, i15*4(src), m3
#define PRECOMPUTE_MSG(dst, off, src, R8, R9, R10, R11, R12, R13, R14, R15) \
MOVQ 0*4(src), R8; \
MOVQ 2*4(src), R9; \
MOVQ 4*4(src), R10; \
MOVQ 6*4(src), R11; \
MOVQ 8*4(src), R12; \
MOVQ 10*4(src), R13; \
MOVQ 12*4(src), R14; \
MOVQ 14*4(src), R15; \
\
MOVL R8, 0*4+off+0(dst); \
MOVL R8, 9*4+off+64(dst); \
MOVL R8, 5*4+off+128(dst); \
MOVL R8, 14*4+off+192(dst); \
MOVL R8, 4*4+off+256(dst); \
MOVL R8, 2*4+off+320(dst); \
MOVL R8, 8*4+off+384(dst); \
MOVL R8, 12*4+off+448(dst); \
MOVL R8, 3*4+off+512(dst); \
MOVL R8, 15*4+off+576(dst); \
SHRQ $32, R8; \
MOVL R8, 4*4+off+0(dst); \
MOVL R8, 8*4+off+64(dst); \
MOVL R8, 14*4+off+128(dst); \
MOVL R8, 5*4+off+192(dst); \
MOVL R8, 12*4+off+256(dst); \
MOVL R8, 11*4+off+320(dst); \
MOVL R8, 1*4+off+384(dst); \
MOVL R8, 6*4+off+448(dst); \
MOVL R8, 10*4+off+512(dst); \
MOVL R8, 3*4+off+576(dst); \
\
MOVL R9, 1*4+off+0(dst); \
MOVL R9, 13*4+off+64(dst); \
MOVL R9, 6*4+off+128(dst); \
MOVL R9, 8*4+off+192(dst); \
MOVL R9, 2*4+off+256(dst); \
MOVL R9, 0*4+off+320(dst); \
MOVL R9, 14*4+off+384(dst); \
MOVL R9, 11*4+off+448(dst); \
MOVL R9, 12*4+off+512(dst); \
MOVL R9, 4*4+off+576(dst); \
SHRQ $32, R9; \
MOVL R9, 5*4+off+0(dst); \
MOVL R9, 15*4+off+64(dst); \
MOVL R9, 9*4+off+128(dst); \
MOVL R9, 1*4+off+192(dst); \
MOVL R9, 11*4+off+256(dst); \
MOVL R9, 7*4+off+320(dst); \
MOVL R9, 13*4+off+384(dst); \
MOVL R9, 3*4+off+448(dst); \
MOVL R9, 6*4+off+512(dst); \
MOVL R9, 10*4+off+576(dst); \
\
MOVL R10, 2*4+off+0(dst); \
MOVL R10, 1*4+off+64(dst); \
MOVL R10, 15*4+off+128(dst); \
MOVL R10, 10*4+off+192(dst); \
MOVL R10, 6*4+off+256(dst); \
MOVL R10, 8*4+off+320(dst); \
MOVL R10, 3*4+off+384(dst); \
MOVL R10, 13*4+off+448(dst); \
MOVL R10, 14*4+off+512(dst); \
MOVL R10, 5*4+off+576(dst); \
SHRQ $32, R10; \
MOVL R10, 6*4+off+0(dst); \
MOVL R10, 11*4+off+64(dst); \
MOVL R10, 2*4+off+128(dst); \
MOVL R10, 9*4+off+192(dst); \
MOVL R10, 1*4+off+256(dst); \
MOVL R10, 13*4+off+320(dst); \
MOVL R10, 4*4+off+384(dst); \
MOVL R10, 8*4+off+448(dst); \
MOVL R10, 15*4+off+512(dst); \
MOVL R10, 7*4+off+576(dst); \
\
MOVL R11, 3*4+off+0(dst); \
MOVL R11, 7*4+off+64(dst); \
MOVL R11, 13*4+off+128(dst); \
MOVL R11, 12*4+off+192(dst); \
MOVL R11, 10*4+off+256(dst); \
MOVL R11, 1*4+off+320(dst); \
MOVL R11, 9*4+off+384(dst); \
MOVL R11, 14*4+off+448(dst); \
MOVL R11, 0*4+off+512(dst); \
MOVL R11, 6*4+off+576(dst); \
SHRQ $32, R11; \
MOVL R11, 7*4+off+0(dst); \
MOVL R11, 14*4+off+64(dst); \
MOVL R11, 10*4+off+128(dst); \
MOVL R11, 0*4+off+192(dst); \
MOVL R11, 5*4+off+256(dst); \
MOVL R11, 9*4+off+320(dst); \
MOVL R11, 12*4+off+384(dst); \
MOVL R11, 1*4+off+448(dst); \
MOVL R11, 13*4+off+512(dst); \
MOVL R11, 2*4+off+576(dst); \
\
MOVL R12, 8*4+off+0(dst); \
MOVL R12, 5*4+off+64(dst); \
MOVL R12, 4*4+off+128(dst); \
MOVL R12, 15*4+off+192(dst); \
MOVL R12, 14*4+off+256(dst); \
MOVL R12, 3*4+off+320(dst); \
MOVL R12, 11*4+off+384(dst); \
MOVL R12, 10*4+off+448(dst); \
MOVL R12, 7*4+off+512(dst); \
MOVL R12, 1*4+off+576(dst); \
SHRQ $32, R12; \
MOVL R12, 12*4+off+0(dst); \
MOVL R12, 2*4+off+64(dst); \
MOVL R12, 11*4+off+128(dst); \
MOVL R12, 4*4+off+192(dst); \
MOVL R12, 0*4+off+256(dst); \
MOVL R12, 15*4+off+320(dst); \
MOVL R12, 10*4+off+384(dst); \
MOVL R12, 7*4+off+448(dst); \
MOVL R12, 5*4+off+512(dst); \
MOVL R12, 9*4+off+576(dst); \
\
MOVL R13, 9*4+off+0(dst); \
MOVL R13, 4*4+off+64(dst); \
MOVL R13, 8*4+off+128(dst); \
MOVL R13, 13*4+off+192(dst); \
MOVL R13, 3*4+off+256(dst); \
MOVL R13, 5*4+off+320(dst); \
MOVL R13, 7*4+off+384(dst); \
MOVL R13, 15*4+off+448(dst); \
MOVL R13, 11*4+off+512(dst); \
MOVL R13, 0*4+off+576(dst); \
SHRQ $32, R13; \
MOVL R13, 13*4+off+0(dst); \
MOVL R13, 10*4+off+64(dst); \
MOVL R13, 0*4+off+128(dst); \
MOVL R13, 3*4+off+192(dst); \
MOVL R13, 9*4+off+256(dst); \
MOVL R13, 6*4+off+320(dst); \
MOVL R13, 15*4+off+384(dst); \
MOVL R13, 4*4+off+448(dst); \
MOVL R13, 2*4+off+512(dst); \
MOVL R13, 12*4+off+576(dst); \
\
MOVL R14, 10*4+off+0(dst); \
MOVL R14, 12*4+off+64(dst); \
MOVL R14, 1*4+off+128(dst); \
MOVL R14, 6*4+off+192(dst); \
MOVL R14, 13*4+off+256(dst); \
MOVL R14, 4*4+off+320(dst); \
MOVL R14, 0*4+off+384(dst); \
MOVL R14, 2*4+off+448(dst); \
MOVL R14, 8*4+off+512(dst); \
MOVL R14, 14*4+off+576(dst); \
SHRQ $32, R14; \
MOVL R14, 14*4+off+0(dst); \
MOVL R14, 3*4+off+64(dst); \
MOVL R14, 7*4+off+128(dst); \
MOVL R14, 2*4+off+192(dst); \
MOVL R14, 15*4+off+256(dst); \
MOVL R14, 12*4+off+320(dst); \
MOVL R14, 6*4+off+384(dst); \
MOVL R14, 0*4+off+448(dst); \
MOVL R14, 9*4+off+512(dst); \
MOVL R14, 11*4+off+576(dst); \
\
MOVL R15, 11*4+off+0(dst); \
MOVL R15, 0*4+off+64(dst); \
MOVL R15, 12*4+off+128(dst); \
MOVL R15, 7*4+off+192(dst); \
MOVL R15, 8*4+off+256(dst); \
MOVL R15, 14*4+off+320(dst); \
MOVL R15, 2*4+off+384(dst); \
MOVL R15, 5*4+off+448(dst); \
MOVL R15, 1*4+off+512(dst); \
MOVL R15, 13*4+off+576(dst); \
SHRQ $32, R15; \
MOVL R15, 15*4+off+0(dst); \
MOVL R15, 6*4+off+64(dst); \
MOVL R15, 3*4+off+128(dst); \
MOVL R15, 11*4+off+192(dst); \
MOVL R15, 7*4+off+256(dst); \
MOVL R15, 10*4+off+320(dst); \
MOVL R15, 5*4+off+384(dst); \
MOVL R15, 9*4+off+448(dst); \
MOVL R15, 4*4+off+512(dst); \
MOVL R15, 8*4+off+576(dst)
#define BLAKE2s_SSE2() \
PRECOMPUTE_MSG(BP, 16, SI, R8, R9, R10, R11, R12, R13, R14, R15); \
ROUND_SSE2(X4, X5, X6, X7, 16(BP), 32(BP), 48(BP), 64(BP), X8); \
ROUND_SSE2(X4, X5, X6, X7, 16+64(BP), 32+64(BP), 48+64(BP), 64+64(BP), X8); \
ROUND_SSE2(X4, X5, X6, X7, 16+128(BP), 32+128(BP), 48+128(BP), 64+128(BP), X8); \
ROUND_SSE2(X4, X5, X6, X7, 16+192(BP), 32+192(BP), 48+192(BP), 64+192(BP), X8); \
ROUND_SSE2(X4, X5, X6, X7, 16+256(BP), 32+256(BP), 48+256(BP), 64+256(BP), X8); \
ROUND_SSE2(X4, X5, X6, X7, 16+320(BP), 32+320(BP), 48+320(BP), 64+320(BP), X8); \
ROUND_SSE2(X4, X5, X6, X7, 16+384(BP), 32+384(BP), 48+384(BP), 64+384(BP), X8); \
ROUND_SSE2(X4, X5, X6, X7, 16+448(BP), 32+448(BP), 48+448(BP), 64+448(BP), X8); \
ROUND_SSE2(X4, X5, X6, X7, 16+512(BP), 32+512(BP), 48+512(BP), 64+512(BP), X8); \
ROUND_SSE2(X4, X5, X6, X7, 16+576(BP), 32+576(BP), 48+576(BP), 64+576(BP), X8)
#define BLAKE2s_SSSE3() \
PRECOMPUTE_MSG(BP, 16, SI, R8, R9, R10, R11, R12, R13, R14, R15); \
ROUND_SSSE3(X4, X5, X6, X7, 16(BP), 32(BP), 48(BP), 64(BP), X8, X13, X14); \
ROUND_SSSE3(X4, X5, X6, X7, 16+64(BP), 32+64(BP), 48+64(BP), 64+64(BP), X8, X13, X14); \
ROUND_SSSE3(X4, X5, X6, X7, 16+128(BP), 32+128(BP), 48+128(BP), 64+128(BP), X8, X13, X14); \
ROUND_SSSE3(X4, X5, X6, X7, 16+192(BP), 32+192(BP), 48+192(BP), 64+192(BP), X8, X13, X14); \
ROUND_SSSE3(X4, X5, X6, X7, 16+256(BP), 32+256(BP), 48+256(BP), 64+256(BP), X8, X13, X14); \
ROUND_SSSE3(X4, X5, X6, X7, 16+320(BP), 32+320(BP), 48+320(BP), 64+320(BP), X8, X13, X14); \
ROUND_SSSE3(X4, X5, X6, X7, 16+384(BP), 32+384(BP), 48+384(BP), 64+384(BP), X8, X13, X14); \
ROUND_SSSE3(X4, X5, X6, X7, 16+448(BP), 32+448(BP), 48+448(BP), 64+448(BP), X8, X13, X14); \
ROUND_SSSE3(X4, X5, X6, X7, 16+512(BP), 32+512(BP), 48+512(BP), 64+512(BP), X8, X13, X14); \
ROUND_SSSE3(X4, X5, X6, X7, 16+576(BP), 32+576(BP), 48+576(BP), 64+576(BP), X8, X13, X14)
#define BLAKE2s_SSE4() \
LOAD_MSG_SSE4(X8, X9, X10, X11, SI, 0, 2, 4, 6, 1, 3, 5, 7, 8, 10, 12, 14, 9, 11, 13, 15); \
ROUND_SSSE3(X4, X5, X6, X7, X8, X9, X10, X11, X8, X13, X14); \
LOAD_MSG_SSE4(X8, X9, X10, X11, SI, 14, 4, 9, 13, 10, 8, 15, 6, 1, 0, 11, 5, 12, 2, 7, 3); \
ROUND_SSSE3(X4, X5, X6, X7, X8, X9, X10, X11, X8, X13, X14); \
LOAD_MSG_SSE4(X8, X9, X10, X11, SI, 11, 12, 5, 15, 8, 0, 2, 13, 10, 3, 7, 9, 14, 6, 1, 4); \
ROUND_SSSE3(X4, X5, X6, X7, X8, X9, X10, X11, X8, X13, X14); \
LOAD_MSG_SSE4(X8, X9, X10, X11, SI, 7, 3, 13, 11, 9, 1, 12, 14, 2, 5, 4, 15, 6, 10, 0, 8); \
ROUND_SSSE3(X4, X5, X6, X7, X8, X9, X10, X11, X8, X13, X14); \
LOAD_MSG_SSE4(X8, X9, X10, X11, SI, 9, 5, 2, 10, 0, 7, 4, 15, 14, 11, 6, 3, 1, 12, 8, 13); \
ROUND_SSSE3(X4, X5, X6, X7, X8, X9, X10, X11, X8, X13, X14); \
LOAD_MSG_SSE4(X8, X9, X10, X11, SI, 2, 6, 0, 8, 12, 10, 11, 3, 4, 7, 15, 1, 13, 5, 14, 9); \
ROUND_SSSE3(X4, X5, X6, X7, X8, X9, X10, X11, X8, X13, X14); \
LOAD_MSG_SSE4(X8, X9, X10, X11, SI, 12, 1, 14, 4, 5, 15, 13, 10, 0, 6, 9, 8, 7, 3, 2, 11); \
ROUND_SSSE3(X4, X5, X6, X7, X8, X9, X10, X11, X8, X13, X14); \
LOAD_MSG_SSE4(X8, X9, X10, X11, SI, 13, 7, 12, 3, 11, 14, 1, 9, 5, 15, 8, 2, 0, 4, 6, 10); \
ROUND_SSSE3(X4, X5, X6, X7, X8, X9, X10, X11, X8, X13, X14); \
LOAD_MSG_SSE4(X8, X9, X10, X11, SI, 6, 14, 11, 0, 15, 9, 3, 8, 12, 13, 1, 10, 2, 7, 4, 5); \
ROUND_SSSE3(X4, X5, X6, X7, X8, X9, X10, X11, X8, X13, X14); \
LOAD_MSG_SSE4(X8, X9, X10, X11, SI, 10, 8, 7, 1, 2, 4, 6, 5, 15, 9, 3, 13, 11, 14, 12, 0); \
ROUND_SSSE3(X4, X5, X6, X7, X8, X9, X10, X11, X8, X13, X14)
#define HASH_BLOCKS(h, c, flag, blocks_base, blocks_len, BLAKE2s_FUNC) \
MOVQ h, AX; \
MOVQ c, BX; \
MOVL flag, CX; \
MOVQ blocks_base, SI; \
MOVQ blocks_len, DX; \
\
MOVQ SP, BP; \
ADDQ $15, BP; \
ANDQ $~15, BP; \
\
MOVQ 0(BX), R9; \
MOVQ R9, 0(BP); \
MOVQ CX, 8(BP); \
\
MOVOU 0(AX), X0; \
MOVOU 16(AX), X1; \
MOVOU iv0<>(SB), X2; \
MOVOU iv1<>(SB), X3 \
\
MOVOU counter<>(SB), X12; \
MOVOU rol16<>(SB), X13; \
MOVOU rol8<>(SB), X14; \
MOVO 0(BP), X15; \
\
loop: \
MOVO X0, X4; \
MOVO X1, X5; \
MOVO X2, X6; \
MOVO X3, X7; \
\
PADDQ X12, X15; \
PXOR X15, X7; \
\
BLAKE2s_FUNC(); \
\
PXOR X4, X0; \
PXOR X5, X1; \
PXOR X6, X0; \
PXOR X7, X1; \
\
LEAQ 64(SI), SI; \
SUBQ $64, DX; \
JNE loop; \
\
MOVO X15, 0(BP); \
MOVQ 0(BP), R9; \
MOVQ R9, 0(BX); \
\
MOVOU X0, 0(AX); \
MOVOU X1, 16(AX)
// func hashBlocksSSE2(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte)
TEXT ·hashBlocksSSE2(SB), 0, $672-48 // frame = 656 + 16 byte alignment
HASH_BLOCKS(h+0(FP), c+8(FP), flag+16(FP), blocks_base+24(FP), blocks_len+32(FP), BLAKE2s_SSE2)
RET
// func hashBlocksSSSE3(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte)
TEXT ·hashBlocksSSSE3(SB), 0, $672-48 // frame = 656 + 16 byte alignment
HASH_BLOCKS(h+0(FP), c+8(FP), flag+16(FP), blocks_base+24(FP), blocks_len+32(FP), BLAKE2s_SSSE3)
RET
// func hashBlocksSSE4(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte)
TEXT ·hashBlocksSSE4(SB), 0, $32-48 // frame = 16 + 16 byte alignment
HASH_BLOCKS(h+0(FP), c+8(FP), flag+16(FP), blocks_base+24(FP), blocks_len+32(FP), BLAKE2s_SSE4)
RET
+178
View File
@@ -0,0 +1,178 @@
// Copyright 2016 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.
package blake2s
import (
"math/bits"
)
// the precomputed values for BLAKE2s
// there are 10 16-byte arrays - one for each round
// the entries are calculated from the sigma constants.
