Commit 43ec1b12 authored by Filippo Valsorda's avatar Filippo Valsorda

src/vendor,crypto/tls: update to latest x/crypto and use new X25519 API

Change-Id: Icd5006e37861d892a5f3d4397c3826179c1b12ad
Reviewed-on: https://go-review.googlesource.com/c/go/+/206657
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: default avatarKatie Hockman <katie@golang.org>
parent 995ade86
...@@ -7,14 +7,14 @@ package tls ...@@ -7,14 +7,14 @@ package tls
import ( import (
"crypto/elliptic" "crypto/elliptic"
"crypto/hmac" "crypto/hmac"
"crypto/subtle"
"errors" "errors"
"golang.org/x/crypto/cryptobyte"
"golang.org/x/crypto/curve25519"
"golang.org/x/crypto/hkdf"
"hash" "hash"
"io" "io"
"math/big" "math/big"
"golang.org/x/crypto/cryptobyte"
"golang.org/x/crypto/curve25519"
"golang.org/x/crypto/hkdf"
) )
// This file contains the functions necessary to compute the TLS 1.3 key // This file contains the functions necessary to compute the TLS 1.3 key
...@@ -111,12 +111,15 @@ type ecdheParameters interface { ...@@ -111,12 +111,15 @@ type ecdheParameters interface {
func generateECDHEParameters(rand io.Reader, curveID CurveID) (ecdheParameters, error) { func generateECDHEParameters(rand io.Reader, curveID CurveID) (ecdheParameters, error) {
if curveID == X25519 { if curveID == X25519 {
p := &x25519Parameters{} privateKey := make([]byte, curve25519.ScalarSize)
if _, err := io.ReadFull(rand, p.privateKey[:]); err != nil { if _, err := io.ReadFull(rand, privateKey); err != nil {
return nil, err
}
publicKey, err := curve25519.X25519(privateKey, curve25519.Basepoint)
if err != nil {
return nil, err return nil, err
} }
curve25519.ScalarBaseMult(&p.publicKey, &p.privateKey) return &x25519Parameters{privateKey: privateKey, publicKey: publicKey}, nil
return p, nil
} }
curve, ok := curveForCurveID(curveID) curve, ok := curveForCurveID(curveID)
...@@ -178,8 +181,8 @@ func (p *nistParameters) SharedKey(peerPublicKey []byte) []byte { ...@@ -178,8 +181,8 @@ func (p *nistParameters) SharedKey(peerPublicKey []byte) []byte {
} }
type x25519Parameters struct { type x25519Parameters struct {
privateKey [32]byte privateKey []byte
publicKey [32]byte publicKey []byte
} }
func (p *x25519Parameters) CurveID() CurveID { func (p *x25519Parameters) CurveID() CurveID {
...@@ -191,19 +194,9 @@ func (p *x25519Parameters) PublicKey() []byte { ...@@ -191,19 +194,9 @@ func (p *x25519Parameters) PublicKey() []byte {
} }
func (p *x25519Parameters) SharedKey(peerPublicKey []byte) []byte { func (p *x25519Parameters) SharedKey(peerPublicKey []byte) []byte {
if len(peerPublicKey) != 32 { sharedKey, err := curve25519.X25519(p.privateKey, peerPublicKey)
return nil if err != nil {
}
var theirPublicKey, sharedKey [32]byte
copy(theirPublicKey[:], peerPublicKey)
curve25519.ScalarMult(&sharedKey, &p.privateKey, &theirPublicKey)
// Check for low-order inputs. See RFC 8422, Section 5.11.
var allZeroes [32]byte
if subtle.ConstantTimeCompare(allZeroes[:], sharedKey[:]) == 1 {
return nil return nil
} }
return sharedKey
return sharedKey[:]
} }
...@@ -3,7 +3,7 @@ module std ...@@ -3,7 +3,7 @@ module std
go 1.14 go 1.14
require ( require (
golang.org/x/crypto v0.0.0-20190611184440-5c40567a22f8 golang.org/x/crypto v0.0.0-20191111213947-16651526fdb4
golang.org/x/net v0.0.0-20191105084925-a882066a44e0 golang.org/x/net v0.0.0-20191105084925-a882066a44e0
golang.org/x/sys v0.0.0-20190529130038-5219a1e1c5f8 // indirect golang.org/x/sys v0.0.0-20190529130038-5219a1e1c5f8 // indirect
golang.org/x/text v0.3.3-0.20191031172631-4b67af870c6f // indirect golang.org/x/text v0.3.3-0.20191031172631-4b67af870c6f // indirect
......
golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w= golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
golang.org/x/crypto v0.0.0-20190611184440-5c40567a22f8 h1:1wopBVtVdWnn03fZelqdXTqk7U7zPQCb+T4rbU9ZEoU= golang.org/x/crypto v0.0.0-20191111213947-16651526fdb4 h1:AGVXd+IAyeAb3FuQvYDYQ9+WR2JHm0+C0oYJaU1C4rs=
golang.org/x/crypto v0.0.0-20190611184440-5c40567a22f8/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI= golang.org/x/crypto v0.0.0-20191111213947-16651526fdb4/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg= golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
golang.org/x/net v0.0.0-20191105084925-a882066a44e0 h1:QPlSTtPE2k6PZPasQUbzuK3p9JbS+vMXYVto8g/yrsg= golang.org/x/net v0.0.0-20191105084925-a882066a44e0 h1:QPlSTtPE2k6PZPasQUbzuK3p9JbS+vMXYVto8g/yrsg=
golang.org/x/net v0.0.0-20191105084925-a882066a44e0/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s= golang.org/x/net v0.0.0-20191105084925-a882066a44e0/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
......
...@@ -3,29 +3,15 @@ ...@@ -3,29 +3,15 @@
// license that can be found in the LICENSE file. // license that can be found in the LICENSE file.
// +build go1.11 // +build go1.11
// +build !gccgo // +build !gccgo,!appengine
package chacha20 package chacha20
const ( const bufSize = 256
haveAsm = true
bufSize = 256
)
//go:noescape //go:noescape
func xorKeyStreamVX(dst, src []byte, key *[8]uint32, nonce *[3]uint32, counter *uint32) func xorKeyStreamVX(dst, src []byte, key *[8]uint32, nonce *[3]uint32, counter *uint32)
func (c *Cipher) xorKeyStreamAsm(dst, src []byte) { func (c *Cipher) xorKeyStreamBlocks(dst, src []byte) {
xorKeyStreamVX(dst, src, &c.key, &c.nonce, &c.counter)
if len(src) >= bufSize {
xorKeyStreamVX(dst, src, &c.key, &c.nonce, &c.counter)
}
if len(src)%bufSize != 0 {
i := len(src) - len(src)%bufSize
c.buf = [bufSize]byte{}
copy(c.buf[:], src[i:])
xorKeyStreamVX(c.buf[:], c.buf[:], &c.key, &c.nonce, &c.counter)
c.len = bufSize - copy(dst[i:], c.buf[:len(src)%bufSize])
}
} }
This diff is collapsed.
...@@ -2,15 +2,12 @@ ...@@ -2,15 +2,12 @@
// Use of this source code is governed by a BSD-style // Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file. // license that can be found in the LICENSE file.
// +build !ppc64le,!arm64,!s390x arm64,!go1.11 gccgo appengine // +build !arm64,!s390x,!ppc64le arm64,!go1.11 gccgo appengine
package chacha20 package chacha20
const ( const bufSize = blockSize
bufSize = 64
haveAsm = false
)
func (*Cipher) xorKeyStreamAsm(dst, src []byte) { func (s *Cipher) xorKeyStreamBlocks(dst, src []byte) {
panic("not implemented") s.xorKeyStreamBlocksGeneric(dst, src)
} }
// Copyright 2019 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.
// +build !gccgo,!appengine
package chacha20
const bufSize = 256
//go:noescape
func chaCha20_ctr32_vsx(out, inp *byte, len int, key *[8]uint32, counter *uint32)
func (c *Cipher) xorKeyStreamBlocks(dst, src []byte) {
chaCha20_ctr32_vsx(&dst[0], &src[0], len(src), &c.key, &c.counter)
}
// Copyright 2019 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.
// Based on CRYPTOGAMS code with the following comment:
// # ====================================================================
// # Written by Andy Polyakov <appro@openssl.org> for the OpenSSL
// # project. The module is, however, dual licensed under OpenSSL and
// # CRYPTOGAMS licenses depending on where you obtain it. For further
// # details see http://www.openssl.org/~appro/cryptogams/.
// # ====================================================================
// Code for the perl script that generates the ppc64 assembler
// can be found in the cryptogams repository at the link below. It is based on
// the original from openssl.
// https://github.com/dot-asm/cryptogams/commit/a60f5b50ed908e91
// The differences in this and the original implementation are
// due to the calling conventions and initialization of constants.
