Moving to glide

8 files changed, 2618 insertions, 0 deletions

diff --git a/vendor/golang.org/x/crypto/poly1305/const_amd64.s b/vendor/golang.org/x/crypto/poly1305/const_amd64.s
new file mode 100644
index 000000000..8e861f337
--- /dev/null
+++ b/vendor/golang.org/x/crypto/poly1305/const_amd64.s
@@ -0,0 +1,45 @@
+// 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: http://bench.cr.yp.to/supercop.html
+
+// +build amd64,!gccgo,!appengine
+
+DATA ·SCALE(SB)/8, $0x37F4000000000000
+GLOBL ·SCALE(SB), 8, $8
+DATA ·TWO32(SB)/8, $0x41F0000000000000
+GLOBL ·TWO32(SB), 8, $8
+DATA ·TWO64(SB)/8, $0x43F0000000000000
+GLOBL ·TWO64(SB), 8, $8
+DATA ·TWO96(SB)/8, $0x45F0000000000000
+GLOBL ·TWO96(SB), 8, $8
+DATA ·ALPHA32(SB)/8, $0x45E8000000000000
+GLOBL ·ALPHA32(SB), 8, $8
+DATA ·ALPHA64(SB)/8, $0x47E8000000000000
+GLOBL ·ALPHA64(SB), 8, $8
+DATA ·ALPHA96(SB)/8, $0x49E8000000000000
+GLOBL ·ALPHA96(SB), 8, $8
+DATA ·ALPHA130(SB)/8, $0x4C08000000000000
+GLOBL ·ALPHA130(SB), 8, $8
+DATA ·DOFFSET0(SB)/8, $0x4330000000000000
+GLOBL ·DOFFSET0(SB), 8, $8
+DATA ·DOFFSET1(SB)/8, $0x4530000000000000
+GLOBL ·DOFFSET1(SB), 8, $8
+DATA ·DOFFSET2(SB)/8, $0x4730000000000000
+GLOBL ·DOFFSET2(SB), 8, $8
+DATA ·DOFFSET3(SB)/8, $0x4930000000000000
+GLOBL ·DOFFSET3(SB), 8, $8
+DATA ·DOFFSET3MINUSTWO128(SB)/8, $0x492FFFFE00000000
+GLOBL ·DOFFSET3MINUSTWO128(SB), 8, $8
+DATA ·HOFFSET0(SB)/8, $0x43300001FFFFFFFB
+GLOBL ·HOFFSET0(SB), 8, $8
+DATA ·HOFFSET1(SB)/8, $0x45300001FFFFFFFE
+GLOBL ·HOFFSET1(SB), 8, $8
+DATA ·HOFFSET2(SB)/8, $0x47300001FFFFFFFE
+GLOBL ·HOFFSET2(SB), 8, $8
+DATA ·HOFFSET3(SB)/8, $0x49300003FFFFFFFE
+GLOBL ·HOFFSET3(SB), 8, $8
+DATA ·ROUNDING(SB)/2, $0x137f
+GLOBL ·ROUNDING(SB), 8, $2
diff --git a/vendor/golang.org/x/crypto/poly1305/poly1305.go b/vendor/golang.org/x/crypto/poly1305/poly1305.go
new file mode 100644
index 000000000..4a5f826f7
--- /dev/null
+++ b/vendor/golang.org/x/crypto/poly1305/poly1305.go
@@ -0,0 +1,32 @@
+// 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 poly1305 implements Poly1305 one-time message authentication code as specified in http://cr.yp.to/mac/poly1305-20050329.pdf.
+
+Poly1305 is a fast, one-time authentication function. It is infeasible for an
+attacker to generate an authenticator for a message without the key. However, a
+key must only be used for a single message. Authenticating two different
+messages with the same key allows an attacker to forge authenticators for other
+messages with the same key.
+
+Poly1305 was originally coupled with AES in order to make Poly1305-AES. AES was
+used with a fixed key in order to generate one-time keys from an nonce.
+However, in this package AES isn't used and the one-time key is specified
+directly.
+*/
+package poly1305 // import "golang.org/x/crypto/poly1305"
+
+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.
+func Verify(mac *[16]byte, m []byte, key *[32]byte) bool {
+	var tmp [16]byte
+	Sum(&tmp, m, key)
+	return subtle.ConstantTimeCompare(tmp[:], mac[:]) == 1
+}
diff --git a/vendor/golang.org/x/crypto/poly1305/poly1305_amd64.s b/vendor/golang.org/x/crypto/poly1305/poly1305_amd64.s
new file mode 100644
index 000000000..f8d4ee928
--- /dev/null
+++ b/vendor/golang.org/x/crypto/poly1305/poly1305_amd64.s
@@ -0,0 +1,497 @@
+// 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: http://bench.cr.yp.to/supercop.html
+
+// +build amd64,!gccgo,!appengine
+
+// func poly1305(out *[16]byte, m *byte, mlen uint64, key *[32]key)
+TEXT ·poly1305(SB),0,$224-32
+	MOVQ out+0(FP),DI
+	MOVQ m+8(FP),SI
+	MOVQ mlen+16(FP),DX
+	MOVQ key+24(FP),CX
+
+	MOVQ SP,R11
+	MOVQ $31,R9
+	NOTQ R9
+	ANDQ R9,SP
+	ADDQ $32,SP
+
+	MOVQ R11,32(SP)
+	MOVQ R12,40(SP)
+	MOVQ R13,48(SP)
+	MOVQ R14,56(SP)
+	MOVQ R15,64(SP)
+	MOVQ BX,72(SP)
+	MOVQ BP,80(SP)
+	FLDCW ·ROUNDING(SB)
+	MOVL 0(CX),R8
+	MOVL 4(CX),R9
+	MOVL 8(CX),AX
+	MOVL 12(CX),R10
+	MOVQ DI,88(SP)
+	MOVQ CX,96(SP)
+	MOVL $0X43300000,108(SP)
+	MOVL $0X45300000,116(SP)
+	MOVL $0X47300000,124(SP)
+	MOVL $0X49300000,132(SP)
+	ANDL $0X0FFFFFFF,R8
+	ANDL $0X0FFFFFFC,R9
+	ANDL $0X0FFFFFFC,AX
+	ANDL $0X0FFFFFFC,R10
+	MOVL R8,104(SP)
+	MOVL R9,112(SP)
+	MOVL AX,120(SP)
+	MOVL R10,128(SP)
+	FMOVD 104(SP), F0
+	FSUBD ·DOFFSET0(SB), F0
+	FMOVD 112(SP), F0
+	FSUBD ·DOFFSET1(SB), F0
+	FMOVD 120(SP), F0
+	FSUBD ·DOFFSET2(SB), F0
+	FMOVD 128(SP), F0
+	FSUBD ·DOFFSET3(SB), F0
+	FXCHD F0, F3
+	FMOVDP F0, 136(SP)
+	FXCHD F0, F1
+	FMOVD F0, 144(SP)
+	FMULD ·SCALE(SB), F0
+	FMOVDP F0, 152(SP)
+	FMOVD F0, 160(SP)
+	FMULD ·SCALE(SB), F0
+	FMOVDP F0, 168(SP)
+	FMOVD F0, 176(SP)
+	FMULD ·SCALE(SB), F0
+	FMOVDP F0, 184(SP)
+	FLDZ
+	FLDZ
+	FLDZ
+	FLDZ
+	CMPQ DX,$16
+	JB ADDATMOST15BYTES
+	INITIALATLEAST16BYTES:
+	MOVL 12(SI),DI
+	MOVL 8(SI),CX
+	MOVL 4(SI),R8
+	MOVL 0(SI),R9
+	MOVL DI,128(SP)
+	MOVL CX,120(SP)
+	MOVL R8,112(SP)
+	MOVL R9,104(SP)
+	ADDQ $16,SI
+	SUBQ $16,DX
+	FXCHD F0, F3
+	FADDD 128(SP), F0
+	FSUBD ·DOFFSET3MINUSTWO128(SB), F0
+	FXCHD F0, F1
+	FADDD 112(SP), F0
+	FSUBD ·DOFFSET1(SB), F0
+	FXCHD F0, F2
+	FADDD 120(SP), F0
+	FSUBD ·DOFFSET2(SB), F0
+	FXCHD F0, F3
+	FADDD 104(SP), F0
+	FSUBD ·DOFFSET0(SB), F0
+	CMPQ DX,$16
+	JB MULTIPLYADDATMOST15BYTES
+	MULTIPLYADDATLEAST16BYTES:
+	MOVL 12(SI),DI
+	MOVL 8(SI),CX
+	MOVL 4(SI),R8
+	MOVL 0(SI),R9
+	MOVL DI,128(SP)
+	MOVL CX,120(SP)
+	MOVL R8,112(SP)
+	MOVL R9,104(SP)
+	ADDQ $16,SI
+	SUBQ $16,DX
+	FMOVD ·ALPHA130(SB), F0
+	FADDD F2,F0
+	FSUBD ·ALPHA130(SB), F0
+	FSUBD F0,F2
+	FMULD ·SCALE(SB), F0
+	FMOVD ·ALPHA32(SB), F0
+	FADDD F2,F0
+	FSUBD ·ALPHA32(SB), F0
+	FSUBD F0,F2
+	FXCHD F0, F2
+	FADDDP F0,F1
+	FMOVD ·ALPHA64(SB), F0
+	FADDD F4,F0
+	FSUBD ·ALPHA64(SB), F0
+	FSUBD F0,F4
+	FMOVD ·ALPHA96(SB), F0
+	FADDD F6,F0
+	FSUBD ·ALPHA96(SB), F0
+	FSUBD F0,F6
+	FXCHD F0, F6
+	FADDDP F0,F1
+	FXCHD F0, F3
+	FADDDP F0,F5
+	FXCHD F0, F3
+	FADDDP F0,F1
+	FMOVD 176(SP), F0
+	FMULD F3,F0
+	FMOVD 160(SP), F0
+	FMULD F4,F0
+	FMOVD 144(SP), F0
+	FMULD F5,F0
+	FMOVD 136(SP), F0
+	FMULDP F0,F6
+	FMOVD 160(SP), F0
+	FMULD F4,F0
+	FADDDP F0,F3
+	FMOVD 144(SP), F0
+	FMULD F4,F0
+	FADDDP F0,F2
+	FMOVD 136(SP), F0
+	FMULD F4,F0
+	FADDDP F0,F1
+	FMOVD 184(SP), F0
+	FMULDP F0,F4
+	FXCHD F0, F3
+	FADDDP F0,F5
+	FMOVD 144(SP), F0
+	FMULD F4,F0
+	FADDDP F0,F2
+	FMOVD 136(SP), F0
+	FMULD F4,F0
+	FADDDP F0,F1
+	FMOVD 184(SP), F0
+	FMULD F4,F0
+	FADDDP F0,F3
+	FMOVD 168(SP), F0
+	FMULDP F0,F4
+	FXCHD F0, F3
+	FADDDP F0,F4