var precomputed = [10][16]byte{
{0, 2, 4, 6, 1, 3, 5, 7, 8, 10, 12, 14, 9, 11, 13, 15},
{14, 4, 9, 13, 10, 8, 15, 6, 1, 0, 11, 5, 12, 2, 7, 3},
{11, 12, 5, 15, 8, 0, 2, 13, 10, 3, 7, 9, 14, 6, 1, 4},
{7, 3, 13, 11, 9, 1, 12, 14, 2, 5, 4, 15, 6, 10, 0, 8},
{9, 5, 2, 10, 0, 7, 4, 15, 14, 11, 6, 3, 1, 12, 8, 13},
{2, 6, 0, 8, 12, 10, 11, 3, 4, 7, 15, 1, 13, 5, 14, 9},
{12, 1, 14, 4, 5, 15, 13, 10, 0, 6, 9, 8, 7, 3, 2, 11},
{13, 7, 12, 3, 11, 14, 1, 9, 5, 15, 8, 2, 0, 4, 6, 10},
{6, 14, 11, 0, 15, 9, 3, 8, 12, 13, 1, 10, 2, 7, 4, 5},
{10, 8, 7, 1, 2, 4, 6, 5, 15, 9, 3, 13, 11, 14, 12, 0},
}
func hashBlocksGeneric(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte) {
var m [16]uint32
c0, c1 := c[0], c[1]
for i := 0; i < len(blocks); {
c0 += BlockSize
if c0 < BlockSize {
c1++
}
v0, v1, v2, v3, v4, v5, v6, v7 := h[0], h[1], h[2], h[3], h[4], h[5], h[6], h[7]
v8, v9, v10, v11, v12, v13, v14, v15 := iv[0], iv[1], iv[2], iv[3], iv[4], iv[5], iv[6], iv[7]
v12 ^= c0
v13 ^= c1
v14 ^= flag
for j := range m {
m[j] = uint32(blocks[i]) | uint32(blocks[i+1])<<8 | uint32(blocks[i+2])<<16 | uint32(blocks[i+3])<<24
i += 4
}
for k := range precomputed {
s := &(precomputed[k])
v0 += m[s[0]]
v0 += v4
v12 ^= v0
v12 = bits.RotateLeft32(v12, -16)
v8 += v12
v4 ^= v8
v4 = bits.RotateLeft32(v4, -12)
v1 += m[s[1]]
v1 += v5
v13 ^= v1
v13 = bits.RotateLeft32(v13, -16)
v9 += v13
v5 ^= v9
v5 = bits.RotateLeft32(v5, -12)
v2 += m[s[2]]
v2 += v6
v14 ^= v2
v14 = bits.RotateLeft32(v14, -16)
v10 += v14
v6 ^= v10
v6 = bits.RotateLeft32(v6, -12)
v3 += m[s[3]]
v3 += v7
v15 ^= v3
v15 = bits.RotateLeft32(v15, -16)
v11 += v15
v7 ^= v11
v7 = bits.RotateLeft32(v7, -12)
v0 += m[s[4]]
v0 += v4
v12 ^= v0
v12 = bits.RotateLeft32(v12, -8)
v8 += v12
v4 ^= v8
v4 = bits.RotateLeft32(v4, -7)
v1 += m[s[5]]
v1 += v5
v13 ^= v1
v13 = bits.RotateLeft32(v13, -8)
v9 += v13
v5 ^= v9
v5 = bits.RotateLeft32(v5, -7)
v2 += m[s[6]]
v2 += v6
v14 ^= v2
v14 = bits.RotateLeft32(v14, -8)
v10 += v14
v6 ^= v10
v6 = bits.RotateLeft32(v6, -7)
v3 += m[s[7]]
v3 += v7
v15 ^= v3
v15 = bits.RotateLeft32(v15, -8)
v11 += v15
v7 ^= v11
v7 = bits.RotateLeft32(v7, -7)
v0 += m[s[8]]
v0 += v5
v15 ^= v0
v15 = bits.RotateLeft32(v15, -16)
v10 += v15
v5 ^= v10
v5 = bits.RotateLeft32(v5, -12)
v1 += m[s[9]]
v1 += v6
v12 ^= v1
v12 = bits.RotateLeft32(v12, -16)
v11 += v12
v6 ^= v11
v6 = bits.RotateLeft32(v6, -12)
v2 += m[s[10]]
v2 += v7
v13 ^= v2
v13 = bits.RotateLeft32(v13, -16)
v8 += v13
v7 ^= v8
v7 = bits.RotateLeft32(v7, -12)
v3 += m[s[11]]
v3 += v4
v14 ^= v3
v14 = bits.RotateLeft32(v14, -16)
v9 += v14
v4 ^= v9
v4 = bits.RotateLeft32(v4, -12)
v0 += m[s[12]]
v0 += v5
v15 ^= v0
v15 = bits.RotateLeft32(v15, -8)
v10 += v15
v5 ^= v10
v5 = bits.RotateLeft32(v5, -7)
v1 += m[s[13]]
v1 += v6
v12 ^= v1
v12 = bits.RotateLeft32(v12, -8)
v11 += v12
v6 ^= v11
v6 = bits.RotateLeft32(v6, -7)
v2 += m[s[14]]
v2 += v7
v13 ^= v2
v13 = bits.RotateLeft32(v13, -8)
v8 += v13
v7 ^= v8
v7 = bits.RotateLeft32(v7, -7)
v3 += m[s[15]]
v3 += v4
v14 ^= v3
v14 = bits.RotateLeft32(v14, -8)
v9 += v14
v4 ^= v9
v4 = bits.RotateLeft32(v4, -7)
}
h[0] ^= v0 ^ v8
h[1] ^= v1 ^ v9
h[2] ^= v2 ^ v10
h[3] ^= v3 ^ v11
h[4] ^= v4 ^ v12
h[5] ^= v5 ^ v13
h[6] ^= v6 ^ v14
h[7] ^= v7 ^ v15
}
c[0], c[1] = c0, c1
}
+18
View File
@@ -0,0 +1,18 @@
// Copyright 2016 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.
//go:build (!amd64 && !386) || !gc || purego
// +build !amd64,!386 !gc purego
package blake2s
var (
useSSE4 = false
useSSSE3 = false
useSSE2 = false
)
func hashBlocks(h *[8]uint32, c *[2]uint32, flag uint32, blocks []byte) {
hashBlocksGeneric(h, c, flag, blocks)
}
+178
View File
@@ -0,0 +1,178 @@
// Copyright 2017 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.
package blake2s
import (
"encoding/binary"
"errors"
"io"
)
// XOF defines the interface to hash functions that
// support arbitrary-length output.
type XOF interface {
// Write absorbs more data into the hash's state. It panics if called
// after Read.
io.Writer
// Read reads more output from the hash. It returns io.EOF if the limit
// has been reached.
io.Reader
// Clone returns a copy of the XOF in its current state.
Clone() XOF
// Reset resets the XOF to its initial state.
Reset()
}
// OutputLengthUnknown can be used as the size argument to NewXOF to indicate
// the length of the output is not known in advance.
const OutputLengthUnknown = 0
// magicUnknownOutputLength is a magic value for the output size that indicates
// an unknown number of output bytes.
const magicUnknownOutputLength = 65535
// maxOutputLength is the absolute maximum number of bytes to produce when the
// number of output bytes is unknown.
const maxOutputLength = (1 << 32) * 32
// NewXOF creates a new variable-output-length hash. The hash either produce a
// known number of bytes (1 <= size < 65535), or an unknown number of bytes
// (size == OutputLengthUnknown). In the latter case, an absolute limit of
// 128GiB applies.
//
// A non-nil key turns the hash into a MAC. The key must between
// zero and 32 bytes long.
func NewXOF(size uint16, key []byte) (XOF, error) {
if len(key) > Size {
return nil, errKeySize
}
if size == magicUnknownOutputLength {
// 2^16-1 indicates an unknown number of bytes and thus isn't a
// valid length.
return nil, errors.New("blake2s: XOF length too large")
}
if size == OutputLengthUnknown {
size = magicUnknownOutputLength
}
x := &xof{
d: digest{
size: Size,
keyLen: len(key),
},
length: size,
}
copy(x.d.key[:], key)
x.Reset()
return x, nil
}
type xof struct {
d digest
length uint16
remaining uint64
cfg, root, block [Size]byte
offset int
nodeOffset uint32
readMode bool
}
func (x *xof) Write(p []byte) (n int, err error) {
if x.readMode {
panic("blake2s: write to XOF after read")
}
return x.d.Write(p)
}
func (x *xof) Clone() XOF {
clone := *x
return &clone
}
func (x *xof) Reset() {
x.cfg[0] = byte(Size)
binary.LittleEndian.PutUint32(x.cfg[4:], uint32(Size)) // leaf length
binary.LittleEndian.PutUint16(x.cfg[12:], x.length) // XOF length
x.cfg[15] = byte(Size) // inner hash size
x.d.Reset()
x.d.h[3] ^= uint32(x.length)
x.remaining = uint64(x.length)
if x.remaining == magicUnknownOutputLength {
x.remaining = maxOutputLength
}
x.offset, x.nodeOffset = 0, 0
x.readMode = false
}
func (x *xof) Read(p []byte) (n int, err error) {
if !x.readMode {
x.d.finalize(&x.root)
x.readMode = true
}
if x.remaining == 0 {
return 0, io.EOF
}
n = len(p)
if uint64(n) > x.remaining {
n = int(x.remaining)
p = p[:n]
}
if x.offset > 0 {
blockRemaining := Size - x.offset
if n < blockRemaining {
x.offset += copy(p, x.block[x.offset:])
x.remaining -= uint64(n)
return
}
copy(p, x.block[x.offset:])
p = p[blockRemaining:]
x.offset = 0
x.remaining -= uint64(blockRemaining)
}
for len(p) >= Size {
binary.LittleEndian.PutUint32(x.cfg[8:], x.nodeOffset)
x.nodeOffset++
x.d.initConfig(&x.cfg)
x.d.Write(x.root[:])
x.d.finalize(&x.block)
copy(p, x.block[:])
p = p[Size:]
x.remaining -= uint64(Size)
}
if todo := len(p); todo > 0 {
if x.remaining < uint64(Size) {
x.cfg[0] = byte(x.remaining)
}
binary.LittleEndian.PutUint32(x.cfg[8:], x.nodeOffset)
x.nodeOffset++
x.d.initConfig(&x.cfg)
x.d.Write(x.root[:])
x.d.finalize(&x.block)
x.offset = copy(p, x.block[:todo])
x.remaining -= uint64(todo)
}
return
}
func (d *digest) initConfig(cfg *[Size]byte) {
d.offset, d.c[0], d.c[1] = 0, 0, 0
for i := range d.h {
d.h[i] = iv[i] ^ binary.LittleEndian.Uint32(cfg[i*4:])
}
}
+22
View File
@@ -0,0 +1,22 @@
// Copyright 2017 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.
//go:build go1.9
// +build go1.9
package blake2s
import (
"crypto"
"hash"
)
func init() {
newHash256 := func() hash.Hash {
h, _ := New256(nil)
return h
}
crypto.RegisterHash(crypto.BLAKE2s_256, newHash256)
}
+98
View File
@@ -0,0 +1,98 @@
// Copyright 2016 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.
// Package chacha20poly1305 implements the ChaCha20-Poly1305 AEAD and its
// extended nonce variant XChaCha20-Poly1305, as specified in RFC 8439 and
// draft-irtf-cfrg-xchacha-01.
package chacha20poly1305 // import "golang.org/x/crypto/chacha20poly1305"
import (
"crypto/cipher"
"errors"
)
const (
// KeySize is the size of the key used by this AEAD, in bytes.
KeySize = 32
// NonceSize is the size of the nonce used with the standard variant of this
// AEAD, in bytes.
//
// Note that this is too short to be safely generated at random if the same
// key is reused more than 2³² times.
NonceSize = 12
// NonceSizeX is the size of the nonce used with the XChaCha20-Poly1305
// variant of this AEAD, in bytes.
NonceSizeX = 24
// Overhead is the size of the Poly1305 authentication tag, and the
// difference between a ciphertext length and its plaintext.
Overhead = 16
)
type chacha20poly1305 struct {
key [KeySize]byte
}
// New returns a ChaCha20-Poly1305 AEAD that uses the given 256-bit key.
func New(key []byte) (cipher.AEAD, error) {
if len(key) != KeySize {
return nil, errors.New("chacha20poly1305: bad key length")
}
ret := new(chacha20poly1305)
copy(ret.key[:], key)
return ret, nil
}
func (c *chacha20poly1305) NonceSize() int {
return NonceSize
}
func (c *chacha20poly1305) Overhead() int {
return Overhead
}
func (c *chacha20poly1305) Seal(dst, nonce, plaintext, additionalData []byte) []byte {
if len(nonce) != NonceSize {
panic("chacha20poly1305: bad nonce length passed to Seal")
}
if uint64(len(plaintext)) > (1<<38)-64 {
panic("chacha20poly1305: plaintext too large")
}
return c.seal(dst, nonce, plaintext, additionalData)
}
var errOpen = errors.New("chacha20poly1305: message authentication failed")
func (c *chacha20poly1305) Open(dst, nonce, ciphertext, additionalData []byte) ([]byte, error) {
if len(nonce) != NonceSize {
panic("chacha20poly1305: bad nonce length passed to Open")
}
if len(ciphertext) < 16 {
return nil, errOpen
}
if uint64(len(ciphertext)) > (1<<38)-48 {
panic("chacha20poly1305: ciphertext too large")
}
return c.open(dst, nonce, ciphertext, additionalData)
}
// sliceForAppend takes a slice and a requested number of bytes. It returns a
// slice with the contents of the given slice followed by that many bytes and a
// second slice that aliases into it and contains only the extra bytes. If the
// original slice has sufficient capacity then no allocation is performed.
func sliceForAppend(in []byte, n int) (head, tail []byte) {
if total := len(in) + n; cap(in) >= total {
head = in[:total]
} else {
head = make([]byte, total)
copy(head, in)
}
tail = head[len(in):]
return
}
+87
View File
@@ -0,0 +1,87 @@
// Copyright 2016 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.
//go:build gc && !purego
// +build gc,!purego
package chacha20poly1305
import (
"encoding/binary"
"golang.org/x/crypto/internal/alias"
"golang.org/x/sys/cpu"
)
//go:noescape
func chacha20Poly1305Open(dst []byte, key []uint32, src, ad []byte) bool
//go:noescape
func chacha20Poly1305Seal(dst []byte, key []uint32, src, ad []byte)
var (
useAVX2 = cpu.X86.HasAVX2 && cpu.X86.HasBMI2
)
// setupState writes a ChaCha20 input matrix to state. See
// https://tools.ietf.org/html/rfc7539#section-2.3.
func setupState(state *[16]uint32, key *[32]byte, nonce []byte) {
state[0] = 0x61707865
state[1] = 0x3320646e
state[2] = 0x79622d32
state[3] = 0x6b206574
state[4] = binary.LittleEndian.Uint32(key[0:4])
state[5] = binary.LittleEndian.Uint32(key[4:8])
state[6] = binary.LittleEndian.Uint32(key[8:12])
state[7] = binary.LittleEndian.Uint32(key[12:16])
state[8] = binary.LittleEndian.Uint32(key[16:20])
state[9] = binary.LittleEndian.Uint32(key[20:24])
state[10] = binary.LittleEndian.Uint32(key[24:28])
state[11] = binary.LittleEndian.Uint32(key[28:32])
state[12] = 0
state[13] = binary.LittleEndian.Uint32(nonce[0:4])
state[14] = binary.LittleEndian.Uint32(nonce[4:8])
state[15] = binary.LittleEndian.Uint32(nonce[8:12])
}
func (c *chacha20poly1305) seal(dst, nonce, plaintext, additionalData []byte) []byte {
if !cpu.X86.HasSSSE3 {
return c.sealGeneric(dst, nonce, plaintext, additionalData)
}
var state [16]uint32
setupState(&state, &c.key, nonce)
ret, out := sliceForAppend(dst, len(plaintext)+16)
if alias.InexactOverlap(out, plaintext) {
panic("chacha20poly1305: invalid buffer overlap")
}
chacha20Poly1305Seal(out[:], state[:], plaintext, additionalData)
return ret
}
func (c *chacha20poly1305) open(dst, nonce, ciphertext, additionalData []byte) ([]byte, error) {
if !cpu.X86.HasSSSE3 {
return c.openGeneric(dst, nonce, ciphertext, additionalData)
}
var state [16]uint32
setupState(&state, &c.key, nonce)
ciphertext = ciphertext[:len(ciphertext)-16]
ret, out := sliceForAppend(dst, len(ciphertext))
if alias.InexactOverlap(out, ciphertext) {
panic("chacha20poly1305: invalid buffer overlap")
}
if !chacha20Poly1305Open(out, state[:], ciphertext, additionalData) {
for i := range out {
out[i] = 0
}
return nil, errOpen
}
return ret, nil
}
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,81 @@
// Copyright 2016 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.
package chacha20poly1305
import (
"encoding/binary"
"golang.org/x/crypto/chacha20"
"golang.org/x/crypto/internal/alias"
"golang.org/x/crypto/internal/poly1305"
)
func writeWithPadding(p *poly1305.MAC, b []byte) {
p.Write(b)
if rem := len(b) % 16; rem != 0 {
var buf [16]byte
padLen := 16 - rem
p.Write(buf[:padLen])
}
}
func writeUint64(p *poly1305.MAC, n int) {
var buf [8]byte
binary.LittleEndian.PutUint64(buf[:], uint64(n))
p.Write(buf[:])
}
func (c *chacha20poly1305) sealGeneric(dst, nonce, plaintext, additionalData []byte) []byte {
ret, out := sliceForAppend(dst, len(plaintext)+poly1305.TagSize)
ciphertext, tag := out[:len(plaintext)], out[len(plaintext):]
if alias.InexactOverlap(out, plaintext) {
panic("chacha20poly1305: invalid buffer overlap")
}
var polyKey [32]byte
s, _ := chacha20.NewUnauthenticatedCipher(c.key[:], nonce)
s.XORKeyStream(polyKey[:], polyKey[:])
s.SetCounter(1) // set the counter to 1, skipping 32 bytes
s.XORKeyStream(ciphertext, plaintext)
p := poly1305.New(&polyKey)
writeWithPadding(p, additionalData)
writeWithPadding(p, ciphertext)
writeUint64(p, len(additionalData))
writeUint64(p, len(plaintext))
p.Sum(tag[:0])
return ret
}
func (c *chacha20poly1305) openGeneric(dst, nonce, ciphertext, additionalData []byte) ([]byte, error) {
tag := ciphertext[len(ciphertext)-16:]
ciphertext = ciphertext[:len(ciphertext)-16]
var polyKey [32]byte
s, _ := chacha20.NewUnauthenticatedCipher(c.key[:], nonce)
s.XORKeyStream(polyKey[:], polyKey[:])
s.SetCounter(1) // set the counter to 1, skipping 32 bytes
p := poly1305.New(&polyKey)
writeWithPadding(p, additionalData)
writeWithPadding(p, ciphertext)
writeUint64(p, len(additionalData))
writeUint64(p, len(ciphertext))
ret, out := sliceForAppend(dst, len(ciphertext))
if alias.InexactOverlap(out, ciphertext) {
panic("chacha20poly1305: invalid buffer overlap")
}
if !p.Verify(tag) {
for i := range out {
out[i] = 0
}
return nil, errOpen
}
s.XORKeyStream(out, ciphertext)
return ret, nil
}
+16
View File
@@ -0,0 +1,16 @@
// Copyright 2016 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.
//go:build !amd64 || !gc || purego
// +build !amd64 !gc purego
package chacha20poly1305
func (c *chacha20poly1305) seal(dst, nonce, plaintext, additionalData []byte) []byte {
return c.sealGeneric(dst, nonce, plaintext, additionalData)
}
func (c *chacha20poly1305) open(dst, nonce, ciphertext, additionalData []byte) ([]byte, error) {
return c.openGeneric(dst, nonce, ciphertext, additionalData)
}
+86
View File
@@ -0,0 +1,86 @@
// Copyright 2018 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.
package chacha20poly1305
import (
"crypto/cipher"
"errors"
"golang.org/x/crypto/chacha20"
)
type xchacha20poly1305 struct {
key [KeySize]byte
}
// NewX returns a XChaCha20-Poly1305 AEAD that uses the given 256-bit key.