// +build !gccgo,!appengine
#include "textflag.h"
#define OUT R3
#define INP R4
#define LEN R5
#define KEY R6
#define CNT R7
#define TMP R15
#define CONSTBASE R16
#define BLOCKS R17
DATA consts<>+0x00(SB)/8, $0x3320646e61707865
DATA consts<>+0x08(SB)/8, $0x6b20657479622d32
DATA consts<>+0x10(SB)/8, $0x0000000000000001
DATA consts<>+0x18(SB)/8, $0x0000000000000000
DATA consts<>+0x20(SB)/8, $0x0000000000000004
DATA consts<>+0x28(SB)/8, $0x0000000000000000
DATA consts<>+0x30(SB)/8, $0x0a0b08090e0f0c0d
DATA consts<>+0x38(SB)/8, $0x0203000106070405
DATA consts<>+0x40(SB)/8, $0x090a0b080d0e0f0c
DATA consts<>+0x48(SB)/8, $0x0102030005060704
DATA consts<>+0x50(SB)/8, $0x6170786561707865
DATA consts<>+0x58(SB)/8, $0x6170786561707865
DATA consts<>+0x60(SB)/8, $0x3320646e3320646e
DATA consts<>+0x68(SB)/8, $0x3320646e3320646e
DATA consts<>+0x70(SB)/8, $0x79622d3279622d32
DATA consts<>+0x78(SB)/8, $0x79622d3279622d32
DATA consts<>+0x80(SB)/8, $0x6b2065746b206574
DATA consts<>+0x88(SB)/8, $0x6b2065746b206574
DATA consts<>+0x90(SB)/8, $0x0000000100000000
DATA consts<>+0x98(SB)/8, $0x0000000300000002
GLOBL consts<>(SB), RODATA, $0xa0
//func chaCha20_ctr32_vsx(out, inp *byte, len int, key *[8]uint32, counter *uint32)
TEXT ·chaCha20_ctr32_vsx(SB),NOSPLIT,$64-40
MOVD out+0(FP), OUT
MOVD inp+8(FP), INP
MOVD len+16(FP), LEN
MOVD key+24(FP), KEY
MOVD counter+32(FP), CNT
// Addressing for constants
MOVD $consts<>+0x00(SB), CONSTBASE
MOVD $16, R8
MOVD $32, R9
MOVD $48, R10
MOVD $64, R11
SRD $6, LEN, BLOCKS
// V16
LXVW4X (CONSTBASE)(R0), VS48
ADD $80,CONSTBASE
// Load key into V17,V18
LXVW4X (KEY)(R0), VS49
LXVW4X (KEY)(R8), VS50
// Load CNT, NONCE into V19
LXVW4X (CNT)(R0), VS51
// Clear V27
VXOR V27, V27, V27
// V28
LXVW4X (CONSTBASE)(R11), VS60
// splat slot from V19 -> V26
VSPLTW $0, V19, V26
VSLDOI $4, V19, V27, V19
VSLDOI $12, V27, V19, V19
VADDUWM V26, V28, V26
MOVD $10, R14
MOVD R14, CTR
loop_outer_vsx:
// V0, V1, V2, V3
LXVW4X (R0)(CONSTBASE), VS32
LXVW4X (R8)(CONSTBASE), VS33
LXVW4X (R9)(CONSTBASE), VS34
LXVW4X (R10)(CONSTBASE), VS35
// splat values from V17, V18 into V4-V11
VSPLTW $0, V17, V4
VSPLTW $1, V17, V5
VSPLTW $2, V17, V6
VSPLTW $3, V17, V7
VSPLTW $0, V18, V8
VSPLTW $1, V18, V9
VSPLTW $2, V18, V10
VSPLTW $3, V18, V11
// VOR
VOR V26, V26, V12
// splat values from V19 -> V13, V14, V15
VSPLTW $1, V19, V13
VSPLTW $2, V19, V14
VSPLTW $3, V19, V15
// splat const values
VSPLTISW $-16, V27
VSPLTISW $12, V28
VSPLTISW $8, V29
VSPLTISW $7, V30
loop_vsx:
VADDUWM V0, V4, V0
VADDUWM V1, V5, V1
VADDUWM V2, V6, V2
VADDUWM V3, V7, V3
VXOR V12, V0, V12
VXOR V13, V1, V13
VXOR V14, V2, V14
VXOR V15, V3, V15
VRLW V12, V27, V12
VRLW V13, V27, V13
VRLW V14, V27, V14
VRLW V15, V27, V15
VADDUWM V8, V12, V8
VADDUWM V9, V13, V9
VADDUWM V10, V14, V10
VADDUWM V11, V15, V11
VXOR V4, V8, V4
VXOR V5, V9, V5
VXOR V6, V10, V6
VXOR V7, V11, V7
VRLW V4, V28, V4
VRLW V5, V28, V5
VRLW V6, V28, V6
VRLW V7, V28, V7
VADDUWM V0, V4, V0
VADDUWM V1, V5, V1
VADDUWM V2, V6, V2
VADDUWM V3, V7, V3
VXOR V12, V0, V12
VXOR V13, V1, V13
VXOR V14, V2, V14
VXOR V15, V3, V15
VRLW V12, V29, V12
VRLW V13, V29, V13
VRLW V14, V29, V14
VRLW V15, V29, V15
VADDUWM V8, V12, V8
VADDUWM V9, V13, V9
VADDUWM V10, V14, V10
VADDUWM V11, V15, V11
VXOR V4, V8, V4
VXOR V5, V9, V5
VXOR V6, V10, V6
VXOR V7, V11, V7
VRLW V4, V30, V4
VRLW V5, V30, V5
VRLW V6, V30, V6
VRLW V7, V30, V7
VADDUWM V0, V5, V0
VADDUWM V1, V6, V1
VADDUWM V2, V7, V2
VADDUWM V3, V4, V3
VXOR V15, V0, V15
VXOR V12, V1, V12
VXOR V13, V2, V13
VXOR V14, V3, V14
VRLW V15, V27, V15
VRLW V12, V27, V12
VRLW V13, V27, V13
VRLW V14, V27, V14
VADDUWM V10, V15, V10
VADDUWM V11, V12, V11
VADDUWM V8, V13, V8
VADDUWM V9, V14, V9
VXOR V5, V10, V5
VXOR V6, V11, V6
VXOR V7, V8, V7
VXOR V4, V9, V4
VRLW V5, V28, V5
VRLW V6, V28, V6
VRLW V7, V28, V7
VRLW V4, V28, V4
VADDUWM V0, V5, V0
VADDUWM V1, V6, V1
VADDUWM V2, V7, V2
VADDUWM V3, V4, V3
VXOR V15, V0, V15
VXOR V12, V1, V12
VXOR V13, V2, V13
VXOR V14, V3, V14
VRLW V15, V29, V15
VRLW V12, V29, V12
VRLW V13, V29, V13
VRLW V14, V29, V14
VADDUWM V10, V15, V10
VADDUWM V11, V12, V11
VADDUWM V8, V13, V8
VADDUWM V9, V14, V9
VXOR V5, V10, V5
VXOR V6, V11, V6
VXOR V7, V8, V7
VXOR V4, V9, V4
VRLW V5, V30, V5
VRLW V6, V30, V6
VRLW V7, V30, V7
VRLW V4, V30, V4
BC 16, LT, loop_vsx
VADDUWM V12, V26, V12
WORD $0x13600F8C // VMRGEW V0, V1, V27
WORD $0x13821F8C // VMRGEW V2, V3, V28
WORD $0x10000E8C // VMRGOW V0, V1, V0
WORD $0x10421E8C // VMRGOW V2, V3, V2
WORD $0x13A42F8C // VMRGEW V4, V5, V29
WORD $0x13C63F8C // VMRGEW V6, V7, V30
XXPERMDI VS32, VS34, $0, VS33
XXPERMDI VS32, VS34, $3, VS35
XXPERMDI VS59, VS60, $0, VS32
XXPERMDI VS59, VS60, $3, VS34
WORD $0x10842E8C // VMRGOW V4, V5, V4
WORD $0x10C63E8C // VMRGOW V6, V7, V6
WORD $0x13684F8C // VMRGEW V8, V9, V27
WORD $0x138A5F8C // VMRGEW V10, V11, V28
XXPERMDI VS36, VS38, $0, VS37
XXPERMDI VS36, VS38, $3, VS39
XXPERMDI VS61, VS62, $0, VS36
XXPERMDI VS61, VS62, $3, VS38
WORD $0x11084E8C // VMRGOW V8, V9, V8
WORD $0x114A5E8C // VMRGOW V10, V11, V10
WORD $0x13AC6F8C // VMRGEW V12, V13, V29
WORD $0x13CE7F8C // VMRGEW V14, V15, V30
XXPERMDI VS40, VS42, $0, VS41
XXPERMDI VS40, VS42, $3, VS43
XXPERMDI VS59, VS60, $0, VS40
XXPERMDI VS59, VS60, $3, VS42
WORD $0x118C6E8C // VMRGOW V12, V13, V12
WORD $0x11CE7E8C // VMRGOW V14, V15, V14
VSPLTISW $4, V27
VADDUWM V26, V27, V26
XXPERMDI VS44, VS46, $0, VS45
XXPERMDI VS44, VS46, $3, VS47
XXPERMDI VS61, VS62, $0, VS44
XXPERMDI VS61, VS62, $3, VS46
VADDUWM V0, V16, V0
VADDUWM V4, V17, V4
VADDUWM V8, V18, V8
VADDUWM V12, V19, V12
CMPU LEN, $64
BLT tail_vsx
// Bottom of loop
LXVW4X (INP)(R0), VS59
LXVW4X (INP)(R8), VS60
LXVW4X (INP)(R9), VS61
LXVW4X (INP)(R10), VS62
VXOR V27, V0, V27
VXOR V28, V4, V28
VXOR V29, V8, V29
VXOR V30, V12, V30
STXVW4X VS59, (OUT)(R0)
STXVW4X VS60, (OUT)(R8)
ADD $64, INP
STXVW4X VS61, (OUT)(R9)
ADD $-64, LEN
STXVW4X VS62, (OUT)(R10)
ADD $64, OUT
BEQ done_vsx
VADDUWM V1, V16, V0
VADDUWM V5, V17, V4
VADDUWM V9, V18, V8
VADDUWM V13, V19, V12
CMPU LEN, $64
BLT tail_vsx
LXVW4X (INP)(R0), VS59
LXVW4X (INP)(R8), VS60
LXVW4X (INP)(R9), VS61
LXVW4X (INP)(R10), VS62
VXOR V27, V0, V27
VXOR V28, V4, V28
VXOR V29, V8, V29
VXOR V30, V12, V30
STXVW4X VS59, (OUT)(R0)
STXVW4X VS60, (OUT)(R8)
ADD $64, INP
STXVW4X VS61, (OUT)(R9)
ADD $-64, LEN
STXVW4X VS62, (OUT)(V10)
ADD $64, OUT
BEQ done_vsx
VADDUWM V2, V16, V0
VADDUWM V6, V17, V4
VADDUWM V10, V18, V8
VADDUWM V14, V19, V12
CMPU LEN, $64
BLT tail_vsx
LXVW4X (INP)(R0), VS59
LXVW4X (INP)(R8), VS60
LXVW4X (INP)(R9), VS61
LXVW4X (INP)(R10), VS62
VXOR V27, V0, V27
VXOR V28, V4, V28
VXOR V29, V8, V29
VXOR V30, V12, V30
STXVW4X VS59, (OUT)(R0)
STXVW4X VS60, (OUT)(R8)
ADD $64, INP
STXVW4X VS61, (OUT)(R9)
ADD $-64, LEN
STXVW4X VS62, (OUT)(R10)
ADD $64, OUT
BEQ done_vsx
VADDUWM V3, V16, V0
VADDUWM V7, V17, V4
VADDUWM V11, V18, V8
VADDUWM V15, V19, V12
CMPU LEN, $64
BLT tail_vsx
LXVW4X (INP)(R0), VS59
LXVW4X (INP)(R8), VS60
LXVW4X (INP)(R9), VS61
LXVW4X (INP)(R10), VS62
VXOR V27, V0, V27
VXOR V28, V4, V28
VXOR V29, V8, V29
VXOR V30, V12, V30
STXVW4X VS59, (OUT)(R0)
STXVW4X VS60, (OUT)(R8)
ADD $64, INP
STXVW4X VS61, (OUT)(R9)
ADD $-64, LEN
STXVW4X VS62, (OUT)(R10)
ADD $64, OUT
MOVD $10, R14
MOVD R14, CTR
BNE loop_outer_vsx
done_vsx:
// Increment counter by number of 64 byte blocks
MOVD (CNT), R14
ADD BLOCKS, R14
MOVD R14, (CNT)
RET
tail_vsx:
ADD $32, R1, R11
MOVD LEN, CTR
// Save values on stack to copy from
STXVW4X VS32, (R11)(R0)
STXVW4X VS36, (R11)(R8)
STXVW4X VS40, (R11)(R9)
STXVW4X VS44, (R11)(R10)
ADD $-1, R11, R12
ADD $-1, INP
ADD $-1, OUT
looptail_vsx:
// Copying the result to OUT
// in bytes.