+	FMOVD 136(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F1
+	FXCHD F0, F3
+	FMOVD 184(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F3
+	FXCHD F0, F1
+	FMOVD 168(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F1
+	FMOVD 152(SP), F0
+	FMULDP F0,F5
+	FXCHD F0, F4
+	FADDDP F0,F1
+	CMPQ DX,$16
+	FXCHD F0, F2
+	FMOVD 128(SP), F0
+	FSUBD ·DOFFSET3MINUSTWO128(SB), F0
+	FADDDP F0,F1
+	FXCHD F0, F1
+	FMOVD 120(SP), F0
+	FSUBD ·DOFFSET2(SB), F0
+	FADDDP F0,F1
+	FXCHD F0, F3
+	FMOVD 112(SP), F0
+	FSUBD ·DOFFSET1(SB), F0
+	FADDDP F0,F1
+	FXCHD F0, F2
+	FMOVD 104(SP), F0
+	FSUBD ·DOFFSET0(SB), F0
+	FADDDP F0,F1
+	JAE MULTIPLYADDATLEAST16BYTES
+	MULTIPLYADDATMOST15BYTES:
+	FMOVD ·ALPHA130(SB), F0
+	FADDD F2,F0
+	FSUBD ·ALPHA130(SB), F0
+	FSUBD F0,F2
+	FMULD ·SCALE(SB), F0
+	FMOVD ·ALPHA32(SB), F0
+	FADDD F2,F0
+	FSUBD ·ALPHA32(SB), F0
+	FSUBD F0,F2
+	FMOVD ·ALPHA64(SB), F0
+	FADDD F5,F0
+	FSUBD ·ALPHA64(SB), F0
+	FSUBD F0,F5
+	FMOVD ·ALPHA96(SB), F0
+	FADDD F7,F0
+	FSUBD ·ALPHA96(SB), F0
+	FSUBD F0,F7
+	FXCHD F0, F7
+	FADDDP F0,F1
+	FXCHD F0, F5
+	FADDDP F0,F1
+	FXCHD F0, F3
+	FADDDP F0,F5
+	FADDDP F0,F1
+	FMOVD 176(SP), F0
+	FMULD F1,F0
+	FMOVD 160(SP), F0
+	FMULD F2,F0
+	FMOVD 144(SP), F0
+	FMULD F3,F0
+	FMOVD 136(SP), F0
+	FMULDP F0,F4
+	FMOVD 160(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F3
+	FMOVD 144(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F2
+	FMOVD 136(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F1
+	FMOVD 184(SP), F0
+	FMULDP F0,F5
+	FXCHD F0, F4
+	FADDDP F0,F3
+	FMOVD 144(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F2
+	FMOVD 136(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F1
+	FMOVD 184(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F4
+	FMOVD 168(SP), F0
+	FMULDP F0,F5
+	FXCHD F0, F4
+	FADDDP F0,F2
+	FMOVD 136(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F1
+	FMOVD 184(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F4
+	FMOVD 168(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F3
+	FMOVD 152(SP), F0
+	FMULDP F0,F5
+	FXCHD F0, F4
+	FADDDP F0,F1
+	ADDATMOST15BYTES:
+	CMPQ DX,$0
+	JE NOMOREBYTES
+	MOVL $0,0(SP)
+	MOVL $0, 4 (SP)
+	MOVL $0, 8 (SP)
+	MOVL $0, 12 (SP)
+	LEAQ 0(SP),DI
+	MOVQ DX,CX
+	REP; MOVSB
+	MOVB $1,0(DI)
+	MOVL  12 (SP),DI
+	MOVL  8 (SP),SI
+	MOVL  4 (SP),DX
+	MOVL 0(SP),CX
+	MOVL DI,128(SP)
+	MOVL SI,120(SP)
+	MOVL DX,112(SP)
+	MOVL CX,104(SP)
+	FXCHD F0, F3
+	FADDD 128(SP), F0
+	FSUBD ·DOFFSET3(SB), F0
+	FXCHD F0, F2
+	FADDD 120(SP), F0
+	FSUBD ·DOFFSET2(SB), F0
+	FXCHD F0, F1
+	FADDD 112(SP), F0
+	FSUBD ·DOFFSET1(SB), F0
+	FXCHD F0, F3
+	FADDD 104(SP), F0
+	FSUBD ·DOFFSET0(SB), F0
+	FMOVD ·ALPHA130(SB), F0
+	FADDD F3,F0
+	FSUBD ·ALPHA130(SB), F0
+	FSUBD F0,F3
+	FMULD ·SCALE(SB), F0
+	FMOVD ·ALPHA32(SB), F0
+	FADDD F2,F0
+	FSUBD ·ALPHA32(SB), F0
+	FSUBD F0,F2
+	FMOVD ·ALPHA64(SB), F0
+	FADDD F6,F0
+	FSUBD ·ALPHA64(SB), F0
+	FSUBD F0,F6
+	FMOVD ·ALPHA96(SB), F0
+	FADDD F5,F0
+	FSUBD ·ALPHA96(SB), F0
+	FSUBD F0,F5
+	FXCHD F0, F4
+	FADDDP F0,F3
+	FXCHD F0, F6
+	FADDDP F0,F1
+	FXCHD F0, F3
+	FADDDP F0,F5
+	FXCHD F0, F3
+	FADDDP F0,F1
+	FMOVD 176(SP), F0
+	FMULD F3,F0
+	FMOVD 160(SP), F0
+	FMULD F4,F0
+	FMOVD 144(SP), F0
+	FMULD F5,F0
+	FMOVD 136(SP), F0
+	FMULDP F0,F6
+	FMOVD 160(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F3
+	FMOVD 144(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F2
+	FMOVD 136(SP), F0
+	FMULD F5,F0
+	FADDDP F0,F1
+	FMOVD 184(SP), F0
+	FMULDP F0,F5
+	FXCHD F0, F4
+	FADDDP F0,F5
+	FMOVD 144(SP), F0
+	FMULD F6,F0
+	FADDDP F0,F2
+	FMOVD 136(SP), F0
+	FMULD F6,F0
+	FADDDP F0,F1
+	FMOVD 184(SP), F0
+	FMULD F6,F0
+	FADDDP F0,F4
+	FMOVD 168(SP), F0
+	FMULDP F0,F6
+	FXCHD F0, F5
+	FADDDP F0,F4
+	FMOVD 136(SP), F0
+	FMULD F2,F0
+	FADDDP F0,F1
+	FMOVD 184(SP), F0
+	FMULD F2,F0
+	FADDDP F0,F5
+	FMOVD 168(SP), F0
+	FMULD F2,F0
+	FADDDP F0,F3
+	FMOVD 152(SP), F0
+	FMULDP F0,F2
+	FXCHD F0, F1
+	FADDDP F0,F3
+	FXCHD F0, F3
+	FXCHD F0, F2
+	NOMOREBYTES:
+	MOVL $0,R10
+	FMOVD ·ALPHA130(SB), F0
+	FADDD F4,F0
+	FSUBD ·ALPHA130(SB), F0
+	FSUBD F0,F4
+	FMULD ·SCALE(SB), F0
+	FMOVD ·ALPHA32(SB), F0
+	FADDD F2,F0
+	FSUBD ·ALPHA32(SB), F0
+	FSUBD F0,F2
+	FMOVD ·ALPHA64(SB), F0
+	FADDD F4,F0
+	FSUBD ·ALPHA64(SB), F0
+	FSUBD F0,F4
+	FMOVD ·ALPHA96(SB), F0
+	FADDD F6,F0
+	FSUBD ·ALPHA96(SB), F0
+	FXCHD F0, F6
+	FSUBD F6,F0
+	FXCHD F0, F4
+	FADDDP F0,F3
+	FXCHD F0, F4
+	FADDDP F0,F1
+	FXCHD F0, F2
+	FADDDP F0,F3
+	FXCHD F0, F4
+	FADDDP F0,F3
+	FXCHD F0, F3
+	FADDD ·HOFFSET0(SB), F0
+	FXCHD F0, F3
+	FADDD ·HOFFSET1(SB), F0
+	FXCHD F0, F1
+	FADDD ·HOFFSET2(SB), F0
+	FXCHD F0, F2
+	FADDD ·HOFFSET3(SB), F0
+	FXCHD F0, F3
+	FMOVDP F0, 104(SP)
+	FMOVDP F0, 112(SP)
+	FMOVDP F0, 120(SP)
+	FMOVDP F0, 128(SP)
+	MOVL 108(SP),DI
+	ANDL $63,DI
+	MOVL 116(SP),SI
+	ANDL $63,SI
+	MOVL 124(SP),DX
+	ANDL $63,DX
+	MOVL 132(SP),CX
+	ANDL $63,CX
+	MOVL 112(SP),R8
+	ADDL DI,R8
+	MOVQ R8,112(SP)
+	MOVL 120(SP),DI
+	ADCL SI,DI
+	MOVQ DI,120(SP)
+	MOVL 128(SP),DI
+	ADCL DX,DI
+	MOVQ DI,128(SP)
+	MOVL R10,DI
+	ADCL CX,DI
+	MOVQ DI,136(SP)
+	MOVQ $5,DI
+	MOVL 104(SP),SI
+	ADDL SI,DI
+	MOVQ DI,104(SP)
+	MOVL R10,DI
+	MOVQ 112(SP),DX
+	ADCL DX,DI
+	MOVQ DI,112(SP)
+	MOVL R10,DI
+	MOVQ 120(SP),CX
+	ADCL CX,DI
+	MOVQ DI,120(SP)
+	MOVL R10,DI
+	MOVQ 128(SP),R8
+	ADCL R8,DI
+	MOVQ DI,128(SP)
+	MOVQ $0XFFFFFFFC,DI
+	MOVQ 136(SP),R9
+	ADCL R9,DI
+	SARL $16,DI
+	MOVQ DI,R9
+	XORL $0XFFFFFFFF,R9
+	ANDQ DI,SI
+	MOVQ 104(SP),AX
+	ANDQ R9,AX
+	ORQ AX,SI
+	ANDQ DI,DX
+	MOVQ 112(SP),AX
+	ANDQ R9,AX
+	ORQ AX,DX
+	ANDQ DI,CX
+	MOVQ 120(SP),AX
+	ANDQ R9,AX
+	ORQ AX,CX
+	ANDQ DI,R8
+	MOVQ 128(SP),DI
+	ANDQ R9,DI
+	ORQ DI,R8
+	MOVQ 88(SP),DI
+	MOVQ 96(SP),R9
+	ADDL 16(R9),SI
+	ADCL 20(R9),DX
+	ADCL 24(R9),CX
+	ADCL 28(R9),R8
+	MOVL SI,0(DI)
+	MOVL DX,4(DI)
+	MOVL CX,8(DI)
+	MOVL R8,12(DI)
+	MOVQ 32(SP),R11
+	MOVQ 40(SP),R12
+	MOVQ 48(SP),R13
+	MOVQ 56(SP),R14
+	MOVQ 64(SP),R15
+	MOVQ 72(SP),BX
+	MOVQ 80(SP),BP
+	MOVQ R11,SP
+	RET
diff --git a/vendor/golang.org/x/crypto/poly1305/poly1305_arm.s b/vendor/golang.org/x/crypto/poly1305/poly1305_arm.s
new file mode 100644
index 000000000..c15386744
--- /dev/null
+++ b/vendor/golang.org/x/crypto/poly1305/poly1305_arm.s
@@ -0,0 +1,379 @@
+// Copyright 2015 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// This code was translated into a form compatible with 5a from the public
+// domain source by Andrew Moon: github.com/floodyberry/poly1305-opt/blob/master/app/extensions/poly1305.