//
// XChaCha20-Poly1305 is a ChaCha20-Poly1305 variant that takes a longer nonce,
// suitable to be generated randomly without risk of collisions. It should be
// preferred when nonce uniqueness cannot be trivially ensured, or whenever
// nonces are randomly generated.
func NewX(key []byte) (cipher.AEAD, error) {
if len(key) != KeySize {
return nil, errors.New("chacha20poly1305: bad key length")
}
ret := new(xchacha20poly1305)
copy(ret.key[:], key)
return ret, nil
}
func (*xchacha20poly1305) NonceSize() int {
return NonceSizeX
}
func (*xchacha20poly1305) Overhead() int {
return Overhead
}
func (x *xchacha20poly1305) Seal(dst, nonce, plaintext, additionalData []byte) []byte {
if len(nonce) != NonceSizeX {
panic("chacha20poly1305: bad nonce length passed to Seal")
}
// XChaCha20-Poly1305 technically supports a 64-bit counter, so there is no
// size limit. However, since we reuse the ChaCha20-Poly1305 implementation,
// the second half of the counter is not available. This is unlikely to be
// an issue because the cipher.AEAD API requires the entire message to be in
// memory, and the counter overflows at 256 GB.
if uint64(len(plaintext)) > (1<<38)-64 {
panic("chacha20poly1305: plaintext too large")
}
c := new(chacha20poly1305)
hKey, _ := chacha20.HChaCha20(x.key[:], nonce[0:16])
copy(c.key[:], hKey)
// The first 4 bytes of the final nonce are unused counter space.
cNonce := make([]byte, NonceSize)
copy(cNonce[4:12], nonce[16:24])
return c.seal(dst, cNonce[:], plaintext, additionalData)
}
func (x *xchacha20poly1305) Open(dst, nonce, ciphertext, additionalData []byte) ([]byte, error) {
if len(nonce) != NonceSizeX {
panic("chacha20poly1305: bad nonce length passed to Open")
}
if len(ciphertext) < 16 {
return nil, errOpen
}
if uint64(len(ciphertext)) > (1<<38)-48 {
panic("chacha20poly1305: ciphertext too large")
}
c := new(chacha20poly1305)
hKey, _ := chacha20.HChaCha20(x.key[:], nonce[0:16])
copy(c.key[:], hKey)
// The first 4 bytes of the final nonce are unused counter space.
cNonce := make([]byte, NonceSize)
copy(cNonce[4:12], nonce[16:24])
return c.open(dst, cNonce[:], ciphertext, additionalData)
}
+824
View File
@@ -0,0 +1,824 @@
// Copyright 2017 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.
package cryptobyte
import (
encoding_asn1 "encoding/asn1"
"fmt"
"math/big"
"reflect"
"time"
"golang.org/x/crypto/cryptobyte/asn1"
)
// This file contains ASN.1-related methods for String and Builder.
// Builder
// AddASN1Int64 appends a DER-encoded ASN.1 INTEGER.
func (b *Builder) AddASN1Int64(v int64) {
b.addASN1Signed(asn1.INTEGER, v)
}
// AddASN1Int64WithTag appends a DER-encoded ASN.1 INTEGER with the
// given tag.
func (b *Builder) AddASN1Int64WithTag(v int64, tag asn1.Tag) {
b.addASN1Signed(tag, v)
}
// AddASN1Enum appends a DER-encoded ASN.1 ENUMERATION.
func (b *Builder) AddASN1Enum(v int64) {
b.addASN1Signed(asn1.ENUM, v)
}
func (b *Builder) addASN1Signed(tag asn1.Tag, v int64) {
b.AddASN1(tag, func(c *Builder) {
length := 1
for i := v; i >= 0x80 || i < -0x80; i >>= 8 {
length++
}
for ; length > 0; length-- {
i := v >> uint((length-1)*8) & 0xff
c.AddUint8(uint8(i))
}
})
}
// AddASN1Uint64 appends a DER-encoded ASN.1 INTEGER.
func (b *Builder) AddASN1Uint64(v uint64) {
b.AddASN1(asn1.INTEGER, func(c *Builder) {
length := 1
for i := v; i >= 0x80; i >>= 8 {
length++
}
for ; length > 0; length-- {
i := v >> uint((length-1)*8) & 0xff
c.AddUint8(uint8(i))
}
})
}
// AddASN1BigInt appends a DER-encoded ASN.1 INTEGER.
func (b *Builder) AddASN1BigInt(n *big.Int) {
if b.err != nil {
return
}
b.AddASN1(asn1.INTEGER, func(c *Builder) {
if n.Sign() < 0 {
// A negative number has to be converted to two's-complement form. So we
// invert and subtract 1. If the most-significant-bit isn't set then
// we'll need to pad the beginning with 0xff in order to keep the number
// negative.
nMinus1 := new(big.Int).Neg(n)
nMinus1.Sub(nMinus1, bigOne)
bytes := nMinus1.Bytes()
for i := range bytes {
bytes[i] ^= 0xff
}
if len(bytes) == 0 || bytes[0]&0x80 == 0 {
c.add(0xff)
}
c.add(bytes...)
} else if n.Sign() == 0 {
c.add(0)
} else {
bytes := n.Bytes()
if bytes[0]&0x80 != 0 {
c.add(0)
}
c.add(bytes...)
}
})
}
// AddASN1OctetString appends a DER-encoded ASN.1 OCTET STRING.
func (b *Builder) AddASN1OctetString(bytes []byte) {
b.AddASN1(asn1.OCTET_STRING, func(c *Builder) {
c.AddBytes(bytes)
})
}
const generalizedTimeFormatStr = "20060102150405Z0700"
// AddASN1GeneralizedTime appends a DER-encoded ASN.1 GENERALIZEDTIME.
func (b *Builder) AddASN1GeneralizedTime(t time.Time) {
if t.Year() < 0 || t.Year() > 9999 {
b.err = fmt.Errorf("cryptobyte: cannot represent %v as a GeneralizedTime", t)
return
}
b.AddASN1(asn1.GeneralizedTime, func(c *Builder) {
c.AddBytes([]byte(t.Format(generalizedTimeFormatStr)))
})
}
// AddASN1UTCTime appends a DER-encoded ASN.1 UTCTime.
func (b *Builder) AddASN1UTCTime(t time.Time) {
b.AddASN1(asn1.UTCTime, func(c *Builder) {
// As utilized by the X.509 profile, UTCTime can only
// represent the years 1950 through 2049.
if t.Year() < 1950 || t.Year() >= 2050 {
b.err = fmt.Errorf("cryptobyte: cannot represent %v as a UTCTime", t)
return
}
c.AddBytes([]byte(t.Format(defaultUTCTimeFormatStr)))
})
}
// AddASN1BitString appends a DER-encoded ASN.1 BIT STRING. This does not
// support BIT STRINGs that are not a whole number of bytes.
func (b *Builder) AddASN1BitString(data []byte) {
b.AddASN1(asn1.BIT_STRING, func(b *Builder) {
b.AddUint8(0)
b.AddBytes(data)
})
}
func (b *Builder) addBase128Int(n int64) {
var length int
if n == 0 {
length = 1
} else {
for i := n; i > 0; i >>= 7 {
length++
}
}
for i := length - 1; i >= 0; i-- {
o := byte(n >> uint(i*7))
o &= 0x7f
if i != 0 {
o |= 0x80
}
b.add(o)
}
}
func isValidOID(oid encoding_asn1.ObjectIdentifier) bool {
if len(oid) < 2 {
return false
}
if oid[0] > 2 || (oid[0] <= 1 && oid[1] >= 40) {
return false
}
for _, v := range oid {
if v < 0 {
return false
}
}
return true
}
func (b *Builder) AddASN1ObjectIdentifier(oid encoding_asn1.ObjectIdentifier) {
b.AddASN1(asn1.OBJECT_IDENTIFIER, func(b *Builder) {
if !isValidOID(oid) {
b.err = fmt.Errorf("cryptobyte: invalid OID: %v", oid)
return
}
b.addBase128Int(int64(oid[0])*40 + int64(oid[1]))
for _, v := range oid[2:] {
b.addBase128Int(int64(v))
}
})
}
func (b *Builder) AddASN1Boolean(v bool) {
b.AddASN1(asn1.BOOLEAN, func(b *Builder) {
if v {
b.AddUint8(0xff)
} else {
b.AddUint8(0)
}
})
}
func (b *Builder) AddASN1NULL() {
b.add(uint8(asn1.NULL), 0)
}
// MarshalASN1 calls encoding_asn1.Marshal on its input and appends the result if
// successful or records an error if one occurred.
func (b *Builder) MarshalASN1(v interface{}) {
// NOTE(martinkr): This is somewhat of a hack to allow propagation of
// encoding_asn1.Marshal errors into Builder.err. N.B. if you call MarshalASN1 with a
// value embedded into a struct, its tag information is lost.
if b.err != nil {
return
}
bytes, err := encoding_asn1.Marshal(v)
if err != nil {
b.err = err
return
}
b.AddBytes(bytes)
}
// AddASN1 appends an ASN.1 object. The object is prefixed with the given tag.
// Tags greater than 30 are not supported and result in an error (i.e.
// low-tag-number form only). The child builder passed to the
// BuilderContinuation can be used to build the content of the ASN.1 object.
func (b *Builder) AddASN1(tag asn1.Tag, f BuilderContinuation) {
if b.err != nil {
return
}
// Identifiers with the low five bits set indicate high-tag-number format
// (two or more octets), which we don't support.
if tag&0x1f == 0x1f {
b.err = fmt.Errorf("cryptobyte: high-tag number identifier octects not supported: 0x%x", tag)
return
}
b.AddUint8(uint8(tag))
b.addLengthPrefixed(1, true, f)
}
// String
// ReadASN1Boolean decodes an ASN.1 BOOLEAN and converts it to a boolean
// representation into out and advances. It reports whether the read
// was successful.
func (s *String) ReadASN1Boolean(out *bool) bool {
var bytes String
if !s.ReadASN1(&bytes, asn1.BOOLEAN) || len(bytes) != 1 {
return false
}
switch bytes[0] {
case 0:
*out = false
case 0xff:
*out = true
default:
return false
}
return true
}
// ReadASN1Integer decodes an ASN.1 INTEGER into out and advances. If out does
// not point to an integer, to a big.Int, or to a []byte it panics. Only
// positive and zero values can be decoded into []byte, and they are returned as
// big-endian binary values that share memory with s. Positive values will have
// no leading zeroes, and zero will be returned as a single zero byte.
// ReadASN1Integer reports whether the read was successful.
func (s *String) ReadASN1Integer(out interface{}) bool {
switch out := out.(type) {
case *int, *int8, *int16, *int32, *int64:
var i int64
if !s.readASN1Int64(&i) || reflect.ValueOf(out).Elem().OverflowInt(i) {
return false
}
reflect.ValueOf(out).Elem().SetInt(i)
return true
case *uint, *uint8, *uint16, *uint32, *uint64:
var u uint64
if !s.readASN1Uint64(&u) || reflect.ValueOf(out).Elem().OverflowUint(u) {
return false
}
reflect.ValueOf(out).Elem().SetUint(u)
return true
case *big.Int:
return s.readASN1BigInt(out)
case *[]byte:
return s.readASN1Bytes(out)
default:
panic("out does not point to an integer type")
}
}
func checkASN1Integer(bytes []byte) bool {
if len(bytes) == 0 {
// An INTEGER is encoded with at least one octet.
return false
}
if len(bytes) == 1 {
return true
}
if bytes[0] == 0 && bytes[1]&0x80 == 0 || bytes[0] == 0xff && bytes[1]&0x80 == 0x80 {
// Value is not minimally encoded.
return false
}
return true
}
var bigOne = big.NewInt(1)
func (s *String) readASN1BigInt(out *big.Int) bool {
var bytes String
if !s.ReadASN1(&bytes, asn1.INTEGER) || !checkASN1Integer(bytes) {
return false
}
if bytes[0]&0x80 == 0x80 {
// Negative number.
neg := make([]byte, len(bytes))
for i, b := range bytes {
neg[i] = ^b
}
out.SetBytes(neg)
out.Add(out, bigOne)
out.Neg(out)
} else {
out.SetBytes(bytes)
}
return true
}
func (s *String) readASN1Bytes(out *[]byte) bool {
var bytes String
if !s.ReadASN1(&bytes, asn1.INTEGER) || !checkASN1Integer(bytes) {
return false
}
if bytes[0]&0x80 == 0x80 {
return false
}
for len(bytes) > 1 && bytes[0] == 0 {
bytes = bytes[1:]
}
*out = bytes
return true
}
func (s *String) readASN1Int64(out *int64) bool {
var bytes String
if !s.ReadASN1(&bytes, asn1.INTEGER) || !checkASN1Integer(bytes) || !asn1Signed(out, bytes) {
return false
}
return true
}
func asn1Signed(out *int64, n []byte) bool {
length := len(n)
if length > 8 {
return false
}
for i := 0; i < length; i++ {
*out <<= 8
*out |= int64(n[i])
}
// Shift up and down in order to sign extend the result.
*out <<= 64 - uint8(length)*8
*out >>= 64 - uint8(length)*8
return true
}
func (s *String) readASN1Uint64(out *uint64) bool {
var bytes String
if !s.ReadASN1(&bytes, asn1.INTEGER) || !checkASN1Integer(bytes) || !asn1Unsigned(out, bytes) {
return false
}
return true
}
func asn1Unsigned(out *uint64, n []byte) bool {
length := len(n)
if length > 9 || length == 9 && n[0] != 0 {
// Too large for uint64.
return false
}
if n[0]&0x80 != 0 {
// Negative number.
return false
}
for i := 0; i < length; i++ {
*out <<= 8
*out |= uint64(n[i])
}
return true
}
// ReadASN1Int64WithTag decodes an ASN.1 INTEGER with the given tag into out
// and advances. It reports whether the read was successful and resulted in a
// value that can be represented in an int64.
func (s *String) ReadASN1Int64WithTag(out *int64, tag asn1.Tag) bool {
var bytes String
return s.ReadASN1(&bytes, tag) && checkASN1Integer(bytes) && asn1Signed(out, bytes)
}
// ReadASN1Enum decodes an ASN.1 ENUMERATION into out and advances. It reports
// whether the read was successful.
func (s *String) ReadASN1Enum(out *int) bool {
var bytes String
var i int64
if !s.ReadASN1(&bytes, asn1.ENUM) || !checkASN1Integer(bytes) || !asn1Signed(&i, bytes) {
return false
}
if int64(int(i)) != i {
return false
}
*out = int(i)
return true
}
func (s *String) readBase128Int(out *int) bool {
ret := 0
for i := 0; len(*s) > 0; i++ {
if i == 5 {
return false
}
// Avoid overflowing int on a 32-bit platform.
// We don't want different behavior based on the architecture.
if ret >= 1<<(31-7) {
return false
}
ret <<= 7
b := s.read(1)[0]
// ITU-T X.690, section 8.19.2:
// The subidentifier shall be encoded in the fewest possible octets,
// that is, the leading octet of the subidentifier shall not have the value 0x80.
if i == 0 && b == 0x80 {
return false
}
ret |= int(b & 0x7f)
if b&0x80 == 0 {
*out = ret
return true
}
}
return false // truncated
}
// ReadASN1ObjectIdentifier decodes an ASN.1 OBJECT IDENTIFIER into out and
// advances. It reports whether the read was successful.
func (s *String) ReadASN1ObjectIdentifier(out *encoding_asn1.ObjectIdentifier) bool {
var bytes String
if !s.ReadASN1(&bytes, asn1.OBJECT_IDENTIFIER) || len(bytes) == 0 {
return false
}
// In the worst case, we get two elements from the first byte (which is
// encoded differently) and then every varint is a single byte long.
components := make([]int, len(bytes)+1)
// The first varint is 40*value1 + value2:
// According to this packing, value1 can take the values 0, 1 and 2 only.
// When value1 = 0 or value1 = 1, then value2 is <= 39. When value1 = 2,
// then there are no restrictions on value2.
var v int
if !bytes.readBase128Int(&v) {
return false
}
if v < 80 {
components[0] = v / 40
components[1] = v % 40
} else {
components[0] = 2
components[1] = v - 80
}
i := 2
for ; len(bytes) > 0; i++ {
if !bytes.readBase128Int(&v) {
return false
}
components[i] = v
}
*out = components[:i]
return true
}
// ReadASN1GeneralizedTime decodes an ASN.1 GENERALIZEDTIME into out and
// advances. It reports whether the read was successful.
func (s *String) ReadASN1GeneralizedTime(out *time.Time) bool {
var bytes String
if !s.ReadASN1(&bytes, asn1.GeneralizedTime) {
return false
}
t := string(bytes)
res, err := time.Parse(generalizedTimeFormatStr, t)
if err != nil {
return false
}
if serialized := res.Format(generalizedTimeFormatStr); serialized != t {
return false
}
*out = res
return true
}
const defaultUTCTimeFormatStr = "060102150405Z0700"
// ReadASN1UTCTime decodes an ASN.1 UTCTime into out and advances.
// It reports whether the read was successful.
func (s *String) ReadASN1UTCTime(out *time.Time) bool {
var bytes String
if !s.ReadASN1(&bytes, asn1.UTCTime) {
return false
}
t := string(bytes)
formatStr := defaultUTCTimeFormatStr
var err error
res, err := time.Parse(formatStr, t)
if err != nil {
// Fallback to minute precision if we can't parse second
// precision. If we are following X.509 or X.690 we shouldn't
// support this, but we do.
formatStr = "0601021504Z0700"
res, err = time.Parse(formatStr, t)
}
if err != nil {
return false
}
if serialized := res.Format(formatStr); serialized != t {
return false
}
if res.Year() >= 2050 {
// UTCTime interprets the low order digits 50-99 as 1950-99.
// This only applies to its use in the X.509 profile.
// See https://tools.ietf.org/html/rfc5280#section-4.1.2.5.1
res = res.AddDate(-100, 0, 0)
}
*out = res
return true
}
// ReadASN1BitString decodes an ASN.1 BIT STRING into out and advances.
// It reports whether the read was successful.
func (s *String) ReadASN1BitString(out *encoding_asn1.BitString) bool {
var bytes String
if !s.ReadASN1(&bytes, asn1.BIT_STRING) || len(bytes) == 0 ||
len(bytes)*8/8 != len(bytes) {
return false
}
paddingBits := bytes[0]
bytes = bytes[1:]
if paddingBits > 7 ||
len(bytes) == 0 && paddingBits != 0 ||
len(bytes) > 0 && bytes[len(bytes)-1]&(1<<paddingBits-1) != 0 {
return false
}
out.BitLength = len(bytes)*8 - int(paddingBits)
out.Bytes = bytes
return true
}
// ReadASN1BitStringAsBytes decodes an ASN.1 BIT STRING into out and advances. It is
// an error if the BIT STRING is not a whole number of bytes. It reports
// whether the read was successful.
func (s *String) ReadASN1BitStringAsBytes(out *[]byte) bool {
var bytes String
if !s.ReadASN1(&bytes, asn1.BIT_STRING) || len(bytes) == 0 {
return false
}
paddingBits := bytes[0]
if paddingBits != 0 {
return false
}
*out = bytes[1:]
return true
}
// ReadASN1Bytes reads the contents of a DER-encoded ASN.1 element (not including
// tag and length bytes) into out, and advances. The element must match the
// given tag. It reports whether the read was successful.
func (s *String) ReadASN1Bytes(out *[]byte, tag asn1.Tag) bool {
return s.ReadASN1((*String)(out), tag)
}
// ReadASN1 reads the contents of a DER-encoded ASN.1 element (not including
// tag and length bytes) into out, and advances. The element must match the
// given tag. It reports whether the read was successful.