MOVBZU 1(R12), KEY
MOVBZU 1(INP), TMP
XOR KEY, TMP, KEY
MOVBU KEY, 1(OUT)
BC 16, LT, looptail_vsx
// Clear the stack values
STXVW4X VS48, (R11)(R0)
STXVW4X VS48, (R11)(R8)
STXVW4X VS48, (R11)(R9)
STXVW4X VS48, (R11)(R10)
BR done_vsx
...@@ -2,28 +2,25 @@ ...@@ -2,28 +2,25 @@
// Use of this source code is governed by a BSD-style // Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file. // license that can be found in the LICENSE file.
// +build s390x,!gccgo,!appengine // +build !gccgo,!appengine
package chacha20 package chacha20
import ( import "golang.org/x/sys/cpu"
"golang.org/x/sys/cpu"
)
var haveAsm = cpu.S390X.HasVX var haveAsm = cpu.S390X.HasVX
const bufSize = 256 const bufSize = 256
// xorKeyStreamVX is an assembly implementation of XORKeyStream. It must only // xorKeyStreamVX is an assembly implementation of XORKeyStream. It must only
// be called when the vector facility is available. // be called when the vector facility is available. Implementation in asm_s390x.s.
// Implementation in asm_s390x.s.
//go:noescape //go:noescape
func xorKeyStreamVX(dst, src []byte, key *[8]uint32, nonce *[3]uint32, counter *uint32, buf *[256]byte, len *int) func xorKeyStreamVX(dst, src []byte, key *[8]uint32, nonce *[3]uint32, counter *uint32)
func (c *Cipher) xorKeyStreamAsm(dst, src []byte) { func (c *Cipher) xorKeyStreamBlocks(dst, src []byte) {
xorKeyStreamVX(dst, src, &c.key, &c.nonce, &c.counter, &c.buf, &c.len) if cpu.S390X.HasVX {
xorKeyStreamVX(dst, src, &c.key, &c.nonce, &c.counter)
} else {
c.xorKeyStreamBlocksGeneric(dst, src)
}
} }
// EXRL targets, DO NOT CALL!
func mvcSrcToBuf()
func mvcBufToDst()
...@@ -2,7 +2,7 @@ ...@@ -2,7 +2,7 @@
// Use of this source code is governed by a BSD-style // Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file. // license that can be found in the LICENSE file.
// +build s390x,!gccgo,!appengine // +build !gccgo,!appengine
#include "go_asm.h" #include "go_asm.h"
#include "textflag.h" #include "textflag.h"
...@@ -24,15 +24,6 @@ DATA ·constants<>+0x14(SB)/4, $0x3320646e ...@@ -24,15 +24,6 @@ DATA ·constants<>+0x14(SB)/4, $0x3320646e
DATA ·constants<>+0x18(SB)/4, $0x79622d32 DATA ·constants<>+0x18(SB)/4, $0x79622d32
DATA ·constants<>+0x1c(SB)/4, $0x6b206574 DATA ·constants<>+0x1c(SB)/4, $0x6b206574
// EXRL targets:
TEXT ·mvcSrcToBuf(SB), NOFRAME|NOSPLIT, $0
MVC $1, (R1), (R8)
RET
TEXT ·mvcBufToDst(SB), NOFRAME|NOSPLIT, $0
MVC $1, (R8), (R9)
RET
#define BSWAP V5 #define BSWAP V5
#define J0 V6 #define J0 V6
#define KEY0 V7 #define KEY0 V7
...@@ -144,7 +135,7 @@ TEXT ·mvcBufToDst(SB), NOFRAME|NOSPLIT, $0 ...@@ -144,7 +135,7 @@ TEXT ·mvcBufToDst(SB), NOFRAME|NOSPLIT, $0
VMRHF v, w, c \ // c = {a[2], b[2], c[2], d[2]} VMRHF v, w, c \ // c = {a[2], b[2], c[2], d[2]}
VMRLF v, w, d // d = {a[3], b[3], c[3], d[3]} VMRLF v, w, d // d = {a[3], b[3], c[3], d[3]}
// func xorKeyStreamVX(dst, src []byte, key *[8]uint32, nonce *[3]uint32, counter *uint32, buf *[256]byte, len *int) // func xorKeyStreamVX(dst, src []byte, key *[8]uint32, nonce *[3]uint32, counter *uint32)
TEXT ·xorKeyStreamVX(SB), NOSPLIT, $0 TEXT ·xorKeyStreamVX(SB), NOSPLIT, $0
MOVD constants<>(SB), R1 MOVD constants<>(SB), R1
MOVD dst+0(FP), R2 // R2=&dst[0] MOVD dst+0(FP), R2 // R2=&dst[0]
...@@ -152,25 +143,10 @@ TEXT ·xorKeyStreamVX(SB), NOSPLIT, $0 ...@@ -152,25 +143,10 @@ TEXT ·xorKeyStreamVX(SB), NOSPLIT, $0
MOVD key+48(FP), R5 // R5=key MOVD key+48(FP), R5 // R5=key
MOVD nonce+56(FP), R6 // R6=nonce MOVD nonce+56(FP), R6 // R6=nonce
MOVD counter+64(FP), R7 // R7=counter MOVD counter+64(FP), R7 // R7=counter
MOVD buf+72(FP), R8 // R8=buf
MOVD len+80(FP), R9 // R9=len
// load BSWAP and J0 // load BSWAP and J0
VLM (R1), BSWAP, J0 VLM (R1), BSWAP, J0
// set up tail buffer
ADD $-1, R4, R12
MOVBZ R12, R12
CMPUBEQ R12, $255, aligned
MOVD R4, R1
AND $~255, R1
MOVD $(R3)(R1*1), R1
EXRL mvcSrcToBuf(SB), R12
MOVD $255, R0
SUB R12, R0
MOVD R0, (R9) // update len
aligned:
// setup // setup
MOVD $95, R0 MOVD $95, R0
VLM (R5), KEY0, KEY1 VLM (R5), KEY0, KEY1
...@@ -217,9 +193,7 @@ loop: ...@@ -217,9 +193,7 @@ loop:
// decrement length // decrement length
ADD $-256, R4 ADD $-256, R4
BLT tail
continue:
// rearrange vectors // rearrange vectors
SHUFFLE(X0, X1, X2, X3, M0, M1, M2, M3) SHUFFLE(X0, X1, X2, X3, M0, M1, M2, M3)
ADDV(J0, X0, X1, X2, X3) ADDV(J0, X0, X1, X2, X3)
...@@ -245,16 +219,6 @@ continue: ...@@ -245,16 +219,6 @@ continue:
MOVD $256(R3), R3 MOVD $256(R3), R3
CMPBNE R4, $0, chacha CMPBNE R4, $0, chacha
CMPUBEQ R12, $255, return
EXRL mvcBufToDst(SB), R12 // len was updated during setup
return:
VSTEF $0, CTR, (R7) VSTEF $0, CTR, (R7)
RET RET
tail:
MOVD R2, R9
MOVD R8, R2
MOVD R8, R3
MOVD $0, R4
JMP continue
...@@ -4,9 +4,7 @@ ...@@ -4,9 +4,7 @@
package chacha20 package chacha20
import ( import "runtime"
"runtime"
)
// Platforms that have fast unaligned 32-bit little endian accesses. // Platforms that have fast unaligned 32-bit little endian accesses.
const unaligned = runtime.GOARCH == "386" || const unaligned = runtime.GOARCH == "386" ||
......
...@@ -2,13 +2,13 @@ ...@@ -2,13 +2,13 @@
// Use of this source code is governed by a BSD-style // Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file. // license that can be found in the LICENSE file.
// Package chacha20poly1305 implements the ChaCha20-Poly1305 AEAD as specified in RFC 7539, // Package chacha20poly1305 implements the ChaCha20-Poly1305 AEAD and its
// and its extended nonce variant XChaCha20-Poly1305. // extended nonce variant XChaCha20-Poly1305, as specified in RFC 8439 and
// draft-irtf-cfrg-xchacha-01.
package chacha20poly1305 // import "golang.org/x/crypto/chacha20poly1305" package chacha20poly1305 // import "golang.org/x/crypto/chacha20poly1305"
import ( import (
"crypto/cipher" "crypto/cipher"
"encoding/binary"
"errors" "errors"
) )
...@@ -29,7 +29,7 @@ const ( ...@@ -29,7 +29,7 @@ const (
) )
type chacha20poly1305 struct { type chacha20poly1305 struct {
key [8]uint32 key [KeySize]byte
} }
// New returns a ChaCha20-Poly1305 AEAD that uses the given 256-bit key. // New returns a ChaCha20-Poly1305 AEAD that uses the given 256-bit key.