+
+// +build arm,!gccgo,!appengine
+
+DATA poly1305_init_constants_armv6<>+0x00(SB)/4, $0x3ffffff
+DATA poly1305_init_constants_armv6<>+0x04(SB)/4, $0x3ffff03
+DATA poly1305_init_constants_armv6<>+0x08(SB)/4, $0x3ffc0ff
+DATA poly1305_init_constants_armv6<>+0x0c(SB)/4, $0x3f03fff
+DATA poly1305_init_constants_armv6<>+0x10(SB)/4, $0x00fffff
+GLOBL poly1305_init_constants_armv6<>(SB), 8, $20
+
+// Warning: the linker may use R11 to synthesize certain instructions. Please
+// take care and verify that no synthetic instructions use it.
+
+TEXT poly1305_init_ext_armv6<>(SB),4,$-4
+  MOVM.DB.W [R4-R11], (R13)
+  MOVM.IA.W (R1), [R2-R5]
+  MOVW $poly1305_init_constants_armv6<>(SB), R7
+  MOVW R2, R8
+  MOVW R2>>26, R9
+  MOVW R3>>20, g
+  MOVW R4>>14, R11
+  MOVW R5>>8, R12
+  ORR R3<<6, R9, R9
+  ORR R4<<12, g, g
+  ORR R5<<18, R11, R11
+  MOVM.IA (R7), [R2-R6]
+  AND R8, R2, R2
+  AND R9, R3, R3
+  AND g, R4, R4
+  AND R11, R5, R5
+  AND R12, R6, R6
+  MOVM.IA.W [R2-R6], (R0)
+  EOR R2, R2, R2
+  EOR R3, R3, R3
+  EOR R4, R4, R4
+  EOR R5, R5, R5
+  EOR R6, R6, R6
+  MOVM.IA.W [R2-R6], (R0)
+  MOVM.IA.W (R1), [R2-R5]
+  MOVM.IA [R2-R6], (R0)
+  MOVM.IA.W (R13), [R4-R11]
+  RET
+
+#define MOVW_UNALIGNED(Rsrc, Rdst, Rtmp, offset) \
+  MOVBU (offset+0)(Rsrc), Rtmp; \
+  MOVBU Rtmp, (offset+0)(Rdst); \
+  MOVBU (offset+1)(Rsrc), Rtmp; \
+  MOVBU Rtmp, (offset+1)(Rdst); \
+  MOVBU (offset+2)(Rsrc), Rtmp; \
+  MOVBU Rtmp, (offset+2)(Rdst); \
+  MOVBU (offset+3)(Rsrc), Rtmp; \
+  MOVBU Rtmp, (offset+3)(Rdst)
+
+TEXT poly1305_blocks_armv6<>(SB),4,$-4
+  MOVM.DB.W [R4, R5, R6, R7, R8, R9, g, R11, R14], (R13)
+  SUB $128, R13
+  MOVW R0, 36(R13)
+  MOVW R1, 40(R13)
+  MOVW R2, 44(R13)
+  MOVW R1, R14
+  MOVW R2, R12
+  MOVW 56(R0), R8
+  WORD $0xe1180008 // TST R8, R8 not working see issue 5921
+  EOR R6, R6, R6
+  MOVW.EQ $(1<<24), R6
+  MOVW R6, 32(R13)
+  ADD $64, R13, g
+  MOVM.IA (R0), [R0-R9]
+  MOVM.IA [R0-R4], (g)
+  CMP $16, R12
+  BLO poly1305_blocks_armv6_done
+poly1305_blocks_armv6_mainloop:
+  WORD $0xe31e0003 // TST R14, #3 not working see issue 5921
+  BEQ poly1305_blocks_armv6_mainloop_aligned
+  ADD $48, R13, g
+  MOVW_UNALIGNED(R14, g, R0, 0)
+  MOVW_UNALIGNED(R14, g, R0, 4)
+  MOVW_UNALIGNED(R14, g, R0, 8)
+  MOVW_UNALIGNED(R14, g, R0, 12)
+  MOVM.IA (g), [R0-R3]
+  ADD $16, R14
+  B poly1305_blocks_armv6_mainloop_loaded
+poly1305_blocks_armv6_mainloop_aligned:
+  MOVM.IA.W (R14), [R0-R3]
+poly1305_blocks_armv6_mainloop_loaded:
+  MOVW R0>>26, g
+  MOVW R1>>20, R11
+  MOVW R2>>14, R12
+  MOVW R14, 40(R13)
+  MOVW R3>>8, R4
+  ORR R1<<6, g, g
+  ORR R2<<12, R11, R11
+  ORR R3<<18, R12, R12
+  BIC $0xfc000000, R0, R0
+  BIC $0xfc000000, g, g
+  MOVW 32(R13), R3
+  BIC $0xfc000000, R11, R11
+  BIC $0xfc000000, R12, R12
+  ADD R0, R5, R5
+  ADD g, R6, R6
+  ORR R3, R4, R4
+  ADD R11, R7, R7
+  ADD $64, R13, R14
+  ADD R12, R8, R8
+  ADD R4, R9, R9
+  MOVM.IA (R14), [R0-R4]
+  MULLU R4, R5, (R11, g)
+  MULLU R3, R5, (R14, R12)
+  MULALU R3, R6, (R11, g)
+  MULALU R2, R6, (R14, R12)
+  MULALU R2, R7, (R11, g)
+  MULALU R1, R7, (R14, R12)
+  ADD R4<<2, R4, R4
+  ADD R3<<2, R3, R3
+  MULALU R1, R8, (R11, g)
+  MULALU R0, R8, (R14, R12)
+  MULALU R0, R9, (R11, g)
+  MULALU R4, R9, (R14, R12)
+  MOVW g, 24(R13)
+  MOVW R11, 28(R13)
+  MOVW R12, 16(R13)
+  MOVW R14, 20(R13)
+  MULLU R2, R5, (R11, g)
+  MULLU R1, R5, (R14, R12)
+  MULALU R1, R6, (R11, g)
+  MULALU R0, R6, (R14, R12)
+  MULALU R0, R7, (R11, g)
+  MULALU R4, R7, (R14, R12)
+  ADD R2<<2, R2, R2
+  ADD R1<<2, R1, R1
+  MULALU R4, R8, (R11, g)
+  MULALU R3, R8, (R14, R12)
+  MULALU R3, R9, (R11, g)
+  MULALU R2, R9, (R14, R12)
+  MOVW g, 8(R13)
+  MOVW R11, 12(R13)
+  MOVW R12, 0(R13)
+  MOVW R14, w+4(SP)
+  MULLU R0, R5, (R11, g)
+  MULALU R4, R6, (R11, g)
+  MULALU R3, R7, (R11, g)
+  MULALU R2, R8, (R11, g)
+  MULALU R1, R9, (R11, g)
+  MOVM.IA (R13), [R0-R7]
+  MOVW g>>26, R12
+  MOVW R4>>26, R14
+  ORR R11<<6, R12, R12
+  ORR R5<<6, R14, R14
+  BIC $0xfc000000, g, g
+  BIC $0xfc000000, R4, R4
+  ADD.S R12, R0, R0
+  ADC $0, R1, R1
+  ADD.S R14, R6, R6
+  ADC $0, R7, R7
+  MOVW R0>>26, R12
+  MOVW R6>>26, R14
+  ORR R1<<6, R12, R12
+  ORR R7<<6, R14, R14
+  BIC $0xfc000000, R0, R0
+  BIC $0xfc000000, R6, R6
+  ADD R14<<2, R14, R14
+  ADD.S R12, R2, R2
+  ADC $0, R3, R3
+  ADD R14, g, g
+  MOVW R2>>26, R12
+  MOVW g>>26, R14
+  ORR R3<<6, R12, R12
+  BIC $0xfc000000, g, R5
+  BIC $0xfc000000, R2, R7
+  ADD R12, R4, R4
+  ADD R14, R0, R0
+  MOVW R4>>26, R12
+  BIC $0xfc000000, R4, R8
+  ADD R12, R6, R9
+  MOVW w+44(SP), R12
+  MOVW w+40(SP), R14
+  MOVW R0, R6
+  CMP $32, R12
+  SUB $16, R12, R12
+  MOVW R12, 44(R13)
+  BHS poly1305_blocks_armv6_mainloop
+poly1305_blocks_armv6_done:
+  MOVW 36(R13), R12
+  MOVW R5, 20(R12)
+  MOVW R6, 24(R12)
+  MOVW R7, 28(R12)
+  MOVW R8, 32(R12)
+  MOVW R9, 36(R12)
+  ADD $128, R13, R13
+  MOVM.IA.W (R13), [R4, R5, R6, R7, R8, R9, g, R11, R14]
+  RET
+
+#define MOVHUP_UNALIGNED(Rsrc, Rdst, Rtmp) \
+  MOVBU.P 1(Rsrc), Rtmp; \
+  MOVBU.P Rtmp, 1(Rdst); \
+  MOVBU.P 1(Rsrc), Rtmp; \
+  MOVBU.P Rtmp, 1(Rdst)
+
+#define MOVWP_UNALIGNED(Rsrc, Rdst, Rtmp) \
+  MOVHUP_UNALIGNED(Rsrc, Rdst, Rtmp); \
+  MOVHUP_UNALIGNED(Rsrc, Rdst, Rtmp)
+
+TEXT poly1305_finish_ext_armv6<>(SB),4,$-4
+  MOVM.DB.W [R4, R5, R6, R7, R8, R9, g, R11, R14], (R13)
+  SUB $16, R13, R13
+  MOVW R0, R5
+  MOVW R1, R6
+  MOVW R2, R7
+  MOVW R3, R8
+  AND.S R2, R2, R2
+  BEQ poly1305_finish_ext_armv6_noremaining
+  EOR R0, R0
+  MOVW R13, R9
+  MOVW R0, 0(R13)
+  MOVW R0, 4(R13)
+  MOVW R0, 8(R13)
+  MOVW R0, 12(R13)
+  WORD $0xe3110003 // TST R1, #3 not working see issue 5921
+  BEQ poly1305_finish_ext_armv6_aligned
+  WORD $0xe3120008 // TST R2, #8 not working see issue 5921
+  BEQ poly1305_finish_ext_armv6_skip8
+  MOVWP_UNALIGNED(R1, R9, g)
+  MOVWP_UNALIGNED(R1, R9, g)
+poly1305_finish_ext_armv6_skip8:
+  WORD $0xe3120004 // TST $4, R2 not working see issue 5921
+  BEQ poly1305_finish_ext_armv6_skip4
+  MOVWP_UNALIGNED(R1, R9, g)
+poly1305_finish_ext_armv6_skip4:
+  WORD $0xe3120002 // TST $2, R2 not working see issue 5921
+  BEQ poly1305_finish_ext_armv6_skip2
+  MOVHUP_UNALIGNED(R1, R9, g)
+  B poly1305_finish_ext_armv6_skip2
+poly1305_finish_ext_armv6_aligned:
+  WORD $0xe3120008 // TST R2, #8 not working see issue 5921
+  BEQ poly1305_finish_ext_armv6_skip8_aligned
+  MOVM.IA.W (R1), [g-R11]
+  MOVM.IA.W [g-R11], (R9)
+poly1305_finish_ext_armv6_skip8_aligned:
+  WORD $0xe3120004 // TST $4, R2 not working see issue 5921
+  BEQ poly1305_finish_ext_armv6_skip4_aligned
+  MOVW.P 4(R1), g
+  MOVW.P g, 4(R9)
+poly1305_finish_ext_armv6_skip4_aligned:
+  WORD $0xe3120002 // TST $2, R2 not working see issue 5921
+  BEQ poly1305_finish_ext_armv6_skip2
+  MOVHU.P 2(R1), g
+  MOVH.P g, 2(R9)
+poly1305_finish_ext_armv6_skip2:
+  WORD $0xe3120001 // TST $1, R2 not working see issue 5921
+  BEQ poly1305_finish_ext_armv6_skip1
+  MOVBU.P 1(R1), g
+  MOVBU.P g, 1(R9)
+poly1305_finish_ext_armv6_skip1:
+  MOVW $1, R11
+  MOVBU R11, 0(R9)
+  MOVW R11, 56(R5)
+  MOVW R5, R0
+  MOVW R13, R1
+  MOVW $16, R2
+  BL poly1305_blocks_armv6<>(SB)
+poly1305_finish_ext_armv6_noremaining:
+  MOVW 20(R5), R0
+  MOVW 24(R5), R1
+  MOVW 28(R5), R2
+  MOVW 32(R5), R3
+  MOVW 36(R5), R4
+  MOVW R4>>26, R12
+  BIC $0xfc000000, R4, R4
+  ADD R12<<2, R12, R12
+  ADD R12, R0, R0
+  MOVW R0>>26, R12
+  BIC $0xfc000000, R0, R0
+  ADD R12, R1, R1
+  MOVW R1>>26, R12
+  BIC $0xfc000000, R1, R1
+  ADD R12, R2, R2
+  MOVW R2>>26, R12
+  BIC $0xfc000000, R2, R2
+  ADD R12, R3, R3
+  MOVW R3>>26, R12
+  BIC $0xfc000000, R3, R3
+  ADD R12, R4, R4
+  ADD $5, R0, R6
+  MOVW R6>>26, R12
+  BIC $0xfc000000, R6, R6
+  ADD R12, R1, R7
+  MOVW R7>>26, R12
+  BIC $0xfc000000, R7, R7
+  ADD R12, R2, g
+  MOVW g>>26, R12
+  BIC $0xfc000000, g, g
+  ADD R12, R3, R11
+  MOVW $-(1<<26), R12
+  ADD R11>>26, R12, R12
+  BIC $0xfc000000, R11, R11
+  ADD R12, R4, R14
+  MOVW R14>>31, R12
+  SUB $1, R12
+  AND R12, R6, R6
+  AND R12, R7, R7
+  AND R12, g, g
+  AND R12, R11, R11
+  AND R12, R14, R14
+  MVN R12, R12
+  AND R12, R0, R0
+  AND R12, R1, R1
+  AND R12, R2, R2
+  AND R12, R3, R3
+  AND R12, R4, R4
+  ORR R6, R0, R0
+  ORR R7, R1, R1
+  ORR g, R2, R2
+  ORR R11, R3, R3
+  ORR R14, R4, R4
+  ORR R1<<26, R0, R0
+  MOVW R1>>6, R1
+  ORR R2<<20, R1, R1
+  MOVW R2>>12, R2
+  ORR R3<<14, R2, R2
+  MOVW R3>>18, R3
+  ORR R4<<8, R3, R3
+  MOVW 40(R5), R6
+  MOVW 44(R5), R7
+  MOVW 48(R5), g
+  MOVW 52(R5), R11
+  ADD.S R6, R0, R0
+  ADC.S R7, R1, R1
+  ADC.S g, R2, R2
+  ADC.S R11, R3, R3
+  MOVM.IA [R0-R3], (R8)
+  MOVW R5, R12
+  EOR R0, R0, R0
+  EOR R1, R1, R1
+  EOR R2, R2, R2
+  EOR R3, R3, R3
+  EOR R4, R4, R4
+  EOR R5, R5, R5
+  EOR R6, R6, R6
+  EOR R7, R7, R7
+  MOVM.IA.W [R0-R7], (R12)
+  MOVM.IA [R0-R7], (R12)
+  ADD $16, R13, R13
+  MOVM.IA.W (R13), [R4, R5, R6, R7, R8, R9, g, R11, R14]
+  RET
+
+// func poly1305_auth_armv6(out *[16]byte, m *byte, mlen uint32, key *[32]key)
+TEXT ·poly1305_auth_armv6(SB),0,$280-16
+  MOVW  out+0(FP), R4
+  MOVW  m+4(FP), R5
+  MOVW  mlen+8(FP), R6
+  MOVW  key+12(FP), R7
+
+  MOVW R13, R8
+  BIC $63, R13
+  SUB $64, R13, R13
+  MOVW  R13, R0
+  MOVW  R7, R1
+  BL poly1305_init_ext_armv6<>(SB)
+  BIC.S $15, R6, R2
+  BEQ poly1305_auth_armv6_noblocks
+  MOVW R13, R0
+  MOVW R5, R1
+  ADD R2, R5, R5
+  SUB R2, R6, R6
+  BL poly1305_blocks_armv6<>(SB)
+poly1305_auth_armv6_noblocks:
+  MOVW R13, R0
+  MOVW R5, R1
+  MOVW R6, R2
+  MOVW R4, R3
+  BL poly1305_finish_ext_armv6<>(SB)
+  MOVW R8, R13
+  RET
diff --git a/vendor/golang.org/x/crypto/poly1305/poly1305_test.go b/vendor/golang.org/x/crypto/poly1305/poly1305_test.go
new file mode 100644
index 000000000..b3e92310b
--- /dev/null
+++ b/vendor/golang.org/x/crypto/poly1305/poly1305_test.go
@@ -0,0 +1,86 @@
+// 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 poly1305
+
+import (
+	"bytes"
+	"testing"
+	"unsafe"
+)
+
+var testData = []struct {
+	in, k, correct []byte
+}{
+	{
+		[]byte("Hello world!"),
+		[]byte("this is 32-byte key for Poly1305"),
+		[]byte{0xa6, 0xf7, 0x45, 0x00, 0x8f, 0x81, 0xc9, 0x16, 0xa2, 0x0d, 0xcc, 0x74, 0xee, 0xf2, 0xb2, 0xf0},
+	},
+	{
+		make([]byte, 32),
+		[]byte("this is 32-byte key for Poly1305"),
+		[]byte{0x49, 0xec, 0x78, 0x09, 0x0e, 0x48, 0x1e, 0xc6, 0xc2, 0x6b, 0x33, 0xb9, 0x1c, 0xcc, 0x03, 0x07},
+	},
+	{
+		make([]byte, 2007),
+		[]byte("this is 32-byte key for Poly1305"),
+		[]byte{0xda, 0x84, 0xbc, 0xab, 0x02, 0x67, 0x6c, 0x38, 0xcd, 0xb0, 0x15, 0x60, 0x42, 0x74, 0xc2, 0xaa},
+	},
+	{
+		make([]byte, 2007),
+		make([]byte, 32),
+		make([]byte, 16),
+	},
+}
+
+func testSum(t *testing.T, unaligned bool) {
+	var out [16]byte
+	var key [32]byte
+
+	for i, v := range testData {
+		in := v.in
+		if unaligned {
+			in = unalignBytes(in)
+		}
+		copy(key[:], v.k)
+		Sum(&out, in, &key)
+		if !bytes.Equal(out[:], v.correct) {
+			t.Errorf("%d: expected %x, got %x", i, v.correct, out[:])
+		}
+	}
+}
+
+func TestSum(t *testing.T)          { testSum(t, false) }
+func TestSumUnaligned(t *testing.T) { testSum(t, true) }
+
+func benchmark(b *testing.B, size int, unaligned bool) {
+	var out [16]byte
+	var key [32]byte
+	in := make([]byte, size)
+	if unaligned {
+		in = unalignBytes(in)
+	}
+	b.SetBytes(int64(len(in)))
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		Sum(&out, in, &key)
+	}
+}
+
+func Benchmark64(b *testing.B)          { benchmark(b, 64, false) }
+func Benchmark1K(b *testing.B)          { benchmark(b, 1024, false) }
+func Benchmark64Unaligned(b *testing.B) { benchmark(b, 64, true) }
+func Benchmark1KUnaligned(b *testing.B) { benchmark(b, 1024, true) }
+
+func unalignBytes(in []byte) []byte {
+	out := make([]byte, len(in)+1)
+	if uintptr(unsafe.Pointer(&out[0]))&(unsafe.Alignof(uint32(0))-1) == 0 {
+		out = out[1:]