//
// Tags greater than 30 are not supported (i.e. low-tag-number format only).
func (s *String) ReadASN1(out *String, tag asn1.Tag) bool {
var t asn1.Tag
if !s.ReadAnyASN1(out, &t) || t != tag {
return false
}
return true
}
// ReadASN1Element reads the contents of a DER-encoded ASN.1 element (including
// tag and length bytes) into out, and advances. The element must match the
// given tag. It reports whether the read was successful.
//
// Tags greater than 30 are not supported (i.e. low-tag-number format only).
func (s *String) ReadASN1Element(out *String, tag asn1.Tag) bool {
var t asn1.Tag
if !s.ReadAnyASN1Element(out, &t) || t != tag {
return false
}
return true
}
// ReadAnyASN1 reads the contents of a DER-encoded ASN.1 element (not including
// tag and length bytes) into out, sets outTag to its tag, and advances.
// It reports whether the read was successful.
//
// Tags greater than 30 are not supported (i.e. low-tag-number format only).
func (s *String) ReadAnyASN1(out *String, outTag *asn1.Tag) bool {
return s.readASN1(out, outTag, true /* skip header */)
}
// ReadAnyASN1Element reads the contents of a DER-encoded ASN.1 element
// (including tag and length bytes) into out, sets outTag to is tag, and
// advances. It reports whether the read was successful.
//
// Tags greater than 30 are not supported (i.e. low-tag-number format only).
func (s *String) ReadAnyASN1Element(out *String, outTag *asn1.Tag) bool {
return s.readASN1(out, outTag, false /* include header */)
}
// PeekASN1Tag reports whether the next ASN.1 value on the string starts with
// the given tag.
func (s String) PeekASN1Tag(tag asn1.Tag) bool {
if len(s) == 0 {
return false
}
return asn1.Tag(s[0]) == tag
}
// SkipASN1 reads and discards an ASN.1 element with the given tag. It
// reports whether the operation was successful.
func (s *String) SkipASN1(tag asn1.Tag) bool {
var unused String
return s.ReadASN1(&unused, tag)
}
// ReadOptionalASN1 attempts to read the contents of a DER-encoded ASN.1
// element (not including tag and length bytes) tagged with the given tag into
// out. It stores whether an element with the tag was found in outPresent,
// unless outPresent is nil. It reports whether the read was successful.
func (s *String) ReadOptionalASN1(out *String, outPresent *bool, tag asn1.Tag) bool {
present := s.PeekASN1Tag(tag)
if outPresent != nil {
*outPresent = present
}
if present && !s.ReadASN1(out, tag) {
return false
}
return true
}
// SkipOptionalASN1 advances s over an ASN.1 element with the given tag, or
// else leaves s unchanged. It reports whether the operation was successful.
func (s *String) SkipOptionalASN1(tag asn1.Tag) bool {
if !s.PeekASN1Tag(tag) {
return true
}
var unused String
return s.ReadASN1(&unused, tag)
}
// ReadOptionalASN1Integer attempts to read an optional ASN.1 INTEGER explicitly
// tagged with tag into out and advances. If no element with a matching tag is
// present, it writes defaultValue into out instead. Otherwise, it behaves like
// ReadASN1Integer.
func (s *String) ReadOptionalASN1Integer(out interface{}, tag asn1.Tag, defaultValue interface{}) bool {
var present bool
var i String
if !s.ReadOptionalASN1(&i, &present, tag) {
return false
}
if !present {
switch out.(type) {
case *int, *int8, *int16, *int32, *int64,
*uint, *uint8, *uint16, *uint32, *uint64, *[]byte:
reflect.ValueOf(out).Elem().Set(reflect.ValueOf(defaultValue))
case *big.Int:
if defaultValue, ok := defaultValue.(*big.Int); ok {
out.(*big.Int).Set(defaultValue)
} else {
panic("out points to big.Int, but defaultValue does not")
}
default:
panic("invalid integer type")
}
return true
}
if !i.ReadASN1Integer(out) || !i.Empty() {
return false
}
return true
}
// ReadOptionalASN1OctetString attempts to read an optional ASN.1 OCTET STRING
// explicitly tagged with tag into out and advances. If no element with a
// matching tag is present, it sets "out" to nil instead. It reports
// whether the read was successful.
func (s *String) ReadOptionalASN1OctetString(out *[]byte, outPresent *bool, tag asn1.Tag) bool {
var present bool
var child String
if !s.ReadOptionalASN1(&child, &present, tag) {
return false
}
if outPresent != nil {
*outPresent = present
}
if present {
var oct String
if !child.ReadASN1(&oct, asn1.OCTET_STRING) || !child.Empty() {
return false
}
*out = oct
} else {
*out = nil
}
return true
}
// ReadOptionalASN1Boolean sets *out to the value of the next ASN.1 BOOLEAN or,
// if the next bytes are not an ASN.1 BOOLEAN, to the value of defaultValue.
// It reports whether the operation was successful.
func (s *String) ReadOptionalASN1Boolean(out *bool, defaultValue bool) bool {
var present bool
var child String
if !s.ReadOptionalASN1(&child, &present, asn1.BOOLEAN) {
return false
}
if !present {
*out = defaultValue
return true
}
return s.ReadASN1Boolean(out)
}
func (s *String) readASN1(out *String, outTag *asn1.Tag, skipHeader bool) bool {
if len(*s) < 2 {
return false
}
tag, lenByte := (*s)[0], (*s)[1]
if tag&0x1f == 0x1f {
// ITU-T X.690 section 8.1.2
//
// An identifier octet with a tag part of 0x1f indicates a high-tag-number
// form identifier with two or more octets. We only support tags less than
// 31 (i.e. low-tag-number form, single octet identifier).
return false
}
if outTag != nil {
*outTag = asn1.Tag(tag)
}
// ITU-T X.690 section 8.1.3
//
// Bit 8 of the first length byte indicates whether the length is short- or
// long-form.
var length, headerLen uint32 // length includes headerLen
if lenByte&0x80 == 0 {
// Short-form length (section 8.1.3.4), encoded in bits 1-7.
length = uint32(lenByte) + 2
headerLen = 2
} else {
// Long-form length (section 8.1.3.5). Bits 1-7 encode the number of octets
// used to encode the length.
lenLen := lenByte & 0x7f
var len32 uint32
if lenLen == 0 || lenLen > 4 || len(*s) < int(2+lenLen) {
return false
}
lenBytes := String((*s)[2 : 2+lenLen])
if !lenBytes.readUnsigned(&len32, int(lenLen)) {
return false
}
// ITU-T X.690 section 10.1 (DER length forms) requires encoding the length
// with the minimum number of octets.
if len32 < 128 {
// Length should have used short-form encoding.
return false
}
if len32>>((lenLen-1)*8) == 0 {
// Leading octet is 0. Length should have been at least one byte shorter.
return false
}
headerLen = 2 + uint32(lenLen)
if headerLen+len32 < len32 {
// Overflow.
return false
}
length = headerLen + len32
}
if int(length) < 0 || !s.ReadBytes((*[]byte)(out), int(length)) {
return false
}
if skipHeader && !out.Skip(int(headerLen)) {
panic("cryptobyte: internal error")
}
return true
}
+46
View File
@@ -0,0 +1,46 @@
// Copyright 2017 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.
// Package asn1 contains supporting types for parsing and building ASN.1
// messages with the cryptobyte package.
package asn1 // import "golang.org/x/crypto/cryptobyte/asn1"
// Tag represents an ASN.1 identifier octet, consisting of a tag number
// (indicating a type) and class (such as context-specific or constructed).
//
// Methods in the cryptobyte package only support the low-tag-number form, i.e.
// a single identifier octet with bits 7-8 encoding the class and bits 1-6
// encoding the tag number.
type Tag uint8
const (
classConstructed = 0x20
classContextSpecific = 0x80
)
// Constructed returns t with the constructed class bit set.
func (t Tag) Constructed() Tag { return t | classConstructed }
// ContextSpecific returns t with the context-specific class bit set.
func (t Tag) ContextSpecific() Tag { return t | classContextSpecific }
// The following is a list of standard tag and class combinations.
const (
BOOLEAN = Tag(1)
INTEGER = Tag(2)
BIT_STRING = Tag(3)
OCTET_STRING = Tag(4)
NULL = Tag(5)
OBJECT_IDENTIFIER = Tag(6)
ENUM = Tag(10)
UTF8String = Tag(12)
SEQUENCE = Tag(16 | classConstructed)
SET = Tag(17 | classConstructed)
PrintableString = Tag(19)
T61String = Tag(20)
IA5String = Tag(22)
UTCTime = Tag(23)
GeneralizedTime = Tag(24)
GeneralString = Tag(27)
)
+345
View File
@@ -0,0 +1,345 @@
// Copyright 2017 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.
package cryptobyte
import (
"errors"
"fmt"
)
// A Builder builds byte strings from fixed-length and length-prefixed values.
// Builders either allocate space as needed, or are fixed, which means that
// they write into a given buffer and produce an error if it's exhausted.
//
// The zero value is a usable Builder that allocates space as needed.
//
// Simple values are marshaled and appended to a Builder using methods on the
// Builder. Length-prefixed values are marshaled by providing a
// BuilderContinuation, which is a function that writes the inner contents of
// the value to a given Builder. See the documentation for BuilderContinuation
// for details.
type Builder struct {
err error
result []byte
fixedSize bool
child *Builder
offset int
pendingLenLen int
pendingIsASN1 bool
inContinuation *bool
}
// NewBuilder creates a Builder that appends its output to the given buffer.
// Like append(), the slice will be reallocated if its capacity is exceeded.
// Use Bytes to get the final buffer.
func NewBuilder(buffer []byte) *Builder {
return &Builder{
result: buffer,
}
}
// NewFixedBuilder creates a Builder that appends its output into the given
// buffer. This builder does not reallocate the output buffer. Writes that
// would exceed the buffer's capacity are treated as an error.
func NewFixedBuilder(buffer []byte) *Builder {
return &Builder{
result: buffer,
fixedSize: true,
}
}
// SetError sets the value to be returned as the error from Bytes. Writes
// performed after calling SetError are ignored.
func (b *Builder) SetError(err error) {
b.err = err
}
// Bytes returns the bytes written by the builder or an error if one has
// occurred during building.
func (b *Builder) Bytes() ([]byte, error) {
if b.err != nil {
return nil, b.err
}
return b.result[b.offset:], nil
}
// BytesOrPanic returns the bytes written by the builder or panics if an error
// has occurred during building.
func (b *Builder) BytesOrPanic() []byte {
if b.err != nil {
panic(b.err)
}
return b.result[b.offset:]
}
// AddUint8 appends an 8-bit value to the byte string.
func (b *Builder) AddUint8(v uint8) {
b.add(byte(v))
}
// AddUint16 appends a big-endian, 16-bit value to the byte string.
func (b *Builder) AddUint16(v uint16) {
b.add(byte(v>>8), byte(v))
}
// AddUint24 appends a big-endian, 24-bit value to the byte string. The highest
// byte of the 32-bit input value is silently truncated.
func (b *Builder) AddUint24(v uint32) {
b.add(byte(v>>16), byte(v>>8), byte(v))
}
// AddUint32 appends a big-endian, 32-bit value to the byte string.
func (b *Builder) AddUint32(v uint32) {
b.add(byte(v>>24), byte(v>>16), byte(v>>8), byte(v))
}
// AddUint64 appends a big-endian, 64-bit value to the byte string.
func (b *Builder) AddUint64(v uint64) {
b.add(byte(v>>56), byte(v>>48), byte(v>>40), byte(v>>32), byte(v>>24), byte(v>>16), byte(v>>8), byte(v))
}
// AddBytes appends a sequence of bytes to the byte string.
func (b *Builder) AddBytes(v []byte) {
b.add(v...)
}
// BuilderContinuation is a continuation-passing interface for building
// length-prefixed byte sequences. Builder methods for length-prefixed
// sequences (AddUint8LengthPrefixed etc) will invoke the BuilderContinuation
// supplied to them. The child builder passed to the continuation can be used
// to build the content of the length-prefixed sequence. For example:
//
// parent := cryptobyte.NewBuilder()
// parent.AddUint8LengthPrefixed(func (child *Builder) {
// child.AddUint8(42)
// child.AddUint8LengthPrefixed(func (grandchild *Builder) {
// grandchild.AddUint8(5)
// })
// })
//
// It is an error to write more bytes to the child than allowed by the reserved
// length prefix. After the continuation returns, the child must be considered
// invalid, i.e. users must not store any copies or references of the child
// that outlive the continuation.
//
// If the continuation panics with a value of type BuildError then the inner
// error will be returned as the error from Bytes. If the child panics
// otherwise then Bytes will repanic with the same value.
type BuilderContinuation func(child *Builder)
// BuildError wraps an error. If a BuilderContinuation panics with this value,
// the panic will be recovered and the inner error will be returned from
// Builder.Bytes.
type BuildError struct {
Err error
}
// AddUint8LengthPrefixed adds a 8-bit length-prefixed byte sequence.
func (b *Builder) AddUint8LengthPrefixed(f BuilderContinuation) {
b.addLengthPrefixed(1, false, f)
}
// AddUint16LengthPrefixed adds a big-endian, 16-bit length-prefixed byte sequence.
func (b *Builder) AddUint16LengthPrefixed(f BuilderContinuation) {
b.addLengthPrefixed(2, false, f)
}
// AddUint24LengthPrefixed adds a big-endian, 24-bit length-prefixed byte sequence.
func (b *Builder) AddUint24LengthPrefixed(f BuilderContinuation) {
b.addLengthPrefixed(3, false, f)
}
// AddUint32LengthPrefixed adds a big-endian, 32-bit length-prefixed byte sequence.
func (b *Builder) AddUint32LengthPrefixed(f BuilderContinuation) {
b.addLengthPrefixed(4, false, f)
}
func (b *Builder) callContinuation(f BuilderContinuation, arg *Builder) {
if !*b.inContinuation {
*b.inContinuation = true
defer func() {
*b.inContinuation = false
r := recover()
if r == nil {
return
}
if buildError, ok := r.(BuildError); ok {
b.err = buildError.Err
} else {
panic(r)
}
}()
}
f(arg)
}
func (b *Builder) addLengthPrefixed(lenLen int, isASN1 bool, f BuilderContinuation) {
// Subsequent writes can be ignored if the builder has encountered an error.
if b.err != nil {
return
}
offset := len(b.result)
b.add(make([]byte, lenLen)...)
if b.inContinuation == nil {
b.inContinuation = new(bool)
}
b.child = &Builder{
result: b.result,
fixedSize: b.fixedSize,
offset: offset,
pendingLenLen: lenLen,
pendingIsASN1: isASN1,
inContinuation: b.inContinuation,
}
b.callContinuation(f, b.child)
b.flushChild()
if b.child != nil {
panic("cryptobyte: internal error")
}
}
func (b *Builder) flushChild() {
if b.child == nil {
return
}
b.child.flushChild()
child := b.child
b.child = nil
if child.err != nil {
b.err = child.err
return
}
length := len(child.result) - child.pendingLenLen - child.offset
if length < 0 {
panic("cryptobyte: internal error") // result unexpectedly shrunk
}
if child.pendingIsASN1 {
// For ASN.1, we reserved a single byte for the length. If that turned out
// to be incorrect, we have to move the contents along in order to make
// space.
if child.pendingLenLen != 1 {
panic("cryptobyte: internal error")
}
var lenLen, lenByte uint8
if int64(length) > 0xfffffffe {
b.err = errors.New("pending ASN.1 child too long")
return
} else if length > 0xffffff {
lenLen = 5
lenByte = 0x80 | 4
} else if length > 0xffff {
lenLen = 4
lenByte = 0x80 | 3
} else if length > 0xff {
lenLen = 3
lenByte = 0x80 | 2
} else if length > 0x7f {
lenLen = 2
lenByte = 0x80 | 1
} else {
lenLen = 1
lenByte = uint8(length)
length = 0
}
// Insert the initial length byte, make space for successive length bytes,
// and adjust the offset.
child.result[child.offset] = lenByte
extraBytes := int(lenLen - 1)
if extraBytes != 0 {
child.add(make([]byte, extraBytes)...)
childStart := child.offset + child.pendingLenLen
copy(child.result[childStart+extraBytes:], child.result[childStart:])
}
child.offset++
child.pendingLenLen = extraBytes
}
l := length
for i := child.pendingLenLen - 1; i >= 0; i-- {
child.result[child.offset+i] = uint8(l)
l >>= 8
}
if l != 0 {
b.err = fmt.Errorf("cryptobyte: pending child length %d exceeds %d-byte length prefix", length, child.pendingLenLen)
return
}
if b.fixedSize && &b.result[0] != &child.result[0] {
panic("cryptobyte: BuilderContinuation reallocated a fixed-size buffer")
}
b.result = child.result
}
func (b *Builder) add(bytes ...byte) {
if b.err != nil {
return
}
if b.child != nil {
panic("cryptobyte: attempted write while child is pending")
}
if len(b.result)+len(bytes) < len(bytes) {
b.err = errors.New("cryptobyte: length overflow")
}
if b.fixedSize && len(b.result)+len(bytes) > cap(b.result) {
b.err = errors.New("cryptobyte: Builder is exceeding its fixed-size buffer")
return
}
b.result = append(b.result, bytes...)
}
// Unwrite rolls back non-negative n bytes written directly to the Builder.
// An attempt by a child builder passed to a continuation to unwrite bytes
// from its parent will panic.
func (b *Builder) Unwrite(n int) {
if b.err != nil {
return
}
if b.child != nil {
panic("cryptobyte: attempted unwrite while child is pending")
}
length := len(b.result) - b.pendingLenLen - b.offset
if length < 0 {
panic("cryptobyte: internal error")
}
if n < 0 {
panic("cryptobyte: attempted to unwrite negative number of bytes")
}
if n > length {
panic("cryptobyte: attempted to unwrite more than was written")
}
b.result = b.result[:len(b.result)-n]
}
// A MarshalingValue marshals itself into a Builder.
type MarshalingValue interface {
// Marshal is called by Builder.AddValue. It receives a pointer to a builder
// to marshal itself into. It may return an error that occurred during
// marshaling, such as unset or invalid values.
Marshal(b *Builder) error
}
// AddValue calls Marshal on v, passing a pointer to the builder to append to.
// If Marshal returns an error, it is set on the Builder so that subsequent
// appends don't have an effect.
func (b *Builder) AddValue(v MarshalingValue) {
err := v.Marshal(b)
if err != nil {
b.err = err
}
}
+172
View File
@@ -0,0 +1,172 @@
// Copyright 2017 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.
// Package cryptobyte contains types that help with parsing and constructing
// length-prefixed, binary messages, including ASN.1 DER. (The asn1 subpackage
// contains useful ASN.1 constants.)
