Switch back go upstream bwmarrin/discordgo

Commit https://github.com/bwmarrin/discordgo/commit/ffa9956c9b41e8e2a10c26a254389854e016b006 got merged in.

5 files changed, 618 insertions, 180 deletions

diff --git a/vendor/golang.org/x/crypto/internal/chacha20/chacha_generic.go b/vendor/golang.org/x/crypto/internal/chacha20/chacha_generic.go
index 0f8efdba..6570847f 100644
--- a/vendor/golang.org/x/crypto/internal/chacha20/chacha_generic.go
+++ b/vendor/golang.org/x/crypto/internal/chacha20/chacha_generic.go
@@ -2,197 +2,263 @@
 // Use of this source code is governed by a BSD-style
 // license that can be found in the LICENSE file.
 
-// Package ChaCha20 implements the core ChaCha20 function as specified in https://tools.ietf.org/html/rfc7539#section-2.3.
+// Package ChaCha20 implements the core ChaCha20 function as specified
+// in https://tools.ietf.org/html/rfc7539#section-2.3.
 package chacha20
 
-import "encoding/binary"
-
-const rounds = 20
-
-// core applies the ChaCha20 core function to 16-byte input in, 32-byte key k,
-// and 16-byte constant c, and puts the result into 64-byte array out.
-func core(out *[64]byte, in *[16]byte, k *[32]byte) {
-	j0 := uint32(0x61707865)
-	j1 := uint32(0x3320646e)
-	j2 := uint32(0x79622d32)
-	j3 := uint32(0x6b206574)
-	j4 := binary.LittleEndian.Uint32(k[0:4])
-	j5 := binary.LittleEndian.Uint32(k[4:8])
-	j6 := binary.LittleEndian.Uint32(k[8:12])
-	j7 := binary.LittleEndian.Uint32(k[12:16])
-	j8 := binary.LittleEndian.Uint32(k[16:20])
-	j9 := binary.LittleEndian.Uint32(k[20:24])
-	j10 := binary.LittleEndian.Uint32(k[24:28])
-	j11 := binary.LittleEndian.Uint32(k[28:32])
-	j12 := binary.LittleEndian.Uint32(in[0:4])
-	j13 := binary.LittleEndian.Uint32(in[4:8])
-	j14 := binary.LittleEndian.Uint32(in[8:12])
-	j15 := binary.LittleEndian.Uint32(in[12:16])
-
-	x0, x1, x2, x3, x4, x5, x6, x7 := j0, j1, j2, j3, j4, j5, j6, j7
-	x8, x9, x10, x11, x12, x13, x14, x15 := j8, j9, j10, j11, j12, j13, j14, j15
-
-	for i := 0; i < rounds; i += 2 {
-		x0 += x4
-		x12 ^= x0
-		x12 = (x12 << 16) | (x12 >> (16))
-		x8 += x12
-		x4 ^= x8
-		x4 = (x4 << 12) | (x4 >> (20))
-		x0 += x4
-		x12 ^= x0
-		x12 = (x12 << 8) | (x12 >> (24))
-		x8 += x12
-		x4 ^= x8
-		x4 = (x4 << 7) | (x4 >> (25))
-		x1 += x5
-		x13 ^= x1
-		x13 = (x13 << 16) | (x13 >> 16)
-		x9 += x13
-		x5 ^= x9
-		x5 = (x5 << 12) | (x5 >> 20)
-		x1 += x5
-		x13 ^= x1
-		x13 = (x13 << 8) | (x13 >> 24)
-		x9 += x13
-		x5 ^= x9
-		x5 = (x5 << 7) | (x5 >> 25)
-		x2 += x6
-		x14 ^= x2
-		x14 = (x14 << 16) | (x14 >> 16)
-		x10 += x14
-		x6 ^= x10
-		x6 = (x6 << 12) | (x6 >> 20)
-		x2 += x6
-		x14 ^= x2
-		x14 = (x14 << 8) | (x14 >> 24)
-		x10 += x14
-		x6 ^= x10
-		x6 = (x6 << 7) | (x6 >> 25)
-		x3 += x7
-		x15 ^= x3
-		x15 = (x15 << 16) | (x15 >> 16)
-		x11 += x15
-		x7 ^= x11
-		x7 = (x7 << 12) | (x7 >> 20)
-		x3 += x7
-		x15 ^= x3
-		x15 = (x15 << 8) | (x15 >> 24)
-		x11 += x15
-		x7 ^= x11
-		x7 = (x7 << 7) | (x7 >> 25)
-		x0 += x5
-		x15 ^= x0
-		x15 = (x15 << 16) | (x15 >> 16)
-		x10 += x15
-		x5 ^= x10
-		x5 = (x5 << 12) | (x5 >> 20)
-		x0 += x5
-		x15 ^= x0
-		x15 = (x15 << 8) | (x15 >> 24)
-		x10 += x15
-		x5 ^= x10
-		x5 = (x5 << 7) | (x5 >> 25)
-		x1 += x6
-		x12 ^= x1
-		x12 = (x12 << 16) | (x12 >> 16)
-		x11 += x12
-		x6 ^= x11
-		x6 = (x6 << 12) | (x6 >> 20)
-		x1 += x6
-		x12 ^= x1
-		x12 = (x12 << 8) | (x12 >> 24)
-		x11 += x12
-		x6 ^= x11
-		x6 = (x6 << 7) | (x6 >> 25)
-		x2 += x7
-		x13 ^= x2
-		x13 = (x13 << 16) | (x13 >> 16)
-		x8 += x13
-		x7 ^= x8
-		x7 = (x7 << 12) | (x7 >> 20)
-		x2 += x7
-		x13 ^= x2
-		x13 = (x13 << 8) | (x13 >> 24)
-		x8 += x13
-		x7 ^= x8
-		x7 = (x7 << 7) | (x7 >> 25)
-		x3 += x4
-		x14 ^= x3
-		x14 = (x14 << 16) | (x14 >> 16)
-		x9 += x14
-		x4 ^= x9
-		x4 = (x4 << 12) | (x4 >> 20)
-		x3 += x4
-		x14 ^= x3
-		x14 = (x14 << 8) | (x14 >> 24)
-		x9 += x14
-		x4 ^= x9
-		x4 = (x4 << 7) | (x4 >> 25)
+import (
+	"crypto/cipher"
+	"encoding/binary"
+
+	"golang.org/x/crypto/internal/subtle"
+)
+
+// assert that *Cipher implements cipher.Stream
+var _ cipher.Stream = (*Cipher)(nil)
+
+// Cipher is a stateful instance of ChaCha20 using a particular key
+// and nonce. A *Cipher implements the cipher.Stream interface.
+type Cipher struct {
+	key     [8]uint32
+	counter uint32 // incremented after each block
+	nonce   [3]uint32
+	buf     [bufSize]byte // buffer for unused keystream bytes
+	len     int           // number of unused keystream bytes at end of buf
+}
+
+// New creates a new ChaCha20 stream cipher with the given key and nonce.
+// The initial counter value is set to 0.
+func New(key [8]uint32, nonce [3]uint32) *Cipher {
+	return &Cipher{key: key, nonce: nonce}
+}
+
+// ChaCha20 constants spelling "expand 32-byte k"
+const (
+	j0 uint32 = 0x61707865
+	j1 uint32 = 0x3320646e
+	j2 uint32 = 0x79622d32
+	j3 uint32 = 0x6b206574
+)
+
+func quarterRound(a, b, c, d uint32) (uint32, uint32, uint32, uint32) {
+	a += b
+	d ^= a
+	d = (d << 16) | (d >> 16)
+	c += d
+	b ^= c
+	b = (b << 12) | (b >> 20)
+	a += b
+	d ^= a
+	d = (d << 8) | (d >> 24)
+	c += d
+	b ^= c
+	b = (b << 7) | (b >> 25)
+	return a, b, c, d
+}
+
+// XORKeyStream XORs each byte in the given slice with a byte from the
+// cipher's key stream. Dst and src must overlap entirely or not at all.
+//
+// If len(dst) < len(src), XORKeyStream will panic. It is acceptable
+// to pass a dst bigger than src, and in that case, XORKeyStream will
+// only update dst[:len(src)] and will not touch the rest of dst.
+//
+// Multiple calls to XORKeyStream behave as if the concatenation of
+// the src buffers was passed in a single run. That is, Cipher
+// maintains state and does not reset at each XORKeyStream call.