...@@ -38,14 +38,7 @@ func New(key []byte) (cipher.AEAD, error) { ...@@ -38,14 +38,7 @@ func New(key []byte) (cipher.AEAD, error) {
return nil, errors.New("chacha20poly1305: bad key length") return nil, errors.New("chacha20poly1305: bad key length")
} }
ret := new(chacha20poly1305) ret := new(chacha20poly1305)
ret.key[0] = binary.LittleEndian.Uint32(key[0:4]) copy(ret.key[:], key)
ret.key[1] = binary.LittleEndian.Uint32(key[4:8])
ret.key[2] = binary.LittleEndian.Uint32(key[8:12])
ret.key[3] = binary.LittleEndian.Uint32(key[12:16])
ret.key[4] = binary.LittleEndian.Uint32(key[16:20])
ret.key[5] = binary.LittleEndian.Uint32(key[20:24])
ret.key[6] = binary.LittleEndian.Uint32(key[24:28])
ret.key[7] = binary.LittleEndian.Uint32(key[28:32])
return ret, nil return ret, nil
} }
......
...@@ -25,23 +25,23 @@ var ( ...@@ -25,23 +25,23 @@ var (
// setupState writes a ChaCha20 input matrix to state. See // setupState writes a ChaCha20 input matrix to state. See
// https://tools.ietf.org/html/rfc7539#section-2.3. // https://tools.ietf.org/html/rfc7539#section-2.3.
func setupState(state *[16]uint32, key *[8]uint32, nonce []byte) { func setupState(state *[16]uint32, key *[32]byte, nonce []byte) {
state[0] = 0x61707865 state[0] = 0x61707865
state[1] = 0x3320646e state[1] = 0x3320646e
state[2] = 0x79622d32 state[2] = 0x79622d32
state[3] = 0x6b206574 state[3] = 0x6b206574
state[4] = key[0] state[4] = binary.LittleEndian.Uint32(key[0:4])
state[5] = key[1] state[5] = binary.LittleEndian.Uint32(key[4:8])
state[6] = key[2] state[6] = binary.LittleEndian.Uint32(key[8:12])
state[7] = key[3] state[7] = binary.LittleEndian.Uint32(key[12:16])
state[8] = key[4] state[8] = binary.LittleEndian.Uint32(key[16:20])
state[9] = key[5] state[9] = binary.LittleEndian.Uint32(key[20:24])
state[10] = key[6] state[10] = binary.LittleEndian.Uint32(key[24:28])
state[11] = key[7] state[11] = binary.LittleEndian.Uint32(key[28:32])
state[12] = 0 state[12] = 0
state[13] = binary.LittleEndian.Uint32(nonce[:4]) state[13] = binary.LittleEndian.Uint32(nonce[0:4])
state[14] = binary.LittleEndian.Uint32(nonce[4:8]) state[14] = binary.LittleEndian.Uint32(nonce[4:8])
state[15] = binary.LittleEndian.Uint32(nonce[8:12]) state[15] = binary.LittleEndian.Uint32(nonce[8:12])
} }
......
...@@ -7,7 +7,7 @@ package chacha20poly1305 ...@@ -7,7 +7,7 @@ package chacha20poly1305
import ( import (
"encoding/binary" "encoding/binary"
"golang.org/x/crypto/internal/chacha20" "golang.org/x/crypto/chacha20"
"golang.org/x/crypto/internal/subtle" "golang.org/x/crypto/internal/subtle"
"golang.org/x/crypto/poly1305" "golang.org/x/crypto/poly1305"
) )
...@@ -22,14 +22,10 @@ func (c *chacha20poly1305) sealGeneric(dst, nonce, plaintext, additionalData []b ...@@ -22,14 +22,10 @@ func (c *chacha20poly1305) sealGeneric(dst, nonce, plaintext, additionalData []b
panic("chacha20poly1305: invalid buffer overlap") panic("chacha20poly1305: invalid buffer overlap")
} }
var polyKey [32]byte var polyKey, discardBuf [32]byte
s := chacha20.New(c.key, [3]uint32{ s, _ := chacha20.NewUnauthenticatedCipher(c.key[:], nonce)
binary.LittleEndian.Uint32(nonce[0:4]),
binary.LittleEndian.Uint32(nonce[4:8]),
binary.LittleEndian.Uint32(nonce[8:12]),
})
s.XORKeyStream(polyKey[:], polyKey[:]) s.XORKeyStream(polyKey[:], polyKey[:])
s.Advance() // skip the next 32 bytes s.XORKeyStream(discardBuf[:], discardBuf[:]) // skip the next 32 bytes
s.XORKeyStream(out, plaintext) s.XORKeyStream(out, plaintext)
polyInput := make([]byte, roundTo16(len(additionalData))+roundTo16(len(plaintext))+8+8) polyInput := make([]byte, roundTo16(len(additionalData))+roundTo16(len(plaintext))+8+8)
...@@ -50,14 +46,10 @@ func (c *chacha20poly1305) openGeneric(dst, nonce, ciphertext, additionalData [] ...@@ -50,14 +46,10 @@ func (c *chacha20poly1305) openGeneric(dst, nonce, ciphertext, additionalData []
copy(tag[:], ciphertext[len(ciphertext)-16:]) copy(tag[:], ciphertext[len(ciphertext)-16:])
ciphertext = ciphertext[:len(ciphertext)-16] ciphertext = ciphertext[:len(ciphertext)-16]
var polyKey [32]byte var polyKey, discardBuf [32]byte
s := chacha20.New(c.key, [3]uint32{ s, _ := chacha20.NewUnauthenticatedCipher(c.key[:], nonce)
binary.LittleEndian.Uint32(nonce[0:4]),
binary.LittleEndian.Uint32(nonce[4:8]),
binary.LittleEndian.Uint32(nonce[8:12]),
})
s.XORKeyStream(polyKey[:], polyKey[:]) s.XORKeyStream(polyKey[:], polyKey[:])
s.Advance() // skip the next 32 bytes s.XORKeyStream(discardBuf[:], discardBuf[:]) // skip the next 32 bytes
polyInput := make([]byte, roundTo16(len(additionalData))+roundTo16(len(ciphertext))+8+8) polyInput := make([]byte, roundTo16(len(additionalData))+roundTo16(len(ciphertext))+8+8)
copy(polyInput, additionalData) copy(polyInput, additionalData)
......
...@@ -6,14 +6,13 @@ package chacha20poly1305 ...@@ -6,14 +6,13 @@ package chacha20poly1305
import ( import (
"crypto/cipher" "crypto/cipher"
"encoding/binary"
"errors" "errors"
"golang.org/x/crypto/internal/chacha20" "golang.org/x/crypto/chacha20"
) )
type xchacha20poly1305 struct { type xchacha20poly1305 struct {
key [8]uint32 key [KeySize]byte
} }
// NewX returns a XChaCha20-Poly1305 AEAD that uses the given 256-bit key. // NewX returns a XChaCha20-Poly1305 AEAD that uses the given 256-bit key.
...@@ -27,14 +26,7 @@ func NewX(key []byte) (cipher.AEAD, error) { ...@@ -27,14 +26,7 @@ func NewX(key []byte) (cipher.AEAD, error) {
return nil, errors.New("chacha20poly1305: bad key length") return nil, errors.New("chacha20poly1305: bad key length")
} }
ret := new(xchacha20poly1305) ret := new(xchacha20poly1305)
ret.key[0] = binary.LittleEndian.Uint32(key[0:4]) copy(ret.key[:], key)
ret.key[1] = binary.LittleEndian.Uint32(key[4:8])
ret.key[2] = binary.LittleEndian.Uint32(key[8:12])
ret.key[3] = binary.LittleEndian.Uint32(key[12:16])
ret.key[4] = binary.LittleEndian.Uint32(key[16:20])
ret.key[5] = binary.LittleEndian.Uint32(key[20:24])
ret.key[6] = binary.LittleEndian.Uint32(key[24:28])
ret.key[7] = binary.LittleEndian.Uint32(key[28:32])
return ret, nil return ret, nil
} }
...@@ -60,15 +52,10 @@ func (x *xchacha20poly1305) Seal(dst, nonce, plaintext, additionalData []byte) [ ...@@ -60,15 +52,10 @@ func (x *xchacha20poly1305) Seal(dst, nonce, plaintext, additionalData []byte) [
panic("chacha20poly1305: plaintext too large") panic("chacha20poly1305: plaintext too large")
} }
hNonce := [4]uint32{ c := new(chacha20poly1305)
binary.LittleEndian.Uint32(nonce[0:4]), hKey, _ := chacha20.HChaCha20(x.key[:], nonce[0:16])
binary.LittleEndian.Uint32(nonce[4:8]), copy(c.key[:], hKey)
binary.LittleEndian.Uint32(nonce[8:12]),
binary.LittleEndian.Uint32(nonce[12:16]),
}
c := &chacha20poly1305{
key: chacha20.HChaCha20(&x.key, &hNonce),
}
// The first 4 bytes of the final nonce are unused counter space. // The first 4 bytes of the final nonce are unused counter space.
cNonce := make([]byte, NonceSize) cNonce := make([]byte, NonceSize)
copy(cNonce[4:12], nonce[16:24]) copy(cNonce[4:12], nonce[16:24])
...@@ -87,15 +74,10 @@ func (x *xchacha20poly1305) Open(dst, nonce, ciphertext, additionalData []byte) ...@@ -87,15 +74,10 @@ func (x *xchacha20poly1305) Open(dst, nonce, ciphertext, additionalData []byte)
panic("chacha20poly1305: ciphertext too large") panic("chacha20poly1305: ciphertext too large")
} }
hNonce := [4]uint32{ c := new(chacha20poly1305)
binary.LittleEndian.Uint32(nonce[0:4]), hKey, _ := chacha20.HChaCha20(x.key[:], nonce[0:16])
binary.LittleEndian.Uint32(nonce[4:8]), copy(c.key[:], hKey)
binary.LittleEndian.Uint32(nonce[8:12]),
binary.LittleEndian.Uint32(nonce[12:16]),
}
c := &chacha20poly1305{
key: chacha20.HChaCha20(&x.key, &hNonce),
}
// The first 4 bytes of the final nonce are unused counter space. // The first 4 bytes of the final nonce are unused counter space.
cNonce := make([]byte, NonceSize) cNonce := make([]byte, NonceSize)
copy(cNonce[4:12], nonce[16:24]) copy(cNonce[4:12], nonce[16:24])
......
// Copyright 2012 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.