//
// The String type is for parsing. It wraps a []byte slice and provides helper
// functions for consuming structures, value by value.
//
// The Builder type is for constructing messages. It providers helper functions
// for appending values and also for appending length-prefixed submessages
// without having to worry about calculating the length prefix ahead of time.
//
// See the documentation and examples for the Builder and String types to get
// started.
package cryptobyte // import "golang.org/x/crypto/cryptobyte"
// String represents a string of bytes. It provides methods for parsing
// fixed-length and length-prefixed values from it.
type String []byte
// read advances a String by n bytes and returns them. If less than n bytes
// remain, it returns nil.
func (s *String) read(n int) []byte {
if len(*s) < n || n < 0 {
return nil
}
v := (*s)[:n]
*s = (*s)[n:]
return v
}
// Skip advances the String by n byte and reports whether it was successful.
func (s *String) Skip(n int) bool {
return s.read(n) != nil
}
// ReadUint8 decodes an 8-bit value into out and advances over it.
// It reports whether the read was successful.
func (s *String) ReadUint8(out *uint8) bool {
v := s.read(1)
if v == nil {
return false
}
*out = uint8(v[0])
return true
}
// ReadUint16 decodes a big-endian, 16-bit value into out and advances over it.
// It reports whether the read was successful.
func (s *String) ReadUint16(out *uint16) bool {
v := s.read(2)
if v == nil {
return false
}
*out = uint16(v[0])<<8 | uint16(v[1])
return true
}
// ReadUint24 decodes a big-endian, 24-bit value into out and advances over it.
// It reports whether the read was successful.
func (s *String) ReadUint24(out *uint32) bool {
v := s.read(3)
if v == nil {
return false
}
*out = uint32(v[0])<<16 | uint32(v[1])<<8 | uint32(v[2])
return true
}
// ReadUint32 decodes a big-endian, 32-bit value into out and advances over it.
// It reports whether the read was successful.
func (s *String) ReadUint32(out *uint32) bool {
v := s.read(4)
if v == nil {
return false
}
*out = uint32(v[0])<<24 | uint32(v[1])<<16 | uint32(v[2])<<8 | uint32(v[3])
return true
}
// ReadUint64 decodes a big-endian, 64-bit value into out and advances over it.
// It reports whether the read was successful.
func (s *String) ReadUint64(out *uint64) bool {
v := s.read(8)
if v == nil {
return false
}
*out = uint64(v[0])<<56 | uint64(v[1])<<48 | uint64(v[2])<<40 | uint64(v[3])<<32 | uint64(v[4])<<24 | uint64(v[5])<<16 | uint64(v[6])<<8 | uint64(v[7])
return true
}
func (s *String) readUnsigned(out *uint32, length int) bool {
v := s.read(length)
if v == nil {
return false
}
var result uint32
for i := 0; i < length; i++ {
result <<= 8
result |= uint32(v[i])
}
*out = result
return true
}
func (s *String) readLengthPrefixed(lenLen int, outChild *String) bool {
lenBytes := s.read(lenLen)
if lenBytes == nil {
return false
}
var length uint32
for _, b := range lenBytes {
length = length << 8
length = length | uint32(b)
}
v := s.read(int(length))
if v == nil {
return false
}
*outChild = v
return true
}
// ReadUint8LengthPrefixed reads the content of an 8-bit length-prefixed value
// into out and advances over it. It reports whether the read was successful.
func (s *String) ReadUint8LengthPrefixed(out *String) bool {
return s.readLengthPrefixed(1, out)
}
// ReadUint16LengthPrefixed reads the content of a big-endian, 16-bit
// length-prefixed value into out and advances over it. It reports whether the
// read was successful.
func (s *String) ReadUint16LengthPrefixed(out *String) bool {
return s.readLengthPrefixed(2, out)
}
// ReadUint24LengthPrefixed reads the content of a big-endian, 24-bit
// length-prefixed value into out and advances over it. It reports whether
// the read was successful.
func (s *String) ReadUint24LengthPrefixed(out *String) bool {
return s.readLengthPrefixed(3, out)
}
// ReadBytes reads n bytes into out and advances over them. It reports
// whether the read was successful.
func (s *String) ReadBytes(out *[]byte, n int) bool {
v := s.read(n)
if v == nil {
return false
}
*out = v
return true
}
// CopyBytes copies len(out) bytes into out and advances over them. It reports
// whether the copy operation was successful
func (s *String) CopyBytes(out []byte) bool {
n := len(out)
v := s.read(n)
if v == nil {
return false
}
return copy(out, v) == n
}
// Empty reports whether the string does not contain any bytes.
func (s String) Empty() bool {
return len(s) == 0
}
+124
View File
@@ -0,0 +1,124 @@
// Copyright 2010 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.
// Package ripemd160 implements the RIPEMD-160 hash algorithm.
//
// Deprecated: RIPEMD-160 is a legacy hash and should not be used for new
// applications. Also, this package does not and will not provide an optimized
// implementation. Instead, use a modern hash like SHA-256 (from crypto/sha256).
package ripemd160 // import "golang.org/x/crypto/ripemd160"
// RIPEMD-160 is designed by Hans Dobbertin, Antoon Bosselaers, and Bart
// Preneel with specifications available at:
// http://homes.esat.kuleuven.be/~cosicart/pdf/AB-9601/AB-9601.pdf.
import (
"crypto"
"hash"
)
func init() {
crypto.RegisterHash(crypto.RIPEMD160, New)
}
// The size of the checksum in bytes.
const Size = 20
// The block size of the hash algorithm in bytes.
const BlockSize = 64
const (
_s0 = 0x67452301
_s1 = 0xefcdab89
_s2 = 0x98badcfe
_s3 = 0x10325476
_s4 = 0xc3d2e1f0
)
// digest represents the partial evaluation of a checksum.
type digest struct {
s [5]uint32 // running context
x [BlockSize]byte // temporary buffer
nx int // index into x
tc uint64 // total count of bytes processed
}
func (d *digest) Reset() {
d.s[0], d.s[1], d.s[2], d.s[3], d.s[4] = _s0, _s1, _s2, _s3, _s4
d.nx = 0
d.tc = 0
}
// New returns a new hash.Hash computing the checksum.
func New() hash.Hash {
result := new(digest)
result.Reset()
return result
}
func (d *digest) Size() int { return Size }
func (d *digest) BlockSize() int { return BlockSize }
func (d *digest) Write(p []byte) (nn int, err error) {
nn = len(p)
d.tc += uint64(nn)
if d.nx > 0 {
n := len(p)
if n > BlockSize-d.nx {
n = BlockSize - d.nx
}
for i := 0; i < n; i++ {
d.x[d.nx+i] = p[i]
}
d.nx += n
if d.nx == BlockSize {
_Block(d, d.x[0:])
d.nx = 0
}
p = p[n:]
}
n := _Block(d, p)
p = p[n:]
if len(p) > 0 {
d.nx = copy(d.x[:], p)
}
return
}
func (d0 *digest) Sum(in []byte) []byte {
// Make a copy of d0 so that caller can keep writing and summing.
d := *d0
// Padding. Add a 1 bit and 0 bits until 56 bytes mod 64.
tc := d.tc
var tmp [64]byte
tmp[0] = 0x80
if tc%64 < 56 {
d.Write(tmp[0 : 56-tc%64])
} else {
d.Write(tmp[0 : 64+56-tc%64])
}
// Length in bits.
tc <<= 3
for i := uint(0); i < 8; i++ {
tmp[i] = byte(tc >> (8 * i))
}
d.Write(tmp[0:8])
if d.nx != 0 {
panic("d.nx != 0")
}
var digest [Size]byte
for i, s := range d.s {
digest[i*4] = byte(s)
digest[i*4+1] = byte(s >> 8)
digest[i*4+2] = byte(s >> 16)
digest[i*4+3] = byte(s >> 24)
}
return append(in, digest[:]...)
}
+165
View File
@@ -0,0 +1,165 @@
// Copyright 2010 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.
// RIPEMD-160 block step.
// In its own file so that a faster assembly or C version
// can be substituted easily.
package ripemd160
import (
"math/bits"
)
// work buffer indices and roll amounts for one line
var _n = [80]uint{
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15,
7, 4, 13, 1, 10, 6, 15, 3, 12, 0, 9, 5, 2, 14, 11, 8,
3, 10, 14, 4, 9, 15, 8, 1, 2, 7, 0, 6, 13, 11, 5, 12,
1, 9, 11, 10, 0, 8, 12, 4, 13, 3, 7, 15, 14, 5, 6, 2,
4, 0, 5, 9, 7, 12, 2, 10, 14, 1, 3, 8, 11, 6, 15, 13,
}
var _r = [80]uint{
11, 14, 15, 12, 5, 8, 7, 9, 11, 13, 14, 15, 6, 7, 9, 8,
7, 6, 8, 13, 11, 9, 7, 15, 7, 12, 15, 9, 11, 7, 13, 12,
11, 13, 6, 7, 14, 9, 13, 15, 14, 8, 13, 6, 5, 12, 7, 5,
11, 12, 14, 15, 14, 15, 9, 8, 9, 14, 5, 6, 8, 6, 5, 12,
9, 15, 5, 11, 6, 8, 13, 12, 5, 12, 13, 14, 11, 8, 5, 6,
}
// same for the other parallel one
var n_ = [80]uint{
5, 14, 7, 0, 9, 2, 11, 4, 13, 6, 15, 8, 1, 10, 3, 12,
6, 11, 3, 7, 0, 13, 5, 10, 14, 15, 8, 12, 4, 9, 1, 2,
15, 5, 1, 3, 7, 14, 6, 9, 11, 8, 12, 2, 10, 0, 4, 13,
8, 6, 4, 1, 3, 11, 15, 0, 5, 12, 2, 13, 9, 7, 10, 14,
12, 15, 10, 4, 1, 5, 8, 7, 6, 2, 13, 14, 0, 3, 9, 11,
}
var r_ = [80]uint{
8, 9, 9, 11, 13, 15, 15, 5, 7, 7, 8, 11, 14, 14, 12, 6,
9, 13, 15, 7, 12, 8, 9, 11, 7, 7, 12, 7, 6, 15, 13, 11,
9, 7, 15, 11, 8, 6, 6, 14, 12, 13, 5, 14, 13, 13, 7, 5,
15, 5, 8, 11, 14, 14, 6, 14, 6, 9, 12, 9, 12, 5, 15, 8,
8, 5, 12, 9, 12, 5, 14, 6, 8, 13, 6, 5, 15, 13, 11, 11,
}
func _Block(md *digest, p []byte) int {
n := 0
var x [16]uint32
var alpha, beta uint32
for len(p) >= BlockSize {
a, b, c, d, e := md.s[0], md.s[1], md.s[2], md.s[3], md.s[4]
aa, bb, cc, dd, ee := a, b, c, d, e
j := 0
for i := 0; i < 16; i++ {
x[i] = uint32(p[j]) | uint32(p[j+1])<<8 | uint32(p[j+2])<<16 | uint32(p[j+3])<<24
j += 4
}
// round 1
i := 0
for i < 16 {
alpha = a + (b ^ c ^ d) + x[_n[i]]
s := int(_r[i])
alpha = bits.RotateLeft32(alpha, s) + e
beta = bits.RotateLeft32(c, 10)
a, b, c, d, e = e, alpha, b, beta, d
// parallel line
alpha = aa + (bb ^ (cc | ^dd)) + x[n_[i]] + 0x50a28be6
s = int(r_[i])
alpha = bits.RotateLeft32(alpha, s) + ee
beta = bits.RotateLeft32(cc, 10)
aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
i++
}
// round 2
for i < 32 {
alpha = a + (b&c | ^b&d) + x[_n[i]] + 0x5a827999
s := int(_r[i])
alpha = bits.RotateLeft32(alpha, s) + e
beta = bits.RotateLeft32(c, 10)
a, b, c, d, e = e, alpha, b, beta, d
// parallel line
alpha = aa + (bb&dd | cc&^dd) + x[n_[i]] + 0x5c4dd124
s = int(r_[i])
alpha = bits.RotateLeft32(alpha, s) + ee
beta = bits.RotateLeft32(cc, 10)
aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
i++
}
// round 3
for i < 48 {
alpha = a + (b | ^c ^ d) + x[_n[i]] + 0x6ed9eba1
s := int(_r[i])
alpha = bits.RotateLeft32(alpha, s) + e
beta = bits.RotateLeft32(c, 10)
a, b, c, d, e = e, alpha, b, beta, d
// parallel line
alpha = aa + (bb | ^cc ^ dd) + x[n_[i]] + 0x6d703ef3
s = int(r_[i])
alpha = bits.RotateLeft32(alpha, s) + ee
beta = bits.RotateLeft32(cc, 10)
aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
i++
}
// round 4
for i < 64 {
alpha = a + (b&d | c&^d) + x[_n[i]] + 0x8f1bbcdc
s := int(_r[i])
alpha = bits.RotateLeft32(alpha, s) + e
beta = bits.RotateLeft32(c, 10)
a, b, c, d, e = e, alpha, b, beta, d
// parallel line
alpha = aa + (bb&cc | ^bb&dd) + x[n_[i]] + 0x7a6d76e9
s = int(r_[i])
alpha = bits.RotateLeft32(alpha, s) + ee
beta = bits.RotateLeft32(cc, 10)
aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
i++
}
// round 5
for i < 80 {
alpha = a + (b ^ (c | ^d)) + x[_n[i]] + 0xa953fd4e
s := int(_r[i])
alpha = bits.RotateLeft32(alpha, s) + e
beta = bits.RotateLeft32(c, 10)
a, b, c, d, e = e, alpha, b, beta, d
// parallel line
alpha = aa + (bb ^ cc ^ dd) + x[n_[i]]
s = int(r_[i])
alpha = bits.RotateLeft32(alpha, s) + ee
beta = bits.RotateLeft32(cc, 10)
aa, bb, cc, dd, ee = ee, alpha, bb, beta, dd
i++
}
// combine results
dd += c + md.s[1]
md.s[1] = md.s[2] + d + ee
md.s[2] = md.s[3] + e + aa
md.s[3] = md.s[4] + a + bb
md.s[4] = md.s[0] + b + cc
md.s[0] = dd
p = p[BlockSize:]
n += BlockSize
}
return n
}
+62
View File
@@ -0,0 +1,62 @@
// Copyright 2014 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.
// Package sha3 implements the SHA-3 fixed-output-length hash functions and
// the SHAKE variable-output-length hash functions defined by FIPS-202.
//
// Both types of hash function use the "sponge" construction and the Keccak
// permutation. For a detailed specification see http://keccak.noekeon.org/
//
// # Guidance
//
// If you aren't sure what function you need, use SHAKE256 with at least 64
// bytes of output. The SHAKE instances are faster than the SHA3 instances;
// the latter have to allocate memory to conform to the hash.Hash interface.
//
// If you need a secret-key MAC (message authentication code), prepend the
// secret key to the input, hash with SHAKE256 and read at least 32 bytes of
// output.
//
// # Security strengths
//
// The SHA3-x (x equals 224, 256, 384, or 512) functions have a security
// strength against preimage attacks of x bits. Since they only produce "x"
// bits of output, their collision-resistance is only "x/2" bits.
//
// The SHAKE-256 and -128 functions have a generic security strength of 256 and
// 128 bits against all attacks, provided that at least 2x bits of their output
// is used. Requesting more than 64 or 32 bytes of output, respectively, does
// not increase the collision-resistance of the SHAKE functions.
//
// # The sponge construction
//
// A sponge builds a pseudo-random function from a public pseudo-random
// permutation, by applying the permutation to a state of "rate + capacity"
// bytes, but hiding "capacity" of the bytes.
//
// A sponge starts out with a zero state. To hash an input using a sponge, up
// to "rate" bytes of the input are XORed into the sponge's state. The sponge
// is then "full" and the permutation is applied to "empty" it. This process is
// repeated until all the input has been "absorbed". The input is then padded.
// The digest is "squeezed" from the sponge in the same way, except that output
// is copied out instead of input being XORed in.
//
// A sponge is parameterized by its generic security strength, which is equal
// to half its capacity; capacity + rate is equal to the permutation's width.
// Since the KeccakF-1600 permutation is 1600 bits (200 bytes) wide, this means
// that the security strength of a sponge instance is equal to (1600 - bitrate) / 2.
//
// # Recommendations
//
// The SHAKE functions are recommended for most new uses. They can produce
// output of arbitrary length. SHAKE256, with an output length of at least
// 64 bytes, provides 256-bit security against all attacks. The Keccak team
// recommends it for most applications upgrading from SHA2-512. (NIST chose a
// much stronger, but much slower, sponge instance for SHA3-512.)
//
// The SHA-3 functions are "drop-in" replacements for the SHA-2 functions.
// They produce output of the same length, with the same security strengths
// against all attacks. This means, in particular, that SHA3-256 only has
// 128-bit collision resistance, because its output length is 32 bytes.
package sha3 // import "golang.org/x/crypto/sha3"
+97
View File
@@ -0,0 +1,97 @@
// Copyright 2014 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.
package sha3
// This file provides functions for creating instances of the SHA-3
// and SHAKE hash functions, as well as utility functions for hashing
// bytes.
import (
"hash"
)
// New224 creates a new SHA3-224 hash.
// Its generic security strength is 224 bits against preimage attacks,
// and 112 bits against collision attacks.
func New224() hash.Hash {
if h := new224Asm(); h != nil {
return h
}
return &state{rate: 144, outputLen: 28, dsbyte: 0x06}
}
// New256 creates a new SHA3-256 hash.
// Its generic security strength is 256 bits against preimage attacks,
// and 128 bits against collision attacks.
func New256() hash.Hash {
if h := new256Asm(); h != nil {
return h
}
return &state{rate: 136, outputLen: 32, dsbyte: 0x06}
}
// New384 creates a new SHA3-384 hash.
// Its generic security strength is 384 bits against preimage attacks,
// and 192 bits against collision attacks.
func New384() hash.Hash {
if h := new384Asm(); h != nil {
return h
}
return &state{rate: 104, outputLen: 48, dsbyte: 0x06}
}
// New512 creates a new SHA3-512 hash.
// Its generic security strength is 512 bits against preimage attacks,
// and 256 bits against collision attacks.
func New512() hash.Hash {
if h := new512Asm(); h != nil {
return h
}
return &state{rate: 72, outputLen: 64, dsbyte: 0x06}
}
// NewLegacyKeccak256 creates a new Keccak-256 hash.
//
// Only use this function if you require compatibility with an existing cryptosystem
// that uses non-standard padding. All other users should use New256 instead.
func NewLegacyKeccak256() hash.Hash { return &state{rate: 136, outputLen: 32, dsbyte: 0x01} }
// NewLegacyKeccak512 creates a new Keccak-512 hash.
