Update deps

This commit is contained in:
Frank Denis 2019-11-14 22:34:38 -05:00
parent d063a7959e
commit 568376ea13
39 changed files with 1832 additions and 1636 deletions

6
go.mod
View File

@ -21,9 +21,9 @@ require (
github.com/kardianos/service v1.0.0
github.com/miekg/dns v1.1.22
github.com/powerman/check v1.2.1
golang.org/x/crypto v0.0.0-20191029031824-8986dd9e96cf
golang.org/x/net v0.0.0-20191101175033-0deb6923b6d9
golang.org/x/sys v0.0.0-20191104094858-e8c54fb511f6
golang.org/x/crypto v0.0.0-20191112222119-e1110fd1c708
golang.org/x/net v0.0.0-20191112182307-2180aed22343
golang.org/x/sys v0.0.0-20191113165036-4c7a9d0fe056
gopkg.in/natefinch/lumberjack.v2 v2.0.0
gopkg.in/yaml.v2 v2.2.2 // indirect
)

12
go.sum
View File

@ -63,14 +63,14 @@ golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACk
golang.org/x/crypto v0.0.0-20190909091759-094676da4a83 h1:mgAKeshyNqWKdENOnQsg+8dRTwZFIwFaO3HNl52sweA=
golang.org/x/crypto v0.0.0-20190909091759-094676da4a83/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI=
golang.org/x/crypto v0.0.0-20190923035154-9ee001bba392/go.mod h1:/lpIB1dKB+9EgE3H3cr1v9wB50oz8l4C4h62xy7jSTY=
golang.org/x/crypto v0.0.0-20191029031824-8986dd9e96cf h1:fnPsqIDRbCSgumaMCRpoIoF2s4qxv0xSSS0BVZUE/ss=
golang.org/x/crypto v0.0.0-20191029031824-8986dd9e96cf/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
golang.org/x/crypto v0.0.0-20191112222119-e1110fd1c708 h1:pXVtWnwHkrWD9ru3sDxY/qFK/bfc0egRovX91EjWjf4=
golang.org/x/crypto v0.0.0-20191112222119-e1110fd1c708/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
golang.org/x/net v0.0.0-20190311183353-d8887717615a/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
golang.org/x/net v0.0.0-20190620200207-3b0461eec859/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
golang.org/x/net v0.0.0-20190923162816-aa69164e4478/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
golang.org/x/net v0.0.0-20191101175033-0deb6923b6d9 h1:DPz9iiH3YoKiKhX/ijjoZvT0VFwK2c6CWYWQ7Zyr8TU=
golang.org/x/net v0.0.0-20191101175033-0deb6923b6d9/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
golang.org/x/net v0.0.0-20191112182307-2180aed22343 h1:00ohfJ4K98s3m6BGUoBd8nyfp4Yl0GoIKvw5abItTjI=
golang.org/x/net v0.0.0-20191112182307-2180aed22343/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
golang.org/x/sync v0.0.0-20190423024810-112230192c58 h1:8gQV6CLnAEikrhgkHFbMAEhagSSnXWGV915qUMm9mrU=
golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
golang.org/x/sys v0.0.0-20190204203706-41f3e6584952/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
@ -82,8 +82,8 @@ golang.org/x/sys v0.0.0-20190909082730-f460065e899a/go.mod h1:h1NjWce9XRLGQEsW7w
golang.org/x/sys v0.0.0-20190922100055-0a153f010e69/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/sys v0.0.0-20190924154521-2837fb4f24fe/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/sys v0.0.0-20191029155521-f43be2a4598c/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/sys v0.0.0-20191104094858-e8c54fb511f6 h1:ZJUmhYTp8GbGC0ViZRc2U+MIYQ8xx9MscsdXnclfIhw=
golang.org/x/sys v0.0.0-20191104094858-e8c54fb511f6/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/sys v0.0.0-20191113165036-4c7a9d0fe056 h1:dHtDnRWQtSx0Hjq9kvKFpBh9uPPKfQN70NZZmvssGwk=
golang.org/x/sys v0.0.0-20191113165036-4c7a9d0fe056/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/text v0.3.0 h1:g61tztE5qeGQ89tm6NTjjM9VPIm088od1l6aSorWRWg=
golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
golang.org/x/text v0.3.2 h1:tW2bmiBqwgJj/UpqtC8EpXEZVYOwU0yG4iWbprSVAcs=

View File

@ -1,8 +0,0 @@
// 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
#define REDMASK51 0x0007FFFFFFFFFFFF

View File

@ -1,20 +0,0 @@
// 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

View File

@ -1,65 +0,0 @@
// 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

View File

@ -1,834 +1,95 @@
// Copyright 2013 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
// license that can be found in the LICENSE file.
// We have an implementation in amd64 assembly so this code is only run on
// non-amd64 platforms. The amd64 assembly does not support gccgo.
// +build !amd64 gccgo appengine
package curve25519
// Package curve25519 provides an implementation of the X25519 function, which
// performs scalar multiplication on the elliptic curve known as Curve25519.
// See RFC 7748.
package curve25519 // import "golang.org/x/crypto/curve25519"
import (
"encoding/binary"
"crypto/subtle"
"fmt"
)
// This code is a port of the public domain, "ref10" implementation of
// curve25519 from SUPERCOP 20130419 by D. J. Bernstein.
// ScalarMult sets dst to the product scalar * point.
//
// Deprecated: when provided a low-order point, ScalarMult will set dst to all
// zeroes, irrespective of the scalar. Instead, use the X25519 function, which
// will return an error.
func ScalarMult(dst, scalar, point *[32]byte) {
scalarMult(dst, scalar, point)
}
// fieldElement represents an element of the field GF(2^255 - 19). An element
// t, entries t[0]...t[9], represents the integer t[0]+2^26 t[1]+2^51 t[2]+2^77
// t[3]+2^102 t[4]+...+2^230 t[9]. Bounds on each t[i] vary depending on
// context.
type fieldElement [10]int32
// ScalarBaseMult sets dst to the product scalar * base where base is the
// standard generator.
//
// It is recommended to use the X25519 function with Basepoint instead, as
// copying into fixed size arrays can lead to unexpected bugs.
func ScalarBaseMult(dst, scalar *[32]byte) {
ScalarMult(dst, scalar, &basePoint)
}
func feZero(fe *fieldElement) {
for i := range fe {
fe[i] = 0
const (
// ScalarSize is the size of the scalar input to X25519.
ScalarSize = 32
// PointSize is the size of the point input to X25519.
PointSize = 32
)
// Basepoint is the canonical Curve25519 generator.
var Basepoint []byte
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}
func init() { Basepoint = basePoint[:] }
func checkBasepoint() {
if subtle.ConstantTimeCompare(Basepoint, []byte{
0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
}) != 1 {
panic("curve25519: global Basepoint value was modified")
}
}
func feOne(fe *fieldElement) {
feZero(fe)
fe[0] = 1
// X25519 returns the result of the scalar multiplication (scalar * point),
// according to RFC 7748, Section 5. scalar, point and the return value are
// slices of 32 bytes.
//
// scalar can be generated at random, for example with crypto/rand. point should
// be either Basepoint or the output of another X25519 call.
//
// If point is Basepoint (but not if it's a different slice with the same
// contents) a precomputed implementation might be used for performance.
func X25519(scalar, point []byte) ([]byte, error) {
// Outline the body of function, to let the allocation be inlined in the
// caller, and possibly avoid escaping to the heap.
var dst [32]byte
return x25519(&dst, scalar, point)
}
func feAdd(dst, a, b *fieldElement) {
for i := range dst {
dst[i] = a[i] + b[i]
func x25519(dst *[32]byte, scalar, point []byte) ([]byte, error) {
var in [32]byte
if l := len(scalar); l != 32 {
return nil, fmt.Errorf("bad scalar length: %d, expected %d", l, 32)
}
}
func feSub(dst, a, b *fieldElement) {
for i := range dst {
dst[i] = a[i] - b[i]
}
}
func feCopy(dst, src *fieldElement) {
for i := range dst {
dst[i] = src[i]
}
}
// feCSwap replaces (f,g) with (g,f) if b == 1; replaces (f,g) with (f,g) if b == 0.
//
// Preconditions: b in {0,1}.
func feCSwap(f, g *fieldElement, b int32) {
b = -b
for i := range f {
t := b & (f[i] ^ g[i])
f[i] ^= t
g[i] ^= t
}
}
// load3 reads a 24-bit, little-endian value from in.
func load3(in []byte) int64 {
var r int64
r = int64(in[0])
r |= int64(in[1]) << 8
r |= int64(in[2]) << 16
return r
}
// load4 reads a 32-bit, little-endian value from in.
func load4(in []byte) int64 {
return int64(binary.LittleEndian.Uint32(in))
}
func feFromBytes(dst *fieldElement, src *[32]byte) {
h0 := load4(src[:])
h1 := load3(src[4:]) << 6
h2 := load3(src[7:]) << 5
h3 := load3(src[10:]) << 3
h4 := load3(src[13:]) << 2
h5 := load4(src[16:])
h6 := load3(src[20:]) << 7
h7 := load3(src[23:]) << 5
h8 := load3(src[26:]) << 4
h9 := (load3(src[29:]) & 0x7fffff) << 2
var carry [10]int64
carry[9] = (h9 + 1<<24) >> 25
h0 += carry[9] * 19
h9 -= carry[9] << 25
carry[1] = (h1 + 1<<24) >> 25
h2 += carry[1]
h1 -= carry[1] << 25
carry[3] = (h3 + 1<<24) >> 25
h4 += carry[3]
h3 -= carry[3] << 25
carry[5] = (h5 + 1<<24) >> 25
h6 += carry[5]
h5 -= carry[5] << 25
carry[7] = (h7 + 1<<24) >> 25
h8 += carry[7]
h7 -= carry[7] << 25
carry[0] = (h0 + 1<<25) >> 26
h1 += carry[0]
h0 -= carry[0] << 26
carry[2] = (h2 + 1<<25) >> 26
h3 += carry[2]
h2 -= carry[2] << 26
carry[4] = (h4 + 1<<25) >> 26
h5 += carry[4]
h4 -= carry[4] << 26
carry[6] = (h6 + 1<<25) >> 26
h7 += carry[6]
h6 -= carry[6] << 26
carry[8] = (h8 + 1<<25) >> 26
h9 += carry[8]
h8 -= carry[8] << 26
dst[0] = int32(h0)
dst[1] = int32(h1)
dst[2] = int32(h2)
dst[3] = int32(h3)
dst[4] = int32(h4)
dst[5] = int32(h5)
dst[6] = int32(h6)
dst[7] = int32(h7)
dst[8] = int32(h8)
dst[9] = int32(h9)
}
// feToBytes marshals h to s.
// Preconditions:
// |h| bounded by 1.1*2^25,1.1*2^24,1.1*2^25,1.1*2^24,etc.
//
// Write p=2^255-19; q=floor(h/p).
// Basic claim: q = floor(2^(-255)(h + 19 2^(-25)h9 + 2^(-1))).
//
// Proof:
// Have |h|<=p so |q|<=1 so |19^2 2^(-255) q|<1/4.
// Also have |h-2^230 h9|<2^230 so |19 2^(-255)(h-2^230 h9)|<1/4.
//
// Write y=2^(-1)-19^2 2^(-255)q-19 2^(-255)(h-2^230 h9).
// Then 0<y<1.
//
// Write r=h-pq.
// Have 0<=r<=p-1=2^255-20.
// Thus 0<=r+19(2^-255)r<r+19(2^-255)2^255<=2^255-1.
//
// Write x=r+19(2^-255)r+y.
// Then 0<x<2^255 so floor(2^(-255)x) = 0 so floor(q+2^(-255)x) = q.
//
// Have q+2^(-255)x = 2^(-255)(h + 19 2^(-25) h9 + 2^(-1))
// so floor(2^(-255)(h + 19 2^(-25) h9 + 2^(-1))) = q.
func feToBytes(s *[32]byte, h *fieldElement) {
var carry [10]int32
q := (19*h[9] + (1 << 24)) >> 25
q = (h[0] + q) >> 26
q = (h[1] + q) >> 25
q = (h[2] + q) >> 26
q = (h[3] + q) >> 25
q = (h[4] + q) >> 26
q = (h[5] + q) >> 25
q = (h[6] + q) >> 26
q = (h[7] + q) >> 25
q = (h[8] + q) >> 26
q = (h[9] + q) >> 25
// Goal: Output h-(2^255-19)q, which is between 0 and 2^255-20.
h[0] += 19 * q
// Goal: Output h-2^255 q, which is between 0 and 2^255-20.
carry[0] = h[0] >> 26
h[1] += carry[0]
h[0] -= carry[0] << 26
carry[1] = h[1] >> 25
h[2] += carry[1]
h[1] -= carry[1] << 25
carry[2] = h[2] >> 26
h[3] += carry[2]
h[2] -= carry[2] << 26
carry[3] = h[3] >> 25
h[4] += carry[3]
h[3] -= carry[3] << 25
carry[4] = h[4] >> 26
h[5] += carry[4]
h[4] -= carry[4] << 26
carry[5] = h[5] >> 25
h[6] += carry[5]
h[5] -= carry[5] << 25
carry[6] = h[6] >> 26
h[7] += carry[6]
h[6] -= carry[6] << 26
carry[7] = h[7] >> 25
h[8] += carry[7]
h[7] -= carry[7] << 25
carry[8] = h[8] >> 26
h[9] += carry[8]
h[8] -= carry[8] << 26
carry[9] = h[9] >> 25
h[9] -= carry[9] << 25
// h10 = carry9
// Goal: Output h[0]+...+2^255 h10-2^255 q, which is between 0 and 2^255-20.
// Have h[0]+...+2^230 h[9] between 0 and 2^255-1;
// evidently 2^255 h10-2^255 q = 0.
// Goal: Output h[0]+...+2^230 h[9].
s[0] = byte(h[0] >> 0)
s[1] = byte(h[0] >> 8)
s[2] = byte(h[0] >> 16)
s[3] = byte((h[0] >> 24) | (h[1] << 2))
s[4] = byte(h[1] >> 6)
s[5] = byte(h[1] >> 14)
s[6] = byte((h[1] >> 22) | (h[2] << 3))
s[7] = byte(h[2] >> 5)
s[8] = byte(h[2] >> 13)
s[9] = byte((h[2] >> 21) | (h[3] << 5))
s[10] = byte(h[3] >> 3)
s[11] = byte(h[3] >> 11)
s[12] = byte((h[3] >> 19) | (h[4] << 6))
s[13] = byte(h[4] >> 2)
s[14] = byte(h[4] >> 10)
s[15] = byte(h[4] >> 18)
s[16] = byte(h[5] >> 0)
s[17] = byte(h[5] >> 8)
s[18] = byte(h[5] >> 16)
s[19] = byte((h[5] >> 24) | (h[6] << 1))
s[20] = byte(h[6] >> 7)
s[21] = byte(h[6] >> 15)
s[22] = byte((h[6] >> 23) | (h[7] << 3))
s[23] = byte(h[7] >> 5)
s[24] = byte(h[7] >> 13)
s[25] = byte((h[7] >> 21) | (h[8] << 4))
s[26] = byte(h[8] >> 4)
s[27] = byte(h[8] >> 12)
s[28] = byte((h[8] >> 20) | (h[9] << 6))
s[29] = byte(h[9] >> 2)
s[30] = byte(h[9] >> 10)
s[31] = byte(h[9] >> 18)
}
// feMul calculates h = f * g
// Can overlap h with f or g.
//
// Preconditions:
// |f| bounded by 1.1*2^26,1.1*2^25,1.1*2^26,1.1*2^25,etc.
// |g| bounded by 1.1*2^26,1.1*2^25,1.1*2^26,1.1*2^25,etc.
//
// Postconditions:
// |h| bounded by 1.1*2^25,1.1*2^24,1.1*2^25,1.1*2^24,etc.
//
// Notes on implementation strategy:
//
// Using schoolbook multiplication.
// Karatsuba would save a little in some cost models.
//
// Most multiplications by 2 and 19 are 32-bit precomputations;
// cheaper than 64-bit postcomputations.
//
// There is one remaining multiplication by 19 in the carry chain;
// one *19 precomputation can be merged into this,
// but the resulting data flow is considerably less clean.
//
// There are 12 carries below.
// 10 of them are 2-way parallelizable and vectorizable.
// Can get away with 11 carries, but then data flow is much deeper.
//
// With tighter constraints on inputs can squeeze carries into int32.
func feMul(h, f, g *fieldElement) {
f0 := f[0]
f1 := f[1]
f2 := f[2]
f3 := f[3]
f4 := f[4]
f5 := f[5]
f6 := f[6]
f7 := f[7]
f8 := f[8]
f9 := f[9]
g0 := g[0]
g1 := g[1]
g2 := g[2]
g3 := g[3]
g4 := g[4]
g5 := g[5]
g6 := g[6]
g7 := g[7]
g8 := g[8]
g9 := g[9]
g1_19 := 19 * g1 // 1.4*2^29
g2_19 := 19 * g2 // 1.4*2^30; still ok
g3_19 := 19 * g3
g4_19 := 19 * g4
g5_19 := 19 * g5
g6_19 := 19 * g6
g7_19 := 19 * g7
g8_19 := 19 * g8
g9_19 := 19 * g9
f1_2 := 2 * f1
f3_2 := 2 * f3
f5_2 := 2 * f5
f7_2 := 2 * f7
f9_2 := 2 * f9
f0g0 := int64(f0) * int64(g0)
f0g1 := int64(f0) * int64(g1)
f0g2 := int64(f0) * int64(g2)
f0g3 := int64(f0) * int64(g3)
f0g4 := int64(f0) * int64(g4)
f0g5 := int64(f0) * int64(g5)
f0g6 := int64(f0) * int64(g6)
f0g7 := int64(f0) * int64(g7)
f0g8 := int64(f0) * int64(g8)
f0g9 := int64(f0) * int64(g9)
f1g0 := int64(f1) * int64(g0)
f1g1_2 := int64(f1_2) * int64(g1)
f1g2 := int64(f1) * int64(g2)
f1g3_2 := int64(f1_2) * int64(g3)
f1g4 := int64(f1) * int64(g4)
f1g5_2 := int64(f1_2) * int64(g5)
f1g6 := int64(f1) * int64(g6)
f1g7_2 := int64(f1_2) * int64(g7)
f1g8 := int64(f1) * int64(g8)
f1g9_38 := int64(f1_2) * int64(g9_19)
f2g0 := int64(f2) * int64(g0)
f2g1 := int64(f2) * int64(g1)
f2g2 := int64(f2) * int64(g2)
f2g3 := int64(f2) * int64(g3)
f2g4 := int64(f2) * int64(g4)
f2g5 := int64(f2) * int64(g5)
f2g6 := int64(f2) * int64(g6)
f2g7 := int64(f2) * int64(g7)
f2g8_19 := int64(f2) * int64(g8_19)
f2g9_19 := int64(f2) * int64(g9_19)
f3g0 := int64(f3) * int64(g0)
f3g1_2 := int64(f3_2) * int64(g1)
f3g2 := int64(f3) * int64(g2)
f3g3_2 := int64(f3_2) * int64(g3)
f3g4 := int64(f3) * int64(g4)
f3g5_2 := int64(f3_2) * int64(g5)
f3g6 := int64(f3) * int64(g6)
f3g7_38 := int64(f3_2) * int64(g7_19)
f3g8_19 := int64(f3) * int64(g8_19)
f3g9_38 := int64(f3_2) * int64(g9_19)
f4g0 := int64(f4) * int64(g0)
f4g1 := int64(f4) * int64(g1)
f4g2 := int64(f4) * int64(g2)
f4g3 := int64(f4) * int64(g3)
f4g4 := int64(f4) * int64(g4)
f4g5 := int64(f4) * int64(g5)
f4g6_19 := int64(f4) * int64(g6_19)
f4g7_19 := int64(f4) * int64(g7_19)
f4g8_19 := int64(f4) * int64(g8_19)
f4g9_19 := int64(f4) * int64(g9_19)
f5g0 := int64(f5) * int64(g0)
f5g1_2 := int64(f5_2) * int64(g1)
f5g2 := int64(f5) * int64(g2)
f5g3_2 := int64(f5_2) * int64(g3)
f5g4 := int64(f5) * int64(g4)
f5g5_38 := int64(f5_2) * int64(g5_19)
f5g6_19 := int64(f5) * int64(g6_19)
f5g7_38 := int64(f5_2) * int64(g7_19)
f5g8_19 := int64(f5) * int64(g8_19)
f5g9_38 := int64(f5_2) * int64(g9_19)
f6g0 := int64(f6) * int64(g0)
f6g1 := int64(f6) * int64(g1)
f6g2 := int64(f6) * int64(g2)
f6g3 := int64(f6) * int64(g3)
f6g4_19 := int64(f6) * int64(g4_19)
f6g5_19 := int64(f6) * int64(g5_19)
f6g6_19 := int64(f6) * int64(g6_19)
f6g7_19 := int64(f6) * int64(g7_19)
f6g8_19 := int64(f6) * int64(g8_19)
f6g9_19 := int64(f6) * int64(g9_19)
f7g0 := int64(f7) * int64(g0)
f7g1_2 := int64(f7_2) * int64(g1)
f7g2 := int64(f7) * int64(g2)
f7g3_38 := int64(f7_2) * int64(g3_19)
f7g4_19 := int64(f7) * int64(g4_19)
f7g5_38 := int64(f7_2) * int64(g5_19)
f7g6_19 := int64(f7) * int64(g6_19)
f7g7_38 := int64(f7_2) * int64(g7_19)
f7g8_19 := int64(f7) * int64(g8_19)
f7g9_38 := int64(f7_2) * int64(g9_19)
f8g0 := int64(f8) * int64(g0)
f8g1 := int64(f8) * int64(g1)
f8g2_19 := int64(f8) * int64(g2_19)
f8g3_19 := int64(f8) * int64(g3_19)
f8g4_19 := int64(f8) * int64(g4_19)
f8g5_19 := int64(f8) * int64(g5_19)
f8g6_19 := int64(f8) * int64(g6_19)
f8g7_19 := int64(f8) * int64(g7_19)
f8g8_19 := int64(f8) * int64(g8_19)
f8g9_19 := int64(f8) * int64(g9_19)
f9g0 := int64(f9) * int64(g0)
f9g1_38 := int64(f9_2) * int64(g1_19)
f9g2_19 := int64(f9) * int64(g2_19)
f9g3_38 := int64(f9_2) * int64(g3_19)
f9g4_19 := int64(f9) * int64(g4_19)
f9g5_38 := int64(f9_2) * int64(g5_19)
f9g6_19 := int64(f9) * int64(g6_19)
f9g7_38 := int64(f9_2) * int64(g7_19)
f9g8_19 := int64(f9) * int64(g8_19)
f9g9_38 := int64(f9_2) * int64(g9_19)
h0 := f0g0 + f1g9_38 + f2g8_19 + f3g7_38 + f4g6_19 + f5g5_38 + f6g4_19 + f7g3_38 + f8g2_19 + f9g1_38
h1 := f0g1 + f1g0 + f2g9_19 + f3g8_19 + f4g7_19 + f5g6_19 + f6g5_19 + f7g4_19 + f8g3_19 + f9g2_19
h2 := f0g2 + f1g1_2 + f2g0 + f3g9_38 + f4g8_19 + f5g7_38 + f6g6_19 + f7g5_38 + f8g4_19 + f9g3_38
h3 := f0g3 + f1g2 + f2g1 + f3g0 + f4g9_19 + f5g8_19 + f6g7_19 + f7g6_19 + f8g5_19 + f9g4_19
h4 := f0g4 + f1g3_2 + f2g2 + f3g1_2 + f4g0 + f5g9_38 + f6g8_19 + f7g7_38 + f8g6_19 + f9g5_38
h5 := f0g5 + f1g4 + f2g3 + f3g2 + f4g1 + f5g0 + f6g9_19 + f7g8_19 + f8g7_19 + f9g6_19
h6 := f0g6 + f1g5_2 + f2g4 + f3g3_2 + f4g2 + f5g1_2 + f6g0 + f7g9_38 + f8g8_19 + f9g7_38
h7 := f0g7 + f1g6 + f2g5 + f3g4 + f4g3 + f5g2 + f6g1 + f7g0 + f8g9_19 + f9g8_19
h8 := f0g8 + f1g7_2 + f2g6 + f3g5_2 + f4g4 + f5g3_2 + f6g2 + f7g1_2 + f8g0 + f9g9_38
h9 := f0g9 + f1g8 + f2g7 + f3g6 + f4g5 + f5g4 + f6g3 + f7g2 + f8g1 + f9g0
var carry [10]int64
// |h0| <= (1.1*1.1*2^52*(1+19+19+19+19)+1.1*1.1*2^50*(38+38+38+38+38))
// i.e. |h0| <= 1.2*2^59; narrower ranges for h2, h4, h6, h8
// |h1| <= (1.1*1.1*2^51*(1+1+19+19+19+19+19+19+19+19))
// i.e. |h1| <= 1.5*2^58; narrower ranges for h3, h5, h7, h9
carry[0] = (h0 + (1 << 25)) >> 26
h1 += carry[0]
h0 -= carry[0] << 26
carry[4] = (h4 + (1 << 25)) >> 26
h5 += carry[4]
h4 -= carry[4] << 26
// |h0| <= 2^25
// |h4| <= 2^25
// |h1| <= 1.51*2^58
// |h5| <= 1.51*2^58
carry[1] = (h1 + (1 << 24)) >> 25
h2 += carry[1]
h1 -= carry[1] << 25
carry[5] = (h5 + (1 << 24)) >> 25
h6 += carry[5]
h5 -= carry[5] << 25
// |h1| <= 2^24; from now on fits into int32
// |h5| <= 2^24; from now on fits into int32
// |h2| <= 1.21*2^59
// |h6| <= 1.21*2^59
carry[2] = (h2 + (1 << 25)) >> 26
h3 += carry[2]
h2 -= carry[2] << 26
carry[6] = (h6 + (1 << 25)) >> 26
h7 += carry[6]
h6 -= carry[6] << 26
// |h2| <= 2^25; from now on fits into int32 unchanged
// |h6| <= 2^25; from now on fits into int32 unchanged
// |h3| <= 1.51*2^58
// |h7| <= 1.51*2^58
carry[3] = (h3 + (1 << 24)) >> 25
h4 += carry[3]
h3 -= carry[3] << 25
carry[7] = (h7 + (1 << 24)) >> 25
h8 += carry[7]
h7 -= carry[7] << 25
// |h3| <= 2^24; from now on fits into int32 unchanged
// |h7| <= 2^24; from now on fits into int32 unchanged
// |h4| <= 1.52*2^33
// |h8| <= 1.52*2^33
carry[4] = (h4 + (1 << 25)) >> 26
h5 += carry[4]
h4 -= carry[4] << 26
carry[8] = (h8 + (1 << 25)) >> 26
h9 += carry[8]
h8 -= carry[8] << 26
// |h4| <= 2^25; from now on fits into int32 unchanged
// |h8| <= 2^25; from now on fits into int32 unchanged
// |h5| <= 1.01*2^24
// |h9| <= 1.51*2^58
carry[9] = (h9 + (1 << 24)) >> 25
h0 += carry[9] * 19
h9 -= carry[9] << 25
// |h9| <= 2^24; from now on fits into int32 unchanged
// |h0| <= 1.8*2^37
carry[0] = (h0 + (1 << 25)) >> 26
h1 += carry[0]
h0 -= carry[0] << 26
// |h0| <= 2^25; from now on fits into int32 unchanged
// |h1| <= 1.01*2^24
h[0] = int32(h0)
h[1] = int32(h1)
h[2] = int32(h2)
h[3] = int32(h3)
h[4] = int32(h4)
h[5] = int32(h5)
h[6] = int32(h6)
h[7] = int32(h7)
h[8] = int32(h8)
h[9] = int32(h9)
}
// feSquare calculates h = f*f. Can overlap h with f.
//
// Preconditions:
// |f| bounded by 1.1*2^26,1.1*2^25,1.1*2^26,1.1*2^25,etc.
//
// Postconditions:
// |h| bounded by 1.1*2^25,1.1*2^24,1.1*2^25,1.1*2^24,etc.
func feSquare(h, f *fieldElement) {
f0 := f[0]
f1 := f[1]
f2 := f[2]
f3 := f[3]
f4 := f[4]
f5 := f[5]
f6 := f[6]
f7 := f[7]
f8 := f[8]
f9 := f[9]
f0_2 := 2 * f0
f1_2 := 2 * f1
f2_2 := 2 * f2
f3_2 := 2 * f3
f4_2 := 2 * f4
f5_2 := 2 * f5
f6_2 := 2 * f6
f7_2 := 2 * f7
f5_38 := 38 * f5 // 1.31*2^30
f6_19 := 19 * f6 // 1.31*2^30
f7_38 := 38 * f7 // 1.31*2^30
f8_19 := 19 * f8 // 1.31*2^30
f9_38 := 38 * f9 // 1.31*2^30
f0f0 := int64(f0) * int64(f0)
f0f1_2 := int64(f0_2) * int64(f1)
f0f2_2 := int64(f0_2) * int64(f2)
f0f3_2 := int64(f0_2) * int64(f3)
f0f4_2 := int64(f0_2) * int64(f4)
f0f5_2 := int64(f0_2) * int64(f5)
f0f6_2 := int64(f0_2) * int64(f6)
f0f7_2 := int64(f0_2) * int64(f7)
f0f8_2 := int64(f0_2) * int64(f8)
f0f9_2 := int64(f0_2) * int64(f9)
f1f1_2 := int64(f1_2) * int64(f1)
f1f2_2 := int64(f1_2) * int64(f2)
f1f3_4 := int64(f1_2) * int64(f3_2)
f1f4_2 := int64(f1_2) * int64(f4)
f1f5_4 := int64(f1_2) * int64(f5_2)
f1f6_2 := int64(f1_2) * int64(f6)
f1f7_4 := int64(f1_2) * int64(f7_2)
f1f8_2 := int64(f1_2) * int64(f8)
f1f9_76 := int64(f1_2) * int64(f9_38)
f2f2 := int64(f2) * int64(f2)
f2f3_2 := int64(f2_2) * int64(f3)
f2f4_2 := int64(f2_2) * int64(f4)
f2f5_2 := int64(f2_2) * int64(f5)
f2f6_2 := int64(f2_2) * int64(f6)
f2f7_2 := int64(f2_2) * int64(f7)
f2f8_38 := int64(f2_2) * int64(f8_19)
f2f9_38 := int64(f2) * int64(f9_38)
f3f3_2 := int64(f3_2) * int64(f3)
f3f4_2 := int64(f3_2) * int64(f4)
f3f5_4 := int64(f3_2) * int64(f5_2)
f3f6_2 := int64(f3_2) * int64(f6)
f3f7_76 := int64(f3_2) * int64(f7_38)
f3f8_38 := int64(f3_2) * int64(f8_19)
f3f9_76 := int64(f3_2) * int64(f9_38)
f4f4 := int64(f4) * int64(f4)
f4f5_2 := int64(f4_2) * int64(f5)
f4f6_38 := int64(f4_2) * int64(f6_19)
f4f7_38 := int64(f4) * int64(f7_38)
f4f8_38 := int64(f4_2) * int64(f8_19)
f4f9_38 := int64(f4) * int64(f9_38)
f5f5_38 := int64(f5) * int64(f5_38)
f5f6_38 := int64(f5_2) * int64(f6_19)
f5f7_76 := int64(f5_2) * int64(f7_38)
f5f8_38 := int64(f5_2) * int64(f8_19)
f5f9_76 := int64(f5_2) * int64(f9_38)
f6f6_19 := int64(f6) * int64(f6_19)
f6f7_38 := int64(f6) * int64(f7_38)
f6f8_38 := int64(f6_2) * int64(f8_19)
f6f9_38 := int64(f6) * int64(f9_38)
f7f7_38 := int64(f7) * int64(f7_38)
f7f8_38 := int64(f7_2) * int64(f8_19)
f7f9_76 := int64(f7_2) * int64(f9_38)
f8f8_19 := int64(f8) * int64(f8_19)
f8f9_38 := int64(f8) * int64(f9_38)
f9f9_38 := int64(f9) * int64(f9_38)
h0 := f0f0 + f1f9_76 + f2f8_38 + f3f7_76 + f4f6_38 + f5f5_38
h1 := f0f1_2 + f2f9_38 + f3f8_38 + f4f7_38 + f5f6_38
h2 := f0f2_2 + f1f1_2 + f3f9_76 + f4f8_38 + f5f7_76 + f6f6_19
h3 := f0f3_2 + f1f2_2 + f4f9_38 + f5f8_38 + f6f7_38
h4 := f0f4_2 + f1f3_4 + f2f2 + f5f9_76 + f6f8_38 + f7f7_38
h5 := f0f5_2 + f1f4_2 + f2f3_2 + f6f9_38 + f7f8_38
h6 := f0f6_2 + f1f5_4 + f2f4_2 + f3f3_2 + f7f9_76 + f8f8_19
h7 := f0f7_2 + f1f6_2 + f2f5_2 + f3f4_2 + f8f9_38
h8 := f0f8_2 + f1f7_4 + f2f6_2 + f3f5_4 + f4f4 + f9f9_38
h9 := f0f9_2 + f1f8_2 + f2f7_2 + f3f6_2 + f4f5_2
var carry [10]int64
carry[0] = (h0 + (1 << 25)) >> 26
h1 += carry[0]
h0 -= carry[0] << 26
carry[4] = (h4 + (1 << 25)) >> 26
h5 += carry[4]
h4 -= carry[4] << 26
carry[1] = (h1 + (1 << 24)) >> 25
h2 += carry[1]
h1 -= carry[1] << 25
carry[5] = (h5 + (1 << 24)) >> 25
h6 += carry[5]
h5 -= carry[5] << 25
carry[2] = (h2 + (1 << 25)) >> 26
h3 += carry[2]
h2 -= carry[2] << 26
carry[6] = (h6 + (1 << 25)) >> 26
h7 += carry[6]
h6 -= carry[6] << 26
carry[3] = (h3 + (1 << 24)) >> 25
h4 += carry[3]
h3 -= carry[3] << 25
carry[7] = (h7 + (1 << 24)) >> 25
h8 += carry[7]
h7 -= carry[7] << 25
carry[4] = (h4 + (1 << 25)) >> 26
h5 += carry[4]
h4 -= carry[4] << 26
carry[8] = (h8 + (1 << 25)) >> 26
h9 += carry[8]
h8 -= carry[8] << 26
carry[9] = (h9 + (1 << 24)) >> 25
h0 += carry[9] * 19
h9 -= carry[9] << 25
carry[0] = (h0 + (1 << 25)) >> 26
h1 += carry[0]
h0 -= carry[0] << 26
h[0] = int32(h0)
h[1] = int32(h1)
h[2] = int32(h2)
h[3] = int32(h3)
h[4] = int32(h4)
h[5] = int32(h5)
h[6] = int32(h6)
h[7] = int32(h7)
h[8] = int32(h8)
h[9] = int32(h9)
}
// feMul121666 calculates h = f * 121666. Can overlap h with f.
//
// Preconditions:
// |f| bounded by 1.1*2^26,1.1*2^25,1.1*2^26,1.1*2^25,etc.
//
// Postconditions:
// |h| bounded by 1.1*2^25,1.1*2^24,1.1*2^25,1.1*2^24,etc.
func feMul121666(h, f *fieldElement) {
h0 := int64(f[0]) * 121666
h1 := int64(f[1]) * 121666
h2 := int64(f[2]) * 121666
h3 := int64(f[3]) * 121666
h4 := int64(f[4]) * 121666
h5 := int64(f[5]) * 121666
h6 := int64(f[6]) * 121666
h7 := int64(f[7]) * 121666
h8 := int64(f[8]) * 121666
h9 := int64(f[9]) * 121666
var carry [10]int64
carry[9] = (h9 + (1 << 24)) >> 25
h0 += carry[9] * 19
h9 -= carry[9] << 25
carry[1] = (h1 + (1 << 24)) >> 25
h2 += carry[1]
h1 -= carry[1] << 25
carry[3] = (h3 + (1 << 24)) >> 25
h4 += carry[3]
h3 -= carry[3] << 25
carry[5] = (h5 + (1 << 24)) >> 25
h6 += carry[5]
h5 -= carry[5] << 25
carry[7] = (h7 + (1 << 24)) >> 25
h8 += carry[7]
h7 -= carry[7] << 25
carry[0] = (h0 + (1 << 25)) >> 26
h1 += carry[0]
h0 -= carry[0] << 26
carry[2] = (h2 + (1 << 25)) >> 26
h3 += carry[2]
h2 -= carry[2] << 26
carry[4] = (h4 + (1 << 25)) >> 26
h5 += carry[4]
h4 -= carry[4] << 26
carry[6] = (h6 + (1 << 25)) >> 26
h7 += carry[6]
h6 -= carry[6] << 26
carry[8] = (h8 + (1 << 25)) >> 26
h9 += carry[8]
h8 -= carry[8] << 26
h[0] = int32(h0)
h[1] = int32(h1)
h[2] = int32(h2)
h[3] = int32(h3)
h[4] = int32(h4)
h[5] = int32(h5)
h[6] = int32(h6)
h[7] = int32(h7)
h[8] = int32(h8)
h[9] = int32(h9)
}
// feInvert sets out = z^-1.
func feInvert(out, z *fieldElement) {
var t0, t1, t2, t3 fieldElement
var i int
feSquare(&t0, z)
for i = 1; i < 1; i++ {
feSquare(&t0, &t0)
}
feSquare(&t1, &t0)
for i = 1; i < 2; i++ {
feSquare(&t1, &t1)
}
feMul(&t1, z, &t1)
feMul(&t0, &t0, &t1)
feSquare(&t2, &t0)
for i = 1; i < 1; i++ {
feSquare(&t2, &t2)
}
feMul(&t1, &t1, &t2)
feSquare(&t2, &t1)
for i = 1; i < 5; i++ {
feSquare(&t2, &t2)
}
feMul(&t1, &t2, &t1)
feSquare(&t2, &t1)
for i = 1; i < 10; i++ {
feSquare(&t2, &t2)
}
feMul(&t2, &t2, &t1)
feSquare(&t3, &t2)
for i = 1; i < 20; i++ {
feSquare(&t3, &t3)
}
feMul(&t2, &t3, &t2)
feSquare(&t2, &t2)
for i = 1; i < 10; i++ {
feSquare(&t2, &t2)
}
feMul(&t1, &t2, &t1)
feSquare(&t2, &t1)
for i = 1; i < 50; i++ {
feSquare(&t2, &t2)
}
feMul(&t2, &t2, &t1)
feSquare(&t3, &t2)
for i = 1; i < 100; i++ {
feSquare(&t3, &t3)
}
feMul(&t2, &t3, &t2)
feSquare(&t2, &t2)
for i = 1; i < 50; i++ {
feSquare(&t2, &t2)
}
feMul(&t1, &t2, &t1)
feSquare(&t1, &t1)
for i = 1; i < 5; i++ {
feSquare(&t1, &t1)
}
feMul(out, &t1, &t0)
}
func scalarMult(out, in, base *[32]byte) {
var e [32]byte
copy(e[:], in[:])
e[0] &= 248
e[31] &= 127
e[31] |= 64
var x1, x2, z2, x3, z3, tmp0, tmp1 fieldElement
feFromBytes(&x1, base)
feOne(&x2)
feCopy(&x3, &x1)
feOne(&z3)
swap := int32(0)
for pos := 254; pos >= 0; pos-- {
b := e[pos/8] >> uint(pos&7)
b &= 1
swap ^= int32(b)
feCSwap(&x2, &x3, swap)
feCSwap(&z2, &z3, swap)
swap = int32(b)
feSub(&tmp0, &x3, &z3)
feSub(&tmp1, &x2, &z2)
feAdd(&x2, &x2, &z2)
feAdd(&z2, &x3, &z3)
feMul(&z3, &tmp0, &x2)
feMul(&z2, &z2, &tmp1)
feSquare(&tmp0, &tmp1)
feSquare(&tmp1, &x2)
feAdd(&x3, &z3, &z2)
feSub(&z2, &z3, &z2)
feMul(&x2, &tmp1, &tmp0)
feSub(&tmp1, &tmp1, &tmp0)
feSquare(&z2, &z2)
feMul121666(&z3, &tmp1)
feSquare(&x3, &x3)
feAdd(&tmp0, &tmp0, &z3)
feMul(&z3, &x1, &z2)
feMul(&z2, &tmp1, &tmp0)
}
feCSwap(&x2, &x3, swap)
feCSwap(&z2, &z3, swap)
feInvert(&z2, &z2)
feMul(&x2, &x2, &z2)
feToBytes(out, &x2)
if l := len(point); l != 32 {
return nil, fmt.Errorf("bad point length: %d, expected %d", l, 32)
}
copy(in[:], scalar)
if &point[0] == &Basepoint[0] {
checkBasepoint()
ScalarBaseMult(dst, &in)
} else {
var base, zero [32]byte
copy(base[:], point)
ScalarMult(dst, &in, &base)
if subtle.ConstantTimeCompare(dst[:], zero[:]) == 1 {
return nil, fmt.Errorf("bad input point: low order point")
}
}
return dst[:], nil
}

View File

@ -2,7 +2,7 @@
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// +build amd64,!gccgo,!appengine
// +build amd64,!gccgo,!appengine,!purego
package curve25519

View File

@ -5,9 +5,84 @@
// 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
// +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)
TEXT ·ladderstep(SB),0,$296-8
@ -1375,3 +1450,344 @@ TEXT ·ladderstep(SB),0,$296-8
MOVQ AX,104(DI)
MOVQ R10,112(DI)
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

View File

@ -0,0 +1,828 @@
// Copyright 2013 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package curve25519
import "encoding/binary"
// This code is a port of the public domain, "ref10" implementation of
// curve25519 from SUPERCOP 20130419 by D. J. Bernstein.
// fieldElement represents an element of the field GF(2^255 - 19). An element
// t, entries t[0]...t[9], represents the integer t[0]+2^26 t[1]+2^51 t[2]+2^77
// t[3]+2^102 t[4]+...+2^230 t[9]. Bounds on each t[i] vary depending on
// context.
type fieldElement [10]int32
func feZero(fe *fieldElement) {
for i := range fe {
fe[i] = 0
}
}
func feOne(fe *fieldElement) {
feZero(fe)
fe[0] = 1
}
func feAdd(dst, a, b *fieldElement) {
for i := range dst {
dst[i] = a[i] + b[i]
}
}
func feSub(dst, a, b *fieldElement) {
for i := range dst {
dst[i] = a[i] - b[i]
}
}
func feCopy(dst, src *fieldElement) {
for i := range dst {
dst[i] = src[i]
}
}
// feCSwap replaces (f,g) with (g,f) if b == 1; replaces (f,g) with (f,g) if b == 0.
//
// Preconditions: b in {0,1}.
func feCSwap(f, g *fieldElement, b int32) {
b = -b
for i := range f {
t := b & (f[i] ^ g[i])
f[i] ^= t
g[i] ^= t
}
}
// load3 reads a 24-bit, little-endian value from in.
func load3(in []byte) int64 {
var r int64
r = int64(in[0])
r |= int64(in[1]) << 8
r |= int64(in[2]) << 16
return r
}
// load4 reads a 32-bit, little-endian value from in.
func load4(in []byte) int64 {
return int64(binary.LittleEndian.Uint32(in))
}
func feFromBytes(dst *fieldElement, src *[32]byte) {
h0 := load4(src[:])
h1 := load3(src[4:]) << 6
h2 := load3(src[7:]) << 5
h3 := load3(src[10:]) << 3
h4 := load3(src[13:]) << 2
h5 := load4(src[16:])
h6 := load3(src[20:]) << 7
h7 := load3(src[23:]) << 5
h8 := load3(src[26:]) << 4
h9 := (load3(src[29:]) & 0x7fffff) << 2
var carry [10]int64
carry[9] = (h9 + 1<<24) >> 25
h0 += carry[9] * 19
h9 -= carry[9] << 25
carry[1] = (h1 + 1<<24) >> 25
h2 += carry[1]
h1 -= carry[1] << 25
carry[3] = (h3 + 1<<24) >> 25
h4 += carry[3]
h3 -= carry[3] << 25
carry[5] = (h5 + 1<<24) >> 25
h6 += carry[5]
h5 -= carry[5] << 25
carry[7] = (h7 + 1<<24) >> 25
h8 += carry[7]
h7 -= carry[7] << 25
carry[0] = (h0 + 1<<25) >> 26
h1 += carry[0]
h0 -= carry[0] << 26
carry[2] = (h2 + 1<<25) >> 26
h3 += carry[2]
h2 -= carry[2] << 26
carry[4] = (h4 + 1<<25) >> 26
h5 += carry[4]
h4 -= carry[4] << 26
carry[6] = (h6 + 1<<25) >> 26
h7 += carry[6]
h6 -= carry[6] << 26
carry[8] = (h8 + 1<<25) >> 26
h9 += carry[8]
h8 -= carry[8] << 26
dst[0] = int32(h0)
dst[1] = int32(h1)
dst[2] = int32(h2)
dst[3] = int32(h3)
dst[4] = int32(h4)
dst[5] = int32(h5)
dst[6] = int32(h6)
dst[7] = int32(h7)
dst[8] = int32(h8)
dst[9] = int32(h9)
}
// feToBytes marshals h to s.
// Preconditions:
// |h| bounded by 1.1*2^25,1.1*2^24,1.1*2^25,1.1*2^24,etc.
//
// Write p=2^255-19; q=floor(h/p).
// Basic claim: q = floor(2^(-255)(h + 19 2^(-25)h9 + 2^(-1))).
//
// Proof:
// Have |h|<=p so |q|<=1 so |19^2 2^(-255) q|<1/4.
// Also have |h-2^230 h9|<2^230 so |19 2^(-255)(h-2^230 h9)|<1/4.
//
// Write y=2^(-1)-19^2 2^(-255)q-19 2^(-255)(h-2^230 h9).
// Then 0<y<1.
//
// Write r=h-pq.
// Have 0<=r<=p-1=2^255-20.
// Thus 0<=r+19(2^-255)r<r+19(2^-255)2^255<=2^255-1.
//
// Write x=r+19(2^-255)r+y.
// Then 0<x<2^255 so floor(2^(-255)x) = 0 so floor(q+2^(-255)x) = q.
//
// Have q+2^(-255)x = 2^(-255)(h + 19 2^(-25) h9 + 2^(-1))
// so floor(2^(-255)(h + 19 2^(-25) h9 + 2^(-1))) = q.
func feToBytes(s *[32]byte, h *fieldElement) {
var carry [10]int32
q := (19*h[9] + (1 << 24)) >> 25
q = (h[0] + q) >> 26
q = (h[1] + q) >> 25
q = (h[2] + q) >> 26
q = (h[3] + q) >> 25
q = (h[4] + q) >> 26
q = (h[5] + q) >> 25
q = (h[6] + q) >> 26
q = (h[7] + q) >> 25
q = (h[8] + q) >> 26
q = (h[9] + q) >> 25
// Goal: Output h-(2^255-19)q, which is between 0 and 2^255-20.
h[0] += 19 * q
// Goal: Output h-2^255 q, which is between 0 and 2^255-20.
carry[0] = h[0] >> 26
h[1] += carry[0]
h[0] -= carry[0] << 26
carry[1] = h[1] >> 25
h[2] += carry[1]
h[1] -= carry[1] << 25
carry[2] = h[2] >> 26
h[3] += carry[2]
h[2] -= carry[2] << 26
carry[3] = h[3] >> 25
h[4] += carry[3]
h[3] -= carry[3] << 25
carry[4] = h[4] >> 26
h[5] += carry[4]
h[4] -= carry[4] << 26
carry[5] = h[5] >> 25
h[6] += carry[5]
h[5] -= carry[5] << 25
carry[6] = h[6] >> 26
h[7] += carry[6]
h[6] -= carry[6] << 26
carry[7] = h[7] >> 25
h[8] += carry[7]
h[7] -= carry[7] << 25
carry[8] = h[8] >> 26
h[9] += carry[8]
h[8] -= carry[8] << 26
carry[9] = h[9] >> 25
h[9] -= carry[9] << 25
// h10 = carry9
// Goal: Output h[0]+...+2^255 h10-2^255 q, which is between 0 and 2^255-20.
// Have h[0]+...+2^230 h[9] between 0 and 2^255-1;
// evidently 2^255 h10-2^255 q = 0.
// Goal: Output h[0]+...+2^230 h[9].
s[0] = byte(h[0] >> 0)
s[1] = byte(h[0] >> 8)
s[2] = byte(h[0] >> 16)
s[3] = byte((h[0] >> 24) | (h[1] << 2))
s[4] = byte(h[1] >> 6)
s[5] = byte(h[1] >> 14)
s[6] = byte((h[1] >> 22) | (h[2] << 3))
s[7] = byte(h[2] >> 5)
s[8] = byte(h[2] >> 13)
s[9] = byte((h[2] >> 21) | (h[3] << 5))
s[10] = byte(h[3] >> 3)
s[11] = byte(h[3] >> 11)
s[12] = byte((h[3] >> 19) | (h[4] << 6))
s[13] = byte(h[4] >> 2)
s[14] = byte(h[4] >> 10)
s[15] = byte(h[4] >> 18)
s[16] = byte(h[5] >> 0)
s[17] = byte(h[5] >> 8)
s[18] = byte(h[5] >> 16)
s[19] = byte((h[5] >> 24) | (h[6] << 1))
s[20] = byte(h[6] >> 7)
s[21] = byte(h[6] >> 15)
s[22] = byte((h[6] >> 23) | (h[7] << 3))
s[23] = byte(h[7] >> 5)
s[24] = byte(h[7] >> 13)
s[25] = byte((h[7] >> 21) | (h[8] << 4))
s[26] = byte(h[8] >> 4)
s[27] = byte(h[8] >> 12)
s[28] = byte((h[8] >> 20) | (h[9] << 6))
s[29] = byte(h[9] >> 2)
s[30] = byte(h[9] >> 10)
s[31] = byte(h[9] >> 18)
}
// feMul calculates h = f * g
// Can overlap h with f or g.
//
// Preconditions:
// |f| bounded by 1.1*2^26,1.1*2^25,1.1*2^26,1.1*2^25,etc.
// |g| bounded by 1.1*2^26,1.1*2^25,1.1*2^26,1.1*2^25,etc.
//
// Postconditions:
// |h| bounded by 1.1*2^25,1.1*2^24,1.1*2^25,1.1*2^24,etc.
//
// Notes on implementation strategy:
//
// Using schoolbook multiplication.
// Karatsuba would save a little in some cost models.
//
// Most multiplications by 2 and 19 are 32-bit precomputations;
// cheaper than 64-bit postcomputations.
//
// There is one remaining multiplication by 19 in the carry chain;
// one *19 precomputation can be merged into this,
// but the resulting data flow is considerably less clean.
//
// There are 12 carries below.
// 10 of them are 2-way parallelizable and vectorizable.
// Can get away with 11 carries, but then data flow is much deeper.
//
// With tighter constraints on inputs can squeeze carries into int32.
func feMul(h, f, g *fieldElement) {
f0 := f[0]
f1 := f[1]
f2 := f[2]
f3 := f[3]
f4 := f[4]
f5 := f[5]
f6 := f[6]
f7 := f[7]
f8 := f[8]
f9 := f[9]
g0 := g[0]
g1 := g[1]
g2 := g[2]
g3 := g[3]
g4 := g[4]
g5 := g[5]
g6 := g[6]
g7 := g[7]
g8 := g[8]
g9 := g[9]
g1_19 := 19 * g1 // 1.4*2^29
g2_19 := 19 * g2 // 1.4*2^30; still ok
g3_19 := 19 * g3
g4_19 := 19 * g4
g5_19 := 19 * g5
g6_19 := 19 * g6
g7_19 := 19 * g7
g8_19 := 19 * g8
g9_19 := 19 * g9
f1_2 := 2 * f1
f3_2 := 2 * f3
f5_2 := 2 * f5
f7_2 := 2 * f7
f9_2 := 2 * f9
f0g0 := int64(f0) * int64(g0)
f0g1 := int64(f0) * int64(g1)
f0g2 := int64(f0) * int64(g2)
f0g3 := int64(f0) * int64(g3)
f0g4 := int64(f0) * int64(g4)
f0g5 := int64(f0) * int64(g5)
f0g6 := int64(f0) * int64(g6)
f0g7 := int64(f0) * int64(g7)
f0g8 := int64(f0) * int64(g8)
f0g9 := int64(f0) * int64(g9)
f1g0 := int64(f1) * int64(g0)
f1g1_2 := int64(f1_2) * int64(g1)
f1g2 := int64(f1) * int64(g2)
f1g3_2 := int64(f1_2) * int64(g3)
f1g4 := int64(f1) * int64(g4)
f1g5_2 := int64(f1_2) * int64(g5)
f1g6 := int64(f1) * int64(g6)
f1g7_2 := int64(f1_2) * int64(g7)
f1g8 := int64(f1) * int64(g8)
f1g9_38 := int64(f1_2) * int64(g9_19)
f2g0 := int64(f2) * int64(g0)
f2g1 := int64(f2) * int64(g1)
f2g2 := int64(f2) * int64(g2)
f2g3 := int64(f2) * int64(g3)
f2g4 := int64(f2) * int64(g4)
f2g5 := int64(f2) * int64(g5)
f2g6 := int64(f2) * int64(g6)
f2g7 := int64(f2) * int64(g7)
f2g8_19 := int64(f2) * int64(g8_19)
f2g9_19 := int64(f2) * int64(g9_19)
f3g0 := int64(f3) * int64(g0)
f3g1_2 := int64(f3_2) * int64(g1)
f3g2 := int64(f3) * int64(g2)
f3g3_2 := int64(f3_2) * int64(g3)
f3g4 := int64(f3) * int64(g4)
f3g5_2 := int64(f3_2) * int64(g5)
f3g6 := int64(f3) * int64(g6)
f3g7_38 := int64(f3_2) * int64(g7_19)
f3g8_19 := int64(f3) * int64(g8_19)
f3g9_38 := int64(f3_2) * int64(g9_19)
f4g0 := int64(f4) * int64(g0)
f4g1 := int64(f4) * int64(g1)
f4g2 := int64(f4) * int64(g2)
f4g3 := int64(f4) * int64(g3)
f4g4 := int64(f4) * int64(g4)
f4g5 := int64(f4) * int64(g5)
f4g6_19 := int64(f4) * int64(g6_19)
f4g7_19 := int64(f4) * int64(g7_19)
f4g8_19 := int64(f4) * int64(g8_19)
f4g9_19 := int64(f4) * int64(g9_19)
f5g0 := int64(f5) * int64(g0)
f5g1_2 := int64(f5_2) * int64(g1)
f5g2 := int64(f5) * int64(g2)
f5g3_2 := int64(f5_2) * int64(g3)
f5g4 := int64(f5) * int64(g4)
f5g5_38 := int64(f5_2) * int64(g5_19)
f5g6_19 := int64(f5) * int64(g6_19)
f5g7_38 := int64(f5_2) * int64(g7_19)
f5g8_19 := int64(f5) * int64(g8_19)
f5g9_38 := int64(f5_2) * int64(g9_19)
f6g0 := int64(f6) * int64(g0)
f6g1 := int64(f6) * int64(g1)
f6g2 := int64(f6) * int64(g2)
f6g3 := int64(f6) * int64(g3)
f6g4_19 := int64(f6) * int64(g4_19)
f6g5_19 := int64(f6) * int64(g5_19)
f6g6_19 := int64(f6) * int64(g6_19)
f6g7_19 := int64(f6) * int64(g7_19)
f6g8_19 := int64(f6) * int64(g8_19)
f6g9_19 := int64(f6) * int64(g9_19)
f7g0 := int64(f7) * int64(g0)
f7g1_2 := int64(f7_2) * int64(g1)
f7g2 := int64(f7) * int64(g2)
f7g3_38 := int64(f7_2) * int64(g3_19)
f7g4_19 := int64(f7) * int64(g4_19)
f7g5_38 := int64(f7_2) * int64(g5_19)
f7g6_19 := int64(f7) * int64(g6_19)
f7g7_38 := int64(f7_2) * int64(g7_19)
f7g8_19 := int64(f7) * int64(g8_19)
f7g9_38 := int64(f7_2) * int64(g9_19)
f8g0 := int64(f8) * int64(g0)
f8g1 := int64(f8) * int64(g1)
f8g2_19 := int64(f8) * int64(g2_19)
f8g3_19 := int64(f8) * int64(g3_19)
f8g4_19 := int64(f8) * int64(g4_19)
f8g5_19 := int64(f8) * int64(g5_19)
f8g6_19 := int64(f8) * int64(g6_19)
f8g7_19 := int64(f8) * int64(g7_19)
f8g8_19 := int64(f8) * int64(g8_19)
f8g9_19 := int64(f8) * int64(g9_19)
f9g0 := int64(f9) * int64(g0)
f9g1_38 := int64(f9_2) * int64(g1_19)
f9g2_19 := int64(f9) * int64(g2_19)
f9g3_38 := int64(f9_2) * int64(g3_19)
f9g4_19 := int64(f9) * int64(g4_19)
f9g5_38 := int64(f9_2) * int64(g5_19)
f9g6_19 := int64(f9) * int64(g6_19)
f9g7_38 := int64(f9_2) * int64(g7_19)
f9g8_19 := int64(f9) * int64(g8_19)
f9g9_38 := int64(f9_2) * int64(g9_19)
h0 := f0g0 + f1g9_38 + f2g8_19 + f3g7_38 + f4g6_19 + f5g5_38 + f6g4_19 + f7g3_38 + f8g2_19 + f9g1_38
h1 := f0g1 + f1g0 + f2g9_19 + f3g8_19 + f4g7_19 + f5g6_19 + f6g5_19 + f7g4_19 + f8g3_19 + f9g2_19
h2 := f0g2 + f1g1_2 + f2g0 + f3g9_38 + f4g8_19 + f5g7_38 + f6g6_19 + f7g5_38 + f8g4_19 + f9g3_38
h3 := f0g3 + f1g2 + f2g1 + f3g0 + f4g9_19 + f5g8_19 + f6g7_19 + f7g6_19 + f8g5_19 + f9g4_19
h4 := f0g4 + f1g3_2 + f2g2 + f3g1_2 + f4g0 + f5g9_38 + f6g8_19 + f7g7_38 + f8g6_19 + f9g5_38
h5 := f0g5 + f1g4 + f2g3 + f3g2 + f4g1 + f5g0 + f6g9_19 + f7g8_19 + f8g7_19 + f9g6_19
h6 := f0g6 + f1g5_2 + f2g4 + f3g3_2 + f4g2 + f5g1_2 + f6g0 + f7g9_38 + f8g8_19 + f9g7_38
h7 := f0g7 + f1g6 + f2g5 + f3g4 + f4g3 + f5g2 + f6g1 + f7g0 + f8g9_19 + f9g8_19
h8 := f0g8 + f1g7_2 + f2g6 + f3g5_2 + f4g4 + f5g3_2 + f6g2 + f7g1_2 + f8g0 + f9g9_38
h9 := f0g9 + f1g8 + f2g7 + f3g6 + f4g5 + f5g4 + f6g3 + f7g2 + f8g1 + f9g0
var carry [10]int64
// |h0| <= (1.1*1.1*2^52*(1+19+19+19+19)+1.1*1.1*2^50*(38+38+38+38+38))
// i.e. |h0| <= 1.2*2^59; narrower ranges for h2, h4, h6, h8
// |h1| <= (1.1*1.1*2^51*(1+1+19+19+19+19+19+19+19+19))
// i.e. |h1| <= 1.5*2^58; narrower ranges for h3, h5, h7, h9
carry[0] = (h0 + (1 << 25)) >> 26
h1 += carry[0]
h0 -= carry[0] << 26
carry[4] = (h4 + (1 << 25)) >> 26
h5 += carry[4]
h4 -= carry[4] << 26
// |h0| <= 2^25
// |h4| <= 2^25
// |h1| <= 1.51*2^58
// |h5| <= 1.51*2^58
carry[1] = (h1 + (1 << 24)) >> 25
h2 += carry[1]
h1 -= carry[1] << 25
carry[5] = (h5 + (1 << 24)) >> 25
h6 += carry[5]
h5 -= carry[5] << 25
// |h1| <= 2^24; from now on fits into int32
// |h5| <= 2^24; from now on fits into int32
// |h2| <= 1.21*2^59
// |h6| <= 1.21*2^59
carry[2] = (h2 + (1 << 25)) >> 26
h3 += carry[2]
h2 -= carry[2] << 26
carry[6] = (h6 + (1 << 25)) >> 26
h7 += carry[6]
h6 -= carry[6] << 26
// |h2| <= 2^25; from now on fits into int32 unchanged
// |h6| <= 2^25; from now on fits into int32 unchanged
// |h3| <= 1.51*2^58
// |h7| <= 1.51*2^58
carry[3] = (h3 + (1 << 24)) >> 25
h4 += carry[3]
h3 -= carry[3] << 25
carry[7] = (h7 + (1 << 24)) >> 25
h8 += carry[7]
h7 -= carry[7] << 25
// |h3| <= 2^24; from now on fits into int32 unchanged
// |h7| <= 2^24; from now on fits into int32 unchanged
// |h4| <= 1.52*2^33
// |h8| <= 1.52*2^33
carry[4] = (h4 + (1 << 25)) >> 26
h5 += carry[4]
h4 -= carry[4] << 26
carry[8] = (h8 + (1 << 25)) >> 26
h9 += carry[8]
h8 -= carry[8] << 26
// |h4| <= 2^25; from now on fits into int32 unchanged
// |h8| <= 2^25; from now on fits into int32 unchanged
// |h5| <= 1.01*2^24
// |h9| <= 1.51*2^58
carry[9] = (h9 + (1 << 24)) >> 25
h0 += carry[9] * 19
h9 -= carry[9] << 25
// |h9| <= 2^24; from now on fits into int32 unchanged
// |h0| <= 1.8*2^37
carry[0] = (h0 + (1 << 25)) >> 26
h1 += carry[0]
h0 -= carry[0] << 26
// |h0| <= 2^25; from now on fits into int32 unchanged
// |h1| <= 1.01*2^24
h[0] = int32(h0)
h[1] = int32(h1)
h[2] = int32(h2)
h[3] = int32(h3)
h[4] = int32(h4)
h[5] = int32(h5)
h[6] = int32(h6)
h[7] = int32(h7)
h[8] = int32(h8)
h[9] = int32(h9)
}
// feSquare calculates h = f*f. Can overlap h with f.
//
// Preconditions:
// |f| bounded by 1.1*2^26,1.1*2^25,1.1*2^26,1.1*2^25,etc.
//
// Postconditions:
// |h| bounded by 1.1*2^25,1.1*2^24,1.1*2^25,1.1*2^24,etc.
func feSquare(h, f *fieldElement) {
f0 := f[0]
f1 := f[1]
f2 := f[2]
f3 := f[3]
f4 := f[4]
f5 := f[5]
f6 := f[6]
f7 := f[7]
f8 := f[8]
f9 := f[9]
f0_2 := 2 * f0
f1_2 := 2 * f1
f2_2 := 2 * f2
f3_2 := 2 * f3
f4_2 := 2 * f4
f5_2 := 2 * f5
f6_2 := 2 * f6
f7_2 := 2 * f7
f5_38 := 38 * f5 // 1.31*2^30
f6_19 := 19 * f6 // 1.31*2^30
f7_38 := 38 * f7 // 1.31*2^30
f8_19 := 19 * f8 // 1.31*2^30
f9_38 := 38 * f9 // 1.31*2^30
f0f0 := int64(f0) * int64(f0)
f0f1_2 := int64(f0_2) * int64(f1)
f0f2_2 := int64(f0_2) * int64(f2)
f0f3_2 := int64(f0_2) * int64(f3)
f0f4_2 := int64(f0_2) * int64(f4)
f0f5_2 := int64(f0_2) * int64(f5)
f0f6_2 := int64(f0_2) * int64(f6)
f0f7_2 := int64(f0_2) * int64(f7)
f0f8_2 := int64(f0_2) * int64(f8)
f0f9_2 := int64(f0_2) * int64(f9)
f1f1_2 := int64(f1_2) * int64(f1)
f1f2_2 := int64(f1_2) * int64(f2)
f1f3_4 := int64(f1_2) * int64(f3_2)
f1f4_2 := int64(f1_2) * int64(f4)
f1f5_4 := int64(f1_2) * int64(f5_2)
f1f6_2 := int64(f1_2) * int64(f6)
f1f7_4 := int64(f1_2) * int64(f7_2)
f1f8_2 := int64(f1_2) * int64(f8)
f1f9_76 := int64(f1_2) * int64(f9_38)
f2f2 := int64(f2) * int64(f2)
f2f3_2 := int64(f2_2) * int64(f3)
f2f4_2 := int64(f2_2) * int64(f4)
f2f5_2 := int64(f2_2) * int64(f5)
f2f6_2 := int64(f2_2) * int64(f6)
f2f7_2 := int64(f2_2) * int64(f7)
f2f8_38 := int64(f2_2) * int64(f8_19)
f2f9_38 := int64(f2) * int64(f9_38)
f3f3_2 := int64(f3_2) * int64(f3)
f3f4_2 := int64(f3_2) * int64(f4)
f3f5_4 := int64(f3_2) * int64(f5_2)
f3f6_2 := int64(f3_2) * int64(f6)
f3f7_76 := int64(f3_2) * int64(f7_38)
f3f8_38 := int64(f3_2) * int64(f8_19)
f3f9_76 := int64(f3_2) * int64(f9_38)
f4f4 := int64(f4) * int64(f4)
f4f5_2 := int64(f4_2) * int64(f5)
f4f6_38 := int64(f4_2) * int64(f6_19)
f4f7_38 := int64(f4) * int64(f7_38)
f4f8_38 := int64(f4_2) * int64(f8_19)
f4f9_38 := int64(f4) * int64(f9_38)
f5f5_38 := int64(f5) * int64(f5_38)
f5f6_38 := int64(f5_2) * int64(f6_19)
f5f7_76 := int64(f5_2) * int64(f7_38)
f5f8_38 := int64(f5_2) * int64(f8_19)
f5f9_76 := int64(f5_2) * int64(f9_38)
f6f6_19 := int64(f6) * int64(f6_19)
f6f7_38 := int64(f6) * int64(f7_38)
f6f8_38 := int64(f6_2) * int64(f8_19)
f6f9_38 := int64(f6) * int64(f9_38)
f7f7_38 := int64(f7) * int64(f7_38)
f7f8_38 := int64(f7_2) * int64(f8_19)
f7f9_76 := int64(f7_2) * int64(f9_38)
f8f8_19 := int64(f8) * int64(f8_19)
f8f9_38 := int64(f8) * int64(f9_38)
f9f9_38 := int64(f9) * int64(f9_38)
h0 := f0f0 + f1f9_76 + f2f8_38 + f3f7_76 + f4f6_38 + f5f5_38
h1 := f0f1_2 + f2f9_38 + f3f8_38 + f4f7_38 + f5f6_38
h2 := f0f2_2 + f1f1_2 + f3f9_76 + f4f8_38 + f5f7_76 + f6f6_19
h3 := f0f3_2 + f1f2_2 + f4f9_38 + f5f8_38 + f6f7_38
h4 := f0f4_2 + f1f3_4 + f2f2 + f5f9_76 + f6f8_38 + f7f7_38
h5 := f0f5_2 + f1f4_2 + f2f3_2 + f6f9_38 + f7f8_38
h6 := f0f6_2 + f1f5_4 + f2f4_2 + f3f3_2 + f7f9_76 + f8f8_19
h7 := f0f7_2 + f1f6_2 + f2f5_2 + f3f4_2 + f8f9_38
h8 := f0f8_2 + f1f7_4 + f2f6_2 + f3f5_4 + f4f4 + f9f9_38
h9 := f0f9_2 + f1f8_2 + f2f7_2 + f3f6_2 + f4f5_2
var carry [10]int64
carry[0] = (h0 + (1 << 25)) >> 26
h1 += carry[0]
h0 -= carry[0] << 26
carry[4] = (h4 + (1 << 25)) >> 26
h5 += carry[4]
h4 -= carry[4] << 26
carry[1] = (h1 + (1 << 24)) >> 25
h2 += carry[1]
h1 -= carry[1] << 25
carry[5] = (h5 + (1 << 24)) >> 25
h6 += carry[5]
h5 -= carry[5] << 25
carry[2] = (h2 + (1 << 25)) >> 26
h3 += carry[2]
h2 -= carry[2] << 26
carry[6] = (h6 + (1 << 25)) >> 26
h7 += carry[6]
h6 -= carry[6] << 26
carry[3] = (h3 + (1 << 24)) >> 25
h4 += carry[3]
h3 -= carry[3] << 25
carry[7] = (h7 + (1 << 24)) >> 25
h8 += carry[7]
h7 -= carry[7] << 25
carry[4] = (h4 + (1 << 25)) >> 26
h5 += carry[4]
h4 -= carry[4] << 26
carry[8] = (h8 + (1 << 25)) >> 26
h9 += carry[8]
h8 -= carry[8] << 26
carry[9] = (h9 + (1 << 24)) >> 25
h0 += carry[9] * 19
h9 -= carry[9] << 25
carry[0] = (h0 + (1 << 25)) >> 26
h1 += carry[0]
h0 -= carry[0] << 26
h[0] = int32(h0)
h[1] = int32(h1)
h[2] = int32(h2)
h[3] = int32(h3)
h[4] = int32(h4)
h[5] = int32(h5)
h[6] = int32(h6)
h[7] = int32(h7)
h[8] = int32(h8)
h[9] = int32(h9)
}
// feMul121666 calculates h = f * 121666. Can overlap h with f.
//
// Preconditions:
// |f| bounded by 1.1*2^26,1.1*2^25,1.1*2^26,1.1*2^25,etc.
//
// Postconditions:
// |h| bounded by 1.1*2^25,1.1*2^24,1.1*2^25,1.1*2^24,etc.
func feMul121666(h, f *fieldElement) {
h0 := int64(f[0]) * 121666
h1 := int64(f[1]) * 121666
h2 := int64(f[2]) * 121666
h3 := int64(f[3]) * 121666
h4 := int64(f[4]) * 121666
h5 := int64(f[5]) * 121666
h6 := int64(f[6]) * 121666
h7 := int64(f[7]) * 121666
h8 := int64(f[8]) * 121666
h9 := int64(f[9]) * 121666
var carry [10]int64
carry[9] = (h9 + (1 << 24)) >> 25
h0 += carry[9] * 19
h9 -= carry[9] << 25
carry[1] = (h1 + (1 << 24)) >> 25
h2 += carry[1]
h1 -= carry[1] << 25
carry[3] = (h3 + (1 << 24)) >> 25
h4 += carry[3]
h3 -= carry[3] << 25
carry[5] = (h5 + (1 << 24)) >> 25
h6 += carry[5]
h5 -= carry[5] << 25
carry[7] = (h7 + (1 << 24)) >> 25
h8 += carry[7]
h7 -= carry[7] << 25
carry[0] = (h0 + (1 << 25)) >> 26
h1 += carry[0]
h0 -= carry[0] << 26
carry[2] = (h2 + (1 << 25)) >> 26
h3 += carry[2]
h2 -= carry[2] << 26
carry[4] = (h4 + (1 << 25)) >> 26
h5 += carry[4]
h4 -= carry[4] << 26
carry[6] = (h6 + (1 << 25)) >> 26
h7 += carry[6]
h6 -= carry[6] << 26
carry[8] = (h8 + (1 << 25)) >> 26
h9 += carry[8]
h8 -= carry[8] << 26
h[0] = int32(h0)
h[1] = int32(h1)
h[2] = int32(h2)
h[3] = int32(h3)
h[4] = int32(h4)
h[5] = int32(h5)
h[6] = int32(h6)
h[7] = int32(h7)
h[8] = int32(h8)
h[9] = int32(h9)
}
// feInvert sets out = z^-1.
func feInvert(out, z *fieldElement) {
var t0, t1, t2, t3 fieldElement
var i int
feSquare(&t0, z)
for i = 1; i < 1; i++ {
feSquare(&t0, &t0)
}
feSquare(&t1, &t0)
for i = 1; i < 2; i++ {
feSquare(&t1, &t1)
}
feMul(&t1, z, &t1)
feMul(&t0, &t0, &t1)
feSquare(&t2, &t0)
for i = 1; i < 1; i++ {
feSquare(&t2, &t2)
}
feMul(&t1, &t1, &t2)
feSquare(&t2, &t1)
for i = 1; i < 5; i++ {
feSquare(&t2, &t2)
}
feMul(&t1, &t2, &t1)
feSquare(&t2, &t1)
for i = 1; i < 10; i++ {
feSquare(&t2, &t2)
}
feMul(&t2, &t2, &t1)
feSquare(&t3, &t2)
for i = 1; i < 20; i++ {
feSquare(&t3, &t3)
}
feMul(&t2, &t3, &t2)
feSquare(&t2, &t2)
for i = 1; i < 10; i++ {
feSquare(&t2, &t2)
}
feMul(&t1, &t2, &t1)
feSquare(&t2, &t1)
for i = 1; i < 50; i++ {
feSquare(&t2, &t2)
}
feMul(&t2, &t2, &t1)
feSquare(&t3, &t2)
for i = 1; i < 100; i++ {
feSquare(&t3, &t3)
}
feMul(&t2, &t3, &t2)
feSquare(&t2, &t2)
for i = 1; i < 50; i++ {
feSquare(&t2, &t2)
}
feMul(&t1, &t2, &t1)
feSquare(&t1, &t1)
for i = 1; i < 5; i++ {
feSquare(&t1, &t1)
}
feMul(out, &t1, &t0)
}
func scalarMultGeneric(out, in, base *[32]byte) {
var e [32]byte
copy(e[:], in[:])
e[0] &= 248
e[31] &= 127
e[31] |= 64
var x1, x2, z2, x3, z3, tmp0, tmp1 fieldElement
feFromBytes(&x1, base)
feOne(&x2)
feCopy(&x3, &x1)
feOne(&z3)
swap := int32(0)
for pos := 254; pos >= 0; pos-- {
b := e[pos/8] >> uint(pos&7)
b &= 1
swap ^= int32(b)
feCSwap(&x2, &x3, swap)
feCSwap(&z2, &z3, swap)
swap = int32(b)
feSub(&tmp0, &x3, &z3)
feSub(&tmp1, &x2, &z2)
feAdd(&x2, &x2, &z2)
feAdd(&z2, &x3, &z3)
feMul(&z3, &tmp0, &x2)
feMul(&z2, &z2, &tmp1)
feSquare(&tmp0, &tmp1)
feSquare(&tmp1, &x2)
feAdd(&x3, &z3, &z2)
feSub(&z2, &z3, &z2)
feMul(&x2, &tmp1, &tmp0)
feSub(&tmp1, &tmp1, &tmp0)
feSquare(&z2, &z2)
feMul121666(&z3, &tmp1)
feSquare(&x3, &x3)
feAdd(&tmp0, &tmp0, &z3)
feMul(&z3, &x1, &z2)
feMul(&z2, &tmp1, &tmp0)
}
feCSwap(&x2, &x3, swap)
feCSwap(&z2, &z3, swap)
feInvert(&z2, &z2)
feMul(&x2, &x2, &z2)
feToBytes(out, &x2)
}

View File

@ -0,0 +1,11 @@
// 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 !amd64 gccgo appengine purego
package curve25519
func scalarMult(out, in, base *[32]byte) {
scalarMultGeneric(out, in, base)
}

View File

@ -1,23 +0,0 @@
// 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)
}

View File

@ -1,73 +0,0 @@
// 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

View File

@ -1,169 +0,0 @@
// 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

View File

@ -1,132 +0,0 @@
// 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

39
vendor/golang.org/x/crypto/poly1305/bits_compat.go generated vendored Normal file
View File

@ -0,0 +1,39 @@
// 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
}

21
vendor/golang.org/x/crypto/poly1305/bits_go1.13.go generated vendored Normal file
View File

@ -0,0 +1,21 @@
// 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)
}

View File

@ -22,8 +22,14 @@ import "crypto/subtle"
// TagSize is the size, in bytes, of a poly1305 authenticator.
const TagSize = 16
// Verify returns true if mac is a valid authenticator for m with the given
// key.
// Sum generates an authenticator for msg using a one-time key and puts the
// 16-byte result into out. Authenticating two different messages with the same
// key allows an attacker to forge messages at will.
func Sum(out *[16]byte, m []byte, key *[32]byte) {
sum(out, m, key)
}
// Verify returns true if mac is a valid authenticator for m with the given key.
func Verify(mac *[16]byte, m []byte, key *[32]byte) bool {
var tmp [16]byte
Sum(&tmp, m, key)

View File

@ -7,62 +7,52 @@
package poly1305
//go:noescape
func initialize(state *[7]uint64, key *[32]byte)
func update(state *macState, msg []byte)
//go:noescape
func update(state *[7]uint64, msg []byte)
//go:noescape
func finalize(tag *[TagSize]byte, state *[7]uint64)
// Sum generates an authenticator for m using a one-time key and puts the
// 16-byte result into out. Authenticating two different messages with the same
// key allows an attacker to forge messages at will.
func Sum(out *[16]byte, m []byte, key *[32]byte) {
func sum(out *[16]byte, m []byte, key *[32]byte) {
h := newMAC(key)
h.Write(m)
h.Sum(out)
}
func newMAC(key *[32]byte) (h mac) {
initialize(&h.state, key)
initialize(key, &h.r, &h.s)
return
}
type mac struct {
state [7]uint64 // := uint64{ h0, h1, h2, r0, r1, pad0, pad1 }
// mac is a wrapper for macGeneric that redirects calls that would have gone to
// updateGeneric to update.
//
// Its Write and Sum methods are otherwise identical to the macGeneric ones, but
// using function pointers would carry a major performance cost.
type mac struct{ macGeneric }
buffer [TagSize]byte
offset int
}
func (h *mac) Write(p []byte) (n int, err error) {
n = len(p)
func (h *mac) Write(p []byte) (int, error) {
nn := len(p)
if h.offset > 0 {
remaining := TagSize - h.offset
if n < remaining {
h.offset += copy(h.buffer[h.offset:], p)
return n, nil
n := copy(h.buffer[h.offset:], p)
if h.offset+n < TagSize {
h.offset += n
return nn, nil
}
copy(h.buffer[h.offset:], p[:remaining])
p = p[remaining:]
p = p[n:]
h.offset = 0
update(&h.state, h.buffer[:])
update(&h.macState, h.buffer[:])
}
if nn := len(p) - (len(p) % TagSize); nn > 0 {
update(&h.state, p[:nn])
p = p[nn:]
if n := len(p) - (len(p) % TagSize); n > 0 {
update(&h.macState, p[:n])
p = p[n:]
}
if len(p) > 0 {
h.offset += copy(h.buffer[h.offset:], p)
}
return n, nil
return nn, nil
}
func (h *mac) Sum(out *[16]byte) {
state := h.state
state := h.macState
if h.offset > 0 {
update(&state, h.buffer[:h.offset])
}
finalize(out, &state)
finalize(out, &state.h, &state.s)
}

View File

@ -54,10 +54,6 @@
ADCQ t3, h1; \
ADCQ $0, h2
DATA ·poly1305Mask<>+0x00(SB)/8, $0x0FFFFFFC0FFFFFFF
DATA ·poly1305Mask<>+0x08(SB)/8, $0x0FFFFFFC0FFFFFFC
GLOBL ·poly1305Mask<>(SB), RODATA, $16
// func update(state *[7]uint64, msg []byte)
TEXT ·update(SB), $0-32
MOVQ state+0(FP), DI
@ -110,39 +106,3 @@ done:
MOVQ R9, 8(DI)
MOVQ R10, 16(DI)
RET
// func initialize(state *[7]uint64, key *[32]byte)
TEXT ·initialize(SB), $0-16
MOVQ state+0(FP), DI
MOVQ key+8(FP), SI
// state[0...7] is initialized with zero
MOVOU 0(SI), X0
MOVOU 16(SI), X1
MOVOU ·poly1305Mask<>(SB), X2
PAND X2, X0
MOVOU X0, 24(DI)
MOVOU X1, 40(DI)
RET
// func finalize(tag *[TagSize]byte, state *[7]uint64)
TEXT ·finalize(SB), $0-16
MOVQ tag+0(FP), DI
MOVQ state+8(FP), SI
MOVQ 0(SI), AX
MOVQ 8(SI), BX
MOVQ 16(SI), CX
MOVQ AX, R8
MOVQ BX, R9
SUBQ $0xFFFFFFFFFFFFFFFB, AX
SBBQ $0xFFFFFFFFFFFFFFFF, BX
SBBQ $3, CX
CMOVQCS R8, AX
CMOVQCS R9, BX
ADDQ 40(SI), AX
ADCQ 48(SI), BX
MOVQ AX, 0(DI)
MOVQ BX, 8(DI)
RET

View File

@ -6,14 +6,11 @@
package poly1305
// This function is implemented in sum_arm.s
// poly1305_auth_armv6 is implemented in sum_arm.s
//go:noescape
func poly1305_auth_armv6(out *[16]byte, m *byte, mlen uint32, key *[32]byte)
// Sum generates an authenticator for m using a one-time key and puts the
// 16-byte result into out. Authenticating two different messages with the same
// key allows an attacker to forge messages at will.
func Sum(out *[16]byte, m []byte, key *[32]byte) {
func sum(out *[16]byte, m []byte, key *[32]byte) {
var mPtr *byte
if len(m) > 0 {
mPtr = &m[0]

View File

@ -2,18 +2,29 @@
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// This file provides the generic implementation of Sum and MAC. Other files
// might provide optimized assembly implementations of some of this code.
package poly1305
import "encoding/binary"
const (
msgBlock = uint32(1 << 24)
finalBlock = uint32(0)
)
// Poly1305 [RFC 7539] is a relatively simple algorithm: the authentication tag
// for a 64 bytes message is approximately
//
// s + m[0:16] * r⁴ + m[16:32] * r³ + m[32:48] * r² + m[48:64] * r mod 2¹³⁰ - 5
//
// for some secret r and s. It can be computed sequentially like
//
// for len(msg) > 0:
// h += read(msg, 16)
// h *= r
// h %= 2¹³⁰ - 5
// return h + s
//
// All the complexity is about doing performant constant-time math on numbers
// larger than any available numeric type.
// sumGeneric generates an authenticator for msg using a one-time key and
// puts the 16-byte result into out. This is the generic implementation of
// Sum and should be called if no assembly implementation is available.
func sumGeneric(out *[TagSize]byte, msg []byte, key *[32]byte) {
h := newMACGeneric(key)
h.Write(msg)
@ -21,152 +32,276 @@ func sumGeneric(out *[TagSize]byte, msg []byte, key *[32]byte) {
}
func newMACGeneric(key *[32]byte) (h macGeneric) {
h.r[0] = binary.LittleEndian.Uint32(key[0:]) & 0x3ffffff
h.r[1] = (binary.LittleEndian.Uint32(key[3:]) >> 2) & 0x3ffff03
h.r[2] = (binary.LittleEndian.Uint32(key[6:]) >> 4) & 0x3ffc0ff
h.r[3] = (binary.LittleEndian.Uint32(key[9:]) >> 6) & 0x3f03fff
h.r[4] = (binary.LittleEndian.Uint32(key[12:]) >> 8) & 0x00fffff
h.s[0] = binary.LittleEndian.Uint32(key[16:])
h.s[1] = binary.LittleEndian.Uint32(key[20:])
h.s[2] = binary.LittleEndian.Uint32(key[24:])
h.s[3] = binary.LittleEndian.Uint32(key[28:])
initialize(key, &h.r, &h.s)
return
}
// macState holds numbers in saturated 64-bit little-endian limbs. That is,
// the value of [x0, x1, x2] is x[0] + x[1] * 2⁶⁴ + x[2] * 2¹²⁸.
type macState struct {
// h is the main accumulator. It is to be interpreted modulo 2¹³⁰ - 5, but
// can grow larger during and after rounds.
h [3]uint64
// r and s are the private key components.
r [2]uint64
s [2]uint64
}
type macGeneric struct {
h, r [5]uint32
s [4]uint32
macState
buffer [TagSize]byte
offset int
}
func (h *macGeneric) Write(p []byte) (n int, err error) {
n = len(p)
// Write splits the incoming message into TagSize chunks, and passes them to
// update. It buffers incomplete chunks.
func (h *macGeneric) Write(p []byte) (int, error) {
nn := len(p)
if h.offset > 0 {
remaining := TagSize - h.offset
if n < remaining {
h.offset += copy(h.buffer[h.offset:], p)
return n, nil
n := copy(h.buffer[h.offset:], p)
if h.offset+n < TagSize {
h.offset += n
return nn, nil
}
copy(h.buffer[h.offset:], p[:remaining])
p = p[remaining:]
p = p[n:]
h.offset = 0
updateGeneric(h.buffer[:], msgBlock, &(h.h), &(h.r))
updateGeneric(&h.macState, h.buffer[:])
}
if nn := len(p) - (len(p) % TagSize); nn > 0 {
updateGeneric(p, msgBlock, &(h.h), &(h.r))
p = p[nn:]
if n := len(p) - (len(p) % TagSize); n > 0 {
updateGeneric(&h.macState, p[:n])
p = p[n:]
}
if len(p) > 0 {
h.offset += copy(h.buffer[h.offset:], p)
}
return n, nil
return nn, nil
}
func (h *macGeneric) Sum(out *[16]byte) {
H, R := h.h, h.r
// Sum flushes the last incomplete chunk from the buffer, if any, and generates
// the MAC output. It does not modify its state, in order to allow for multiple
// calls to Sum, even if no Write is allowed after Sum.
func (h *macGeneric) Sum(out *[TagSize]byte) {
state := h.macState
if h.offset > 0 {
var buffer [TagSize]byte
copy(buffer[:], h.buffer[:h.offset])
buffer[h.offset] = 1 // invariant: h.offset < TagSize
updateGeneric(buffer[:], finalBlock, &H, &R)
updateGeneric(&state, h.buffer[:h.offset])
}
finalizeGeneric(out, &H, &(h.s))
finalize(out, &state.h, &state.s)
}
func updateGeneric(msg []byte, flag uint32, h, r *[5]uint32) {
h0, h1, h2, h3, h4 := h[0], h[1], h[2], h[3], h[4]
r0, r1, r2, r3, r4 := uint64(r[0]), uint64(r[1]), uint64(r[2]), uint64(r[3]), uint64(r[4])
R1, R2, R3, R4 := r1*5, r2*5, r3*5, r4*5
// [rMask0, rMask1] is the specified Poly1305 clamping mask in little-endian. It
// clears some bits of the secret coefficient to make it possible to implement
// multiplication more efficiently.
const (
rMask0 = 0x0FFFFFFC0FFFFFFF
rMask1 = 0x0FFFFFFC0FFFFFFC
)
for len(msg) >= TagSize {
// h += msg
h0 += binary.LittleEndian.Uint32(msg[0:]) & 0x3ffffff
h1 += (binary.LittleEndian.Uint32(msg[3:]) >> 2) & 0x3ffffff
h2 += (binary.LittleEndian.Uint32(msg[6:]) >> 4) & 0x3ffffff
h3 += (binary.LittleEndian.Uint32(msg[9:]) >> 6) & 0x3ffffff
h4 += (binary.LittleEndian.Uint32(msg[12:]) >> 8) | flag
func initialize(key *[32]byte, r, s *[2]uint64) {
r[0] = binary.LittleEndian.Uint64(key[0:8]) & rMask0
r[1] = binary.LittleEndian.Uint64(key[8:16]) & rMask1
s[0] = binary.LittleEndian.Uint64(key[16:24])
s[1] = binary.LittleEndian.Uint64(key[24:32])
}
// h *= r
d0 := (uint64(h0) * r0) + (uint64(h1) * R4) + (uint64(h2) * R3) + (uint64(h3) * R2) + (uint64(h4) * R1)
d1 := (d0 >> 26) + (uint64(h0) * r1) + (uint64(h1) * r0) + (uint64(h2) * R4) + (uint64(h3) * R3) + (uint64(h4) * R2)
d2 := (d1 >> 26) + (uint64(h0) * r2) + (uint64(h1) * r1) + (uint64(h2) * r0) + (uint64(h3) * R4) + (uint64(h4) * R3)
d3 := (d2 >> 26) + (uint64(h0) * r3) + (uint64(h1) * r2) + (uint64(h2) * r1) + (uint64(h3) * r0) + (uint64(h4) * R4)
d4 := (d3 >> 26) + (uint64(h0) * r4) + (uint64(h1) * r3) + (uint64(h2) * r2) + (uint64(h3) * r1) + (uint64(h4) * r0)
// uint128 holds a 128-bit number as two 64-bit limbs, for use with the
// bits.Mul64 and bits.Add64 intrinsics.
type uint128 struct {
lo, hi uint64
}
// h %= p
h0 = uint32(d0) & 0x3ffffff
h1 = uint32(d1) & 0x3ffffff
h2 = uint32(d2) & 0x3ffffff
h3 = uint32(d3) & 0x3ffffff
h4 = uint32(d4) & 0x3ffffff
func mul64(a, b uint64) uint128 {
hi, lo := bitsMul64(a, b)
return uint128{lo, hi}
}
h0 += uint32(d4>>26) * 5
h1 += h0 >> 26
h0 = h0 & 0x3ffffff
func add128(a, b uint128) uint128 {
lo, c := bitsAdd64(a.lo, b.lo, 0)
hi, c := bitsAdd64(a.hi, b.hi, c)
if c != 0 {
panic("poly1305: unexpected overflow")
}
return uint128{lo, hi}
}
msg = msg[TagSize:]
func shiftRightBy2(a uint128) uint128 {
a.lo = a.lo>>2 | (a.hi&3)<<62
a.hi = a.hi >> 2
return a
}
// updateGeneric absorbs msg into the state.h accumulator. For each chunk m of
// 128 bits of message, it computes
//
// h₊ = (h + m) * r mod 2¹³⁰ - 5
//
// If the msg length is not a multiple of TagSize, it assumes the last
// incomplete chunk is the final one.
func updateGeneric(state *macState, msg []byte) {
h0, h1, h2 := state.h[0], state.h[1], state.h[2]
r0, r1 := state.r[0], state.r[1]
for len(msg) > 0 {
var c uint64
// For the first step, h + m, we use a chain of bits.Add64 intrinsics.
// The resulting value of h might exceed 2¹³⁰ - 5, but will be partially
// reduced at the end of the multiplication below.
//
// The spec requires us to set a bit just above the message size, not to
// hide leading zeroes. For full chunks, that's 1 << 128, so we can just
// add 1 to the most significant (2¹²⁸) limb, h2.
if len(msg) >= TagSize {
h0, c = bitsAdd64(h0, binary.LittleEndian.Uint64(msg[0:8]), 0)
h1, c = bitsAdd64(h1, binary.LittleEndian.Uint64(msg[8:16]), c)
h2 += c + 1
msg = msg[TagSize:]
} else {
var buf [TagSize]byte
copy(buf[:], msg)
buf[len(msg)] = 1
h0, c = bitsAdd64(h0, binary.LittleEndian.Uint64(buf[0:8]), 0)
h1, c = bitsAdd64(h1, binary.LittleEndian.Uint64(buf[8:16]), c)
h2 += c
msg = nil
}
// Multiplication of big number limbs is similar to elementary school
// columnar multiplication. Instead of digits, there are 64-bit limbs.
//
// We are multiplying a 3 limbs number, h, by a 2 limbs number, r.
//
// h2 h1 h0 x
// r1 r0 =
// ----------------
// h2r0 h1r0 h0r0 <-- individual 128-bit products
// + h2r1 h1r1 h0r1
// ------------------------
// m3 m2 m1 m0 <-- result in 128-bit overlapping limbs
// ------------------------
// m3.hi m2.hi m1.hi m0.hi <-- carry propagation
// + m3.lo m2.lo m1.lo m0.lo
// -------------------------------
// t4 t3 t2 t1 t0 <-- final result in 64-bit limbs
//
// The main difference from pen-and-paper multiplication is that we do
// carry propagation in a separate step, as if we wrote two digit sums
// at first (the 128-bit limbs), and then carried the tens all at once.
h0r0 := mul64(h0, r0)
h1r0 := mul64(h1, r0)
h2r0 := mul64(h2, r0)
h0r1 := mul64(h0, r1)
h1r1 := mul64(h1, r1)
h2r1 := mul64(h2, r1)
// Since h2 is known to be at most 7 (5 + 1 + 1), and r0 and r1 have their
// top 4 bits cleared by rMask{0,1}, we know that their product is not going
// to overflow 64 bits, so we can ignore the high part of the products.
//
// This also means that the product doesn't have a fifth limb (t4).
if h2r0.hi != 0 {
panic("poly1305: unexpected overflow")
}
if h2r1.hi != 0 {
panic("poly1305: unexpected overflow")
}
m0 := h0r0
m1 := add128(h1r0, h0r1) // These two additions don't overflow thanks again
m2 := add128(h2r0, h1r1) // to the 4 masked bits at the top of r0 and r1.
m3 := h2r1
t0 := m0.lo
t1, c := bitsAdd64(m1.lo, m0.hi, 0)
t2, c := bitsAdd64(m2.lo, m1.hi, c)
t3, _ := bitsAdd64(m3.lo, m2.hi, c)
// Now we have the result as 4 64-bit limbs, and we need to reduce it
// modulo 2¹³⁰ - 5. The special shape of this Crandall prime lets us do
// a cheap partial reduction according to the reduction identity
//
// c * 2¹³⁰ + n = c * 5 + n mod 2¹³⁰ - 5
//
// because 2¹³⁰ = 5 mod 2¹³⁰ - 5. Partial reduction since the result is
// likely to be larger than 2¹³⁰ - 5, but still small enough to fit the
// assumptions we make about h in the rest of the code.
//
// See also https://speakerdeck.com/gtank/engineering-prime-numbers?slide=23
// We split the final result at the 2¹³⁰ mark into h and cc, the carry.
// Note that the carry bits are effectively shifted left by 2, in other
// words, cc = c * 4 for the c in the reduction identity.
h0, h1, h2 = t0, t1, t2&maskLow2Bits
cc := uint128{t2 & maskNotLow2Bits, t3}
// To add c * 5 to h, we first add cc = c * 4, and then add (cc >> 2) = c.
h0, c = bitsAdd64(h0, cc.lo, 0)
h1, c = bitsAdd64(h1, cc.hi, c)
h2 += c
cc = shiftRightBy2(cc)
h0, c = bitsAdd64(h0, cc.lo, 0)
h1, c = bitsAdd64(h1, cc.hi, c)
h2 += c
// h2 is at most 3 + 1 + 1 = 5, making the whole of h at most
//
// 5 * 2¹²⁸ + (2¹²⁸ - 1) = 6 * 2¹²⁸ - 1
}
h[0], h[1], h[2], h[3], h[4] = h0, h1, h2, h3, h4
state.h[0], state.h[1], state.h[2] = h0, h1, h2
}
func finalizeGeneric(out *[TagSize]byte, h *[5]uint32, s *[4]uint32) {
h0, h1, h2, h3, h4 := h[0], h[1], h[2], h[3], h[4]
const (
maskLow2Bits uint64 = 0x0000000000000003
maskNotLow2Bits uint64 = ^maskLow2Bits
)
// h %= p reduction
h2 += h1 >> 26
h1 &= 0x3ffffff
h3 += h2 >> 26
h2 &= 0x3ffffff
h4 += h3 >> 26
h3 &= 0x3ffffff
h0 += 5 * (h4 >> 26)
h4 &= 0x3ffffff
h1 += h0 >> 26
h0 &= 0x3ffffff
// select64 returns x if v == 1 and y if v == 0, in constant time.
func select64(v, x, y uint64) uint64 { return ^(v-1)&x | (v-1)&y }
// h - p
t0 := h0 + 5
t1 := h1 + (t0 >> 26)
t2 := h2 + (t1 >> 26)
t3 := h3 + (t2 >> 26)
t4 := h4 + (t3 >> 26) - (1 << 26)
t0 &= 0x3ffffff
t1 &= 0x3ffffff
t2 &= 0x3ffffff
t3 &= 0x3ffffff
// [p0, p1, p2] is 2¹³⁰ - 5 in little endian order.
const (
p0 = 0xFFFFFFFFFFFFFFFB
p1 = 0xFFFFFFFFFFFFFFFF
p2 = 0x0000000000000003
)
// select h if h < p else h - p
t_mask := (t4 >> 31) - 1
h_mask := ^t_mask
h0 = (h0 & h_mask) | (t0 & t_mask)
h1 = (h1 & h_mask) | (t1 & t_mask)
h2 = (h2 & h_mask) | (t2 & t_mask)
h3 = (h3 & h_mask) | (t3 & t_mask)
h4 = (h4 & h_mask) | (t4 & t_mask)
// finalize completes the modular reduction of h and computes
//
// out = h + s mod 2¹²⁸
//
func finalize(out *[TagSize]byte, h *[3]uint64, s *[2]uint64) {
h0, h1, h2 := h[0], h[1], h[2]
// h %= 2^128
h0 |= h1 << 26
h1 = ((h1 >> 6) | (h2 << 20))
h2 = ((h2 >> 12) | (h3 << 14))
h3 = ((h3 >> 18) | (h4 << 8))
// After the partial reduction in updateGeneric, h might be more than
// 2¹³⁰ - 5, but will be less than 2 * (2¹³⁰ - 5). To complete the reduction
// in constant time, we compute t = h - (2¹³⁰ - 5), and select h as the
// result if the subtraction underflows, and t otherwise.
// s: the s part of the key
// tag = (h + s) % (2^128)
t := uint64(h0) + uint64(s[0])
h0 = uint32(t)
t = uint64(h1) + uint64(s[1]) + (t >> 32)
h1 = uint32(t)
t = uint64(h2) + uint64(s[2]) + (t >> 32)
h2 = uint32(t)
t = uint64(h3) + uint64(s[3]) + (t >> 32)
h3 = uint32(t)
hMinusP0, b := bitsSub64(h0, p0, 0)
hMinusP1, b := bitsSub64(h1, p1, b)
_, b = bitsSub64(h2, p2, b)
binary.LittleEndian.PutUint32(out[0:], h0)
binary.LittleEndian.PutUint32(out[4:], h1)
binary.LittleEndian.PutUint32(out[8:], h2)
binary.LittleEndian.PutUint32(out[12:], h3)
// h = h if h < p else h - p
h0 = select64(b, h0, hMinusP0)
h1 = select64(b, h1, hMinusP1)
// Finally, we compute the last Poly1305 step
//
// tag = h + s mod 2¹²⁸
//
// by just doing a wide addition with the 128 low bits of h and discarding
// the overflow.
h0, c := bitsAdd64(h0, s[0], 0)
h1, _ = bitsAdd64(h1, s[1], c)
binary.LittleEndian.PutUint64(out[0:8], h0)
binary.LittleEndian.PutUint64(out[8:16], h1)
}

View File

@ -6,10 +6,7 @@
package poly1305
// Sum generates an authenticator for msg using a one-time key and puts the
// 16-byte result into out. Authenticating two different messages with the same
// key allows an attacker to forge messages at will.
func Sum(out *[TagSize]byte, msg []byte, key *[32]byte) {
func sum(out *[TagSize]byte, msg []byte, key *[32]byte) {
h := newMAC(key)
h.Write(msg)
h.Sum(out)

View File

@ -7,62 +7,52 @@
package poly1305
//go:noescape
func initialize(state *[7]uint64, key *[32]byte)
func update(state *macState, msg []byte)
//go:noescape
func update(state *[7]uint64, msg []byte)
//go:noescape
func finalize(tag *[TagSize]byte, state *[7]uint64)
// Sum generates an authenticator for m using a one-time key and puts the
// 16-byte result into out. Authenticating two different messages with the same
// key allows an attacker to forge messages at will.
func Sum(out *[16]byte, m []byte, key *[32]byte) {
func sum(out *[16]byte, m []byte, key *[32]byte) {
h := newMAC(key)
h.Write(m)
h.Sum(out)
}
func newMAC(key *[32]byte) (h mac) {
initialize(&h.state, key)
initialize(key, &h.r, &h.s)
return
}
type mac struct {
state [7]uint64 // := uint64{ h0, h1, h2, r0, r1, pad0, pad1 }
// mac is a wrapper for macGeneric that redirects calls that would have gone to
// updateGeneric to update.
//
// Its Write and Sum methods are otherwise identical to the macGeneric ones, but
// using function pointers would carry a major performance cost.
type mac struct{ macGeneric }
buffer [TagSize]byte
offset int
}
func (h *mac) Write(p []byte) (n int, err error) {
n = len(p)
func (h *mac) Write(p []byte) (int, error) {
nn := len(p)
if h.offset > 0 {
remaining := TagSize - h.offset
if n < remaining {
h.offset += copy(h.buffer[h.offset:], p)
return n, nil
n := copy(h.buffer[h.offset:], p)
if h.offset+n < TagSize {
h.offset += n
return nn, nil
}
copy(h.buffer[h.offset:], p[:remaining])
p = p[remaining:]
p = p[n:]
h.offset = 0
update(&h.state, h.buffer[:])
update(&h.macState, h.buffer[:])
}
if nn := len(p) - (len(p) % TagSize); nn > 0 {
update(&h.state, p[:nn])
p = p[nn:]
if n := len(p) - (len(p) % TagSize); n > 0 {
update(&h.macState, p[:n])
p = p[n:]
}
if len(p) > 0 {
h.offset += copy(h.buffer[h.offset:], p)
}
return n, nil
return nn, nil
}
func (h *mac) Sum(out *[16]byte) {
state := h.state
state := h.macState
if h.offset > 0 {
update(&state, h.buffer[:h.offset])
}
finalize(out, &state)
finalize(out, &state.h, &state.s)
}

View File

@ -58,7 +58,6 @@ DATA ·poly1305Mask<>+0x08(SB)/8, $0x0FFFFFFC0FFFFFFC
GLOBL ·poly1305Mask<>(SB), RODATA, $16
// func update(state *[7]uint64, msg []byte)
TEXT ·update(SB), $0-32
MOVD state+0(FP), R3
MOVD msg_base+8(FP), R4
@ -180,68 +179,3 @@ done:
MOVD R9, 8(R3)
MOVD R10, 16(R3)
RET
// func initialize(state *[7]uint64, key *[32]byte)
TEXT ·initialize(SB), $0-16
MOVD state+0(FP), R3
MOVD key+8(FP), R4
// state[0...7] is initialized with zero
// Load key
MOVD 0(R4), R5
MOVD 8(R4), R6
MOVD 16(R4), R7
MOVD 24(R4), R8
// Address of key mask
MOVD $·poly1305Mask<>(SB), R9
// Save original key in state
MOVD R7, 40(R3)
MOVD R8, 48(R3)
// Get mask
MOVD (R9), R7
MOVD 8(R9), R8
// And with key
AND R5, R7, R5
AND R6, R8, R6
// Save masked key in state
MOVD R5, 24(R3)
MOVD R6, 32(R3)
RET
// func finalize(tag *[TagSize]byte, state *[7]uint64)
TEXT ·finalize(SB), $0-16
MOVD tag+0(FP), R3
MOVD state+8(FP), R4
// Get h0, h1, h2 from state
MOVD 0(R4), R5
MOVD 8(R4), R6
MOVD 16(R4), R7
// Save h0, h1
MOVD R5, R8
MOVD R6, R9
MOVD $3, R20
MOVD $-1, R21
SUBC $-5, R5
SUBE R21, R6
SUBE R20, R7
MOVD $0, R21
SUBZE R21
// Check for carry
CMP $0, R21
ISEL $2, R5, R8, R5
ISEL $2, R6, R9, R6
MOVD 40(R4), R8
MOVD 48(R4), R9
ADDC R8, R5
ADDE R9, R6
MOVD R5, 0(R3)
MOVD R6, 8(R3)
RET

View File

@ -22,10 +22,7 @@ func poly1305vx(out *[16]byte, m *byte, mlen uint64, key *[32]byte)
//go:noescape
func poly1305vmsl(out *[16]byte, m *byte, mlen uint64, key *[32]byte)
// Sum generates an authenticator for m using a one-time key and puts the
// 16-byte result into out. Authenticating two different messages with the same
// key allows an attacker to forge messages at will.
func Sum(out *[16]byte, m []byte, key *[32]byte) {
func sum(out *[16]byte, m []byte, key *[32]byte) {
if cpu.S390X.HasVX {
var mPtr *byte
if len(m) > 0 {

View File

@ -252,7 +252,7 @@ func ConfigureServer(s *http.Server, conf *Server) error {
}
}
if !haveRequired {
return fmt.Errorf("http2: TLSConfig.CipherSuites is missing an HTTP/2-required AES_128_GCM_SHA256 cipher.")
return fmt.Errorf("http2: TLSConfig.CipherSuites is missing an HTTP/2-required AES_128_GCM_SHA256 cipher (need at least one of TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256 or TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256).")
}
}

12
vendor/golang.org/x/net/internal/socket/norace.go generated vendored Normal file
View File

@ -0,0 +1,12 @@
// 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 !race
package socket
func (m *Message) raceRead() {
}
func (m *Message) raceWrite() {
}

37
vendor/golang.org/x/net/internal/socket/race.go generated vendored Normal file
View File

@ -0,0 +1,37 @@
// 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 race
package socket
import (
"runtime"
"unsafe"
)
// This package reads and writes the Message buffers using a
// direct system call, which the race detector can't see.
// These functions tell the race detector what is going on during the syscall.
func (m *Message) raceRead() {
for _, b := range m.Buffers {
if len(b) > 0 {
runtime.RaceReadRange(unsafe.Pointer(&b[0]), len(b))
}
}
if b := m.OOB; len(b) > 0 {
runtime.RaceReadRange(unsafe.Pointer(&b[0]), len(b))
}
}
func (m *Message) raceWrite() {
for _, b := range m.Buffers {
if len(b) > 0 {
runtime.RaceWriteRange(unsafe.Pointer(&b[0]), len(b))
}
}
if b := m.OOB; len(b) > 0 {
runtime.RaceWriteRange(unsafe.Pointer(&b[0]), len(b))
}
}

View File

@ -13,6 +13,9 @@ import (
)
func (c *Conn) recvMsgs(ms []Message, flags int) (int, error) {
for i := range ms {
ms[i].raceWrite()
}
hs := make(mmsghdrs, len(ms))
var parseFn func([]byte, string) (net.Addr, error)
if c.network != "tcp" {
@ -43,6 +46,9 @@ func (c *Conn) recvMsgs(ms []Message, flags int) (int, error) {
}
func (c *Conn) sendMsgs(ms []Message, flags int) (int, error) {
for i := range ms {
ms[i].raceRead()
}
hs := make(mmsghdrs, len(ms))
var marshalFn func(net.Addr) []byte
if c.network != "tcp" {

View File

@ -12,6 +12,7 @@ import (
)
func (c *Conn) recvMsg(m *Message, flags int) error {
m.raceWrite()
var h msghdr
vs := make([]iovec, len(m.Buffers))
var sa []byte
@ -48,6 +49,7 @@ func (c *Conn) recvMsg(m *Message, flags int) error {
}
func (c *Conn) sendMsg(m *Message, flags int) error {
m.raceRead()
var h msghdr
vs := make([]iovec, len(m.Buffers))
var sa []byte

9
vendor/golang.org/x/sys/cpu/cpu_linux_noinit.go generated vendored Normal file
View File

@ -0,0 +1,9 @@
// 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 linux,!arm,!arm64,!ppc64,!ppc64le,!s390x
package cpu
func doinit() {}

View File

@ -7,5 +7,3 @@
package cpu
const cacheLineSize = 32
func doinit() {}

View File

@ -7,5 +7,3 @@
package cpu
const cacheLineSize = 32
func doinit() {}

View File

@ -7,5 +7,3 @@
package cpu
const cacheLineSize = 64
func doinit() {}

9
vendor/golang.org/x/sys/cpu/cpu_riscv64.go generated vendored Normal file
View File

@ -0,0 +1,9 @@
// 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 riscv64
package cpu
const cacheLineSize = 32

View File

@ -11,5 +11,3 @@ package cpu
// rules are good enough.
const cacheLineSize = 0
func doinit() {}

29
vendor/golang.org/x/sys/unix/fdset.go generated vendored Normal file
View File

@ -0,0 +1,29 @@
// 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 aix darwin dragonfly freebsd linux netbsd openbsd solaris
package unix
// Set adds fd to the set fds.
func (fds *FdSet) Set(fd int) {
fds.Bits[fd/NFDBITS] |= (1 << (uintptr(fd) % NFDBITS))
}
// Clear removes fd from the set fds.
func (fds *FdSet) Clear(fd int) {
fds.Bits[fd/NFDBITS] &^= (1 << (uintptr(fd) % NFDBITS))
}
// IsSet returns whether fd is in the set fds.
func (fds *FdSet) IsSet(fd int) bool {
return fds.Bits[fd/NFDBITS]&(1<<(uintptr(fd)%NFDBITS)) != 0
}
// Zero clears the set fds.
func (fds *FdSet) Zero() {
for i := range fds.Bits {
fds.Bits[i] = 0
}
}

8
vendor/golang.org/x/sys/windows/empty.s generated vendored Normal file
View File

@ -0,0 +1,8 @@
// 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 !go1.12
// This file is here to allow bodyless functions with go:linkname for Go 1.11
// and earlier (see https://golang.org/issue/23311).

6
vendor/modules.txt vendored
View File

@ -52,7 +52,7 @@ github.com/powerman/check
# github.com/smartystreets/goconvey v1.6.4
github.com/smartystreets/goconvey/convey/gotest
github.com/smartystreets/goconvey/convey/reporting
# golang.org/x/crypto v0.0.0-20191029031824-8986dd9e96cf
# golang.org/x/crypto v0.0.0-20191112222119-e1110fd1c708
golang.org/x/crypto/curve25519
golang.org/x/crypto/ed25519
golang.org/x/crypto/ed25519/internal/edwards25519
@ -61,7 +61,7 @@ golang.org/x/crypto/nacl/box
golang.org/x/crypto/nacl/secretbox
golang.org/x/crypto/poly1305
golang.org/x/crypto/salsa20/salsa
# golang.org/x/net v0.0.0-20191101175033-0deb6923b6d9
# golang.org/x/net v0.0.0-20191112182307-2180aed22343
golang.org/x/net/bpf
golang.org/x/net/http/httpguts
golang.org/x/net/http2
@ -73,7 +73,7 @@ golang.org/x/net/internal/socks
golang.org/x/net/ipv4
golang.org/x/net/ipv6
golang.org/x/net/proxy
# golang.org/x/sys v0.0.0-20191104094858-e8c54fb511f6
# golang.org/x/sys v0.0.0-20191113165036-4c7a9d0fe056
golang.org/x/sys/cpu
golang.org/x/sys/unix
golang.org/x/sys/windows