// This code was translated into a form compatible with 6a from the public
// domain sources in SUPERCOP: https://bench.cr.yp.to/supercop.html
// +build amd64,!gccgo,!appengine
// These constants cannot be encoded in non-MOVQ immediates.
// We access them directly from memory instead.
DATA ·_121666_213(SB)/8, $996687872
GLOBL ·_121666_213(SB), 8, $8
DATA ·_2P0(SB)/8, $0xFFFFFFFFFFFDA
GLOBL ·_2P0(SB), 8, $8
DATA ·_2P1234(SB)/8, $0xFFFFFFFFFFFFE
GLOBL ·_2P1234(SB), 8, $8
// Copyright 2012 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.
// +build amd64,!gccgo,!appengine
// func cswap(inout *[4][5]uint64, v uint64)
TEXT ·cswap(SB),7,$0
MOVQ inout+0(FP),DI
MOVQ v+8(FP),SI
SUBQ $1, SI
NOTQ SI
MOVQ SI, X15
PSHUFD $0x44, X15, X15
MOVOU 0(DI), X0
MOVOU 16(DI), X2
MOVOU 32(DI), X4
MOVOU 48(DI), X6
MOVOU 64(DI), X8
MOVOU 80(DI), X1
MOVOU 96(DI), X3
MOVOU 112(DI), X5
MOVOU 128(DI), X7
MOVOU 144(DI), X9
MOVO X1, X10
MOVO X3, X11
MOVO X5, X12
MOVO X7, X13
MOVO X9, X14
PXOR X0, X10
PXOR X2, X11
PXOR X4, X12
PXOR X6, X13
PXOR X8, X14
PAND X15, X10
PAND X15, X11
PAND X15, X12
PAND X15, X13
PAND X15, X14
PXOR X10, X0
PXOR X10, X1
PXOR X11, X2
PXOR X11, X3
PXOR X12, X4
PXOR X12, X5
PXOR X13, X6
PXOR X13, X7
PXOR X14, X8
PXOR X14, X9
MOVOU X0, 0(DI)
MOVOU X2, 16(DI)
MOVOU X4, 32(DI)
MOVOU X6, 48(DI)
MOVOU X8, 64(DI)
MOVOU X1, 80(DI)
MOVOU X3, 96(DI)
MOVOU X5, 112(DI)
MOVOU X7, 128(DI)
MOVOU X9, 144(DI)
RET
...@@ -2,7 +2,7 @@ ...@@ -2,7 +2,7 @@
// Use of this source code is governed by a BSD-style // Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file. // license that can be found in the LICENSE file.
// +build amd64,!gccgo,!appengine // +build amd64,!gccgo,!appengine,!purego
package curve25519 package curve25519
......
...@@ -5,9 +5,84 @@ ...@@ -5,9 +5,84 @@
// This code was translated into a form compatible with 6a from the public // This code was translated into a form compatible with 6a from the public
// domain sources in SUPERCOP: https://bench.cr.yp.to/supercop.html // domain sources in SUPERCOP: https://bench.cr.yp.to/supercop.html
// +build amd64,!gccgo,!appengine // +build amd64,!gccgo,!appengine,!purego
#include "const_amd64.h" #define REDMASK51 0x0007FFFFFFFFFFFF
// These constants cannot be encoded in non-MOVQ immediates.
// We access them directly from memory instead.
DATA ·_121666_213(SB)/8, $996687872
GLOBL ·_121666_213(SB), 8, $8
DATA ·_2P0(SB)/8, $0xFFFFFFFFFFFDA
GLOBL ·_2P0(SB), 8, $8
DATA ·_2P1234(SB)/8, $0xFFFFFFFFFFFFE
GLOBL ·_2P1234(SB), 8, $8
// func freeze(inout *[5]uint64)
TEXT ·freeze(SB),7,$0-8
MOVQ inout+0(FP), DI
MOVQ 0(DI),SI
MOVQ 8(DI),DX
MOVQ 16(DI),CX
MOVQ 24(DI),R8
MOVQ 32(DI),R9
MOVQ $REDMASK51,AX
MOVQ AX,R10
SUBQ $18,R10
MOVQ $3,R11
REDUCELOOP:
MOVQ SI,R12
SHRQ $51,R12
ANDQ AX,SI
ADDQ R12,DX
MOVQ DX,R12
SHRQ $51,R12
ANDQ AX,DX
ADDQ R12,CX
MOVQ CX,R12
SHRQ $51,R12
ANDQ AX,CX
ADDQ R12,R8
MOVQ R8,R12
SHRQ $51,R12
ANDQ AX,R8
ADDQ R12,R9
MOVQ R9,R12
SHRQ $51,R12
ANDQ AX,R9
IMUL3Q $19,R12,R12
ADDQ R12,SI
SUBQ $1,R11
JA REDUCELOOP
MOVQ $1,R12
CMPQ R10,SI
CMOVQLT R11,R12
CMPQ AX,DX
CMOVQNE R11,R12
CMPQ AX,CX
CMOVQNE R11,R12
CMPQ AX,R8
CMOVQNE R11,R12
CMPQ AX,R9
CMOVQNE R11,R12
NEGQ R12
ANDQ R12,AX
ANDQ R12,R10
SUBQ R10,SI
SUBQ AX,DX
SUBQ AX,CX
SUBQ AX,R8
SUBQ AX,R9
MOVQ SI,0(DI)
MOVQ DX,8(DI)
MOVQ CX,16(DI)
MOVQ R8,24(DI)
MOVQ R9,32(DI)
RET
// func ladderstep(inout *[5][5]uint64) // func ladderstep(inout *[5][5]uint64)
TEXT ·ladderstep(SB),0,$296-8 TEXT ·ladderstep(SB),0,$296-8
...@@ -1375,3 +1450,344 @@ TEXT ·ladderstep(SB),0,$296-8 ...@@ -1375,3 +1450,344 @@ TEXT ·ladderstep(SB),0,$296-8
MOVQ AX,104(DI) MOVQ AX,104(DI)
MOVQ R10,112(DI) MOVQ R10,112(DI)
RET RET
// func cswap(inout *[4][5]uint64, v uint64)
TEXT ·cswap(SB),7,$0
MOVQ inout+0(FP),DI
MOVQ v+8(FP),SI
SUBQ $1, SI
NOTQ SI
MOVQ SI, X15
PSHUFD $0x44, X15, X15
MOVOU 0(DI), X0
MOVOU 16(DI), X2
MOVOU 32(DI), X4
MOVOU 48(DI), X6
MOVOU 64(DI), X8
MOVOU 80(DI), X1
MOVOU 96(DI), X3
MOVOU 112(DI), X5
MOVOU 128(DI), X7
MOVOU 144(DI), X9
MOVO X1, X10
MOVO X3, X11
MOVO X5, X12
MOVO X7, X13
MOVO X9, X14
PXOR X0, X10
PXOR X2, X11
PXOR X4, X12
PXOR X6, X13
PXOR X8, X14
PAND X15, X10
PAND X15, X11
PAND X15, X12
PAND X15, X13
PAND X15, X14
PXOR X10, X0
PXOR X10, X1
PXOR X11, X2
PXOR X11, X3
PXOR X12, X4
PXOR X12, X5
PXOR X13, X6
PXOR X13, X7
PXOR X14, X8
PXOR X14, X9
MOVOU X0, 0(DI)
MOVOU X2, 16(DI)
MOVOU X4, 32(DI)
MOVOU X6, 48(DI)
MOVOU X8, 64(DI)
MOVOU X1, 80(DI)
MOVOU X3, 96(DI)
MOVOU X5, 112(DI)
MOVOU X7, 128(DI)
MOVOU X9, 144(DI)
RET
// func mul(dest, a, b *[5]uint64)
TEXT ·mul(SB),0,$16-24
MOVQ dest+0(FP), DI
MOVQ a+8(FP), SI
MOVQ b+16(FP), DX
MOVQ DX,CX
MOVQ 24(SI),DX
IMUL3Q $19,DX,AX
MOVQ AX,0(SP)
MULQ 16(CX)
MOVQ AX,R8
MOVQ DX,R9
MOVQ 32(SI),DX
IMUL3Q $19,DX,AX
MOVQ AX,8(SP)
MULQ 8(CX)
ADDQ AX,R8
ADCQ DX,R9
MOVQ 0(SI),AX
MULQ 0(CX)
ADDQ AX,R8
ADCQ DX,R9
MOVQ 0(SI),AX
MULQ 8(CX)
MOVQ AX,R10
MOVQ DX,R11
MOVQ 0(SI),AX
MULQ 16(CX)
MOVQ AX,R12
MOVQ DX,R13
MOVQ 0(SI),AX
MULQ 24(CX)
MOVQ AX,R14
MOVQ DX,R15
MOVQ 0(SI),AX
MULQ 32(CX)
MOVQ AX,BX
MOVQ DX,BP
MOVQ 8(SI),AX
MULQ 0(CX)
ADDQ AX,R10
ADCQ DX,R11
MOVQ 8(SI),AX
MULQ 8(CX)
ADDQ AX,R12
ADCQ DX,R13
MOVQ 8(SI),AX
MULQ 16(CX)
ADDQ AX,R14
ADCQ DX,R15
MOVQ 8(SI),AX
MULQ 24(CX)
ADDQ AX,BX
ADCQ DX,BP
MOVQ 8(SI),DX
IMUL3Q $19,DX,AX
MULQ 32(CX)
ADDQ AX,R8
ADCQ DX,R9
MOVQ 16(SI),AX
MULQ 0(CX)
ADDQ AX,R12
ADCQ DX,R13
MOVQ 16(SI),AX
MULQ 8(CX)