//
// Only use this function if you require compatibility with an existing cryptosystem
// that uses non-standard padding. All other users should use New512 instead.
func NewLegacyKeccak512() hash.Hash { return &state{rate: 72, outputLen: 64, dsbyte: 0x01} }
// Sum224 returns the SHA3-224 digest of the data.
func Sum224(data []byte) (digest [28]byte) {
h := New224()
h.Write(data)
h.Sum(digest[:0])
return
}
// Sum256 returns the SHA3-256 digest of the data.
func Sum256(data []byte) (digest [32]byte) {
h := New256()
h.Write(data)
h.Sum(digest[:0])
return
}
// Sum384 returns the SHA3-384 digest of the data.
func Sum384(data []byte) (digest [48]byte) {
h := New384()
h.Write(data)
h.Sum(digest[:0])
return
}
// Sum512 returns the SHA3-512 digest of the data.
func Sum512(data []byte) (digest [64]byte) {
h := New512()
h.Write(data)
h.Sum(digest[:0])
return
}
+28
View File
@@ -0,0 +1,28 @@
// Copyright 2017 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.
//go:build !gc || purego || !s390x
// +build !gc purego !s390x
package sha3
import (
"hash"
)
// new224Asm returns an assembly implementation of SHA3-224 if available,
// otherwise it returns nil.
func new224Asm() hash.Hash { return nil }
// new256Asm returns an assembly implementation of SHA3-256 if available,
// otherwise it returns nil.
func new256Asm() hash.Hash { return nil }
// new384Asm returns an assembly implementation of SHA3-384 if available,
// otherwise it returns nil.
func new384Asm() hash.Hash { return nil }
// new512Asm returns an assembly implementation of SHA3-512 if available,
// otherwise it returns nil.
func new512Asm() hash.Hash { return nil }
+415
View File
@@ -0,0 +1,415 @@
// Copyright 2014 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.
//go:build !amd64 || purego || !gc
// +build !amd64 purego !gc
package sha3
import "math/bits"
// rc stores the round constants for use in the ι step.
var rc = [24]uint64{
0x0000000000000001,
0x0000000000008082,
0x800000000000808A,
0x8000000080008000,
0x000000000000808B,
0x0000000080000001,
0x8000000080008081,
0x8000000000008009,
0x000000000000008A,
0x0000000000000088,
0x0000000080008009,
0x000000008000000A,
0x000000008000808B,
0x800000000000008B,
0x8000000000008089,
0x8000000000008003,
0x8000000000008002,
0x8000000000000080,
0x000000000000800A,
0x800000008000000A,
0x8000000080008081,
0x8000000000008080,
0x0000000080000001,
0x8000000080008008,
}
// keccakF1600 applies the Keccak permutation to a 1600b-wide
// state represented as a slice of 25 uint64s.
func keccakF1600(a *[25]uint64) {
// Implementation translated from Keccak-inplace.c
// in the keccak reference code.
var t, bc0, bc1, bc2, bc3, bc4, d0, d1, d2, d3, d4 uint64
for i := 0; i < 24; i += 4 {
// Combines the 5 steps in each round into 2 steps.
// Unrolls 4 rounds per loop and spreads some steps across rounds.
// Round 1
bc0 = a[0] ^ a[5] ^ a[10] ^ a[15] ^ a[20]
bc1 = a[1] ^ a[6] ^ a[11] ^ a[16] ^ a[21]
bc2 = a[2] ^ a[7] ^ a[12] ^ a[17] ^ a[22]
bc3 = a[3] ^ a[8] ^ a[13] ^ a[18] ^ a[23]
bc4 = a[4] ^ a[9] ^ a[14] ^ a[19] ^ a[24]
d0 = bc4 ^ (bc1<<1 | bc1>>63)
d1 = bc0 ^ (bc2<<1 | bc2>>63)
d2 = bc1 ^ (bc3<<1 | bc3>>63)
d3 = bc2 ^ (bc4<<1 | bc4>>63)
d4 = bc3 ^ (bc0<<1 | bc0>>63)
bc0 = a[0] ^ d0
t = a[6] ^ d1
bc1 = bits.RotateLeft64(t, 44)
t = a[12] ^ d2
bc2 = bits.RotateLeft64(t, 43)
t = a[18] ^ d3
bc3 = bits.RotateLeft64(t, 21)
t = a[24] ^ d4
bc4 = bits.RotateLeft64(t, 14)
a[0] = bc0 ^ (bc2 &^ bc1) ^ rc[i]
a[6] = bc1 ^ (bc3 &^ bc2)
a[12] = bc2 ^ (bc4 &^ bc3)
a[18] = bc3 ^ (bc0 &^ bc4)
a[24] = bc4 ^ (bc1 &^ bc0)
t = a[10] ^ d0
bc2 = bits.RotateLeft64(t, 3)
t = a[16] ^ d1
bc3 = bits.RotateLeft64(t, 45)
t = a[22] ^ d2
bc4 = bits.RotateLeft64(t, 61)
t = a[3] ^ d3
bc0 = bits.RotateLeft64(t, 28)
t = a[9] ^ d4
bc1 = bits.RotateLeft64(t, 20)
a[10] = bc0 ^ (bc2 &^ bc1)
a[16] = bc1 ^ (bc3 &^ bc2)
a[22] = bc2 ^ (bc4 &^ bc3)
a[3] = bc3 ^ (bc0 &^ bc4)
a[9] = bc4 ^ (bc1 &^ bc0)
t = a[20] ^ d0
bc4 = bits.RotateLeft64(t, 18)
t = a[1] ^ d1
bc0 = bits.RotateLeft64(t, 1)
t = a[7] ^ d2
bc1 = bits.RotateLeft64(t, 6)
t = a[13] ^ d3
bc2 = bits.RotateLeft64(t, 25)
t = a[19] ^ d4
bc3 = bits.RotateLeft64(t, 8)
a[20] = bc0 ^ (bc2 &^ bc1)
a[1] = bc1 ^ (bc3 &^ bc2)
a[7] = bc2 ^ (bc4 &^ bc3)
a[13] = bc3 ^ (bc0 &^ bc4)
a[19] = bc4 ^ (bc1 &^ bc0)
t = a[5] ^ d0
bc1 = bits.RotateLeft64(t, 36)
t = a[11] ^ d1
bc2 = bits.RotateLeft64(t, 10)
t = a[17] ^ d2
bc3 = bits.RotateLeft64(t, 15)
t = a[23] ^ d3
bc4 = bits.RotateLeft64(t, 56)
t = a[4] ^ d4
bc0 = bits.RotateLeft64(t, 27)
a[5] = bc0 ^ (bc2 &^ bc1)
a[11] = bc1 ^ (bc3 &^ bc2)
a[17] = bc2 ^ (bc4 &^ bc3)
a[23] = bc3 ^ (bc0 &^ bc4)
a[4] = bc4 ^ (bc1 &^ bc0)
t = a[15] ^ d0
bc3 = bits.RotateLeft64(t, 41)
t = a[21] ^ d1
bc4 = bits.RotateLeft64(t, 2)
t = a[2] ^ d2
bc0 = bits.RotateLeft64(t, 62)
t = a[8] ^ d3
bc1 = bits.RotateLeft64(t, 55)
t = a[14] ^ d4
bc2 = bits.RotateLeft64(t, 39)
a[15] = bc0 ^ (bc2 &^ bc1)
a[21] = bc1 ^ (bc3 &^ bc2)
a[2] = bc2 ^ (bc4 &^ bc3)
a[8] = bc3 ^ (bc0 &^ bc4)
a[14] = bc4 ^ (bc1 &^ bc0)
// Round 2
bc0 = a[0] ^ a[5] ^ a[10] ^ a[15] ^ a[20]
bc1 = a[1] ^ a[6] ^ a[11] ^ a[16] ^ a[21]
bc2 = a[2] ^ a[7] ^ a[12] ^ a[17] ^ a[22]
bc3 = a[3] ^ a[8] ^ a[13] ^ a[18] ^ a[23]
bc4 = a[4] ^ a[9] ^ a[14] ^ a[19] ^ a[24]
d0 = bc4 ^ (bc1<<1 | bc1>>63)
d1 = bc0 ^ (bc2<<1 | bc2>>63)
d2 = bc1 ^ (bc3<<1 | bc3>>63)
d3 = bc2 ^ (bc4<<1 | bc4>>63)
d4 = bc3 ^ (bc0<<1 | bc0>>63)
bc0 = a[0] ^ d0
t = a[16] ^ d1
bc1 = bits.RotateLeft64(t, 44)
t = a[7] ^ d2
bc2 = bits.RotateLeft64(t, 43)
t = a[23] ^ d3
bc3 = bits.RotateLeft64(t, 21)
t = a[14] ^ d4
bc4 = bits.RotateLeft64(t, 14)
a[0] = bc0 ^ (bc2 &^ bc1) ^ rc[i+1]
a[16] = bc1 ^ (bc3 &^ bc2)
a[7] = bc2 ^ (bc4 &^ bc3)
a[23] = bc3 ^ (bc0 &^ bc4)
a[14] = bc4 ^ (bc1 &^ bc0)
t = a[20] ^ d0
bc2 = bits.RotateLeft64(t, 3)
t = a[11] ^ d1
bc3 = bits.RotateLeft64(t, 45)
t = a[2] ^ d2
bc4 = bits.RotateLeft64(t, 61)
t = a[18] ^ d3
bc0 = bits.RotateLeft64(t, 28)
t = a[9] ^ d4
bc1 = bits.RotateLeft64(t, 20)
a[20] = bc0 ^ (bc2 &^ bc1)
a[11] = bc1 ^ (bc3 &^ bc2)
a[2] = bc2 ^ (bc4 &^ bc3)
a[18] = bc3 ^ (bc0 &^ bc4)
a[9] = bc4 ^ (bc1 &^ bc0)
t = a[15] ^ d0
bc4 = bits.RotateLeft64(t, 18)
t = a[6] ^ d1
bc0 = bits.RotateLeft64(t, 1)
t = a[22] ^ d2
bc1 = bits.RotateLeft64(t, 6)
t = a[13] ^ d3
bc2 = bits.RotateLeft64(t, 25)
t = a[4] ^ d4
bc3 = bits.RotateLeft64(t, 8)
a[15] = bc0 ^ (bc2 &^ bc1)
a[6] = bc1 ^ (bc3 &^ bc2)
a[22] = bc2 ^ (bc4 &^ bc3)
a[13] = bc3 ^ (bc0 &^ bc4)
a[4] = bc4 ^ (bc1 &^ bc0)
t = a[10] ^ d0
bc1 = bits.RotateLeft64(t, 36)
t = a[1] ^ d1
bc2 = bits.RotateLeft64(t, 10)
t = a[17] ^ d2
bc3 = bits.RotateLeft64(t, 15)
t = a[8] ^ d3
bc4 = bits.RotateLeft64(t, 56)
t = a[24] ^ d4
bc0 = bits.RotateLeft64(t, 27)
a[10] = bc0 ^ (bc2 &^ bc1)
a[1] = bc1 ^ (bc3 &^ bc2)
a[17] = bc2 ^ (bc4 &^ bc3)
a[8] = bc3 ^ (bc0 &^ bc4)
a[24] = bc4 ^ (bc1 &^ bc0)
t = a[5] ^ d0
bc3 = bits.RotateLeft64(t, 41)
t = a[21] ^ d1
bc4 = bits.RotateLeft64(t, 2)
t = a[12] ^ d2
bc0 = bits.RotateLeft64(t, 62)
t = a[3] ^ d3
bc1 = bits.RotateLeft64(t, 55)
t = a[19] ^ d4
bc2 = bits.RotateLeft64(t, 39)
a[5] = bc0 ^ (bc2 &^ bc1)
a[21] = bc1 ^ (bc3 &^ bc2)
a[12] = bc2 ^ (bc4 &^ bc3)
a[3] = bc3 ^ (bc0 &^ bc4)
a[19] = bc4 ^ (bc1 &^ bc0)
// Round 3
bc0 = a[0] ^ a[5] ^ a[10] ^ a[15] ^ a[20]
bc1 = a[1] ^ a[6] ^ a[11] ^ a[16] ^ a[21]
bc2 = a[2] ^ a[7] ^ a[12] ^ a[17] ^ a[22]
bc3 = a[3] ^ a[8] ^ a[13] ^ a[18] ^ a[23]
bc4 = a[4] ^ a[9] ^ a[14] ^ a[19] ^ a[24]
d0 = bc4 ^ (bc1<<1 | bc1>>63)
d1 = bc0 ^ (bc2<<1 | bc2>>63)
d2 = bc1 ^ (bc3<<1 | bc3>>63)
d3 = bc2 ^ (bc4<<1 | bc4>>63)
d4 = bc3 ^ (bc0<<1 | bc0>>63)
bc0 = a[0] ^ d0
t = a[11] ^ d1
bc1 = bits.RotateLeft64(t, 44)
t = a[22] ^ d2
bc2 = bits.RotateLeft64(t, 43)
t = a[8] ^ d3
bc3 = bits.RotateLeft64(t, 21)
t = a[19] ^ d4
bc4 = bits.RotateLeft64(t, 14)
a[0] = bc0 ^ (bc2 &^ bc1) ^ rc[i+2]
a[11] = bc1 ^ (bc3 &^ bc2)
a[22] = bc2 ^ (bc4 &^ bc3)
a[8] = bc3 ^ (bc0 &^ bc4)
a[19] = bc4 ^ (bc1 &^ bc0)
t = a[15] ^ d0
bc2 = bits.RotateLeft64(t, 3)
t = a[1] ^ d1
bc3 = bits.RotateLeft64(t, 45)
t = a[12] ^ d2
bc4 = bits.RotateLeft64(t, 61)
t = a[23] ^ d3
bc0 = bits.RotateLeft64(t, 28)
t = a[9] ^ d4
bc1 = bits.RotateLeft64(t, 20)
a[15] = bc0 ^ (bc2 &^ bc1)
a[1] = bc1 ^ (bc3 &^ bc2)
a[12] = bc2 ^ (bc4 &^ bc3)
a[23] = bc3 ^ (bc0 &^ bc4)
a[9] = bc4 ^ (bc1 &^ bc0)
t = a[5] ^ d0
bc4 = bits.RotateLeft64(t, 18)
t = a[16] ^ d1
bc0 = bits.RotateLeft64(t, 1)
t = a[2] ^ d2
bc1 = bits.RotateLeft64(t, 6)
t = a[13] ^ d3
bc2 = bits.RotateLeft64(t, 25)
t = a[24] ^ d4
bc3 = bits.RotateLeft64(t, 8)
a[5] = bc0 ^ (bc2 &^ bc1)
a[16] = bc1 ^ (bc3 &^ bc2)
a[2] = bc2 ^ (bc4 &^ bc3)
a[13] = bc3 ^ (bc0 &^ bc4)
a[24] = bc4 ^ (bc1 &^ bc0)
t = a[20] ^ d0
bc1 = bits.RotateLeft64(t, 36)
t = a[6] ^ d1
bc2 = bits.RotateLeft64(t, 10)
t = a[17] ^ d2
bc3 = bits.RotateLeft64(t, 15)
t = a[3] ^ d3
bc4 = bits.RotateLeft64(t, 56)
t = a[14] ^ d4
bc0 = bits.RotateLeft64(t, 27)
a[20] = bc0 ^ (bc2 &^ bc1)
a[6] = bc1 ^ (bc3 &^ bc2)
a[17] = bc2 ^ (bc4 &^ bc3)
a[3] = bc3 ^ (bc0 &^ bc4)
a[14] = bc4 ^ (bc1 &^ bc0)
t = a[10] ^ d0
bc3 = bits.RotateLeft64(t, 41)
t = a[21] ^ d1
bc4 = bits.RotateLeft64(t, 2)
t = a[7] ^ d2
bc0 = bits.RotateLeft64(t, 62)
t = a[18] ^ d3
bc1 = bits.RotateLeft64(t, 55)
t = a[4] ^ d4
bc2 = bits.RotateLeft64(t, 39)
a[10] = bc0 ^ (bc2 &^ bc1)
a[21] = bc1 ^ (bc3 &^ bc2)
a[7] = bc2 ^ (bc4 &^ bc3)
a[18] = bc3 ^ (bc0 &^ bc4)
a[4] = bc4 ^ (bc1 &^ bc0)
// Round 4
bc0 = a[0] ^ a[5] ^ a[10] ^ a[15] ^ a[20]
bc1 = a[1] ^ a[6] ^ a[11] ^ a[16] ^ a[21]
bc2 = a[2] ^ a[7] ^ a[12] ^ a[17] ^ a[22]
bc3 = a[3] ^ a[8] ^ a[13] ^ a[18] ^ a[23]
bc4 = a[4] ^ a[9] ^ a[14] ^ a[19] ^ a[24]
d0 = bc4 ^ (bc1<<1 | bc1>>63)
d1 = bc0 ^ (bc2<<1 | bc2>>63)
d2 = bc1 ^ (bc3<<1 | bc3>>63)
d3 = bc2 ^ (bc4<<1 | bc4>>63)
d4 = bc3 ^ (bc0<<1 | bc0>>63)
bc0 = a[0] ^ d0
t = a[1] ^ d1
bc1 = bits.RotateLeft64(t, 44)
t = a[2] ^ d2
bc2 = bits.RotateLeft64(t, 43)
t = a[3] ^ d3
bc3 = bits.RotateLeft64(t, 21)
t = a[4] ^ d4
bc4 = bits.RotateLeft64(t, 14)
a[0] = bc0 ^ (bc2 &^ bc1) ^ rc[i+3]
a[1] = bc1 ^ (bc3 &^ bc2)
a[2] = bc2 ^ (bc4 &^ bc3)
a[3] = bc3 ^ (bc0 &^ bc4)
a[4] = bc4 ^ (bc1 &^ bc0)
t = a[5] ^ d0
bc2 = bits.RotateLeft64(t, 3)
t = a[6] ^ d1
bc3 = bits.RotateLeft64(t, 45)
t = a[7] ^ d2
bc4 = bits.RotateLeft64(t, 61)
t = a[8] ^ d3
bc0 = bits.RotateLeft64(t, 28)
t = a[9] ^ d4
bc1 = bits.RotateLeft64(t, 20)
a[5] = bc0 ^ (bc2 &^ bc1)
a[6] = bc1 ^ (bc3 &^ bc2)
a[7] = bc2 ^ (bc4 &^ bc3)
a[8] = bc3 ^ (bc0 &^ bc4)
a[9] = bc4 ^ (bc1 &^ bc0)
t = a[10] ^ d0
bc4 = bits.RotateLeft64(t, 18)
t = a[11] ^ d1
bc0 = bits.RotateLeft64(t, 1)
t = a[12] ^ d2
bc1 = bits.RotateLeft64(t, 6)
t = a[13] ^ d3
bc2 = bits.RotateLeft64(t, 25)
t = a[14] ^ d4
bc3 = bits.RotateLeft64(t, 8)
a[10] = bc0 ^ (bc2 &^ bc1)
a[11] = bc1 ^ (bc3 &^ bc2)
a[12] = bc2 ^ (bc4 &^ bc3)
a[13] = bc3 ^ (bc0 &^ bc4)
a[14] = bc4 ^ (bc1 &^ bc0)
t = a[15] ^ d0
bc1 = bits.RotateLeft64(t, 36)
t = a[16] ^ d1
bc2 = bits.RotateLeft64(t, 10)
t = a[17] ^ d2
bc3 = bits.RotateLeft64(t, 15)
t = a[18] ^ d3
bc4 = bits.RotateLeft64(t, 56)
t = a[19] ^ d4
bc0 = bits.RotateLeft64(t, 27)
a[15] = bc0 ^ (bc2 &^ bc1)
a[16] = bc1 ^ (bc3 &^ bc2)
a[17] = bc2 ^ (bc4 &^ bc3)
a[18] = bc3 ^ (bc0 &^ bc4)
a[19] = bc4 ^ (bc1 &^ bc0)
t = a[20] ^ d0
bc3 = bits.RotateLeft64(t, 41)
t = a[21] ^ d1
bc4 = bits.RotateLeft64(t, 2)
t = a[22] ^ d2
bc0 = bits.RotateLeft64(t, 62)
t = a[23] ^ d3
bc1 = bits.RotateLeft64(t, 55)
t = a[24] ^ d4
bc2 = bits.RotateLeft64(t, 39)
a[20] = bc0 ^ (bc2 &^ bc1)
a[21] = bc1 ^ (bc3 &^ bc2)
a[22] = bc2 ^ (bc4 &^ bc3)
a[23] = bc3 ^ (bc0 &^ bc4)
a[24] = bc4 ^ (bc1 &^ bc0)
}
}
+14
View File
@@ -0,0 +1,14 @@
// Copyright 2015 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.