+	} else {
+		out = out[:len(in)]
+	}
+	copy(out, in)
+	return out
+}
diff --git a/vendor/golang.org/x/crypto/poly1305/sum_amd64.go b/vendor/golang.org/x/crypto/poly1305/sum_amd64.go
new file mode 100644
index 000000000..6775c703f
--- /dev/null
+++ b/vendor/golang.org/x/crypto/poly1305/sum_amd64.go
@@ -0,0 +1,24 @@
+// Copyright 2012 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// +build amd64,!gccgo,!appengine
+
+package poly1305
+
+// This function is implemented in poly1305_amd64.s
+
+//go:noescape
+
+func poly1305(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) {
+	var mPtr *byte
+	if len(m) > 0 {
+		mPtr = &m[0]
+	}
+	poly1305(out, mPtr, uint64(len(m)), key)
+}
diff --git a/vendor/golang.org/x/crypto/poly1305/sum_arm.go b/vendor/golang.org/x/crypto/poly1305/sum_arm.go
new file mode 100644
index 000000000..50b979c24
--- /dev/null
+++ b/vendor/golang.org/x/crypto/poly1305/sum_arm.go
@@ -0,0 +1,24 @@
+// Copyright 2015 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// +build arm,!gccgo,!appengine
+
+package poly1305
+
+// This function is implemented in poly1305_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) {
+	var mPtr *byte
+	if len(m) > 0 {
+		mPtr = &m[0]
+	}
+	poly1305_auth_armv6(out, mPtr, uint32(len(m)), key)
+}
diff --git a/vendor/golang.org/x/crypto/poly1305/sum_ref.go b/vendor/golang.org/x/crypto/poly1305/sum_ref.go
new file mode 100644
index 000000000..0b24fc78b
--- /dev/null
+++ b/vendor/golang.org/x/crypto/poly1305/sum_ref.go
@@ -0,0 +1,1531 @@
+// 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,!arm gccgo appengine
+
+package poly1305
+
+// Based on original, public domain implementation from NaCl by D. J.
+// Bernstein.
+
+import "math"
+
+const (
+	alpham80 = 0.00000000558793544769287109375
+	alpham48 = 24.0
+	alpham16 = 103079215104.0
+	alpha0   = 6755399441055744.0
+	alpha18  = 1770887431076116955136.0
+	alpha32  = 29014219670751100192948224.0
+	alpha50  = 7605903601369376408980219232256.0
+	alpha64  = 124615124604835863084731911901282304.0
+	alpha82  = 32667107224410092492483962313449748299776.0
+	alpha96  = 535217884764734955396857238543560676143529984.0
+	alpha112 = 35076039295941670036888435985190792471742381031424.0
+	alpha130 = 9194973245195333150150082162901855101712434733101613056.0
+	scale    = 0.0000000000000000000000000000000000000036734198463196484624023016788195177431833298649127735047148490821200539357960224151611328125
+	offset0  = 6755408030990331.0
+	offset1  = 29014256564239239022116864.0
+	offset2  = 124615283061160854719918951570079744.0
+	offset3  = 535219245894202480694386063513315216128475136.0
+)
+
+// 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) {
+	r := key
+	s := key[16:]
+	var (
+		y7        float64
+		y6        float64
+		y1        float64
+		y0        float64
+		y5        float64
+		y4        float64
+		x7        float64
+		x6        float64
+		x1        float64
+		x0        float64
+		y3        float64
+		y2        float64
+		x5        float64
+		r3lowx0   float64
+		x4        float64
+		r0lowx6   float64
+		x3        float64
+		r3highx0  float64
+		x2        float64
+		r0highx6  float64
+		r0lowx0   float64
+		sr1lowx6  float64
+		r0highx0  float64
+		sr1highx6 float64
+		sr3low    float64
+		r1lowx0   float64
+		sr2lowx6  float64
+		r1highx0  float64
+		sr2highx6 float64
+		r2lowx0   float64
+		sr3lowx6  float64
+		r2highx0  float64
+		sr3highx6 float64
+		r1highx4  float64
+		r1lowx4   float64
+		r0highx4  float64
+		r0lowx4   float64
+		sr3highx4 float64
+		sr3lowx4  float64
+		sr2highx4 float64
+		sr2lowx4  float64
+		r0lowx2   float64
+		r0highx2  float64
+		r1lowx2   float64
+		r1highx2  float64
+		r2lowx2   float64
+		r2highx2  float64
+		sr3lowx2  float64
+		sr3highx2 float64
+		z0        float64
+		z1        float64
+		z2        float64
+		z3        float64
+		m0        int64
+		m1        int64
+		m2        int64
+		m3        int64
+		m00       uint32
+		m01       uint32
+		m02       uint32
+		m03       uint32
+		m10       uint32
+		m11       uint32
+		m12       uint32
+		m13       uint32
+		m20       uint32
+		m21       uint32
+		m22       uint32
+		m23       uint32
+		m30       uint32
+		m31       uint32
+		m32       uint32
+		m33       uint64
+		lbelow2   int32
+		lbelow3   int32
+		lbelow4   int32
+		lbelow5   int32
+		lbelow6   int32
+		lbelow7   int32
+		lbelow8   int32
+		lbelow9   int32
+		lbelow10  int32
+		lbelow11  int32
+		lbelow12  int32
+		lbelow13  int32
+		lbelow14  int32
+		lbelow15  int32
+		s00       uint32
+		s01       uint32
+		s02       uint32
+		s03       uint32
+		s10       uint32
+		s11       uint32
+		s12       uint32
+		s13       uint32
+		s20       uint32
+		s21       uint32
+		s22       uint32
+		s23       uint32
+		s30       uint32
+		s31       uint32
+		s32       uint32
+		s33       uint32
+		bits32    uint64
+		f         uint64
+		f0        uint64
+		f1        uint64
+		f2        uint64
+		f3        uint64
+		f4        uint64
+		g         uint64
+		g0        uint64
+		g1        uint64
+		g2        uint64
+		g3        uint64
+		g4        uint64
+	)
+
+	var p int32
+
+	l := int32(len(m))
+
+	r00 := uint32(r[0])
+
+	r01 := uint32(r[1])
+
+	r02 := uint32(r[2])
+	r0 := int64(2151)
+
+	r03 := uint32(r[3])
+	r03 &= 15
+	r0 <<= 51
+
+	r10 := uint32(r[4])
+	r10 &= 252
+	r01 <<= 8
+	r0 += int64(r00)
+
+	r11 := uint32(r[5])
+	r02 <<= 16
+	r0 += int64(r01)
+
+	r12 := uint32(r[6])
+	r03 <<= 24
+	r0 += int64(r02)
+
+	r13 := uint32(r[7])
+	r13 &= 15
+	r1 := int64(2215)
+	r0 += int64(r03)
+
+	d0 := r0
+	r1 <<= 51
+	r2 := int64(2279)
+
+	r20 := uint32(r[8])
+	r20 &= 252
+	r11 <<= 8
+	r1 += int64(r10)
+
+	r21 := uint32(r[9])
+	r12 <<= 16
+	r1 += int64(r11)
+
+	r22 := uint32(r[10])
+	r13 <<= 24
+	r1 += int64(r12)
+
+	r23 := uint32(r[11])
+	r23 &= 15
+	r2 <<= 51
+	r1 += int64(r13)
+
+	d1 := r1
+	r21 <<= 8
+	r2 += int64(r20)
+
+	r30 := uint32(r[12])
+	r30 &= 252
+	r22 <<= 16
+	r2 += int64(r21)
+
+	r31 := uint32(r[13])
+	r23 <<= 24
+	r2 += int64(r22)
+
+	r32 := uint32(r[14])
+	r2 += int64(r23)
+	r3 := int64(2343)
+
+	d2 := r2
+	r3 <<= 51
+
+	r33 := uint32(r[15])
+	r33 &= 15
+	r31 <<= 8
+	r3 += int64(r30)
+
+	r32 <<= 16
+	r3 += int64(r31)
+
+	r33 <<= 24
+	r3 += int64(r32)
+
+	r3 += int64(r33)
+	h0 := alpha32 - alpha32
+
+	d3 := r3
+	h1 := alpha32 - alpha32
+
+	h2 := alpha32 - alpha32
+
+	h3 := alpha32 - alpha32
+
+	h4 := alpha32 - alpha32
+
+	r0low := math.Float64frombits(uint64(d0))
+	h5 := alpha32 - alpha32
+
+	r1low := math.Float64frombits(uint64(d1))
+	h6 := alpha32 - alpha32
+
+	r2low := math.Float64frombits(uint64(d2))
+	h7 := alpha32 - alpha32
+
+	r0low -= alpha0
+
+	r1low -= alpha32
+
+	r2low -= alpha64
+
+	r0high := r0low + alpha18
+
+	r3low := math.Float64frombits(uint64(d3))
+
+	r1high := r1low + alpha50
+	sr1low := scale * r1low
+
+	r2high := r2low + alpha82
+	sr2low := scale * r2low
+
+	r0high -= alpha18
+	r0high_stack := r0high
+
+	r3low -= alpha96
+
+	r1high -= alpha50
+	r1high_stack := r1high
+
+	sr1high := sr1low + alpham80
+
+	r0low -= r0high
+
+	r2high -= alpha82
+	sr3low = scale * r3low
+
+	sr2high := sr2low + alpham48
+
+	r1low -= r1high
+	r1low_stack := r1low
+
+	sr1high -= alpham80
+	sr1high_stack := sr1high
+
+	r2low -= r2high
+	r2low_stack := r2low
+
+	sr2high -= alpham48
+	sr2high_stack := sr2high
+
+	r3high := r3low + alpha112
+	r0low_stack := r0low
+
+	sr1low -= sr1high
+	sr1low_stack := sr1low
+
+	sr3high := sr3low + alpham16
+	r2high_stack := r2high
+
+	sr2low -= sr2high
+	sr2low_stack := sr2low
+
+	r3high -= alpha112
+	r3high_stack := r3high
+
+	sr3high -= alpham16
+	sr3high_stack := sr3high
+
+	r3low -= r3high
+	r3low_stack := r3low
+
+	sr3low -= sr3high
+	sr3low_stack := sr3low
+
+	if l < 16 {
+		goto addatmost15bytes
+	}
+
+	m00 = uint32(m[p+0])
+	m0 = 2151
+
+	m0 <<= 51
+	m1 = 2215
+	m01 = uint32(m[p+1])
+
+	m1 <<= 51
+	m2 = 2279
+	m02 = uint32(m[p+2])
+
+	m2 <<= 51
+	m3 = 2343
+	m03 = uint32(m[p+3])
+
+	m10 = uint32(m[p+4])
+	m01 <<= 8
+	m0 += int64(m00)
+
+	m11 = uint32(m[p+5])
+	m02 <<= 16
+	m0 += int64(m01)
+
+	m12 = uint32(m[p+6])
+	m03 <<= 24
+	m0 += int64(m02)
+
+	m13 = uint32(m[p+7])
+	m3 <<= 51
+	m0 += int64(m03)
+
+	m20 = uint32(m[p+8])
+	m11 <<= 8
+	m1 += int64(m10)
+
+	m21 = uint32(m[p+9])
+	m12 <<= 16
+	m1 += int64(m11)
+
+	m22 = uint32(m[p+10])
+	m13 <<= 24
+	m1 += int64(m12)
+
+	m23 = uint32(m[p+11])
+	m1 += int64(m13)
+
+	m30 = uint32(m[p+12])
+	m21 <<= 8
+	m2 += int64(m20)
+
+	m31 = uint32(m[p+13])
+	m22 <<= 16
+	m2 += int64(m21)
+
+	m32 = uint32(m[p+14])
+	m23 <<= 24
+	m2 += int64(m22)
+
+	m33 = uint64(m[p+15])
+	m2 += int64(m23)
+
+	d0 = m0
+	m31 <<= 8
+	m3 += int64(m30)
+
+	d1 = m1
+	m32 <<= 16
+	m3 += int64(m31)
+
+	d2 = m2
+	m33 += 256
+
+	m33 <<= 24
+	m3 += int64(m32)
+
+	m3 += int64(m33)
+	d3 = m3
+
+	p += 16
+	l -= 16
+
+	z0 = math.Float64frombits(uint64(d0))
+
+	z1 = math.Float64frombits(uint64(d1))
+
+	z2 = math.Float64frombits(uint64(d2))
+
+	z3 = math.Float64frombits(uint64(d3))
+
+	z0 -= alpha0
+
+	z1 -= alpha32
+
+	z2 -= alpha64
+
+	z3 -= alpha96
+
+	h0 += z0
+
+	h1 += z1
+
+	h3 += z2
+
+	h5 += z3
+
+	if l < 16 {
+		goto multiplyaddatmost15bytes
+	}
+
+multiplyaddatleast16bytes:
+
+	m2 = 2279
+	m20 = uint32(m[p+8])
+	y7 = h7 + alpha130
+
+	m2 <<= 51
+	m3 = 2343
+	m21 = uint32(m[p+9])
+	y6 = h6 + alpha130
+
+	m3 <<= 51
+	m0 = 2151
+	m22 = uint32(m[p+10])
+	y1 = h1 + alpha32
+
+	m0 <<= 51
+	m1 = 2215
+	m23 = uint32(m[p+11])
+	y0 = h0 + alpha32
+
+	m1 <<= 51
+	m30 = uint32(m[p+12])
+	y7 -= alpha130
+
+	m21 <<= 8
+	m2 += int64(m20)
+	m31 = uint32(m[p+13])
+	y6 -= alpha130
+
+	m22 <<= 16
+	m2 += int64(m21)
+	m32 = uint32(m[p+14])
+	y1 -= alpha32
+
+	m23 <<= 24
+	m2 += int64(m22)
+	m33 = uint64(m[p+15])
+	y0 -= alpha32
+
+	m2 += int64(m23)
+	m00 = uint32(m[p+0])
+	y5 = h5 + alpha96
+
+	m31 <<= 8
+	m3 += int64(m30)
+	m01 = uint32(m[p+1])
+	y4 = h4 + alpha96
+
+	m32 <<= 16
+	m02 = uint32(m[p+2])
+	x7 = h7 - y7
+	y7 *= scale
+
+	m33 += 256
+	m03 = uint32(m[p+3])
+	x6 = h6 - y6
+	y6 *= scale
+
+	m33 <<= 24
+	m3 += int64(m31)
+	m10 = uint32(m[p+4])
+	x1 = h1 - y1
+
+	m01 <<= 8
+	m3 += int64(m32)
+	m11 = uint32(m[p+5])
+	x0 = h0 - y0
+
+	m3 += int64(m33)
+	m0 += int64(m00)
+	m12 = uint32(m[p+6])
+	y5 -= alpha96
+
+	m02 <<= 16
+	m0 += int64(m01)
+	m13 = uint32(m[p+7])
+	y4 -= alpha96
+
+	m03 <<= 24
+	m0 += int64(m02)
+	d2 = m2
+	x1 += y7
+
+	m0 += int64(m03)
+	d3 = m3
+	x0 += y6
+
+	m11 <<= 8
+	m1 += int64(m10)
+	d0 = m0
+	x7 += y5
+
+	m12 <<= 16
+	m1 += int64(m11)
+	x6 += y4
+
+	m13 <<= 24
+	m1 += int64(m12)
+	y3 = h3 + alpha64
+
+	m1 += int64(m13)
+	d1 = m1
+	y2 = h2 + alpha64
+
+	x0 += x1
+
+	x6 += x7
+
+	y3 -= alpha64
+	r3low = r3low_stack
+
+	y2 -= alpha64
+	r0low = r0low_stack
+
+	x5 = h5 - y5
+	r3lowx0 = r3low * x0
+	r3high = r3high_stack
+
+	x4 = h4 - y4
+	r0lowx6 = r0low * x6
+	r0high = r0high_stack
+
+	x3 = h3 - y3
+	r3highx0 = r3high * x0
+	sr1low = sr1low_stack
+
+	x2 = h2 - y2
+	r0highx6 = r0high * x6
+	sr1high = sr1high_stack
+
+	x5 += y3
+	r0lowx0 = r0low * x0
+	r1low = r1low_stack
+
+	h6 = r3lowx0 + r0lowx6
+	sr1lowx6 = sr1low * x6
+	r1high = r1high_stack
+
+	x4 += y2
+	r0highx0 = r0high * x0
+	sr2low = sr2low_stack
+
+	h7 = r3highx0 + r0highx6
+	sr1highx6 = sr1high * x6
+	sr2high = sr2high_stack
+
+	x3 += y1
+	r1lowx0 = r1low * x0
+	r2low = r2low_stack
+
+	h0 = r0lowx0 + sr1lowx6
+	sr2lowx6 = sr2low * x6
+	r2high = r2high_stack
+
+	x2 += y0
+	r1highx0 = r1high * x0
+	sr3low = sr3low_stack
+
+	h1 = r0highx0 + sr1highx6
+	sr2highx6 = sr2high * x6
+	sr3high = sr3high_stack
+
+	x4 += x5
+	r2lowx0 = r2low * x0
+	z2 = math.Float64frombits(uint64(d2))
+
+	h2 = r1lowx0 + sr2lowx6
+	sr3lowx6 = sr3low * x6
+
+	x2 += x3
+	r2highx0 = r2high * x0
+	z3 = math.Float64frombits(uint64(d3))
+
+	h3 = r1highx0 + sr2highx6
+	sr3highx6 = sr3high * x6
+
+	r1highx4 = r1high * x4
+	z2 -= alpha64
+
+	h4 = r2lowx0 + sr3lowx6
+	r1lowx4 = r1low * x4
+
+	r0highx4 = r0high * x4
+	z3 -= alpha96
+
+	h5 = r2highx0 + sr3highx6
+	r0lowx4 = r0low * x4
+
+	h7 += r1highx4
+	sr3highx4 = sr3high * x4
+
+	h6 += r1lowx4
+	sr3lowx4 = sr3low * x4
+
+	h5 += r0highx4
+	sr2highx4 = sr2high * x4
+
+	h4 += r0lowx4
+	sr2lowx4 = sr2low * x4
+
+	h3 += sr3highx4
+	r0lowx2 = r0low * x2
+
+	h2 += sr3lowx4
+	r0highx2 = r0high * x2
+
+	h1 += sr2highx4
+	r1lowx2 = r1low * x2
+
+	h0 += sr2lowx4
+	r1highx2 = r1high * x2
+
+	h2 += r0lowx2
+	r2lowx2 = r2low * x2
+
+	h3 += r0highx2
+	r2highx2 = r2high * x2
+
+	h4 += r1lowx2
+	sr3lowx2 = sr3low * x2
+
+	h5 += r1highx2
+	sr3highx2 = sr3high * x2
+
+	p += 16
+	l -= 16
+	h6 += r2lowx2
+
+	h7 += r2highx2
+
+	z1 = math.Float64frombits(uint64(d1))
+	h0 += sr3lowx2
+
+	z0 = math.Float64frombits(uint64(d0))
+	h1 += sr3highx2
+
+	z1 -= alpha32
+
+	z0 -= alpha0
+
+	h5 += z3
+
+	h3 += z2
+
+	h1 += z1
+
+	h0 += z0
+
+	if l >= 16 {
+		goto multiplyaddatleast16bytes
+	}
+
+multiplyaddatmost15bytes:
+
+	y7 = h7 + alpha130
+
+	y6 = h6 + alpha130
+
+	y1 = h1 + alpha32
+
+	y0 = h0 + alpha32
+
+	y7 -= alpha130
+
+	y6 -= alpha130
+
+	y1 -= alpha32
+
+	y0 -= alpha32
+
+	y5 = h5 + alpha96
+
+	y4 = h4 + alpha96
+
+	x7 = h7 - y7
+	y7 *= scale
+
+	x6 = h6 - y6
+	y6 *= scale
+
+	x1 = h1 - y1
+
+	x0 = h0 - y0
+
+	y5 -= alpha96
+
+	y4 -= alpha96
+
+	x1 += y7
+
+	x0 += y6
+
+	x7 += y5
+
+	x6 += y4
+
+	y3 = h3 + alpha64
+
+	y2 = h2 + alpha64
+
+	x0 += x1
+
+	x6 += x7
+
+	y3 -= alpha64
+	r3low = r3low_stack
+
+	y2 -= alpha64
+	r0low = r0low_stack
+
+	x5 = h5 - y5
+	r3lowx0 = r3low * x0
+	r3high = r3high_stack
+
+	x4 = h4 - y4
+	r0lowx6 = r0low * x6
+	r0high = r0high_stack
+
+	x3 = h3 - y3
+	r3highx0 = r3high * x0
+	sr1low = sr1low_stack
+
+	x2 = h2 - y2
+	r0highx6 = r0high * x6
+	sr1high = sr1high_stack
+
+	x5 += y3
+	r0lowx0 = r0low * x0
+	r1low = r1low_stack
+
+	h6 = r3lowx0 + r0lowx6
+	sr1lowx6 = sr1low * x6
+	r1high = r1high_stack
+
+	x4 += y2
+	r0highx0 = r0high * x0
+	sr2low = sr2low_stack
+
+	h7 = r3highx0 + r0highx6
+	sr1highx6 = sr1high * x6
+	sr2high = sr2high_stack
+
+	x3 += y1
+	r1lowx0 = r1low * x0
+	r2low = r2low_stack
+
+	h0 = r0lowx0 + sr1lowx6
+	sr2lowx6 = sr2low * x6
+	r2high = r2high_stack
+
+	x2 += y0
+	r1highx0 = r1high * x0
+	sr3low = sr3low_stack
+
+	h1 = r0highx0 + sr1highx6
+	sr2highx6 = sr2high * x6
+	sr3high = sr3high_stack
+
+	x4 += x5
+	r2lowx0 = r2low * x0
+
+	h2 = r1lowx0 + sr2lowx6
+	sr3lowx6 = sr3low * x6
+
+	x2 += x3
+	r2highx0 = r2high * x0
+
+	h3 = r1highx0 + sr2highx6
+	sr3highx6 = sr3high * x6
+
+	r1highx4 = r1high * x4
+
+	h4 = r2lowx0 + sr3lowx6
+	r1lowx4 = r1low * x4
+
+	r0highx4 = r0high * x4
+
+	h5 = r2highx0 + sr3highx6
+	r0lowx4 = r0low * x4
+
+	h7 += r1highx4
+	sr3highx4 = sr3high * x4
+
+	h6 += r1lowx4
+	sr3lowx4 = sr3low * x4
+
+	h5 += r0highx4
+	sr2highx4 = sr2high * x4
+
+	h4 += r0lowx4
+	sr2lowx4 = sr2low * x4
+
+	h3 += sr3highx4
+	r0lowx2 = r0low * x2
+
+	h2 += sr3lowx4
+	r0highx2 = r0high * x2
+
+	h1 += sr2highx4
+	r1lowx2 = r1low * x2
+
+	h0 += sr2lowx4
+	r1highx2 = r1high * x2
+
+	h2 += r0lowx2
+	r2lowx2 = r2low * x2
+
+	h3 += r0highx2
+	r2highx2 = r2high * x2
+
+	h4 += r1lowx2
+	sr3lowx2 = sr3low * x2
+
+	h5 += r1highx2
+	sr3highx2 = sr3high * x2
+
+	h6 += r2lowx2
+
+	h7 += r2highx2
+
+	h0 += sr3lowx2
+
+	h1 += sr3highx2
+
+addatmost15bytes:
+
+	if l == 0 {
+		goto nomorebytes
+	}
+
+	lbelow2 = l - 2
+
+	lbelow3 = l - 3
+
+	lbelow2 >>= 31
+	lbelow4 = l - 4
+
+	m00 = uint32(m[p+0])
+	lbelow3 >>= 31
+	p += lbelow2
+
+	m01 = uint32(m[p+1])
+	lbelow4 >>= 31
+	p += lbelow3
+
+	m02 = uint32(m[p+2])
+	p += lbelow4
+	m0 = 2151
+
+	m03 = uint32(m[p+3])
+	m0 <<= 51
+	m1 = 2215
+
+	m0 += int64(m00)
+	m01 &^= uint32(lbelow2)
+
+	m02 &^= uint32(lbelow3)
+	m01 -= uint32(lbelow2)
+
+	m01 <<= 8
+	m03 &^= uint32(lbelow4)
+
+	m0 += int64(m01)
+	lbelow2 -= lbelow3
+
+	m02 += uint32(lbelow2)
+	lbelow3 -= lbelow4
+
+	m02 <<= 16
+	m03 += uint32(lbelow3)
+
+	m03 <<= 24
+	m0 += int64(m02)
+
+	m0 += int64(m03)
+	lbelow5 = l - 5
+
+	lbelow6 = l - 6
+	lbelow7 = l - 7
+
+	lbelow5 >>= 31
+	lbelow8 = l - 8
+
+	lbelow6 >>= 31
+	p += lbelow5
+
+	m10 = uint32(m[p+4])
+	lbelow7 >>= 31
+	p += lbelow6
+
+	m11 = uint32(m[p+5])
+	lbelow8 >>= 31
+	p += lbelow7
+
+	m12 = uint32(m[p+6])
+	m1 <<= 51
+	p += lbelow8
+
+	m13 = uint32(m[p+7])
+	m10 &^= uint32(lbelow5)
+	lbelow4 -= lbelow5
+
+	m10 += uint32(lbelow4)
+	lbelow5 -= lbelow6
+
+	m11 &^= uint32(lbelow6)
+	m11 += uint32(lbelow5)
+
+	m11 <<= 8
+	m1 += int64(m10)
+
+	m1 += int64(m11)
+	m12 &^= uint32(lbelow7)
+
+	lbelow6 -= lbelow7
+	m13 &^= uint32(lbelow8)
+
+	m12 += uint32(lbelow6)
+	lbelow7 -= lbelow8
+
+	m12 <<= 16
+	m13 += uint32(lbelow7)
+
+	m13 <<= 24
+	m1 += int64(m12)
+
+	m1 += int64(m13)
+	m2 = 2279
+
+	lbelow9 = l - 9
+	m3 = 2343
+
+	lbelow10 = l - 10
+	lbelow11 = l - 11
+
+	lbelow9 >>= 31
+	lbelow12 = l - 12
+
+	lbelow10 >>= 31
+	p += lbelow9
+
+	m20 = uint32(m[p+8])
+	lbelow11 >>= 31
+	p += lbelow10
+
+	m21 = uint32(m[p+9])
+	lbelow12 >>= 31
+	p += lbelow11
+
+	m22 = uint32(m[p+10])
+	m2 <<= 51
+	p += lbelow12
+
+	m23 = uint32(m[p+11])
+	m20 &^= uint32(lbelow9)
+	lbelow8 -= lbelow9
+
+	m20 += uint32(lbelow8)
+	lbelow9 -= lbelow10
+
+	m21 &^= uint32(lbelow10)
+	m21 += uint32(lbelow9)
+
+	m21 <<= 8
+	m2 += int64(m20)
+
+	m2 += int64(m21)
+	m22 &^= uint32(lbelow11)
+
+	lbelow10 -= lbelow11
+	m23 &^= uint32(lbelow12)
+
+	m22 += uint32(lbelow10)
+	lbelow11 -= lbelow12
+
+	m22 <<= 16
+	m23 += uint32(lbelow11)
+
+	m23 <<= 24
+	m2 += int64(m22)
+
+	m3 <<= 51
+	lbelow13 = l - 13
+
+	lbelow13 >>= 31
+	lbelow14 = l - 14
+
+	lbelow14 >>= 31
+	p += lbelow13
+	lbelow15 = l - 15
+
+	m30 = uint32(m[p+12])
+	lbelow15 >>= 31
+	p += lbelow14
+
+	m31 = uint32(m[p+13])
+	p += lbelow15
+	m2 += int64(m23)
+
+	m32 = uint32(m[p+14])
+	m30 &^= uint32(lbelow13)
+	lbelow12 -= lbelow13
+
+	m30 += uint32(lbelow12)
+	lbelow13 -= lbelow14
+
+	m3 += int64(m30)
+	m31 &^= uint32(lbelow14)
+
+	m31 += uint32(lbelow13)
+	m32 &^= uint32(lbelow15)
+
+	m31 <<= 8
+	lbelow14 -= lbelow15
+
+	m3 += int64(m31)
+	m32 += uint32(lbelow14)
+	d0 = m0
+
+	m32 <<= 16
+	m33 = uint64(lbelow15 + 1)
+	d1 = m1
+
+	m33 <<= 24
+	m3 += int64(m32)
+	d2 = m2
+
+	m3 += int64(m33)
+	d3 = m3
+
+	z3 = math.Float64frombits(uint64(d3))
+
+	z2 = math.Float64frombits(uint64(d2))
+
+	z1 = math.Float64frombits(uint64(d1))
+
+	z0 = math.Float64frombits(uint64(d0))
+
+	z3 -= alpha96
+
+	z2 -= alpha64
+
+	z1 -= alpha32
+
+	z0 -= alpha0
+
+	h5 += z3
+
+	h3 += z2
+
+	h1 += z1
+
+	h0 += z0
+
+	y7 = h7 + alpha130
+
+	y6 = h6 + alpha130
+
+	y1 = h1 + alpha32
+
+	y0 = h0 + alpha32
+
+	y7 -= alpha130
+
+	y6 -= alpha130
+
+	y1 -= alpha32
+
+	y0 -= alpha32
+
+	y5 = h5 + alpha96
+
+	y4 = h4 + alpha96
+
+	x7 = h7 - y7
+	y7 *= scale
+
+	x6 = h6 - y6
+	y6 *= scale
+
+	x1 = h1 - y1
+
+	x0 = h0 - y0
+
+	y5 -= alpha96
+
+	y4 -= alpha96
+
+	x1 += y7
+
+	x0 += y6
+
+	x7 += y5
+
+	x6 += y4
+
+	y3 = h3 + alpha64
+
+	y2 = h2 + alpha64
+
+	x0 += x1
+
+	x6 += x7
+
+	y3 -= alpha64
+	r3low = r3low_stack
+
+	y2 -= alpha64
+	r0low = r0low_stack
+
+	x5 = h5 - y5
+	r3lowx0 = r3low * x0
+	r3high = r3high_stack
+
+	x4 = h4 - y4
+	r0lowx6 = r0low * x6
+	r0high = r0high_stack
+
+	x3 = h3 - y3
+	r3highx0 = r3high * x0
+	sr1low = sr1low_stack
+
+	x2 = h2 - y2
+	r0highx6 = r0high * x6
+	sr1high = sr1high_stack
+
+	x5 += y3
+	r0lowx0 = r0low * x0
+	r1low = r1low_stack
+
+	h6 = r3lowx0 + r0lowx6
+	sr1lowx6 = sr1low * x6
+	r1high = r1high_stack
+
+	x4 += y2
+	r0highx0 = r0high * x0
+	sr2low = sr2low_stack
+
+	h7 = r3highx0 + r0highx6
+	sr1highx6 = sr1high * x6
+	sr2high = sr2high_stack
+
+	x3 += y1
+	r1lowx0 = r1low * x0
+	r2low = r2low_stack
+
+	h0 = r0lowx0 + sr1lowx6
+	sr2lowx6 = sr2low * x6
+	r2high = r2high_stack
+
+	x2 += y0
+	r1highx0 = r1high * x0
+	sr3low = sr3low_stack
+
+	h1 = r0highx0 + sr1highx6
+	sr2highx6 = sr2high * x6
+	sr3high = sr3high_stack
+
+	x4 += x5
+	r2lowx0 = r2low * x0
+
+	h2 = r1lowx0 + sr2lowx6
+	sr3lowx6 = sr3low * x6
+
+	x2 += x3
+	r2highx0 = r2high * x0
+
+	h3 = r1highx0 + sr2highx6
+	sr3highx6 = sr3high * x6
+
+	r1highx4 = r1high * x4
+
+	h4 = r2lowx0 + sr3lowx6
+	r1lowx4 = r1low * x4
+
+	r0highx4 = r0high * x4
+
+	h5 = r2highx0 + sr3highx6
+	r0lowx4 = r0low * x4
+
+	h7 += r1highx4
+	sr3highx4 = sr3high * x4
+
+	h6 += r1lowx4
+	sr3lowx4 = sr3low * x4
+
+	h5 += r0highx4
+	sr2highx4 = sr2high * x4
+
+	h4 += r0lowx4
+	sr2lowx4 = sr2low * x4
+
+	h3 += sr3highx4
+	r0lowx2 = r0low * x2
+
+	h2 += sr3lowx4
+	r0highx2 = r0high * x2
+
+	h1 += sr2highx4
+	r1lowx2 = r1low * x2
+
+	h0 += sr2lowx4
+	r1highx2 = r1high * x2
+
+	h2 += r0lowx2
+	r2lowx2 = r2low * x2
+
+	h3 += r0highx2
+	r2highx2 = r2high * x2
+
+	h4 += r1lowx2
+	sr3lowx2 = sr3low * x2
+
+	h5 += r1highx2
+	sr3highx2 = sr3high * x2
+
+	h6 += r2lowx2
+
+	h7 += r2highx2
+
+	h0 += sr3lowx2
+
+	h1 += sr3highx2
+
+nomorebytes:
+
+	y7 = h7 + alpha130
+
+	y0 = h0 + alpha32
+
+	y1 = h1 + alpha32
+
+	y2 = h2 + alpha64
+
+	y7 -= alpha130
+
+	y3 = h3 + alpha64
+
+	y4 = h4 + alpha96
+
+	y5 = h5 + alpha96
+
+	x7 = h7 - y7
+	y7 *= scale
+
+	y0 -= alpha32
+
+	y1 -= alpha32
+
+	y2 -= alpha64
+
+	h6 += x7
+
+	y3 -= alpha64
+
+	y4 -= alpha96
+
+	y5 -= alpha96
+
+	y6 = h6 + alpha130
+
+	x0 = h0 - y0
+
+	x1 = h1 - y1
+
+	x2 = h2 - y2
+
+	y6 -= alpha130
+
+	x0 += y7
+
+	x3 = h3 - y3
+
+	x4 = h4 - y4
+
+	x5 = h5 - y5
+
+	x6 = h6 - y6
+
+	y6 *= scale
+
+	x2 += y0
+
+	x3 += y1
+
+	x4 += y2
+
+	x0 += y6
+
+	x5 += y3
+
+	x6 += y4
+
+	x2 += x3
+
+	x0 += x1
+
+	x4 += x5
+
+	x6 += y5
+
+	x2 += offset1
+	d1 = int64(math.Float64bits(x2))
+
+	x0 += offset0
+	d0 = int64(math.Float64bits(x0))
+
+	x4 += offset2
+	d2 = int64(math.Float64bits(x4))
+
+	x6 += offset3
+	d3 = int64(math.Float64bits(x6))
+
+	f0 = uint64(d0)
+
+	f1 = uint64(d1)
+	bits32 = math.MaxUint64
+
+	f2 = uint64(d2)
+	bits32 >>= 32
+
+	f3 = uint64(d3)
+	f = f0 >> 32
+
+	f0 &= bits32
+	f &= 255
+
+	f1 += f
+	g0 = f0 + 5
+
+	g = g0 >> 32
+	g0 &= bits32
+
+	f = f1 >> 32
+	f1 &= bits32
+
+	f &= 255
+	g1 = f1 + g
+
+	g = g1 >> 32
+	f2 += f
+
+	f = f2 >> 32
+	g1 &= bits32
+
+	f2 &= bits32
+	f &= 255
+
+	f3 += f
+	g2 = f2 + g
+
+	g = g2 >> 32
+	g2 &= bits32
+
+	f4 = f3 >> 32
+	f3 &= bits32
+
+	f4 &= 255
+	g3 = f3 + g
+
+	g = g3 >> 32
+	g3 &= bits32
+
+	g4 = f4 + g
+
+	g4 = g4 - 4
+	s00 = uint32(s[0])
+
+	f = uint64(int64(g4) >> 63)
+	s01 = uint32(s[1])
+
+	f0 &= f
+	g0 &^= f
+	s02 = uint32(s[2])
+
+	f1 &= f
+	f0 |= g0
+	s03 = uint32(s[3])
+
+	g1 &^= f
+	f2 &= f
+	s10 = uint32(s[4])
+
+	f3 &= f
+	g2 &^= f
+	s11 = uint32(s[5])
+
+	g3 &^= f
+	f1 |= g1
+	s12 = uint32(s[6])
+
+	f2 |= g2
+	f3 |= g3
+	s13 = uint32(s[7])
+
+	s01 <<= 8
+	f0 += uint64(s00)
+	s20 = uint32(s[8])
+
+	s02 <<= 16
+	f0 += uint64(s01)
+	s21 = uint32(s[9])
+
+	s03 <<= 24
+	f0 += uint64(s02)
+	s22 = uint32(s[10])
+
+	s11 <<= 8
+	f1 += uint64(s10)
+	s23 = uint32(s[11])
+
+	s12 <<= 16
+	f1 += uint64(s11)
+	s30 = uint32(s[12])
+
+	s13 <<= 24
+	f1 += uint64(s12)
+	s31 = uint32(s[13])
+
+	f0 += uint64(s03)
+	f1 += uint64(s13)
+	s32 = uint32(s[14])
+
+	s21 <<= 8
+	f2 += uint64(s20)
+	s33 = uint32(s[15])
+
+	s22 <<= 16
+	f2 += uint64(s21)
+
+	s23 <<= 24
+	f2 += uint64(s22)
+
+	s31 <<= 8
+	f3 += uint64(s30)
+
+	s32 <<= 16
+	f3 += uint64(s31)
+
+	s33 <<= 24
+	f3 += uint64(s32)
+
+	f2 += uint64(s23)
+	f3 += uint64(s33)
+
+	out[0] = byte(f0)
+	f0 >>= 8
+	out[1] = byte(f0)
+	f0 >>= 8
+	out[2] = byte(f0)
+	f0 >>= 8
+	out[3] = byte(f0)
+	f0 >>= 8
+	f1 += f0
+
+	out[4] = byte(f1)
+	f1 >>= 8
+	out[5] = byte(f1)
+	f1 >>= 8
+	out[6] = byte(f1)
+	f1 >>= 8
+	out[7] = byte(f1)
+	f1 >>= 8
+	f2 += f1
+
+	out[8] = byte(f2)
+	f2 >>= 8
+	out[9] = byte(f2)
+	f2 >>= 8
+	out[10] = byte(f2)
+	f2 >>= 8
+	out[11] = byte(f2)
+	f2 >>= 8
+	f3 += f2
+
+	out[12] = byte(f3)
+	f3 >>= 8
+	out[13] = byte(f3)
+	f3 >>= 8
+	out[14] = byte(f3)
+	f3 >>= 8
+	out[15] = byte(f3)
+}