ADDQ AX,R14
ADCQ DX,R15
MOVQ 16(SI),AX
MULQ 16(CX)
ADDQ AX,BX
ADCQ DX,BP
MOVQ 16(SI),DX
IMUL3Q $19,DX,AX
MULQ 24(CX)
ADDQ AX,R8
ADCQ DX,R9
MOVQ 16(SI),DX
IMUL3Q $19,DX,AX
MULQ 32(CX)
ADDQ AX,R10
ADCQ DX,R11
MOVQ 24(SI),AX
MULQ 0(CX)
ADDQ AX,R14
ADCQ DX,R15
MOVQ 24(SI),AX
MULQ 8(CX)
ADDQ AX,BX
ADCQ DX,BP
MOVQ 0(SP),AX
MULQ 24(CX)
ADDQ AX,R10
ADCQ DX,R11
MOVQ 0(SP),AX
MULQ 32(CX)
ADDQ AX,R12
ADCQ DX,R13
MOVQ 32(SI),AX
MULQ 0(CX)
ADDQ AX,BX
ADCQ DX,BP
MOVQ 8(SP),AX
MULQ 16(CX)
ADDQ AX,R10
ADCQ DX,R11
MOVQ 8(SP),AX
MULQ 24(CX)
ADDQ AX,R12
ADCQ DX,R13
MOVQ 8(SP),AX
MULQ 32(CX)
ADDQ AX,R14
ADCQ DX,R15
MOVQ $REDMASK51,SI
SHLQ $13,R8,R9
ANDQ SI,R8
SHLQ $13,R10,R11
ANDQ SI,R10
ADDQ R9,R10
SHLQ $13,R12,R13
ANDQ SI,R12
ADDQ R11,R12
SHLQ $13,R14,R15
ANDQ SI,R14
ADDQ R13,R14
SHLQ $13,BX,BP
ANDQ SI,BX
ADDQ R15,BX
IMUL3Q $19,BP,DX
ADDQ DX,R8
MOVQ R8,DX
SHRQ $51,DX
ADDQ R10,DX
MOVQ DX,CX
SHRQ $51,DX
ANDQ SI,R8
ADDQ R12,DX
MOVQ DX,R9
SHRQ $51,DX
ANDQ SI,CX
ADDQ R14,DX
MOVQ DX,AX
SHRQ $51,DX
ANDQ SI,R9
ADDQ BX,DX
MOVQ DX,R10
SHRQ $51,DX
ANDQ SI,AX
IMUL3Q $19,DX,DX
ADDQ DX,R8
ANDQ SI,R10
MOVQ R8,0(DI)
MOVQ CX,8(DI)
MOVQ R9,16(DI)
MOVQ AX,24(DI)
MOVQ R10,32(DI)
RET
// func square(out, in *[5]uint64)
TEXT ·square(SB),7,$0-16
MOVQ out+0(FP), DI
MOVQ in+8(FP), SI
MOVQ 0(SI),AX
MULQ 0(SI)
MOVQ AX,CX
MOVQ DX,R8
MOVQ 0(SI),AX
SHLQ $1,AX
MULQ 8(SI)
MOVQ AX,R9
MOVQ DX,R10
MOVQ 0(SI),AX
SHLQ $1,AX
MULQ 16(SI)
MOVQ AX,R11
MOVQ DX,R12
MOVQ 0(SI),AX
SHLQ $1,AX
MULQ 24(SI)
MOVQ AX,R13
MOVQ DX,R14
MOVQ 0(SI),AX
SHLQ $1,AX
MULQ 32(SI)
MOVQ AX,R15
MOVQ DX,BX
MOVQ 8(SI),AX
MULQ 8(SI)
ADDQ AX,R11
ADCQ DX,R12
MOVQ 8(SI),AX
SHLQ $1,AX
MULQ 16(SI)
ADDQ AX,R13
ADCQ DX,R14
MOVQ 8(SI),AX
SHLQ $1,AX
MULQ 24(SI)
ADDQ AX,R15
ADCQ DX,BX
MOVQ 8(SI),DX
IMUL3Q $38,DX,AX
MULQ 32(SI)
ADDQ AX,CX
ADCQ DX,R8
MOVQ 16(SI),AX
MULQ 16(SI)
ADDQ AX,R15
ADCQ DX,BX
MOVQ 16(SI),DX
IMUL3Q $38,DX,AX
MULQ 24(SI)
ADDQ AX,CX
ADCQ DX,R8
MOVQ 16(SI),DX
IMUL3Q $38,DX,AX
MULQ 32(SI)
ADDQ AX,R9
ADCQ DX,R10
MOVQ 24(SI),DX
IMUL3Q $19,DX,AX
MULQ 24(SI)
ADDQ AX,R9
ADCQ DX,R10
MOVQ 24(SI),DX
IMUL3Q $38,DX,AX
MULQ 32(SI)
ADDQ AX,R11
ADCQ DX,R12
MOVQ 32(SI),DX
IMUL3Q $19,DX,AX
MULQ 32(SI)
ADDQ AX,R13
ADCQ DX,R14
MOVQ $REDMASK51,SI
SHLQ $13,CX,R8
ANDQ SI,CX
SHLQ $13,R9,R10
ANDQ SI,R9
ADDQ R8,R9
SHLQ $13,R11,R12
ANDQ SI,R11
ADDQ R10,R11
SHLQ $13,R13,R14
ANDQ SI,R13
ADDQ R12,R13
SHLQ $13,R15,BX
ANDQ SI,R15
ADDQ R14,R15
IMUL3Q $19,BX,DX
ADDQ DX,CX
MOVQ CX,DX
SHRQ $51,DX
ADDQ R9,DX
ANDQ SI,CX
MOVQ DX,R8
SHRQ $51,DX
ADDQ R11,DX
ANDQ SI,R8
MOVQ DX,R9
SHRQ $51,DX
ADDQ R13,DX
ANDQ SI,R9
MOVQ DX,AX
SHRQ $51,DX
ADDQ R15,DX
ANDQ SI,AX
MOVQ DX,R10
SHRQ $51,DX
IMUL3Q $19,DX,DX
ADDQ DX,CX
ANDQ SI,R10
MOVQ CX,0(DI)
MOVQ R8,8(DI)
MOVQ R9,16(DI)
MOVQ AX,24(DI)
MOVQ R10,32(DI)
RET
// Copyright 2012 The Go Authors. All rights reserved. // Copyright 2019 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style // Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file. // license that can be found in the LICENSE file.
// This code was translated into a form compatible with 6a from the public // +build !amd64 gccgo appengine purego
// domain sources in SUPERCOP: https://bench.cr.yp.to/supercop.html
#define REDMASK51 0x0007FFFFFFFFFFFF package curve25519
func scalarMult(out, in, base *[32]byte) {
scalarMultGeneric(out, in, base)
}
// Copyright 2012 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 curve25519 provides an implementation of scalar multiplication on
// the elliptic curve known as curve25519. See https://cr.yp.to/ecdh.html
package curve25519 // import "golang.org/x/crypto/curve25519"
// basePoint is the x coordinate of the generator of the curve.
var basePoint = [32]byte{9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}
// ScalarMult sets dst to the product in*base where dst and base are the x
// coordinates of group points and all values are in little-endian form.
func ScalarMult(dst, in, base *[32]byte) {
scalarMult(dst, in, base)
}
// ScalarBaseMult sets dst to the product in*base where dst and base are the x
// coordinates of group points, base is the standard generator and all values
// are in little-endian form.
func ScalarBaseMult(dst, in *[32]byte) {
ScalarMult(dst, in, &basePoint)
}
// Copyright 2012 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.
// This code was translated into a form compatible with 6a from the public
// domain sources in SUPERCOP: https://bench.cr.yp.to/supercop.html
// +build amd64,!gccgo,!appengine
#include "const_amd64.h"
// func freeze(inout *[5]uint64)
TEXT ·freeze(SB),7,$0-8
MOVQ inout+0(FP), DI
MOVQ 0(DI),SI
MOVQ 8(DI),DX
MOVQ 16(DI),CX
MOVQ 24(DI),R8
MOVQ 32(DI),R9
MOVQ $REDMASK51,AX
MOVQ AX,R10
SUBQ $18,R10
MOVQ $3,R11
REDUCELOOP:
MOVQ SI,R12
SHRQ $51,R12
ANDQ AX,SI
ADDQ R12,DX
MOVQ DX,R12
SHRQ $51,R12
ANDQ AX,DX
ADDQ R12,CX
MOVQ CX,R12
SHRQ $51,R12
ANDQ AX,CX
ADDQ R12,R8
MOVQ R8,R12
SHRQ $51,R12
ANDQ AX,R8
ADDQ R12,R9
MOVQ R9,R12
SHRQ $51,R12
ANDQ AX,R9
IMUL3Q $19,R12,R12
ADDQ R12,SI
SUBQ $1,R11
JA REDUCELOOP
MOVQ $1,R12
CMPQ R10,SI
CMOVQLT R11,R12
CMPQ AX,DX
CMOVQNE R11,R12
CMPQ AX,CX
CMOVQNE R11,R12
CMPQ AX,R8
CMOVQNE R11,R12
CMPQ AX,R9
CMOVQNE R11,R12
NEGQ R12
ANDQ R12,AX
ANDQ R12,R10
SUBQ R10,SI
SUBQ AX,DX
SUBQ AX,CX
SUBQ AX,R8
SUBQ AX,R9
MOVQ SI,0(DI)
MOVQ DX,8(DI)
MOVQ CX,16(DI)
MOVQ R8,24(DI)
MOVQ R9,32(DI)
RET
// Copyright 2012 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.