+func (s *Cipher) XORKeyStream(dst, src []byte) {
+	if len(dst) < len(src) {
+		panic("chacha20: output smaller than input")
+	}
+	if subtle.InexactOverlap(dst[:len(src)], src) {
+		panic("chacha20: invalid buffer overlap")
+	}
+
+	// xor src with buffered keystream first
+	if s.len != 0 {
+		buf := s.buf[len(s.buf)-s.len:]
+		if len(src) < len(buf) {
+			buf = buf[:len(src)]
+		}
+		td, ts := dst[:len(buf)], src[:len(buf)] // BCE hint
+		for i, b := range buf {
+			td[i] = ts[i] ^ b
+		}
+		s.len -= len(buf)
+		if s.len != 0 {
+			return
+		}
+		s.buf = [len(s.buf)]byte{} // zero the empty buffer
+		src = src[len(buf):]
+		dst = dst[len(buf):]
+	}
+
+	if len(src) == 0 {
+		return
 	}
+	if haveAsm {
+		if uint64(len(src))+uint64(s.counter)*64 > (1<<38)-64 {
+			panic("chacha20: counter overflow")
+		}
+		s.xorKeyStreamAsm(dst, src)
+		return
+	}
+
+	// set up a 64-byte buffer to pad out the final block if needed
+	// (hoisted out of the main loop to avoid spills)
+	rem := len(src) % 64  // length of final block
+	fin := len(src) - rem // index of final block
+	if rem > 0 {
+		copy(s.buf[len(s.buf)-64:], src[fin:])
+	}
+
+	// pre-calculate most of the first round
+	s1, s5, s9, s13 := quarterRound(j1, s.key[1], s.key[5], s.nonce[0])
+	s2, s6, s10, s14 := quarterRound(j2, s.key[2], s.key[6], s.nonce[1])
+	s3, s7, s11, s15 := quarterRound(j3, s.key[3], s.key[7], s.nonce[2])
+
+	n := len(src)
+	src, dst = src[:n:n], dst[:n:n] // BCE hint
+	for i := 0; i < n; i += 64 {
+		// calculate the remainder of the first round
+		s0, s4, s8, s12 := quarterRound(j0, s.key[0], s.key[4], s.counter)
 
-	x0 += j0
-	x1 += j1
-	x2 += j2
-	x3 += j3
-	x4 += j4
-	x5 += j5
-	x6 += j6
-	x7 += j7
-	x8 += j8
-	x9 += j9
-	x10 += j10
-	x11 += j11
-	x12 += j12
-	x13 += j13
-	x14 += j14
-	x15 += j15
-
-	binary.LittleEndian.PutUint32(out[0:4], x0)
-	binary.LittleEndian.PutUint32(out[4:8], x1)
-	binary.LittleEndian.PutUint32(out[8:12], x2)
-	binary.LittleEndian.PutUint32(out[12:16], x3)
-	binary.LittleEndian.PutUint32(out[16:20], x4)
-	binary.LittleEndian.PutUint32(out[20:24], x5)
-	binary.LittleEndian.PutUint32(out[24:28], x6)
-	binary.LittleEndian.PutUint32(out[28:32], x7)
-	binary.LittleEndian.PutUint32(out[32:36], x8)
-	binary.LittleEndian.PutUint32(out[36:40], x9)
-	binary.LittleEndian.PutUint32(out[40:44], x10)
-	binary.LittleEndian.PutUint32(out[44:48], x11)
-	binary.LittleEndian.PutUint32(out[48:52], x12)
-	binary.LittleEndian.PutUint32(out[52:56], x13)
-	binary.LittleEndian.PutUint32(out[56:60], x14)
-	binary.LittleEndian.PutUint32(out[60:64], x15)
+		// execute the second round
+		x0, x5, x10, x15 := quarterRound(s0, s5, s10, s15)
+		x1, x6, x11, x12 := quarterRound(s1, s6, s11, s12)
+		x2, x7, x8, x13 := quarterRound(s2, s7, s8, s13)
+		x3, x4, x9, x14 := quarterRound(s3, s4, s9, s14)
+
+		// execute the remaining 18 rounds
+		for i := 0; i < 9; i++ {
+			x0, x4, x8, x12 = quarterRound(x0, x4, x8, x12)
+			x1, x5, x9, x13 = quarterRound(x1, x5, x9, x13)
+			x2, x6, x10, x14 = quarterRound(x2, x6, x10, x14)
+			x3, x7, x11, x15 = quarterRound(x3, x7, x11, x15)
+
+			x0, x5, x10, x15 = quarterRound(x0, x5, x10, x15)
+			x1, x6, x11, x12 = quarterRound(x1, x6, x11, x12)
+			x2, x7, x8, x13 = quarterRound(x2, x7, x8, x13)
+			x3, x4, x9, x14 = quarterRound(x3, x4, x9, x14)
+		}
+
+		x0 += j0
+		x1 += j1
+		x2 += j2
+		x3 += j3
+
+		x4 += s.key[0]
+		x5 += s.key[1]
+		x6 += s.key[2]
+		x7 += s.key[3]
+		x8 += s.key[4]
+		x9 += s.key[5]
+		x10 += s.key[6]
+		x11 += s.key[7]
+
+		x12 += s.counter
+		x13 += s.nonce[0]
+		x14 += s.nonce[1]
+		x15 += s.nonce[2]
+
+		// increment the counter
+		s.counter += 1
+		if s.counter == 0 {
+			panic("chacha20: counter overflow")
+		}
+
+		// pad to 64 bytes if needed
+		in, out := src[i:], dst[i:]
+		if i == fin {
+			// src[fin:] has already been copied into s.buf before
+			// the main loop
+			in, out = s.buf[len(s.buf)-64:], s.buf[len(s.buf)-64:]
+		}
+		in, out = in[:64], out[:64] // BCE hint
+
+		// XOR the key stream with the source and write out the result
+		xor(out[0:], in[0:], x0)
+		xor(out[4:], in[4:], x1)
+		xor(out[8:], in[8:], x2)
+		xor(out[12:], in[12:], x3)
+		xor(out[16:], in[16:], x4)
+		xor(out[20:], in[20:], x5)
+		xor(out[24:], in[24:], x6)
+		xor(out[28:], in[28:], x7)
+		xor(out[32:], in[32:], x8)
+		xor(out[36:], in[36:], x9)
+		xor(out[40:], in[40:], x10)
+		xor(out[44:], in[44:], x11)
+		xor(out[48:], in[48:], x12)
+		xor(out[52:], in[52:], x13)
+		xor(out[56:], in[56:], x14)
+		xor(out[60:], in[60:], x15)
+	}
+	// copy any trailing bytes out of the buffer and into dst
+	if rem != 0 {
+		s.len = 64 - rem
+		copy(dst[fin:], s.buf[len(s.buf)-64:])
+	}
+}
+
+// Advance discards bytes in the key stream until the next 64 byte block
+// boundary is reached and updates the counter accordingly. If the key
+// stream is already at a block boundary no bytes will be discarded and
+// the counter will be unchanged.
+func (s *Cipher) Advance() {
+	s.len -= s.len % 64
+	if s.len == 0 {
+		s.buf = [len(s.buf)]byte{}
+	}
 }
 
 // XORKeyStream crypts bytes from in to out using the given key and counters.
 // In and out must overlap entirely or not at all. Counter contains the raw
 // ChaCha20 counter bytes (i.e. block counter followed by nonce).
 func XORKeyStream(out, in []byte, counter *[16]byte, key *[32]byte) {
-	var block [64]byte
-	var counterCopy [16]byte
-	copy(counterCopy[:], counter[:])
-
-	for len(in) >= 64 {
-		core(&block, &counterCopy, key)
-		for i, x := range block {
-			out[i] = in[i] ^ x
-		}
-		u := uint32(1)
-		for i := 0; i < 4; i++ {
-			u += uint32(counterCopy[i])
-			counterCopy[i] = byte(u)
-			u >>= 8
-		}
-		in = in[64:]
-		out = out[64:]
+	s := Cipher{
+		key: [8]uint32{
+			binary.LittleEndian.Uint32(key[0:4]),
+			binary.LittleEndian.Uint32(key[4:8]),
+			binary.LittleEndian.Uint32(key[8:12]),
+			binary.LittleEndian.Uint32(key[12:16]),
+			binary.LittleEndian.Uint32(key[16:20]),
+			binary.LittleEndian.Uint32(key[20:24]),
+			binary.LittleEndian.Uint32(key[24:28]),
+			binary.LittleEndian.Uint32(key[28:32]),
+		},
+		nonce: [3]uint32{
+			binary.LittleEndian.Uint32(counter[4:8]),
+			binary.LittleEndian.Uint32(counter[8:12]),
+			binary.LittleEndian.Uint32(counter[12:16]),
+		},
+		counter: binary.LittleEndian.Uint32(counter[0:4]),
 	}
+	s.XORKeyStream(out, in)
+}
 
-	if len(in) > 0 {
-		core(&block, &counterCopy, key)
-		for i, v := range in {
-			out[i] = v ^ block[i]
-		}
+// HChaCha20 uses the ChaCha20 core to generate a derived key from a key and a
+// nonce. It should only be used as part of the XChaCha20 construction.
+func HChaCha20(key *[8]uint32, nonce *[4]uint32) [8]uint32 {
+	x0, x1, x2, x3 := j0, j1, j2, j3
+	x4, x5, x6, x7 := key[0], key[1], key[2], key[3]
+	x8, x9, x10, x11 := key[4], key[5], key[6], key[7]
+	x12, x13, x14, x15 := nonce[0], nonce[1], nonce[2], nonce[3]
+
+	for i := 0; i < 10; i++ {
+		x0, x4, x8, x12 = quarterRound(x0, x4, x8, x12)
+		x1, x5, x9, x13 = quarterRound(x1, x5, x9, x13)
+		x2, x6, x10, x14 = quarterRound(x2, x6, x10, x14)
+		x3, x7, x11, x15 = quarterRound(x3, x7, x11, x15)
+
+		x0, x5, x10, x15 = quarterRound(x0, x5, x10, x15)
+		x1, x6, x11, x12 = quarterRound(x1, x6, x11, x12)
+		x2, x7, x8, x13 = quarterRound(x2, x7, x8, x13)
+		x3, x4, x9, x14 = quarterRound(x3, x4, x9, x14)
 	}
+
+	var out [8]uint32
+	out[0], out[1], out[2], out[3] = x0, x1, x2, x3
+	out[4], out[5], out[6], out[7] = x12, x13, x14, x15
+	return out
 }
diff --git a/vendor/golang.org/x/crypto/internal/chacha20/chacha_noasm.go b/vendor/golang.org/x/crypto/internal/chacha20/chacha_noasm.go
new file mode 100644
index 00000000..91520d1d
--- /dev/null
+++ b/vendor/golang.org/x/crypto/internal/chacha20/chacha_noasm.go
@@ -0,0 +1,16 @@
+// Copyright 2018 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// +build !s390x gccgo appengine
+
+package chacha20
+
+const (
+	bufSize = 64
+	haveAsm = false
+)
+
+func (*Cipher) xorKeyStreamAsm(dst, src []byte) {
+	panic("not implemented")
+}
diff --git a/vendor/golang.org/x/crypto/internal/chacha20/chacha_s390x.go b/vendor/golang.org/x/crypto/internal/chacha20/chacha_s390x.go
new file mode 100644
index 00000000..0c1c671c
--- /dev/null
+++ b/vendor/golang.org/x/crypto/internal/chacha20/chacha_s390x.go
@@ -0,0 +1,30 @@
+// Copyright 2018 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// +build s390x,!gccgo,!appengine
+
+package chacha20
+
+var haveAsm = hasVectorFacility()
+
+const bufSize = 256
+
+// hasVectorFacility reports whether the machine supports the vector
+// facility (vx).
+// Implementation in asm_s390x.s.
+func hasVectorFacility() bool
+
+// xorKeyStreamVX is an assembly implementation of XORKeyStream. It must only
+// be called when the vector facility is available.
+// Implementation in asm_s390x.s.
+//go:noescape
+func xorKeyStreamVX(dst, src []byte, key *[8]uint32, nonce *[3]uint32, counter *uint32, buf *[256]byte, len *int)
+
+func (c *Cipher) xorKeyStreamAsm(dst, src []byte) {
+	xorKeyStreamVX(dst, src, &c.key, &c.nonce, &c.counter, &c.buf, &c.len)
+}
+
+// EXRL targets, DO NOT CALL!
+func mvcSrcToBuf()
+func mvcBufToDst()
diff --git a/vendor/golang.org/x/crypto/internal/chacha20/chacha_s390x.s b/vendor/golang.org/x/crypto/internal/chacha20/chacha_s390x.s
new file mode 100644
index 00000000..98427c5e
--- /dev/null
+++ b/vendor/golang.org/x/crypto/internal/chacha20/chacha_s390x.s
@@ -0,0 +1,283 @@
+// Copyright 2018 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// +build s390x,!gccgo,!appengine
+
+#include "go_asm.h"
+#include "textflag.h"
+
+// This is an implementation of the ChaCha20 encryption algorithm as
+// specified in RFC 7539. It uses vector instructions to compute
+// 4 keystream blocks in parallel (256 bytes) which are then XORed
+// with the bytes in the input slice.
+
+GLOBL ·constants<>(SB), RODATA|NOPTR, $32
+// BSWAP: swap bytes in each 4-byte element
+DATA ·constants<>+0x00(SB)/4, $0x03020100
+DATA ·constants<>+0x04(SB)/4, $0x07060504
+DATA ·constants<>+0x08(SB)/4, $0x0b0a0908
+DATA ·constants<>+0x0c(SB)/4, $0x0f0e0d0c
+// J0: [j0, j1, j2, j3]
+DATA ·constants<>+0x10(SB)/4, $0x61707865
+DATA ·constants<>+0x14(SB)/4, $0x3320646e
+DATA ·constants<>+0x18(SB)/4, $0x79622d32
+DATA ·constants<>+0x1c(SB)/4, $0x6b206574
+
+// EXRL targets:
+TEXT ·mvcSrcToBuf(SB), NOFRAME|NOSPLIT, $0
+	MVC $1, (R1), (R8)
+	RET
+
+TEXT ·mvcBufToDst(SB), NOFRAME|NOSPLIT, $0
+	MVC $1, (R8), (R9)
+	RET
+
+#define BSWAP V5
+#define J0    V6
+#define KEY0  V7
+#define KEY1  V8
+#define NONCE V9
+#define CTR   V10
+#define M0    V11
+#define M1    V12
+#define M2    V13
+#define M3    V14
+#define INC   V15
+#define X0    V16
+#define X1    V17
+#define X2    V18
+#define X3    V19
+#define X4    V20
+#define X5    V21
+#define X6    V22
+#define X7    V23
+#define X8    V24
+#define X9    V25
+#define X10   V26
+#define X11   V27
+#define X12   V28
+#define X13   V29
+#define X14   V30
+#define X15   V31
+
+#define NUM_ROUNDS 20
+
+#define ROUND4(a0, a1, a2, a3, b0, b1, b2, b3, c0, c1, c2, c3, d0, d1, d2, d3) \
+	VAF    a1, a0, a0  \
+	VAF    b1, b0, b0  \
+	VAF    c1, c0, c0  \
+	VAF    d1, d0, d0  \
+	VX     a0, a2, a2  \
+	VX     b0, b2, b2  \
+	VX     c0, c2, c2  \
+	VX     d0, d2, d2  \
+	VERLLF $16, a2, a2 \
+	VERLLF $16, b2, b2 \
+	VERLLF $16, c2, c2 \
+	VERLLF $16, d2, d2 \
+	VAF    a2, a3, a3  \
+	VAF    b2, b3, b3  \
+	VAF    c2, c3, c3  \
+	VAF    d2, d3, d3  \
+	VX     a3, a1, a1  \
+	VX     b3, b1, b1  \
+	VX     c3, c1, c1  \
+	VX     d3, d1, d1  \
+	VERLLF $12, a1, a1 \
+	VERLLF $12, b1, b1 \
+	VERLLF $12, c1, c1 \
+	VERLLF $12, d1, d1 \
+	VAF    a1, a0, a0  \
+	VAF    b1, b0, b0  \
+	VAF    c1, c0, c0  \
+	VAF    d1, d0, d0  \
+	VX     a0, a2, a2  \
+	VX     b0, b2, b2  \
+	VX     c0, c2, c2  \
+	VX     d0, d2, d2  \
+	VERLLF $8, a2, a2  \
+	VERLLF $8, b2, b2  \
+	VERLLF $8, c2, c2  \
+	VERLLF $8, d2, d2  \
+	VAF    a2, a3, a3  \
+	VAF    b2, b3, b3  \
+	VAF    c2, c3, c3  \
+	VAF    d2, d3, d3  \
+	VX     a3, a1, a1  \
+	VX     b3, b1, b1  \
+	VX     c3, c1, c1  \
+	VX     d3, d1, d1  \
+	VERLLF $7, a1, a1  \
+	VERLLF $7, b1, b1  \
+	VERLLF $7, c1, c1  \
+	VERLLF $7, d1, d1
+
+#define PERMUTE(mask, v0, v1, v2, v3) \
+	VPERM v0, v0, mask, v0 \
+	VPERM v1, v1, mask, v1 \
+	VPERM v2, v2, mask, v2 \
+	VPERM v3, v3, mask, v3
+
+#define ADDV(x, v0, v1, v2, v3) \
+	VAF x, v0, v0 \
+	VAF x, v1, v1 \
+	VAF x, v2, v2 \
+	VAF x, v3, v3
+
+#define XORV(off, dst, src, v0, v1, v2, v3) \
+	VLM  off(src), M0, M3          \
+	PERMUTE(BSWAP, v0, v1, v2, v3) \
+	VX   v0, M0, M0                \
+	VX   v1, M1, M1                \
+	VX   v2, M2, M2                \
+	VX   v3, M3, M3                \
+	VSTM M0, M3, off(dst)
+
+#define SHUFFLE(a, b, c, d, t, u, v, w) \
+	VMRHF a, c, t \ // t = {a[0], c[0], a[1], c[1]}
+	VMRHF b, d, u \ // u = {b[0], d[0], b[1], d[1]}
+	VMRLF a, c, v \ // v = {a[2], c[2], a[3], c[3]}
+	VMRLF b, d, w \ // w = {b[2], d[2], b[3], d[3]}
+	VMRHF t, u, a \ // a = {a[0], b[0], c[0], d[0]}
+	VMRLF t, u, b \ // b = {a[1], b[1], c[1], d[1]}
+	VMRHF v, w, c \ // c = {a[2], b[2], c[2], d[2]}
+	VMRLF v, w, d // d = {a[3], b[3], c[3], d[3]}
+
+// func xorKeyStreamVX(dst, src []byte, key *[8]uint32, nonce *[3]uint32, counter *uint32, buf *[256]byte, len *int)
+TEXT ·xorKeyStreamVX(SB), NOSPLIT, $0
+	MOVD $·constants<>(SB), R1
+	MOVD dst+0(FP), R2         // R2=&dst[0]
+	LMG  src+24(FP), R3, R4    // R3=&src[0] R4=len(src)
+	MOVD key+48(FP), R5        // R5=key
+	MOVD nonce+56(FP), R6      // R6=nonce
+	MOVD counter+64(FP), R7    // R7=counter
+	MOVD buf+72(FP), R8        // R8=buf
+	MOVD len+80(FP), R9        // R9=len
+
+	// load BSWAP and J0
+	VLM (R1), BSWAP, J0
+
+	// set up tail buffer
+	ADD     $-1, R4, R12
+	MOVBZ   R12, R12
+	CMPUBEQ R12, $255, aligned
+	MOVD    R4, R1
+	AND     $~255, R1
+	MOVD    $(R3)(R1*1), R1
+	EXRL    $·mvcSrcToBuf(SB), R12
+	MOVD    $255, R0
+	SUB     R12, R0
+	MOVD    R0, (R9)               // update len
+
+aligned:
+	// setup
+	MOVD  $95, R0
+	VLM   (R5), KEY0, KEY1
+	VLL   R0, (R6), NONCE
+	VZERO M0
+	VLEIB $7, $32, M0
+	VSRLB M0, NONCE, NONCE
+
+	// initialize counter values
+	VLREPF (R7), CTR
+	VZERO  INC
+	VLEIF  $1, $1, INC
+	VLEIF  $2, $2, INC
+	VLEIF  $3, $3, INC
+	VAF    INC, CTR, CTR
+	VREPIF $4, INC
+
+chacha:
+	VREPF $0, J0, X0
+	VREPF $1, J0, X1
+	VREPF $2, J0, X2
+	VREPF $3, J0, X3
+	VREPF $0, KEY0, X4
+	VREPF $1, KEY0, X5
+	VREPF $2, KEY0, X6
+	VREPF $3, KEY0, X7
+	VREPF $0, KEY1, X8
+	VREPF $1, KEY1, X9
+	VREPF $2, KEY1, X10
+	VREPF $3, KEY1, X11
+	VLR   CTR, X12
+	VREPF $1, NONCE, X13
+	VREPF $2, NONCE, X14
+	VREPF $3, NONCE, X15
+
+	MOVD $(NUM_ROUNDS/2), R1
+
+loop:
+	ROUND4(X0, X4, X12,  X8, X1, X5, X13,  X9, X2, X6, X14, X10, X3, X7, X15, X11)
+	ROUND4(X0, X5, X15, X10, X1, X6, X12, X11, X2, X7, X13, X8,  X3, X4, X14, X9)
+
+	ADD $-1, R1
+	BNE loop
+
+	// decrement length
+	ADD $-256, R4
+	BLT tail
+
+continue:
+	// rearrange vectors
+	SHUFFLE(X0, X1, X2, X3, M0, M1, M2, M3)
+	ADDV(J0, X0, X1, X2, X3)
+	SHUFFLE(X4, X5, X6, X7, M0, M1, M2, M3)
+	ADDV(KEY0, X4, X5, X6, X7)
+	SHUFFLE(X8, X9, X10, X11, M0, M1, M2, M3)
+	ADDV(KEY1, X8, X9, X10, X11)
+	VAF CTR, X12, X12
+	SHUFFLE(X12, X13, X14, X15, M0, M1, M2, M3)
+	ADDV(NONCE, X12, X13, X14, X15)
+
+	// increment counters
+	VAF INC, CTR, CTR
+
+	// xor keystream with plaintext
+	XORV(0*64, R2, R3, X0, X4,  X8, X12)
+	XORV(1*64, R2, R3, X1, X5,  X9, X13)
+	XORV(2*64, R2, R3, X2, X6, X10, X14)
+	XORV(3*64, R2, R3, X3, X7, X11, X15)
+
+	// increment pointers
+	MOVD $256(R2), R2
+	MOVD $256(R3), R3
+
+	CMPBNE  R4, $0, chacha
+	CMPUBEQ R12, $255, return
+	EXRL    $·mvcBufToDst(SB), R12 // len was updated during setup
+
+return:
+	VSTEF $0, CTR, (R7)
+	RET
+
+tail:
+	MOVD R2, R9
+	MOVD R8, R2
+	MOVD R8, R3
+	MOVD $0, R4
+	JMP  continue
+
+// func hasVectorFacility() bool
+TEXT ·hasVectorFacility(SB), NOSPLIT, $24-1
+	MOVD  $x-24(SP), R1
+	XC    $24, 0(R1), 0(R1) // clear the storage
+	MOVD  $2, R0            // R0 is the number of double words stored -1
+	WORD  $0xB2B01000       // STFLE 0(R1)
+	XOR   R0, R0            // reset the value of R0
+	MOVBZ z-8(SP), R1
+	AND   $0x40, R1
+	BEQ   novector
+
+vectorinstalled:
+	// check if the vector instruction has been enabled
+	VLEIB  $0, $0xF, V16
+	VLGVB  $0, V16, R1
+	CMPBNE R1, $0xF, novector
+	MOVB   $1, ret+0(FP)      // have vx
+	RET
+
+novector:
+	MOVB $0, ret+0(FP) // no vx
+	RET
diff --git a/vendor/golang.org/x/crypto/internal/chacha20/xor.go b/vendor/golang.org/x/crypto/internal/chacha20/xor.go
new file mode 100644
index 00000000..9c5ba0b3
--- /dev/null
+++ b/vendor/golang.org/x/crypto/internal/chacha20/xor.go
@@ -0,0 +1,43 @@
+// Copyright 2018 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found src the LICENSE file.
+
+package chacha20
+
+import (
+	"runtime"
+)
+
+// Platforms that have fast unaligned 32-bit little endian accesses.
+const unaligned = runtime.GOARCH == "386" ||
+	runtime.GOARCH == "amd64" ||
+	runtime.GOARCH == "arm64" ||
+	runtime.GOARCH == "ppc64le" ||
+	runtime.GOARCH == "s390x"
+
+// xor reads a little endian uint32 from src, XORs it with u and
+// places the result in little endian byte order in dst.
+func xor(dst, src []byte, u uint32) {
+	_, _ = src[3], dst[3] // eliminate bounds checks
+	if unaligned {
+		// The compiler should optimize this code into
+		// 32-bit unaligned little endian loads and stores.
+		// TODO: delete once the compiler does a reliably
+		// good job with the generic code below.
+		// See issue #25111 for more details.
+		v := uint32(src[0])
+		v |= uint32(src[1]) << 8
+		v |= uint32(src[2]) << 16
+		v |= uint32(src[3]) << 24
+		v ^= u
+		dst[0] = byte(v)
+		dst[1] = byte(v >> 8)
+		dst[2] = byte(v >> 16)
+		dst[3] = byte(v >> 24)
+	} else {
+		dst[0] = src[0] ^ byte(u)
+		dst[1] = src[1] ^ byte(u>>8)
+		dst[2] = src[2] ^ byte(u>>16)
+		dst[3] = src[3] ^ byte(u>>24)
+	}
+}