//go:build amd64 && !purego && gc
// +build amd64,!purego,gc
package sha3
// This function is implemented in keccakf_amd64.s.
//go:noescape
func keccakF1600(a *[25]uint64)
+391
View File
@@ -0,0 +1,391 @@
// Copyright 2015 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.
//go:build amd64 && !purego && gc
// +build amd64,!purego,gc
// This code was translated into a form compatible with 6a from the public
// domain sources at https://github.com/gvanas/KeccakCodePackage
// Offsets in state
#define _ba (0*8)
#define _be (1*8)
#define _bi (2*8)
#define _bo (3*8)
#define _bu (4*8)
#define _ga (5*8)
#define _ge (6*8)
#define _gi (7*8)
#define _go (8*8)
#define _gu (9*8)
#define _ka (10*8)
#define _ke (11*8)
#define _ki (12*8)
#define _ko (13*8)
#define _ku (14*8)
#define _ma (15*8)
#define _me (16*8)
#define _mi (17*8)
#define _mo (18*8)
#define _mu (19*8)
#define _sa (20*8)
#define _se (21*8)
#define _si (22*8)
#define _so (23*8)
#define _su (24*8)
// Temporary registers
#define rT1 AX
// Round vars
#define rpState DI
#define rpStack SP
#define rDa BX
#define rDe CX
#define rDi DX
#define rDo R8
#define rDu R9
#define rBa R10
#define rBe R11
#define rBi R12
#define rBo R13
#define rBu R14
#define rCa SI
#define rCe BP
#define rCi rBi
#define rCo rBo
#define rCu R15
#define MOVQ_RBI_RCE MOVQ rBi, rCe
#define XORQ_RT1_RCA XORQ rT1, rCa
#define XORQ_RT1_RCE XORQ rT1, rCe
#define XORQ_RBA_RCU XORQ rBa, rCu
#define XORQ_RBE_RCU XORQ rBe, rCu
#define XORQ_RDU_RCU XORQ rDu, rCu
#define XORQ_RDA_RCA XORQ rDa, rCa
#define XORQ_RDE_RCE XORQ rDe, rCe
#define mKeccakRound(iState, oState, rc, B_RBI_RCE, G_RT1_RCA, G_RT1_RCE, G_RBA_RCU, K_RT1_RCA, K_RT1_RCE, K_RBA_RCU, M_RT1_RCA, M_RT1_RCE, M_RBE_RCU, S_RDU_RCU, S_RDA_RCA, S_RDE_RCE) \
/* Prepare round */ \
MOVQ rCe, rDa; \
ROLQ $1, rDa; \
\
MOVQ _bi(iState), rCi; \
XORQ _gi(iState), rDi; \
XORQ rCu, rDa; \
XORQ _ki(iState), rCi; \
XORQ _mi(iState), rDi; \
XORQ rDi, rCi; \
\
MOVQ rCi, rDe; \
ROLQ $1, rDe; \
\
MOVQ _bo(iState), rCo; \
XORQ _go(iState), rDo; \
XORQ rCa, rDe; \
XORQ _ko(iState), rCo; \
XORQ _mo(iState), rDo; \
XORQ rDo, rCo; \
\
MOVQ rCo, rDi; \
ROLQ $1, rDi; \
\
MOVQ rCu, rDo; \
XORQ rCe, rDi; \
ROLQ $1, rDo; \
\
MOVQ rCa, rDu; \
XORQ rCi, rDo; \
ROLQ $1, rDu; \
\
/* Result b */ \
MOVQ _ba(iState), rBa; \
MOVQ _ge(iState), rBe; \
XORQ rCo, rDu; \
MOVQ _ki(iState), rBi; \
MOVQ _mo(iState), rBo; \
MOVQ _su(iState), rBu; \
XORQ rDe, rBe; \
ROLQ $44, rBe; \
XORQ rDi, rBi; \
XORQ rDa, rBa; \
ROLQ $43, rBi; \
\
MOVQ rBe, rCa; \
MOVQ rc, rT1; \
ORQ rBi, rCa; \
XORQ rBa, rT1; \
XORQ rT1, rCa; \
MOVQ rCa, _ba(oState); \
\
XORQ rDu, rBu; \
ROLQ $14, rBu; \
MOVQ rBa, rCu; \
ANDQ rBe, rCu; \
XORQ rBu, rCu; \
MOVQ rCu, _bu(oState); \
\
XORQ rDo, rBo; \
ROLQ $21, rBo; \
MOVQ rBo, rT1; \
ANDQ rBu, rT1; \
XORQ rBi, rT1; \
MOVQ rT1, _bi(oState); \
\
NOTQ rBi; \
ORQ rBa, rBu; \
ORQ rBo, rBi; \
XORQ rBo, rBu; \
XORQ rBe, rBi; \
MOVQ rBu, _bo(oState); \
MOVQ rBi, _be(oState); \
B_RBI_RCE; \
\
/* Result g */ \
MOVQ _gu(iState), rBe; \
XORQ rDu, rBe; \
MOVQ _ka(iState), rBi; \
ROLQ $20, rBe; \
XORQ rDa, rBi; \
ROLQ $3, rBi; \
MOVQ _bo(iState), rBa; \
MOVQ rBe, rT1; \
ORQ rBi, rT1; \
XORQ rDo, rBa; \
MOVQ _me(iState), rBo; \
MOVQ _si(iState), rBu; \
ROLQ $28, rBa; \
XORQ rBa, rT1; \
MOVQ rT1, _ga(oState); \
G_RT1_RCA; \
\
XORQ rDe, rBo; \
ROLQ $45, rBo; \
MOVQ rBi, rT1; \
ANDQ rBo, rT1; \
XORQ rBe, rT1; \
MOVQ rT1, _ge(oState); \
G_RT1_RCE; \
\
XORQ rDi, rBu; \
ROLQ $61, rBu; \
MOVQ rBu, rT1; \
ORQ rBa, rT1; \
XORQ rBo, rT1; \
MOVQ rT1, _go(oState); \
\
ANDQ rBe, rBa; \
XORQ rBu, rBa; \
MOVQ rBa, _gu(oState); \
NOTQ rBu; \
G_RBA_RCU; \
\
ORQ rBu, rBo; \
XORQ rBi, rBo; \
MOVQ rBo, _gi(oState); \
\
/* Result k */ \
MOVQ _be(iState), rBa; \
MOVQ _gi(iState), rBe; \
MOVQ _ko(iState), rBi; \
MOVQ _mu(iState), rBo; \
MOVQ _sa(iState), rBu; \
XORQ rDi, rBe; \
ROLQ $6, rBe; \
XORQ rDo, rBi; \
ROLQ $25, rBi; \
MOVQ rBe, rT1; \
ORQ rBi, rT1; \
XORQ rDe, rBa; \
ROLQ $1, rBa; \
XORQ rBa, rT1; \
MOVQ rT1, _ka(oState); \
K_RT1_RCA; \
\
XORQ rDu, rBo; \
ROLQ $8, rBo; \
MOVQ rBi, rT1; \
ANDQ rBo, rT1; \
XORQ rBe, rT1; \
MOVQ rT1, _ke(oState); \
K_RT1_RCE; \
\
XORQ rDa, rBu; \
ROLQ $18, rBu; \
NOTQ rBo; \
MOVQ rBo, rT1; \
ANDQ rBu, rT1; \
XORQ rBi, rT1; \
MOVQ rT1, _ki(oState); \
\
MOVQ rBu, rT1; \
ORQ rBa, rT1; \
XORQ rBo, rT1; \
MOVQ rT1, _ko(oState); \
\
ANDQ rBe, rBa; \
XORQ rBu, rBa; \
MOVQ rBa, _ku(oState); \
K_RBA_RCU; \
\
/* Result m */ \
MOVQ _ga(iState), rBe; \
XORQ rDa, rBe; \
MOVQ _ke(iState), rBi; \
ROLQ $36, rBe; \
XORQ rDe, rBi; \
MOVQ _bu(iState), rBa; \
ROLQ $10, rBi; \
MOVQ rBe, rT1; \
MOVQ _mi(iState), rBo; \
ANDQ rBi, rT1; \
XORQ rDu, rBa; \
MOVQ _so(iState), rBu; \
ROLQ $27, rBa; \
XORQ rBa, rT1; \
MOVQ rT1, _ma(oState); \
M_RT1_RCA; \
\
XORQ rDi, rBo; \
ROLQ $15, rBo; \
MOVQ rBi, rT1; \
ORQ rBo, rT1; \
XORQ rBe, rT1; \
MOVQ rT1, _me(oState); \
M_RT1_RCE; \
\
XORQ rDo, rBu; \
ROLQ $56, rBu; \
NOTQ rBo; \
MOVQ rBo, rT1; \
ORQ rBu, rT1; \
XORQ rBi, rT1; \
MOVQ rT1, _mi(oState); \
\
ORQ rBa, rBe; \
XORQ rBu, rBe; \
MOVQ rBe, _mu(oState); \
\
ANDQ rBa, rBu; \
XORQ rBo, rBu; \
MOVQ rBu, _mo(oState); \
M_RBE_RCU; \
\
/* Result s */ \
MOVQ _bi(iState), rBa; \
MOVQ _go(iState), rBe; \
MOVQ _ku(iState), rBi; \
XORQ rDi, rBa; \
MOVQ _ma(iState), rBo; \
ROLQ $62, rBa; \
XORQ rDo, rBe; \
MOVQ _se(iState), rBu; \
ROLQ $55, rBe; \
\
XORQ rDu, rBi; \
MOVQ rBa, rDu; \
XORQ rDe, rBu; \
ROLQ $2, rBu; \
ANDQ rBe, rDu; \
XORQ rBu, rDu; \
MOVQ rDu, _su(oState); \
\
ROLQ $39, rBi; \
S_RDU_RCU; \
NOTQ rBe; \
XORQ rDa, rBo; \
MOVQ rBe, rDa; \
ANDQ rBi, rDa; \
XORQ rBa, rDa; \
MOVQ rDa, _sa(oState); \
S_RDA_RCA; \
\
ROLQ $41, rBo; \
MOVQ rBi, rDe; \
ORQ rBo, rDe; \
XORQ rBe, rDe; \
MOVQ rDe, _se(oState); \
S_RDE_RCE; \
\
MOVQ rBo, rDi; \
MOVQ rBu, rDo; \
ANDQ rBu, rDi; \
ORQ rBa, rDo; \
XORQ rBi, rDi; \
XORQ rBo, rDo; \
MOVQ rDi, _si(oState); \
MOVQ rDo, _so(oState) \
// func keccakF1600(state *[25]uint64)
TEXT ·keccakF1600(SB), 0, $200-8
MOVQ state+0(FP), rpState
// Convert the user state into an internal state
NOTQ _be(rpState)
NOTQ _bi(rpState)
NOTQ _go(rpState)
NOTQ _ki(rpState)
NOTQ _mi(rpState)
NOTQ _sa(rpState)
// Execute the KeccakF permutation
MOVQ _ba(rpState), rCa
MOVQ _be(rpState), rCe
MOVQ _bu(rpState), rCu
XORQ _ga(rpState), rCa
XORQ _ge(rpState), rCe
XORQ _gu(rpState), rCu
XORQ _ka(rpState), rCa
XORQ _ke(rpState), rCe
XORQ _ku(rpState), rCu
XORQ _ma(rpState), rCa
XORQ _me(rpState), rCe
XORQ _mu(rpState), rCu
XORQ _sa(rpState), rCa
XORQ _se(rpState), rCe
MOVQ _si(rpState), rDi
MOVQ _so(rpState), rDo
XORQ _su(rpState), rCu
mKeccakRound(rpState, rpStack, $0x0000000000000001, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpStack, rpState, $0x0000000000008082, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpState, rpStack, $0x800000000000808a, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpStack, rpState, $0x8000000080008000, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpState, rpStack, $0x000000000000808b, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpStack, rpState, $0x0000000080000001, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpState, rpStack, $0x8000000080008081, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpStack, rpState, $0x8000000000008009, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpState, rpStack, $0x000000000000008a, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpStack, rpState, $0x0000000000000088, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpState, rpStack, $0x0000000080008009, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpStack, rpState, $0x000000008000000a, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpState, rpStack, $0x000000008000808b, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpStack, rpState, $0x800000000000008b, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpState, rpStack, $0x8000000000008089, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpStack, rpState, $0x8000000000008003, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpState, rpStack, $0x8000000000008002, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpStack, rpState, $0x8000000000000080, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpState, rpStack, $0x000000000000800a, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpStack, rpState, $0x800000008000000a, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpState, rpStack, $0x8000000080008081, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpStack, rpState, $0x8000000000008080, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpState, rpStack, $0x0000000080000001, MOVQ_RBI_RCE, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBA_RCU, XORQ_RT1_RCA, XORQ_RT1_RCE, XORQ_RBE_RCU, XORQ_RDU_RCU, XORQ_RDA_RCA, XORQ_RDE_RCE)
mKeccakRound(rpStack, rpState, $0x8000000080008008, NOP, NOP, NOP, NOP, NOP, NOP, NOP, NOP, NOP, NOP, NOP, NOP, NOP)
// Revert the internal state to the user state
NOTQ _be(rpState)
NOTQ _bi(rpState)
NOTQ _go(rpState)
NOTQ _ki(rpState)
NOTQ _mi(rpState)
NOTQ _sa(rpState)
RET
+19
View File
@@ -0,0 +1,19 @@
// Copyright 2014 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.
//go:build go1.4
// +build go1.4
package sha3
import (
"crypto"
)
func init() {
crypto.RegisterHash(crypto.SHA3_224, New224)
crypto.RegisterHash(crypto.SHA3_256, New256)
crypto.RegisterHash(crypto.SHA3_384, New384)
crypto.RegisterHash(crypto.SHA3_512, New512)
}
+193
View File
@@ -0,0 +1,193 @@
// Copyright 2014 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.
package sha3
// spongeDirection indicates the direction bytes are flowing through the sponge.
type spongeDirection int
const (
// spongeAbsorbing indicates that the sponge is absorbing input.
spongeAbsorbing spongeDirection = iota
// spongeSqueezing indicates that the sponge is being squeezed.
spongeSqueezing
)
const (
// maxRate is the maximum size of the internal buffer. SHAKE-256
// currently needs the largest buffer.
maxRate = 168
)
type state struct {
// Generic sponge components.
a [25]uint64 // main state of the hash
buf []byte // points into storage
rate int // the number of bytes of state to use
// dsbyte contains the "domain separation" bits and the first bit of
// the padding. Sections 6.1 and 6.2 of [1] separate the outputs of the
// SHA-3 and SHAKE functions by appending bitstrings to the message.
// Using a little-endian bit-ordering convention, these are "01" for SHA-3
// and "1111" for SHAKE, or 00000010b and 00001111b, respectively. Then the
// padding rule from section 5.1 is applied to pad the message to a multiple
// of the rate, which involves adding a "1" bit, zero or more "0" bits, and
// a final "1" bit. We merge the first "1" bit from the padding into dsbyte,
// giving 00000110b (0x06) and 00011111b (0x1f).
// [1] http://csrc.nist.gov/publications/drafts/fips-202/fips_202_draft.pdf
// "Draft FIPS 202: SHA-3 Standard: Permutation-Based Hash and
// Extendable-Output Functions (May 2014)"
dsbyte byte
storage storageBuf
// Specific to SHA-3 and SHAKE.
outputLen int // the default output size in bytes
state spongeDirection // whether the sponge is absorbing or squeezing
}
// BlockSize returns the rate of sponge underlying this hash function.
func (d *state) BlockSize() int { return d.rate }
// Size returns the output size of the hash function in bytes.
func (d *state) Size() int { return d.outputLen }
// Reset clears the internal state by zeroing the sponge state and
// the byte buffer, and setting Sponge.state to absorbing.
func (d *state) Reset() {
// Zero the permutation's state.
for i := range d.a {
d.a[i] = 0
}
d.state = spongeAbsorbing
d.buf = d.storage.asBytes()[:0]
}
func (d *state) clone() *state {
ret := *d
if ret.state == spongeAbsorbing {
ret.buf = ret.storage.asBytes()[:len(ret.buf)]
} else {
ret.buf = ret.storage.asBytes()[d.rate-cap(d.buf) : d.rate]
}
return &ret
}
// permute applies the KeccakF-1600 permutation. It handles
// any input-output buffering.
func (d *state) permute() {
switch d.state {
case spongeAbsorbing:
// If we're absorbing, we need to xor the input into the state
// before applying the permutation.
xorIn(d, d.buf)
d.buf = d.storage.asBytes()[:0]
keccakF1600(&d.a)
case spongeSqueezing:
// If we're squeezing, we need to apply the permutation before
// copying more output.
keccakF1600(&d.a)
d.buf = d.storage.asBytes()[:d.rate]
copyOut(d, d.buf)
}
}
// pads appends the domain separation bits in dsbyte, applies
// the multi-bitrate 10..1 padding rule, and permutes the state.
func (d *state) padAndPermute(dsbyte byte) {
if d.buf == nil {
d.buf = d.storage.asBytes()[:0]
}
// Pad with this instance's domain-separator bits. We know that there's
// at least one byte of space in d.buf because, if it were full,
// permute would have been called to empty it. dsbyte also contains the
// first one bit for the padding. See the comment in the state struct.
d.buf = append(d.buf, dsbyte)
zerosStart := len(d.buf)
d.buf = d.storage.asBytes()[:d.rate]
for i := zerosStart; i < d.rate; i++ {
d.buf[i] = 0
}
// This adds the final one bit for the padding. Because of the way that
// bits are numbered from the LSB upwards, the final bit is the MSB of
// the last byte.
d.buf[d.rate-1] ^= 0x80
// Apply the permutation
d.permute()
d.state = spongeSqueezing
d.buf = d.storage.asBytes()[:d.rate]
copyOut(d, d.buf)
}
// Write absorbs more data into the hash's state. It produces an error
// if more data is written to the ShakeHash after writing
func (d *state) Write(p []byte) (written int, err error) {
if d.state != spongeAbsorbing {
panic("sha3: write to sponge after read")
}
if d.buf == nil {
d.buf = d.storage.asBytes()[:0]
}
written = len(p)
for len(p) > 0 {
if len(d.buf) == 0 && len(p) >= d.rate {
// The fast path; absorb a full "rate" bytes of input and apply the permutation.
xorIn(d, p[:d.rate])
p = p[d.rate:]
keccakF1600(&d.a)
} else {
// The slow path; buffer the input until we can fill the sponge, and then xor it in.
todo := d.rate - len(d.buf)
if todo > len(p) {
todo = len(p)
}
d.buf = append(d.buf, p[:todo]...)
p = p[todo:]
// If the sponge is full, apply the permutation.
if len(d.buf) == d.rate {
d.permute()
}
}
}
return
}
// Read squeezes an arbitrary number of bytes from the sponge.
func (d *state) Read(out []byte) (n int, err error) {
// If we're still absorbing, pad and apply the permutation.
if d.state == spongeAbsorbing {
d.padAndPermute(d.dsbyte)
}
n = len(out)
// Now, do the squeezing.
for len(out) > 0 {
n := copy(out, d.buf)
d.buf = d.buf[n:]
out = out[n:]
// Apply the permutation if we've squeezed the sponge dry.
if len(d.buf) == 0 {
d.permute()
}
}
return
}
// Sum applies padding to the hash state and then squeezes out the desired
// number of output bytes.
func (d *state) Sum(in []byte) []byte {
// Make a copy of the original hash so that caller can keep writing
// and summing.
dup := d.clone()
hash := make([]byte, dup.outputLen)
dup.Read(hash)
return append(in, hash...)
}
+287
View File
@@ -0,0 +1,287 @@
// Copyright 2017 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.
//go:build gc && !purego
// +build gc,!purego
package sha3
// This file contains code for using the 'compute intermediate
// message digest' (KIMD) and 'compute last message digest' (KLMD)
// instructions to compute SHA-3 and SHAKE hashes on IBM Z.
import (
"hash"
"golang.org/x/sys/cpu"
)
// codes represent 7-bit KIMD/KLMD function codes as defined in
// the Principles of Operation.
type code uint64
const (
// function codes for KIMD/KLMD
sha3_224 code = 32
sha3_256 = 33
sha3_384 = 34
sha3_512 = 35
shake_128 = 36
shake_256 = 37
nopad = 0x100
)
// kimd is a wrapper for the 'compute intermediate message digest' instruction.
// src must be a multiple of the rate for the given function code.
//
//go:noescape
func kimd(function code, chain *[200]byte, src []byte)
// klmd is a wrapper for the 'compute last message digest' instruction.
// src padding is handled by the instruction.
//
//go:noescape
func klmd(function code, chain *[200]byte, dst, src []byte)
type asmState struct {
a [200]byte // 1600 bit state
buf []byte // care must be taken to ensure cap(buf) is a multiple of rate
rate int // equivalent to block size
storage [3072]byte // underlying storage for buf
outputLen int // output length if fixed, 0 if not
function code // KIMD/KLMD function code
state spongeDirection // whether the sponge is absorbing or squeezing
}
func newAsmState(function code) *asmState {
var s asmState
s.function = function
switch function {
case sha3_224:
s.rate = 144
s.outputLen = 28
case sha3_256:
s.rate = 136
s.outputLen = 32
case sha3_384:
s.rate = 104
s.outputLen = 48
case sha3_512:
s.rate = 72
s.outputLen = 64
case shake_128:
s.rate = 168
case shake_256:
s.rate = 136
default:
panic("sha3: unrecognized function code")
}
// limit s.buf size to a multiple of s.rate
s.resetBuf()
return &s
}
func (s *asmState) clone() *asmState {
c := *s
c.buf = c.storage[:len(s.buf):cap(s.buf)]
return &c
}
// copyIntoBuf copies b into buf. It will panic if there is not enough space to
// store all of b.
func (s *asmState) copyIntoBuf(b []byte) {
bufLen := len(s.buf)
s.buf = s.buf[:len(s.buf)+len(b)]
copy(s.buf[bufLen:], b)
}
// resetBuf points buf at storage, sets the length to 0 and sets cap to be a
// multiple of the rate.
func (s *asmState) resetBuf() {
max := (cap(s.storage) / s.rate) * s.rate
s.buf = s.storage[:0:max]
}
// Write (via the embedded io.Writer interface) adds more data to the running hash.
// It never returns an error.
func (s *asmState) Write(b []byte) (int, error) {
if s.state != spongeAbsorbing {
panic("sha3: write to sponge after read")
}
length := len(b)
for len(b) > 0 {
if len(s.buf) == 0 && len(b) >= cap(s.buf) {
// Hash the data directly and push any remaining bytes
// into the buffer.
remainder := len(b) % s.rate
kimd(s.function, &s.a, b[:len(b)-remainder])
if remainder != 0 {
s.copyIntoBuf(b[len(b)-remainder:])
}
return length, nil
}
if len(s.buf) == cap(s.buf) {
// flush the buffer
kimd(s.function, &s.a, s.buf)
s.buf = s.buf[:0]
}
// copy as much as we can into the buffer
n := len(b)
if len(b) > cap(s.buf)-len(s.buf) {
n = cap(s.buf) - len(s.buf)
}
s.copyIntoBuf(b[:n])
b = b[n:]
}
return length, nil
}
// Read squeezes an arbitrary number of bytes from the sponge.
func (s *asmState) Read(out []byte) (n int, err error) {
n = len(out)
// need to pad if we were absorbing
if s.state == spongeAbsorbing {
s.state = spongeSqueezing
// write hash directly into out if possible
if len(out)%s.rate == 0 {
klmd(s.function, &s.a, out, s.buf) // len(out) may be 0
s.buf = s.buf[:0]
return
}
// write hash into buffer
max := cap(s.buf)
if max > len(out) {
max = (len(out)/s.rate)*s.rate + s.rate
}
klmd(s.function, &s.a, s.buf[:max], s.buf)
s.buf = s.buf[:max]
}
for len(out) > 0 {
// flush the buffer
if len(s.buf) != 0 {
c := copy(out, s.buf)
out = out[c:]
s.buf = s.buf[c:]
continue
}
// write hash directly into out if possible
if len(out)%s.rate == 0 {
klmd(s.function|nopad, &s.a, out, nil)
return
}
// write hash into buffer
s.resetBuf()
if cap(s.buf) > len(out) {
s.buf = s.buf[:(len(out)/s.rate)*s.rate+s.rate]
}
klmd(s.function|nopad, &s.a, s.buf, nil)
}
return
}
// Sum appends the current hash to b and returns the resulting slice.
// It does not change the underlying hash state.
func (s *asmState) Sum(b []byte) []byte {
if s.outputLen == 0 {
panic("sha3: cannot call Sum on SHAKE functions")
}
// Copy the state to preserve the original.
a := s.a
// Hash the buffer. Note that we don't clear it because we
// aren't updating the state.
klmd(s.function, &a, nil, s.buf)
return append(b, a[:s.outputLen]...)
}
// Reset resets the Hash to its initial state.
func (s *asmState) Reset() {
for i := range s.a {
s.a[i] = 0
}
s.resetBuf()
s.state = spongeAbsorbing
}
// Size returns the number of bytes Sum will return.
func (s *asmState) Size() int {
return s.outputLen
}
// BlockSize returns the hash's underlying block size.
// The Write method must be able to accept any amount
// of data, but it may operate more efficiently if all writes
// are a multiple of the block size.
func (s *asmState) BlockSize() int {
return s.rate
}
// Clone returns a copy of the ShakeHash in its current state.
func (s *asmState) Clone() ShakeHash {
return s.clone()
}
// new224Asm returns an assembly implementation of SHA3-224 if available,
// otherwise it returns nil.
func new224Asm() hash.Hash {
if cpu.S390X.HasSHA3 {
return newAsmState(sha3_224)
}
return nil
}
// new256Asm returns an assembly implementation of SHA3-256 if available,
// otherwise it returns nil.
func new256Asm() hash.Hash {
if cpu.S390X.HasSHA3 {
return newAsmState(sha3_256)
}
return nil
}
// new384Asm returns an assembly implementation of SHA3-384 if available,
// otherwise it returns nil.
func new384Asm() hash.Hash {
if cpu.S390X.HasSHA3 {
return newAsmState(sha3_384)
}
return nil
}
// new512Asm returns an assembly implementation of SHA3-512 if available,
// otherwise it returns nil.
func new512Asm() hash.Hash {
if cpu.S390X.HasSHA3 {
return newAsmState(sha3_512)
}
return nil
}
// newShake128Asm returns an assembly implementation of SHAKE-128 if available,
// otherwise it returns nil.
func newShake128Asm() ShakeHash {
if cpu.S390X.HasSHA3 {
return newAsmState(shake_128)
}
return nil
}
// newShake256Asm returns an assembly implementation of SHAKE-256 if available,
// otherwise it returns nil.
func newShake256Asm() ShakeHash {
if cpu.S390X.HasSHA3 {
return newAsmState(shake_256)
}
return nil
}
+34
View File
@@ -0,0 +1,34 @@
// Copyright 2017 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.
//go:build gc && !purego
// +build gc,!purego
#include "textflag.h"
// func kimd(function code, chain *[200]byte, src []byte)
TEXT ·kimd(SB), NOFRAME|NOSPLIT, $0-40
MOVD function+0(FP), R0
MOVD chain+8(FP), R1
LMG src+16(FP), R2, R3 // R2=base, R3=len
continue:
WORD $0xB93E0002 // KIMD --, R2
BVS continue // continue if interrupted
MOVD $0, R0 // reset R0 for pre-go1.8 compilers
RET
// func klmd(function code, chain *[200]byte, dst, src []byte)
TEXT ·klmd(SB), NOFRAME|NOSPLIT, $0-64
// TODO: SHAKE support
MOVD function+0(FP), R0
MOVD chain+8(FP), R1
LMG dst+16(FP), R2, R3 // R2=base, R3=len
LMG src+40(FP), R4, R5 // R4=base, R5=len
continue:
WORD $0xB93F0024 // KLMD R2, R4
BVS continue // continue if interrupted
MOVD $0, R0 // reset R0 for pre-go1.8 compilers
RET
+173
View File
@@ -0,0 +1,173 @@
// Copyright 2014 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.
package sha3
// This file defines the ShakeHash interface, and provides
// functions for creating SHAKE and cSHAKE instances, as well as utility
// functions for hashing bytes to arbitrary-length output.
//
//
// SHAKE implementation is based on FIPS PUB 202 [1]
// cSHAKE implementations is based on NIST SP 800-185 [2]
//
// [1] https://nvlpubs.nist.gov/nistpubs/FIPS/NIST.FIPS.202.pdf
// [2] https://doi.org/10.6028/NIST.SP.800-185
import (
"encoding/binary"
"io"
)
// ShakeHash defines the interface to hash functions that
// support arbitrary-length output.
type ShakeHash interface {
// Write absorbs more data into the hash's state. It panics if input is
// written to it after output has been read from it.
io.Writer
// Read reads more output from the hash; reading affects the hash's
// state. (ShakeHash.Read is thus very different from Hash.Sum)
// It never returns an error.
io.Reader
// Clone returns a copy of the ShakeHash in its current state.
Clone() ShakeHash
// Reset resets the ShakeHash to its initial state.
Reset()
}
// cSHAKE specific context
type cshakeState struct {
*state // SHA-3 state context and Read/Write operations
// initBlock is the cSHAKE specific initialization set of bytes. It is initialized
// by newCShake function and stores concatenation of N followed by S, encoded
// by the method specified in 3.3 of [1].
// It is stored here in order for Reset() to be able to put context into
// initial state.
initBlock []byte
}
// Consts for configuring initial SHA-3 state
const (
dsbyteShake = 0x1f
dsbyteCShake = 0x04
rate128 = 168
rate256 = 136
)
func bytepad(input []byte, w int) []byte {
// leftEncode always returns max 9 bytes
buf := make([]byte, 0, 9+len(input)+w)
buf = append(buf, leftEncode(uint64(w))...)
buf = append(buf, input...)
padlen := w - (len(buf) % w)
return append(buf, make([]byte, padlen)...)
}
func leftEncode(value uint64) []byte {
var b [9]byte
binary.BigEndian.PutUint64(b[1:], value)
// Trim all but last leading zero bytes
i := byte(1)
for i < 8 && b[i] == 0 {
i++
}
// Prepend number of encoded bytes
b[i-1] = 9 - i
return b[i-1:]
}
func newCShake(N, S []byte, rate int, dsbyte byte) ShakeHash {
c := cshakeState{state: &state{rate: rate, dsbyte: dsbyte}}
// leftEncode returns max 9 bytes
c.initBlock = make([]byte, 0, 9*2+len(N)+len(S))
c.initBlock = append(c.initBlock, leftEncode(uint64(len(N)*8))...)
c.initBlock = append(c.initBlock, N...)
c.initBlock = append(c.initBlock, leftEncode(uint64(len(S)*8))...)
c.initBlock = append(c.initBlock, S...)
c.Write(bytepad(c.initBlock, c.rate))
return &c
}
// Reset resets the hash to initial state.
func (c *cshakeState) Reset() {
c.state.Reset()
c.Write(bytepad(c.initBlock, c.rate))
}
// Clone returns copy of a cSHAKE context within its current state.
func (c *cshakeState) Clone() ShakeHash {
b := make([]byte, len(c.initBlock))
copy(b, c.initBlock)
return &cshakeState{state: c.clone(), initBlock: b}
}
// Clone returns copy of SHAKE context within its current state.
func (c *state) Clone() ShakeHash {
return c.clone()
}
// NewShake128 creates a new SHAKE128 variable-output-length ShakeHash.
// Its generic security strength is 128 bits against all attacks if at
// least 32 bytes of its output are used.
func NewShake128() ShakeHash {
if h := newShake128Asm(); h != nil {
return h
}
return &state{rate: rate128, dsbyte: dsbyteShake}
}
// NewShake256 creates a new SHAKE256 variable-output-length ShakeHash.
// Its generic security strength is 256 bits against all attacks if
// at least 64 bytes of its output are used.
func NewShake256() ShakeHash {
if h := newShake256Asm(); h != nil {
return h
}
return &state{rate: rate256, dsbyte: dsbyteShake}
}
// NewCShake128 creates a new instance of cSHAKE128 variable-output-length ShakeHash,
// a customizable variant of SHAKE128.
// N is used to define functions based on cSHAKE, it can be empty when plain cSHAKE is
// desired. S is a customization byte string used for domain separation - two cSHAKE
// computations on same input with different S yield unrelated outputs.
// When N and S are both empty, this is equivalent to NewShake128.
func NewCShake128(N, S []byte) ShakeHash {
if len(N) == 0 && len(S) == 0 {
return NewShake128()
}
return newCShake(N, S, rate128, dsbyteCShake)
}
// NewCShake256 creates a new instance of cSHAKE256 variable-output-length ShakeHash,
// a customizable variant of SHAKE256.
// N is used to define functions based on cSHAKE, it can be empty when plain cSHAKE is
// desired. S is a customization byte string used for domain separation - two cSHAKE
// computations on same input with different S yield unrelated outputs.
// When N and S are both empty, this is equivalent to NewShake256.
func NewCShake256(N, S []byte) ShakeHash {
if len(N) == 0 && len(S) == 0 {
return NewShake256()
}
return newCShake(N, S, rate256, dsbyteCShake)
}
// ShakeSum128 writes an arbitrary-length digest of data into hash.
func ShakeSum128(hash, data []byte) {
h := NewShake128()
h.Write(data)
h.Read(hash)
}
// ShakeSum256 writes an arbitrary-length digest of data into hash.
func ShakeSum256(hash, data []byte) {
h := NewShake256()
h.Write(data)
h.Read(hash)
}
+20
View File
@@ -0,0 +1,20 @@
// Copyright 2017 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.
//go:build !gc || purego || !s390x
// +build !gc purego !s390x
package sha3
// newShake128Asm returns an assembly implementation of SHAKE-128 if available,
// otherwise it returns nil.
func newShake128Asm() ShakeHash {
return nil
}
// newShake256Asm returns an assembly implementation of SHAKE-256 if available,
// otherwise it returns nil.
func newShake256Asm() ShakeHash {
return nil
}
+24
View File
@@ -0,0 +1,24 @@
// Copyright 2015 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.
//go:build (!amd64 && !386 && !ppc64le) || purego
// +build !amd64,!386,!ppc64le purego
package sha3
// A storageBuf is an aligned array of maxRate bytes.
type storageBuf [maxRate]byte
func (b *storageBuf) asBytes() *[maxRate]byte {
return (*[maxRate]byte)(b)
}
var (
xorIn = xorInGeneric
copyOut = copyOutGeneric
xorInUnaligned = xorInGeneric
copyOutUnaligned = copyOutGeneric
)
const xorImplementationUnaligned = "generic"
+28
View File
@@ -0,0 +1,28 @@
// Copyright 2015 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.
package sha3
import "encoding/binary"
// xorInGeneric xors the bytes in buf into the state; it
// makes no non-portable assumptions about memory layout
// or alignment.
func xorInGeneric(d *state, buf []byte) {
n := len(buf) / 8
for i := 0; i < n; i++ {
a := binary.LittleEndian.Uint64(buf)
d.a[i] ^= a
buf = buf[8:]
}
}
// copyOutGeneric copies uint64s to a byte buffer.
func copyOutGeneric(d *state, b []byte) {
for i := 0; len(b) >= 8; i++ {
binary.LittleEndian.PutUint64(b, d.a[i])
b = b[8:]
}
}
+68
View File
@@ -0,0 +1,68 @@
// Copyright 2015 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.
//go:build (amd64 || 386 || ppc64le) && !purego
// +build amd64 386 ppc64le
// +build !purego
package sha3
import "unsafe"
// A storageBuf is an aligned array of maxRate bytes.
type storageBuf [maxRate / 8]uint64
func (b *storageBuf) asBytes() *[maxRate]byte {
return (*[maxRate]byte)(unsafe.Pointer(b))
}
// xorInUnaligned uses unaligned reads and writes to update d.a to contain d.a
// XOR buf.
func xorInUnaligned(d *state, buf []byte) {
n := len(buf)
bw := (*[maxRate / 8]uint64)(unsafe.Pointer(&buf[0]))[: n/8 : n/8]
if n >= 72 {
d.a[0] ^= bw[0]
d.a[1] ^= bw[1]
d.a[2] ^= bw[2]
d.a[3] ^= bw[3]
d.a[4] ^= bw[4]
d.a[5] ^= bw[5]
d.a[6] ^= bw[6]
d.a[7] ^= bw[7]
d.a[8] ^= bw[8]
}
if n >= 104 {
d.a[9] ^= bw[9]
d.a[10] ^= bw[10]
d.a[11] ^= bw[11]
d.a[12] ^= bw[12]
}
if n >= 136 {
d.a[13] ^= bw[13]
d.a[14] ^= bw[14]
d.a[15] ^= bw[15]
d.a[16] ^= bw[16]
}
if n >= 144 {
d.a[17] ^= bw[17]
}
if n >= 168 {
d.a[18] ^= bw[18]
d.a[19] ^= bw[19]
d.a[20] ^= bw[20]
}
}
func copyOutUnaligned(d *state, buf []byte) {
ab := (*[maxRate]uint8)(unsafe.Pointer(&d.a[0]))
copy(buf, ab[:])
}
var (
xorIn = xorInUnaligned
copyOut = copyOutUnaligned
)
const xorImplementationUnaligned = "unaligned"