// This code was translated into a form compatible with 6a from the public
// domain sources in SUPERCOP: https://bench.cr.yp.to/supercop.html
// +build amd64,!gccgo,!appengine
#include "const_amd64.h"
// func mul(dest, a, b *[5]uint64)
TEXT ·mul(SB),0,$16-24
MOVQ dest+0(FP), DI
MOVQ a+8(FP), SI
MOVQ b+16(FP), DX
MOVQ DX,CX
MOVQ 24(SI),DX
IMUL3Q $19,DX,AX
MOVQ AX,0(SP)
MULQ 16(CX)
MOVQ AX,R8
MOVQ DX,R9
MOVQ 32(SI),DX
IMUL3Q $19,DX,AX
MOVQ AX,8(SP)
MULQ 8(CX)
ADDQ AX,R8
ADCQ DX,R9
MOVQ 0(SI),AX
MULQ 0(CX)
ADDQ AX,R8
ADCQ DX,R9
MOVQ 0(SI),AX
MULQ 8(CX)
MOVQ AX,R10
MOVQ DX,R11
MOVQ 0(SI),AX
MULQ 16(CX)
MOVQ AX,R12
MOVQ DX,R13
MOVQ 0(SI),AX
MULQ 24(CX)
MOVQ AX,R14
MOVQ DX,R15
MOVQ 0(SI),AX
MULQ 32(CX)
MOVQ AX,BX
MOVQ DX,BP
MOVQ 8(SI),AX
MULQ 0(CX)
ADDQ AX,R10
ADCQ DX,R11
MOVQ 8(SI),AX
MULQ 8(CX)
ADDQ AX,R12
ADCQ DX,R13
MOVQ 8(SI),AX
MULQ 16(CX)
ADDQ AX,R14
ADCQ DX,R15
MOVQ 8(SI),AX
MULQ 24(CX)
ADDQ AX,BX
ADCQ DX,BP
MOVQ 8(SI),DX
IMUL3Q $19,DX,AX
MULQ 32(CX)
ADDQ AX,R8
ADCQ DX,R9
MOVQ 16(SI),AX
MULQ 0(CX)
ADDQ AX,R12
ADCQ DX,R13
MOVQ 16(SI),AX
MULQ 8(CX)
ADDQ AX,R14
ADCQ DX,R15
MOVQ 16(SI),AX
MULQ 16(CX)
ADDQ AX,BX
ADCQ DX,BP
MOVQ 16(SI),DX
IMUL3Q $19,DX,AX
MULQ 24(CX)
ADDQ AX,R8
ADCQ DX,R9
MOVQ 16(SI),DX
IMUL3Q $19,DX,AX
MULQ 32(CX)
ADDQ AX,R10
ADCQ DX,R11
MOVQ 24(SI),AX
MULQ 0(CX)
ADDQ AX,R14
ADCQ DX,R15
MOVQ 24(SI),AX
MULQ 8(CX)
ADDQ AX,BX
ADCQ DX,BP
MOVQ 0(SP),AX
MULQ 24(CX)
ADDQ AX,R10
ADCQ DX,R11
MOVQ 0(SP),AX
MULQ 32(CX)
ADDQ AX,R12
ADCQ DX,R13
MOVQ 32(SI),AX
MULQ 0(CX)
ADDQ AX,BX
ADCQ DX,BP
MOVQ 8(SP),AX
MULQ 16(CX)
ADDQ AX,R10
ADCQ DX,R11
MOVQ 8(SP),AX
MULQ 24(CX)
ADDQ AX,R12
ADCQ DX,R13
MOVQ 8(SP),AX
MULQ 32(CX)
ADDQ AX,R14
ADCQ DX,R15
MOVQ $REDMASK51,SI
SHLQ $13,R8,R9
ANDQ SI,R8
SHLQ $13,R10,R11
ANDQ SI,R10
ADDQ R9,R10
SHLQ $13,R12,R13
ANDQ SI,R12
ADDQ R11,R12
SHLQ $13,R14,R15
ANDQ SI,R14
ADDQ R13,R14
SHLQ $13,BX,BP
ANDQ SI,BX
ADDQ R15,BX
IMUL3Q $19,BP,DX
ADDQ DX,R8
MOVQ R8,DX
SHRQ $51,DX
ADDQ R10,DX
MOVQ DX,CX
SHRQ $51,DX
ANDQ SI,R8
ADDQ R12,DX
MOVQ DX,R9
SHRQ $51,DX
ANDQ SI,CX
ADDQ R14,DX
MOVQ DX,AX
SHRQ $51,DX
ANDQ SI,R9
ADDQ BX,DX
MOVQ DX,R10
SHRQ $51,DX
ANDQ SI,AX
IMUL3Q $19,DX,DX
ADDQ DX,R8
ANDQ SI,R10
MOVQ R8,0(DI)
MOVQ CX,8(DI)
MOVQ R9,16(DI)
MOVQ AX,24(DI)
MOVQ R10,32(DI)
RET
// Copyright 2012 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.
// This code was translated into a form compatible with 6a from the public
// domain sources in SUPERCOP: https://bench.cr.yp.to/supercop.html
// +build amd64,!gccgo,!appengine
#include "const_amd64.h"
// func square(out, in *[5]uint64)
TEXT ·square(SB),7,$0-16
MOVQ out+0(FP), DI
MOVQ in+8(FP), SI
MOVQ 0(SI),AX
MULQ 0(SI)
MOVQ AX,CX
MOVQ DX,R8
MOVQ 0(SI),AX
SHLQ $1,AX
MULQ 8(SI)
MOVQ AX,R9
MOVQ DX,R10
MOVQ 0(SI),AX
SHLQ $1,AX
MULQ 16(SI)
MOVQ AX,R11
MOVQ DX,R12
MOVQ 0(SI),AX
SHLQ $1,AX
MULQ 24(SI)
MOVQ AX,R13
MOVQ DX,R14
MOVQ 0(SI),AX
SHLQ $1,AX
MULQ 32(SI)
MOVQ AX,R15
MOVQ DX,BX
MOVQ 8(SI),AX
MULQ 8(SI)
ADDQ AX,R11
ADCQ DX,R12
MOVQ 8(SI),AX
SHLQ $1,AX
MULQ 16(SI)
ADDQ AX,R13
ADCQ DX,R14
MOVQ 8(SI),AX
SHLQ $1,AX
MULQ 24(SI)
ADDQ AX,R15
ADCQ DX,BX
MOVQ 8(SI),DX
IMUL3Q $38,DX,AX
MULQ 32(SI)
ADDQ AX,CX
ADCQ DX,R8
MOVQ 16(SI),AX
MULQ 16(SI)
ADDQ AX,R15
ADCQ DX,BX
MOVQ 16(SI),DX
IMUL3Q $38,DX,AX
MULQ 24(SI)
ADDQ AX,CX
ADCQ DX,R8
MOVQ 16(SI),DX
IMUL3Q $38,DX,AX
MULQ 32(SI)
ADDQ AX,R9
ADCQ DX,R10
MOVQ 24(SI),DX
IMUL3Q $19,DX,AX
MULQ 24(SI)
ADDQ AX,R9
ADCQ DX,R10
MOVQ 24(SI),DX
IMUL3Q $38,DX,AX
MULQ 32(SI)
ADDQ AX,R11
ADCQ DX,R12
MOVQ 32(SI),DX
IMUL3Q $19,DX,AX
MULQ 32(SI)
ADDQ AX,R13
ADCQ DX,R14
MOVQ $REDMASK51,SI
SHLQ $13,CX,R8
ANDQ SI,CX
SHLQ $13,R9,R10
ANDQ SI,R9
ADDQ R8,R9
SHLQ $13,R11,R12
ANDQ SI,R11
ADDQ R10,R11
SHLQ $13,R13,R14
ANDQ SI,R13
ADDQ R12,R13
SHLQ $13,R15,BX
ANDQ SI,R15
ADDQ R14,R15
IMUL3Q $19,BX,DX
ADDQ DX,CX
MOVQ CX,DX
SHRQ $51,DX
ADDQ R9,DX
ANDQ SI,CX
MOVQ DX,R8
SHRQ $51,DX
ADDQ R11,DX
ANDQ SI,R8
MOVQ DX,R9
SHRQ $51,DX
ADDQ R13,DX
ANDQ SI,R9
MOVQ DX,AX
SHRQ $51,DX
ADDQ R15,DX
ANDQ SI,AX
MOVQ DX,R10
SHRQ $51,DX
IMUL3Q $19,DX,DX
ADDQ DX,CX
ANDQ SI,R10
MOVQ CX,0(DI)
MOVQ R8,8(DI)
MOVQ R9,16(DI)
MOVQ AX,24(DI)
MOVQ R10,32(DI)
RET
// 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 ChaCha20 implements the core ChaCha20 function as specified
// in https://tools.ietf.org/html/rfc7539#section-2.3.
package chacha20
import (
"crypto/cipher"
"encoding/binary"
"golang.org/x/crypto/internal/subtle"
)
// assert that *Cipher implements cipher.Stream
var _ cipher.Stream = (*Cipher)(nil)
// Cipher is a stateful instance of ChaCha20 using a particular key
// and nonce. A *Cipher implements the cipher.Stream interface.
type Cipher struct {
key [8]uint32
counter uint32 // incremented after each block
nonce [3]uint32
buf [bufSize]byte // buffer for unused keystream bytes
len int // number of unused keystream bytes at end of buf
}
// New creates a new ChaCha20 stream cipher with the given key and nonce.
// The initial counter value is set to 0.
func New(key [8]uint32, nonce [3]uint32) *Cipher {
return &Cipher{key: key, nonce: nonce}
}
// ChaCha20 constants spelling "expand 32-byte k"
const (
j0 uint32 = 0x61707865
j1 uint32 = 0x3320646e
j2 uint32 = 0x79622d32
j3 uint32 = 0x6b206574
)
func quarterRound(a, b, c, d uint32) (uint32, uint32, uint32, uint32) {
a += b
d ^= a
d = (d << 16) | (d >> 16)
c += d
b ^= c
b = (b << 12) | (b >> 20)
a += b
d ^= a
d = (d << 8) | (d >> 24)
c += d
b ^= c
b = (b << 7) | (b >> 25)
return a, b, c, d
}
// XORKeyStream XORs each byte in the given slice with a byte from the
// cipher's key stream. Dst and src must overlap entirely or not at all.
//
// If len(dst) < len(src), XORKeyStream will panic. It is acceptable
// to pass a dst bigger than src, and in that case, XORKeyStream will
// only update dst[:len(src)] and will not touch the rest of dst.
//
// Multiple calls to XORKeyStream behave as if the concatenation of
// the src buffers was passed in a single run. That is, Cipher
// maintains state and does not reset at each XORKeyStream call.
func (s *Cipher) XORKeyStream(dst, src []byte) {
if len(dst) < len(src) {
panic("chacha20: output smaller than input")
}
if subtle.InexactOverlap(dst[:len(src)], src) {
panic("chacha20: invalid buffer overlap")
}
// xor src with buffered keystream first
if s.len != 0 {
buf := s.buf[len(s.buf)-s.len:]
if len(src) < len(buf) {
buf = buf[:len(src)]
}
td, ts := dst[:len(buf)], src[:len(buf)] // BCE hint
for i, b := range buf {
td[i] = ts[i] ^ b
}
s.len -= len(buf)
if s.len != 0 {
return
}
s.buf = [len(s.buf)]byte{} // zero the empty buffer
src = src[len(buf):]
dst = dst[len(buf):]
}
if len(src) == 0 {
return
}
if haveAsm {
if uint64(len(src))+uint64(s.counter)*64 > (1<<38)-64 {
panic("chacha20: counter overflow")
}
s.xorKeyStreamAsm(dst, src)
return
}
// set up a 64-byte buffer to pad out the final block if needed
// (hoisted out of the main loop to avoid spills)
rem := len(src) % 64 // length of final block
fin := len(src) - rem // index of final block
if rem > 0 {
copy(s.buf[len(s.buf)-64:], src[fin:])
}
// pre-calculate most of the first round
s1, s5, s9, s13 := quarterRound(j1, s.key[1], s.key[5], s.nonce[0])
s2, s6, s10, s14 := quarterRound(j2, s.key[2], s.key[6], s.nonce[1])
s3, s7, s11, s15 := quarterRound(j3, s.key[3], s.key[7], s.nonce[2])
n := len(src)
src, dst = src[:n:n], dst[:n:n] // BCE hint
for i := 0; i < n; i += 64 {
// calculate the remainder of the first round
s0, s4, s8, s12 := quarterRound(j0, s.key[0], s.key[4], s.counter)
// execute the second round
x0, x5, x10, x15 := quarterRound(s0, s5, s10, s15)
x1, x6, x11, x12 := quarterRound(s1, s6, s11, s12)
x2, x7, x8, x13 := quarterRound(s2, s7, s8, s13)
x3, x4, x9, x14 := quarterRound(s3, s4, s9, s14)
// execute the remaining 18 rounds
for i := 0; i < 9; i++ {
x0, x4, x8, x12 = quarterRound(x0, x4, x8, x12)
x1, x5, x9, x13 = quarterRound(x1, x5, x9, x13)
x2, x6, x10, x14 = quarterRound(x2, x6, x10, x14)
x3, x7, x11, x15 = quarterRound(x3, x7, x11, x15)
x0, x5, x10, x15 = quarterRound(x0, x5, x10, x15)
x1, x6, x11, x12 = quarterRound(x1, x6, x11, x12)
x2, x7, x8, x13 = quarterRound(x2, x7, x8, x13)
x3, x4, x9, x14 = quarterRound(x3, x4, x9, x14)
}
x0 += j0
x1 += j1
x2 += j2
x3 += j3
x4 += s.key[0]
x5 += s.key[1]
x6 += s.key[2]
x7 += s.key[3]
x8 += s.key[4]
x9 += s.key[5]
x10 += s.key[6]
x11 += s.key[7]
x12 += s.counter
x13 += s.nonce[0]
x14 += s.nonce[1]
x15 += s.nonce[2]
// increment the counter
s.counter += 1
if s.counter == 0 {
panic("chacha20: counter overflow")
}
// pad to 64 bytes if needed
in, out := src[i:], dst[i:]
if i == fin {
// src[fin:] has already been copied into s.buf before
// the main loop
in, out = s.buf[len(s.buf)-64:], s.buf[len(s.buf)-64:]
}
in, out = in[:64], out[:64] // BCE hint
// XOR the key stream with the source and write out the result
xor(out[0:], in[0:], x0)
xor(out[4:], in[4:], x1)
xor(out[8:], in[8:], x2)
xor(out[12:], in[12:], x3)
xor(out[16:], in[16:], x4)
xor(out[20:], in[20:], x5)
xor(out[24:], in[24:], x6)
xor(out[28:], in[28:], x7)
xor(out[32:], in[32:], x8)
xor(out[36:], in[36:], x9)
xor(out[40:], in[40:], x10)
xor(out[44:], in[44:], x11)
xor(out[48:], in[48:], x12)
xor(out[52:], in[52:], x13)
xor(out[56:], in[56:], x14)
xor(out[60:], in[60:], x15)
}
// copy any trailing bytes out of the buffer and into dst
if rem != 0 {
s.len = 64 - rem
copy(dst[fin:], s.buf[len(s.buf)-64:])
}
}
// Advance discards bytes in the key stream until the next 64 byte block
// boundary is reached and updates the counter accordingly. If the key
// stream is already at a block boundary no bytes will be discarded and
// the counter will be unchanged.
func (s *Cipher) Advance() {
s.len -= s.len % 64
if s.len == 0 {
s.buf = [len(s.buf)]byte{}
}
}
// XORKeyStream crypts bytes from in to out using the given key and counters.
// In and out must overlap entirely or not at all. Counter contains the raw
// ChaCha20 counter bytes (i.e. block counter followed by nonce).
func XORKeyStream(out, in []byte, counter *[16]byte, key *[32]byte) {
s := Cipher{
key: [8]uint32{
binary.LittleEndian.Uint32(key[0:4]),
binary.LittleEndian.Uint32(key[4:8]),
binary.LittleEndian.Uint32(key[8:12]),
binary.LittleEndian.Uint32(key[12:16]),
binary.LittleEndian.Uint32(key[16:20]),
binary.LittleEndian.Uint32(key[20:24]),
binary.LittleEndian.Uint32(key[24:28]),
binary.LittleEndian.Uint32(key[28:32]),
},
nonce: [3]uint32{
binary.LittleEndian.Uint32(counter[4:8]),
binary.LittleEndian.Uint32(counter[8:12]),
binary.LittleEndian.Uint32(counter[12:16]),
},
counter: binary.LittleEndian.Uint32(counter[0:4]),
}
s.XORKeyStream(out, in)
}
// HChaCha20 uses the ChaCha20 core to generate a derived key from a key and a
// nonce. It should only be used as part of the XChaCha20 construction.
func HChaCha20(key *[8]uint32, nonce *[4]uint32) [8]uint32 {
x0, x1, x2, x3 := j0, j1, j2, j3
x4, x5, x6, x7 := key[0], key[1], key[2], key[3]
x8, x9, x10, x11 := key[4], key[5], key[6], key[7]
x12, x13, x14, x15 := nonce[0], nonce[1], nonce[2], nonce[3]
for i := 0; i < 10; i++ {
x0, x4, x8, x12 = quarterRound(x0, x4, x8, x12)
x1, x5, x9, x13 = quarterRound(x1, x5, x9, x13)
x2, x6, x10, x14 = quarterRound(x2, x6, x10, x14)
x3, x7, x11, x15 = quarterRound(x3, x7, x11, x15)
x0, x5, x10, x15 = quarterRound(x0, x5, x10, x15)
x1, x6, x11, x12 = quarterRound(x1, x6, x11, x12)
x2, x7, x8, x13 = quarterRound(x2, x7, x8, x13)
x3, x4, x9, x14 = quarterRound(x3, x4, x9, x14)
}
var out [8]uint32
out[0], out[1], out[2], out[3] = x0, x1, x2, x3
out[4], out[5], out[6], out[7] = x12, x13, x14, x15
return out
}
// Copyright 2019 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.
// +build ppc64le,!gccgo,!appengine
package chacha20
import "encoding/binary"
const (
bufSize = 256
haveAsm = true
)
//go:noescape
func chaCha20_ctr32_vmx(out, inp *byte, len int, key *[8]uint32, counter *uint32)
func (c *Cipher) xorKeyStreamAsm(dst, src []byte) {
if len(src) >= bufSize {
chaCha20_ctr32_vmx(&dst[0], &src[0], len(src)-len(src)%bufSize, &c.key, &c.counter)
}
if len(src)%bufSize != 0 {
chaCha20_ctr32_vmx(&c.buf[0], &c.buf[0], bufSize, &c.key, &c.counter)
start := len(src) - len(src)%bufSize
ts, td, tb := src[start:], dst[start:], c.buf[:]
// Unroll loop to XOR 32 bytes per iteration.
for i := 0; i < len(ts)-32; i += 32 {
td, tb = td[:len(ts)], tb[:len(ts)] // bounds check elimination
s0 := binary.LittleEndian.Uint64(ts[0:8])
s1 := binary.LittleEndian.Uint64(ts[8:16])
s2 := binary.LittleEndian.Uint64(ts[16:24])
s3 := binary.LittleEndian.Uint64(ts[24:32])
b0 := binary.LittleEndian.Uint64(tb[0:8])
b1 := binary.LittleEndian.Uint64(tb[8:16])
b2 := binary.LittleEndian.Uint64(tb[16:24])
b3 := binary.LittleEndian.Uint64(tb[24:32])
binary.LittleEndian.PutUint64(td[0:8], s0^b0)
binary.LittleEndian.PutUint64(td[8:16], s1^b1)
binary.LittleEndian.PutUint64(td[16:24], s2^b2)
binary.LittleEndian.PutUint64(td[24:32], s3^b3)
ts, td, tb = ts[32:], td[32:], tb[32:]
}
td, tb = td[:len(ts)], tb[:len(ts)] // bounds check elimination
for i, v := range ts {
td[i] = tb[i] ^ v
}
c.len = bufSize - (len(src) % bufSize)
}
}
// Copyright 2019 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.
// +build !go1.13
package poly1305
// Generic fallbacks for the math/bits intrinsics, copied from
// src/math/bits/bits.go. They were added in Go 1.12, but Add64 and Sum64 had
// variable time fallbacks until Go 1.13.
func bitsAdd64(x, y, carry uint64) (sum, carryOut uint64) {
sum = x + y + carry
carryOut = ((x & y) | ((x | y) &^ sum)) >> 63
return
}
func bitsSub64(x, y, borrow uint64) (diff, borrowOut uint64) {
diff = x - y - borrow
borrowOut = ((^x & y) | (^(x ^ y) & diff)) >> 63
return
}
func bitsMul64(x, y uint64) (hi, lo uint64) {
const mask32 = 1<<32 - 1
x0 := x & mask32
x1 := x >> 32
y0 := y & mask32
y1 := y >> 32
w0 := x0 * y0
t := x1*y0 + w0>>32
w1 := t & mask32
w2 := t >> 32
w1 += x0 * y1
hi = x1*y1 + w2 + w1>>32
lo = x * y
return
}
// Copyright 2019 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.
// +build go1.13
package poly1305
import "math/bits"
func bitsAdd64(x, y, carry uint64) (sum, carryOut uint64) {
return bits.Add64(x, y, carry)
}
func bitsSub64(x, y, borrow uint64) (diff, borrowOut uint64) {
return bits.Sub64(x, y, borrow)
}
func bitsMul64(x, y uint64) (hi, lo uint64) {
return bits.Mul64(x, y)
}
This diff is collapsed.
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment