forked from loweel/zabov
updated vendor dependencies and golang to 1.15
parent
e2a625a92e
commit
bc2470e344
7
go.mod
7
go.mod
|
@ -1,8 +1,11 @@
|
|||
module zabov
|
||||
|
||||
go 1.13
|
||||
go 1.15
|
||||
|
||||
require (
|
||||
github.com/miekg/dns v1.1.27
|
||||
github.com/golang/snappy v0.0.4
|
||||
github.com/miekg/dns v1.1.43
|
||||
github.com/syndtr/goleveldb v1.0.0
|
||||
golang.org/x/net v0.0.0-20210916014120-12bc252f5db8
|
||||
golang.org/x/sys v0.0.0-20210915083310-ed5796bab164
|
||||
)
|
||||
|
|
36
go.sum
36
go.sum
|
@ -1,35 +1,33 @@
|
|||
github.com/fsnotify/fsnotify v1.4.7/go.mod h1:jwhsz4b93w/PPRr/qN1Yymfu8t87LnFCMoQvtojpjFo=
|
||||
github.com/golang/protobuf v1.2.0/go.mod h1:6lQm79b+lXiMfvg/cZm0SGofjICqVBUtrP5yJMmIC1U=
|
||||
github.com/golang/snappy v0.0.0-20180518054509-2e65f85255db h1:woRePGFeVFfLKN/pOkfl+p/TAqKOfFu+7KPlMVpok/w=
|
||||
github.com/golang/snappy v0.0.0-20180518054509-2e65f85255db/go.mod h1:/XxbfmMg8lxefKM7IXC3fBNl/7bRcc72aCRzEWrmP2Q=
|
||||
github.com/golang/snappy v0.0.4 h1:yAGX7huGHXlcLOEtBnF4w7FQwA26wojNCwOYAEhLjQM=
|
||||
github.com/golang/snappy v0.0.4/go.mod h1:/XxbfmMg8lxefKM7IXC3fBNl/7bRcc72aCRzEWrmP2Q=
|
||||
github.com/hpcloud/tail v1.0.0/go.mod h1:ab1qPbhIpdTxEkNHXyeSf5vhxWSCs/tWer42PpOxQnU=
|
||||
github.com/miekg/dns v1.1.27 h1:aEH/kqUzUxGJ/UHcEKdJY+ugH6WEzsEBBSPa8zuy1aM=
|
||||
github.com/miekg/dns v1.1.27/go.mod h1:KNUDUusw/aVsxyTYZM1oqvCicbwhgbNgztCETuNZ7xM=
|
||||
github.com/miekg/dns v1.1.43 h1:JKfpVSCB84vrAmHzyrsxB5NAr5kLoMXZArPSw7Qlgyg=
|
||||
github.com/miekg/dns v1.1.43/go.mod h1:+evo5L0630/F6ca/Z9+GAqzhjGyn8/c+TBaOyfEl0V4=
|
||||
github.com/onsi/ginkgo v1.6.0/go.mod h1:lLunBs/Ym6LB5Z9jYTR76FiuTmxDTDusOGeTQH+WWjE=
|
||||
github.com/onsi/ginkgo v1.7.0/go.mod h1:lLunBs/Ym6LB5Z9jYTR76FiuTmxDTDusOGeTQH+WWjE=
|
||||
github.com/onsi/gomega v1.4.3/go.mod h1:ex+gbHU/CVuBBDIJjb2X0qEXbFg53c61hWP/1CpauHY=
|
||||
github.com/syndtr/goleveldb v1.0.0 h1:fBdIW9lB4Iz0n9khmH8w27SJ3QEJ7+IgjPEwGSZiFdE=
|
||||
github.com/syndtr/goleveldb v1.0.0/go.mod h1:ZVVdQEZoIme9iO1Ch2Jdy24qqXrMMOU6lpPAyBWyWuQ=
|
||||
golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
|
||||
golang.org/x/crypto v0.0.0-20191011191535-87dc89f01550 h1:ObdrDkeb4kJdCP557AjRjq69pTHfNouLtWZG7j9rPN8=
|
||||
golang.org/x/crypto v0.0.0-20191011191535-87dc89f01550/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI=
|
||||
golang.org/x/mod v0.1.1-0.20191105210325-c90efee705ee/go.mod h1:QqPTAvyqsEbceGzBzNggFXnrqF1CaUcvgkdR5Ot7KZg=
|
||||
golang.org/x/net v0.0.0-20180906233101-161cd47e91fd/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
|
||||
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 h1:l5EDrHhldLYb3ZRHDUhXF7Om7MvYXnkV9/iQNo1lX6g=
|
||||
golang.org/x/net v0.0.0-20190923162816-aa69164e4478/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
|
||||
golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
|
||||
golang.org/x/net v0.0.0-20210916014120-12bc252f5db8 h1:/6y1LfuqNuQdHAm0jjtPtgRcxIxjVZgm5OTu8/QhZvk=
|
||||
golang.org/x/net v0.0.0-20210916014120-12bc252f5db8/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
|
||||
golang.org/x/sync v0.0.0-20180314180146-1d60e4601c6f/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
|
||||
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/sync v0.0.0-20210220032951-036812b2e83c/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
|
||||
golang.org/x/sys v0.0.0-20180909124046-d0be0721c37e/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
|
||||
golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
|
||||
golang.org/x/sys v0.0.0-20190412213103-97732733099d/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
|
||||
golang.org/x/sys v0.0.0-20190924154521-2837fb4f24fe h1:6fAMxZRR6sl1Uq8U61gxU+kPTs2tR8uOySCbBP7BN/M=
|
||||
golang.org/x/sys v0.0.0-20190924154521-2837fb4f24fe/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
|
||||
golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
|
||||
golang.org/x/sys v0.0.0-20210303074136-134d130e1a04/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
|
||||
golang.org/x/sys v0.0.0-20210423082822-04245dca01da/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
|
||||
golang.org/x/sys v0.0.0-20210915083310-ed5796bab164 h1:7ZDGnxgHAMw7thfC5bEos0RDAccZKxioiWBhfIe+tvw=
|
||||
golang.org/x/sys v0.0.0-20210915083310-ed5796bab164/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
|
||||
golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
|
||||
golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
|
||||
golang.org/x/tools v0.0.0-20191216052735-49a3e744a425/go.mod h1:TB2adYChydJhpapKDTa4BR/hXlZSLoq2Wpct/0txZ28=
|
||||
golang.org/x/xerrors v0.0.0-20191011141410-1b5146add898/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
|
||||
golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
|
||||
golang.org/x/text v0.3.6/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
|
||||
golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
|
||||
gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
|
||||
gopkg.in/fsnotify.v1 v1.4.7/go.mod h1:Tz8NjZHkW78fSQdbUxIjBTcgA1z1m8ZHf0WmKUhAMys=
|
||||
gopkg.in/tomb.v1 v1.0.0-20141024135613-dd632973f1e7/go.mod h1:dt/ZhP58zS4L8KSrWDmTeBkI65Dw0HsyUHuEVlX15mw=
|
||||
|
|
|
@ -8,8 +8,11 @@
|
|||
|
||||
# Please keep the list sorted.
|
||||
|
||||
Amazon.com, Inc
|
||||
Damian Gryski <dgryski@gmail.com>
|
||||
Eric Buth <eric@topos.com>
|
||||
Google Inc.
|
||||
Jan Mercl <0xjnml@gmail.com>
|
||||
Klaus Post <klauspost@gmail.com>
|
||||
Rodolfo Carvalho <rhcarvalho@gmail.com>
|
||||
Sebastien Binet <seb.binet@gmail.com>
|
||||
|
|
|
@ -26,9 +26,13 @@
|
|||
|
||||
# Please keep the list sorted.
|
||||
|
||||
Alex Legg <alexlegg@google.com>
|
||||
Damian Gryski <dgryski@gmail.com>
|
||||
Eric Buth <eric@topos.com>
|
||||
Jan Mercl <0xjnml@gmail.com>
|
||||
Jonathan Swinney <jswinney@amazon.com>
|
||||
Kai Backman <kaib@golang.org>
|
||||
Klaus Post <klauspost@gmail.com>
|
||||
Marc-Antoine Ruel <maruel@chromium.org>
|
||||
Nigel Tao <nigeltao@golang.org>
|
||||
Rob Pike <r@golang.org>
|
||||
|
|
|
@ -52,6 +52,8 @@ const (
|
|||
// Otherwise, a newly allocated slice will be returned.
|
||||
//
|
||||
// The dst and src must not overlap. It is valid to pass a nil dst.
|
||||
//
|
||||
// Decode handles the Snappy block format, not the Snappy stream format.
|
||||
func Decode(dst, src []byte) ([]byte, error) {
|
||||
dLen, s, err := decodedLen(src)
|
||||
if err != nil {
|
||||
|
@ -83,6 +85,8 @@ func NewReader(r io.Reader) *Reader {
|
|||
}
|
||||
|
||||
// Reader is an io.Reader that can read Snappy-compressed bytes.
|
||||
//
|
||||
// Reader handles the Snappy stream format, not the Snappy block format.
|
||||
type Reader struct {
|
||||
r io.Reader
|
||||
err error
|
||||
|
@ -114,32 +118,23 @@ func (r *Reader) readFull(p []byte, allowEOF bool) (ok bool) {
|
|||
return true
|
||||
}
|
||||
|
||||
// Read satisfies the io.Reader interface.
|
||||
func (r *Reader) Read(p []byte) (int, error) {
|
||||
if r.err != nil {
|
||||
return 0, r.err
|
||||
}
|
||||
for {
|
||||
if r.i < r.j {
|
||||
n := copy(p, r.decoded[r.i:r.j])
|
||||
r.i += n
|
||||
return n, nil
|
||||
}
|
||||
func (r *Reader) fill() error {
|
||||
for r.i >= r.j {
|
||||
if !r.readFull(r.buf[:4], true) {
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
chunkType := r.buf[0]
|
||||
if !r.readHeader {
|
||||
if chunkType != chunkTypeStreamIdentifier {
|
||||
r.err = ErrCorrupt
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
r.readHeader = true
|
||||
}
|
||||
chunkLen := int(r.buf[1]) | int(r.buf[2])<<8 | int(r.buf[3])<<16
|
||||
if chunkLen > len(r.buf) {
|
||||
r.err = ErrUnsupported
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
|
||||
// The chunk types are specified at
|
||||
|
@ -149,11 +144,11 @@ func (r *Reader) Read(p []byte) (int, error) {
|
|||
// Section 4.2. Compressed data (chunk type 0x00).
|
||||
if chunkLen < checksumSize {
|
||||
r.err = ErrCorrupt
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
buf := r.buf[:chunkLen]
|
||||
if !r.readFull(buf, false) {
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
checksum := uint32(buf[0]) | uint32(buf[1])<<8 | uint32(buf[2])<<16 | uint32(buf[3])<<24
|
||||
buf = buf[checksumSize:]
|
||||
|
@ -161,19 +156,19 @@ func (r *Reader) Read(p []byte) (int, error) {
|
|||
n, err := DecodedLen(buf)
|
||||
if err != nil {
|
||||
r.err = err
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
if n > len(r.decoded) {
|
||||
r.err = ErrCorrupt
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
if _, err := Decode(r.decoded, buf); err != nil {
|
||||
r.err = err
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
if crc(r.decoded[:n]) != checksum {
|
||||
r.err = ErrCorrupt
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
r.i, r.j = 0, n
|
||||
continue
|
||||
|
@ -182,25 +177,25 @@ func (r *Reader) Read(p []byte) (int, error) {
|
|||
// Section 4.3. Uncompressed data (chunk type 0x01).
|
||||
if chunkLen < checksumSize {
|
||||
r.err = ErrCorrupt
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
buf := r.buf[:checksumSize]
|
||||
if !r.readFull(buf, false) {
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
checksum := uint32(buf[0]) | uint32(buf[1])<<8 | uint32(buf[2])<<16 | uint32(buf[3])<<24
|
||||
// Read directly into r.decoded instead of via r.buf.
|
||||
n := chunkLen - checksumSize
|
||||
if n > len(r.decoded) {
|
||||
r.err = ErrCorrupt
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
if !r.readFull(r.decoded[:n], false) {
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
if crc(r.decoded[:n]) != checksum {
|
||||
r.err = ErrCorrupt
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
r.i, r.j = 0, n
|
||||
continue
|
||||
|
@ -209,15 +204,15 @@ func (r *Reader) Read(p []byte) (int, error) {
|
|||
// Section 4.1. Stream identifier (chunk type 0xff).
|
||||
if chunkLen != len(magicBody) {
|
||||
r.err = ErrCorrupt
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
if !r.readFull(r.buf[:len(magicBody)], false) {
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
for i := 0; i < len(magicBody); i++ {
|
||||
if r.buf[i] != magicBody[i] {
|
||||
r.err = ErrCorrupt
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
}
|
||||
continue
|
||||
|
@ -226,12 +221,44 @@ func (r *Reader) Read(p []byte) (int, error) {
|
|||
if chunkType <= 0x7f {
|
||||
// Section 4.5. Reserved unskippable chunks (chunk types 0x02-0x7f).
|
||||
r.err = ErrUnsupported
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
// Section 4.4 Padding (chunk type 0xfe).
|
||||
// Section 4.6. Reserved skippable chunks (chunk types 0x80-0xfd).
|
||||
if !r.readFull(r.buf[:chunkLen], false) {
|
||||
return 0, r.err
|
||||
return r.err
|
||||
}
|
||||
}
|
||||
|
||||
return nil
|
||||
}
|
||||
|
||||
// Read satisfies the io.Reader interface.
|
||||
func (r *Reader) Read(p []byte) (int, error) {
|
||||
if r.err != nil {
|
||||
return 0, r.err
|
||||
}
|
||||
|
||||
if err := r.fill(); err != nil {
|
||||
return 0, err
|
||||
}
|
||||
|
||||
n := copy(p, r.decoded[r.i:r.j])
|
||||
r.i += n
|
||||
return n, nil
|
||||
}
|
||||
|
||||
// ReadByte satisfies the io.ByteReader interface.
|
||||
func (r *Reader) ReadByte() (byte, error) {
|
||||
if r.err != nil {
|
||||
return 0, r.err
|
||||
}
|
||||
|
||||
if err := r.fill(); err != nil {
|
||||
return 0, err
|
||||
}
|
||||
|
||||
c := r.decoded[r.i]
|
||||
r.i++
|
||||
return c, nil
|
||||
}
|
||||
|
|
|
@ -0,0 +1,494 @@
|
|||
// Copyright 2020 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 !appengine
|
||||
// +build gc
|
||||
// +build !noasm
|
||||
|
||||
#include "textflag.h"
|
||||
|
||||
// The asm code generally follows the pure Go code in decode_other.go, except
|
||||
// where marked with a "!!!".
|
||||
|
||||
// func decode(dst, src []byte) int
|
||||
//
|
||||
// All local variables fit into registers. The non-zero stack size is only to
|
||||
// spill registers and push args when issuing a CALL. The register allocation:
|
||||
// - R2 scratch
|
||||
// - R3 scratch
|
||||
// - R4 length or x
|
||||
// - R5 offset
|
||||
// - R6 &src[s]
|
||||
// - R7 &dst[d]
|
||||
// + R8 dst_base
|
||||
// + R9 dst_len
|
||||
// + R10 dst_base + dst_len
|
||||
// + R11 src_base
|
||||
// + R12 src_len
|
||||
// + R13 src_base + src_len
|
||||
// - R14 used by doCopy
|
||||
// - R15 used by doCopy
|
||||
//
|
||||
// The registers R8-R13 (marked with a "+") are set at the start of the
|
||||
// function, and after a CALL returns, and are not otherwise modified.
|
||||
//
|
||||
// The d variable is implicitly R7 - R8, and len(dst)-d is R10 - R7.
|
||||
// The s variable is implicitly R6 - R11, and len(src)-s is R13 - R6.
|
||||
TEXT ·decode(SB), NOSPLIT, $56-56
|
||||
// Initialize R6, R7 and R8-R13.
|
||||
MOVD dst_base+0(FP), R8
|
||||
MOVD dst_len+8(FP), R9
|
||||
MOVD R8, R7
|
||||
MOVD R8, R10
|
||||
ADD R9, R10, R10
|
||||
MOVD src_base+24(FP), R11
|
||||
MOVD src_len+32(FP), R12
|
||||
MOVD R11, R6
|
||||
MOVD R11, R13
|
||||
ADD R12, R13, R13
|
||||
|
||||
loop:
|
||||
// for s < len(src)
|
||||
CMP R13, R6
|
||||
BEQ end
|
||||
|
||||
// R4 = uint32(src[s])
|
||||
//
|
||||
// switch src[s] & 0x03
|
||||
MOVBU (R6), R4
|
||||
MOVW R4, R3
|
||||
ANDW $3, R3
|
||||
MOVW $1, R1
|
||||
CMPW R1, R3
|
||||
BGE tagCopy
|
||||
|
||||
// ----------------------------------------
|
||||
// The code below handles literal tags.
|
||||
|
||||
// case tagLiteral:
|
||||
// x := uint32(src[s] >> 2)
|
||||
// switch
|
||||
MOVW $60, R1
|
||||
LSRW $2, R4, R4
|
||||
CMPW R4, R1
|
||||
BLS tagLit60Plus
|
||||
|
||||
// case x < 60:
|
||||
// s++
|
||||
ADD $1, R6, R6
|
||||
|
||||
doLit:
|
||||
// This is the end of the inner "switch", when we have a literal tag.
|
||||
//
|
||||
// We assume that R4 == x and x fits in a uint32, where x is the variable
|
||||
// used in the pure Go decode_other.go code.
|
||||
|
||||
// length = int(x) + 1
|
||||
//
|
||||
// Unlike the pure Go code, we don't need to check if length <= 0 because
|
||||
// R4 can hold 64 bits, so the increment cannot overflow.
|
||||
ADD $1, R4, R4
|
||||
|
||||
// Prepare to check if copying length bytes will run past the end of dst or
|
||||
// src.
|
||||
//
|
||||
// R2 = len(dst) - d
|
||||
// R3 = len(src) - s
|
||||
MOVD R10, R2
|
||||
SUB R7, R2, R2
|
||||
MOVD R13, R3
|
||||
SUB R6, R3, R3
|
||||
|
||||
// !!! Try a faster technique for short (16 or fewer bytes) copies.
|
||||
//
|
||||
// if length > 16 || len(dst)-d < 16 || len(src)-s < 16 {
|
||||
// goto callMemmove // Fall back on calling runtime·memmove.
|
||||
// }
|
||||
//
|
||||
// The C++ snappy code calls this TryFastAppend. It also checks len(src)-s
|
||||
// against 21 instead of 16, because it cannot assume that all of its input
|
||||
// is contiguous in memory and so it needs to leave enough source bytes to
|
||||
// read the next tag without refilling buffers, but Go's Decode assumes
|
||||
// contiguousness (the src argument is a []byte).
|
||||
CMP $16, R4
|
||||
BGT callMemmove
|
||||
CMP $16, R2
|
||||
BLT callMemmove
|
||||
CMP $16, R3
|
||||
BLT callMemmove
|
||||
|
||||
// !!! Implement the copy from src to dst as a 16-byte load and store.
|
||||
// (Decode's documentation says that dst and src must not overlap.)
|
||||
//
|
||||
// This always copies 16 bytes, instead of only length bytes, but that's
|
||||
// OK. If the input is a valid Snappy encoding then subsequent iterations
|
||||
// will fix up the overrun. Otherwise, Decode returns a nil []byte (and a
|
||||
// non-nil error), so the overrun will be ignored.
|
||||
//
|
||||
// Note that on arm64, it is legal and cheap to issue unaligned 8-byte or
|
||||
// 16-byte loads and stores. This technique probably wouldn't be as
|
||||
// effective on architectures that are fussier about alignment.
|
||||
LDP 0(R6), (R14, R15)
|
||||
STP (R14, R15), 0(R7)
|
||||
|
||||
// d += length
|
||||
// s += length
|
||||
ADD R4, R7, R7
|
||||
ADD R4, R6, R6
|
||||
B loop
|
||||
|
||||
callMemmove:
|
||||
// if length > len(dst)-d || length > len(src)-s { etc }
|
||||
CMP R2, R4
|
||||
BGT errCorrupt
|
||||
CMP R3, R4
|
||||
BGT errCorrupt
|
||||
|
||||
// copy(dst[d:], src[s:s+length])
|
||||
//
|
||||
// This means calling runtime·memmove(&dst[d], &src[s], length), so we push
|
||||
// R7, R6 and R4 as arguments. Coincidentally, we also need to spill those
|
||||
// three registers to the stack, to save local variables across the CALL.
|
||||
MOVD R7, 8(RSP)
|
||||
MOVD R6, 16(RSP)
|
||||
MOVD R4, 24(RSP)
|
||||
MOVD R7, 32(RSP)
|
||||
MOVD R6, 40(RSP)
|
||||
MOVD R4, 48(RSP)
|
||||
CALL runtime·memmove(SB)
|
||||
|
||||
// Restore local variables: unspill registers from the stack and
|
||||
// re-calculate R8-R13.
|
||||
MOVD 32(RSP), R7
|
||||
MOVD 40(RSP), R6
|
||||
MOVD 48(RSP), R4
|
||||
MOVD dst_base+0(FP), R8
|
||||
MOVD dst_len+8(FP), R9
|
||||
MOVD R8, R10
|
||||
ADD R9, R10, R10
|
||||
MOVD src_base+24(FP), R11
|
||||
MOVD src_len+32(FP), R12
|
||||
MOVD R11, R13
|
||||
ADD R12, R13, R13
|
||||
|
||||
// d += length
|
||||
// s += length
|
||||
ADD R4, R7, R7
|
||||
ADD R4, R6, R6
|
||||
B loop
|
||||
|
||||
tagLit60Plus:
|
||||
// !!! This fragment does the
|
||||
//
|
||||
// s += x - 58; if uint(s) > uint(len(src)) { etc }
|
||||
//
|
||||
// checks. In the asm version, we code it once instead of once per switch case.
|
||||
ADD R4, R6, R6
|
||||
SUB $58, R6, R6
|
||||
MOVD R6, R3
|
||||
SUB R11, R3, R3
|
||||
CMP R12, R3
|
||||
BGT errCorrupt
|
||||
|
||||
// case x == 60:
|
||||
MOVW $61, R1
|
||||
CMPW R1, R4
|
||||
BEQ tagLit61
|
||||
BGT tagLit62Plus
|
||||
|
||||
// x = uint32(src[s-1])
|
||||
MOVBU -1(R6), R4
|
||||
B doLit
|
||||
|
||||
tagLit61:
|
||||
// case x == 61:
|
||||
// x = uint32(src[s-2]) | uint32(src[s-1])<<8
|
||||
MOVHU -2(R6), R4
|
||||
B doLit
|
||||
|
||||
tagLit62Plus:
|
||||
CMPW $62, R4
|
||||
BHI tagLit63
|
||||
|
||||
// case x == 62:
|
||||
// x = uint32(src[s-3]) | uint32(src[s-2])<<8 | uint32(src[s-1])<<16
|
||||
MOVHU -3(R6), R4
|
||||
MOVBU -1(R6), R3
|
||||
ORR R3<<16, R4
|
||||
B doLit
|
||||
|
||||
tagLit63:
|
||||
// case x == 63:
|
||||
// x = uint32(src[s-4]) | uint32(src[s-3])<<8 | uint32(src[s-2])<<16 | uint32(src[s-1])<<24
|
||||
MOVWU -4(R6), R4
|
||||
B doLit
|
||||
|
||||
// The code above handles literal tags.
|
||||
// ----------------------------------------
|
||||
// The code below handles copy tags.
|
||||
|
||||
tagCopy4:
|
||||
// case tagCopy4:
|
||||
// s += 5
|
||||
ADD $5, R6, R6
|
||||
|
||||
// if uint(s) > uint(len(src)) { etc }
|
||||
MOVD R6, R3
|
||||
SUB R11, R3, R3
|
||||
CMP R12, R3
|
||||
BGT errCorrupt
|
||||
|
||||
// length = 1 + int(src[s-5])>>2
|
||||
MOVD $1, R1
|
||||
ADD R4>>2, R1, R4
|
||||
|
||||
// offset = int(uint32(src[s-4]) | uint32(src[s-3])<<8 | uint32(src[s-2])<<16 | uint32(src[s-1])<<24)
|
||||
MOVWU -4(R6), R5
|
||||
B doCopy
|
||||
|
||||
tagCopy2:
|
||||
// case tagCopy2:
|
||||
// s += 3
|
||||
ADD $3, R6, R6
|
||||
|
||||
// if uint(s) > uint(len(src)) { etc }
|
||||
MOVD R6, R3
|
||||
SUB R11, R3, R3
|
||||
CMP R12, R3
|
||||
BGT errCorrupt
|
||||
|
||||
// length = 1 + int(src[s-3])>>2
|
||||
MOVD $1, R1
|
||||
ADD R4>>2, R1, R4
|
||||
|
||||
// offset = int(uint32(src[s-2]) | uint32(src[s-1])<<8)
|
||||
MOVHU -2(R6), R5
|
||||
B doCopy
|
||||
|
||||
tagCopy:
|
||||
// We have a copy tag. We assume that:
|
||||
// - R3 == src[s] & 0x03
|
||||
// - R4 == src[s]
|
||||
CMP $2, R3
|
||||
BEQ tagCopy2
|
||||
BGT tagCopy4
|
||||
|
||||
// case tagCopy1:
|
||||
// s += 2
|
||||
ADD $2, R6, R6
|
||||
|
||||
// if uint(s) > uint(len(src)) { etc }
|
||||
MOVD R6, R3
|
||||
SUB R11, R3, R3
|
||||
CMP R12, R3
|
||||
BGT errCorrupt
|
||||
|
||||
// offset = int(uint32(src[s-2])&0xe0<<3 | uint32(src[s-1]))
|
||||
MOVD R4, R5
|
||||
AND $0xe0, R5
|
||||
MOVBU -1(R6), R3
|
||||
ORR R5<<3, R3, R5
|
||||
|
||||
// length = 4 + int(src[s-2])>>2&0x7
|
||||
MOVD $7, R1
|
||||
AND R4>>2, R1, R4
|
||||
ADD $4, R4, R4
|
||||
|
||||
doCopy:
|
||||
// This is the end of the outer "switch", when we have a copy tag.
|
||||
//
|
||||
// We assume that:
|
||||
// - R4 == length && R4 > 0
|
||||
// - R5 == offset
|
||||
|
||||
// if offset <= 0 { etc }
|
||||
MOVD $0, R1
|
||||
CMP R1, R5
|
||||
BLE errCorrupt
|
||||
|
||||
// if d < offset { etc }
|
||||
MOVD R7, R3
|
||||
SUB R8, R3, R3
|
||||
CMP R5, R3
|
||||
BLT errCorrupt
|
||||
|
||||
// if length > len(dst)-d { etc }
|
||||
MOVD R10, R3
|
||||
SUB R7, R3, R3
|
||||
CMP R3, R4
|
||||
BGT errCorrupt
|
||||
|
||||
// forwardCopy(dst[d:d+length], dst[d-offset:]); d += length
|
||||
//
|
||||
// Set:
|
||||
// - R14 = len(dst)-d
|
||||
// - R15 = &dst[d-offset]
|
||||
MOVD R10, R14
|
||||
SUB R7, R14, R14
|
||||
MOVD R7, R15
|
||||
SUB R5, R15, R15
|
||||
|
||||
// !!! Try a faster technique for short (16 or fewer bytes) forward copies.
|
||||
//
|
||||
// First, try using two 8-byte load/stores, similar to the doLit technique
|
||||
// above. Even if dst[d:d+length] and dst[d-offset:] can overlap, this is
|
||||
// still OK if offset >= 8. Note that this has to be two 8-byte load/stores
|
||||
// and not one 16-byte load/store, and the first store has to be before the
|
||||
// second load, due to the overlap if offset is in the range [8, 16).
|
||||
//
|
||||
// if length > 16 || offset < 8 || len(dst)-d < 16 {
|
||||
// goto slowForwardCopy
|
||||
// }
|
||||
// copy 16 bytes
|
||||
// d += length
|
||||
CMP $16, R4
|
||||
BGT slowForwardCopy
|
||||
CMP $8, R5
|
||||
BLT slowForwardCopy
|
||||
CMP $16, R14
|
||||
BLT slowForwardCopy
|
||||
MOVD 0(R15), R2
|
||||
MOVD R2, 0(R7)
|
||||
MOVD 8(R15), R3
|
||||
MOVD R3, 8(R7)
|
||||
ADD R4, R7, R7
|
||||
B loop
|
||||
|
||||
slowForwardCopy:
|
||||
// !!! If the forward copy is longer than 16 bytes, or if offset < 8, we
|
||||
// can still try 8-byte load stores, provided we can overrun up to 10 extra
|
||||
// bytes. As above, the overrun will be fixed up by subsequent iterations
|
||||
// of the outermost loop.
|
||||
//
|
||||
// The C++ snappy code calls this technique IncrementalCopyFastPath. Its
|
||||
// commentary says:
|
||||
//
|
||||
// ----
|
||||
//
|
||||
// The main part of this loop is a simple copy of eight bytes at a time
|
||||
// until we've copied (at least) the requested amount of bytes. However,
|
||||
// if d and d-offset are less than eight bytes apart (indicating a
|
||||
// repeating pattern of length < 8), we first need to expand the pattern in
|
||||
// order to get the correct results. For instance, if the buffer looks like
|
||||
// this, with the eight-byte <d-offset> and <d> patterns marked as
|
||||
// intervals:
|
||||
//
|
||||
// abxxxxxxxxxxxx
|
||||
// [------] d-offset
|
||||
// [------] d
|
||||
//
|
||||
// a single eight-byte copy from <d-offset> to <d> will repeat the pattern
|
||||
// once, after which we can move <d> two bytes without moving <d-offset>:
|
||||
//
|
||||
// ababxxxxxxxxxx
|
||||
// [------] d-offset
|
||||
// [------] d
|
||||
//
|
||||
// and repeat the exercise until the two no longer overlap.
|
||||
//
|
||||
// This allows us to do very well in the special case of one single byte
|
||||
// repeated many times, without taking a big hit for more general cases.
|
||||
//
|
||||
// The worst case of extra writing past the end of the match occurs when
|
||||
// offset == 1 and length == 1; the last copy will read from byte positions
|
||||
// [0..7] and write to [4..11], whereas it was only supposed to write to
|
||||
// position 1. Thus, ten excess bytes.
|
||||
//
|
||||
// ----
|
||||
//
|
||||
// That "10 byte overrun" worst case is confirmed by Go's
|
||||
// TestSlowForwardCopyOverrun, which also tests the fixUpSlowForwardCopy
|
||||
// and finishSlowForwardCopy algorithm.
|
||||
//
|
||||
// if length > len(dst)-d-10 {
|
||||
// goto verySlowForwardCopy
|
||||
// }
|
||||
SUB $10, R14, R14
|
||||
CMP R14, R4
|
||||
BGT verySlowForwardCopy
|
||||
|
||||
makeOffsetAtLeast8:
|
||||
// !!! As above, expand the pattern so that offset >= 8 and we can use
|
||||
// 8-byte load/stores.
|
||||
//
|
||||
// for offset < 8 {
|
||||
// copy 8 bytes from dst[d-offset:] to dst[d:]
|
||||
// length -= offset
|
||||
// d += offset
|
||||
// offset += offset
|
||||
// // The two previous lines together means that d-offset, and therefore
|
||||
// // R15, is unchanged.
|
||||
// }
|
||||
CMP $8, R5
|
||||
BGE fixUpSlowForwardCopy
|
||||
MOVD (R15), R3
|
||||
MOVD R3, (R7)
|
||||
SUB R5, R4, R4
|
||||
ADD R5, R7, R7
|
||||
ADD R5, R5, R5
|
||||
B makeOffsetAtLeast8
|
||||
|
||||
fixUpSlowForwardCopy:
|
||||
// !!! Add length (which might be negative now) to d (implied by R7 being
|
||||
// &dst[d]) so that d ends up at the right place when we jump back to the
|
||||
// top of the loop. Before we do that, though, we save R7 to R2 so that, if
|
||||
// length is positive, copying the remaining length bytes will write to the
|
||||
// right place.
|
||||
MOVD R7, R2
|
||||
ADD R4, R7, R7
|
||||
|
||||
finishSlowForwardCopy:
|
||||
// !!! Repeat 8-byte load/stores until length <= 0. Ending with a negative
|
||||
// length means that we overrun, but as above, that will be fixed up by
|
||||
// subsequent iterations of the outermost loop.
|
||||
MOVD $0, R1
|
||||
CMP R1, R4
|
||||
BLE loop
|
||||
MOVD (R15), R3
|
||||
MOVD R3, (R2)
|
||||
ADD $8, R15, R15
|
||||
ADD $8, R2, R2
|
||||
SUB $8, R4, R4
|
||||
B finishSlowForwardCopy
|
||||
|
||||
verySlowForwardCopy:
|
||||
// verySlowForwardCopy is a simple implementation of forward copy. In C
|
||||
// parlance, this is a do/while loop instead of a while loop, since we know
|
||||
// that length > 0. In Go syntax:
|
||||
//
|
||||
// for {
|
||||
// dst[d] = dst[d - offset]
|
||||
// d++
|
||||
// length--
|
||||
// if length == 0 {
|
||||
// break
|
||||
// }
|
||||
// }
|
||||
MOVB (R15), R3
|
||||
MOVB R3, (R7)
|
||||
ADD $1, R15, R15
|
||||
ADD $1, R7, R7
|
||||
SUB $1, R4, R4
|
||||
CBNZ R4, verySlowForwardCopy
|
||||
B loop
|
||||
|
||||
// The code above handles copy tags.
|
||||
// ----------------------------------------
|
||||
|
||||
end:
|
||||
// This is the end of the "for s < len(src)".
|
||||
//
|
||||
// if d != len(dst) { etc }
|
||||
CMP R10, R7
|
||||
BNE errCorrupt
|
||||
|
||||
// return 0
|
||||
MOVD $0, ret+48(FP)
|
||||
RET
|
||||
|
||||
errCorrupt:
|
||||
// return decodeErrCodeCorrupt
|
||||
MOVD $1, R2
|
||||
MOVD R2, ret+48(FP)
|
||||
RET
|
|
@ -5,6 +5,7 @@
|
|||
// +build !appengine
|
||||
// +build gc
|
||||
// +build !noasm
|
||||
// +build amd64 arm64
|
||||
|
||||
package snappy
|
||||
|
|
@ -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 appengine !gc noasm
|
||||
// +build !amd64,!arm64 appengine !gc noasm
|
||||
|
||||
package snappy
|
||||
|
||||
|
@ -85,14 +85,28 @@ func decode(dst, src []byte) int {
|
|||
if offset <= 0 || d < offset || length > len(dst)-d {
|
||||
return decodeErrCodeCorrupt
|
||||
}
|
||||
// Copy from an earlier sub-slice of dst to a later sub-slice. Unlike
|
||||
// the built-in copy function, this byte-by-byte copy always runs
|
||||
// Copy from an earlier sub-slice of dst to a later sub-slice.
|
||||
// If no overlap, use the built-in copy:
|
||||
if offset >= length {
|
||||
copy(dst[d:d+length], dst[d-offset:])
|
||||
d += length
|
||||
continue
|
||||
}
|
||||
|
||||
// Unlike the built-in copy function, this byte-by-byte copy always runs
|
||||
// forwards, even if the slices overlap. Conceptually, this is:
|
||||
//
|
||||
// d += forwardCopy(dst[d:d+length], dst[d-offset:])
|
||||
for end := d + length; d != end; d++ {
|
||||
dst[d] = dst[d-offset]
|
||||
//
|
||||
// We align the slices into a and b and show the compiler they are the same size.
|
||||
// This allows the loop to run without bounds checks.
|
||||
a := dst[d : d+length]
|
||||
b := dst[d-offset:]
|
||||
b = b[:len(a)]
|
||||
for i := range a {
|
||||
a[i] = b[i]
|
||||
}
|
||||
d += length
|
||||
}
|
||||
if d != len(dst) {
|
||||
return decodeErrCodeCorrupt
|
||||
|
|
|
@ -15,6 +15,8 @@ import (
|
|||
// Otherwise, a newly allocated slice will be returned.
|
||||
//
|
||||
// The dst and src must not overlap. It is valid to pass a nil dst.
|
||||
//
|
||||
// Encode handles the Snappy block format, not the Snappy stream format.
|
||||
func Encode(dst, src []byte) []byte {
|
||||
if n := MaxEncodedLen(len(src)); n < 0 {
|
||||
panic(ErrTooLarge)
|
||||
|
@ -139,6 +141,8 @@ func NewBufferedWriter(w io.Writer) *Writer {
|
|||
}
|
||||
|
||||
// Writer is an io.Writer that can write Snappy-compressed bytes.
|
||||
//
|
||||
// Writer handles the Snappy stream format, not the Snappy block format.
|
||||
type Writer struct {
|
||||
w io.Writer
|
||||
err error
|
||||
|
|
|
@ -0,0 +1,722 @@
|
|||
// Copyright 2020 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 !appengine
|
||||
// +build gc
|
||||
// +build !noasm
|
||||
|
||||
#include "textflag.h"
|
||||
|
||||
// The asm code generally follows the pure Go code in encode_other.go, except
|
||||
// where marked with a "!!!".
|
||||
|
||||
// ----------------------------------------------------------------------------
|
||||
|
||||
// func emitLiteral(dst, lit []byte) int
|
||||
//
|
||||
// All local variables fit into registers. The register allocation:
|
||||
// - R3 len(lit)
|
||||
// - R4 n
|
||||
// - R6 return value
|
||||
// - R8 &dst[i]
|
||||
// - R10 &lit[0]
|
||||
//
|
||||
// The 32 bytes of stack space is to call runtime·memmove.
|
||||
//
|
||||
// The unusual register allocation of local variables, such as R10 for the
|
||||
// source pointer, matches the allocation used at the call site in encodeBlock,
|
||||
// which makes it easier to manually inline this function.
|
||||
TEXT ·emitLiteral(SB), NOSPLIT, $32-56
|
||||
MOVD dst_base+0(FP), R8
|
||||
MOVD lit_base+24(FP), R10
|
||||
MOVD lit_len+32(FP), R3
|
||||
MOVD R3, R6
|
||||
MOVW R3, R4
|
||||
SUBW $1, R4, R4
|
||||
|
||||
CMPW $60, R4
|
||||
BLT oneByte
|
||||
CMPW $256, R4
|
||||
BLT twoBytes
|
||||
|
||||
threeBytes:
|
||||
MOVD $0xf4, R2
|
||||
MOVB R2, 0(R8)
|
||||
MOVW R4, 1(R8)
|
||||
ADD $3, R8, R8
|
||||
ADD $3, R6, R6
|
||||
B memmove
|
||||
|
||||
twoBytes:
|
||||
MOVD $0xf0, R2
|
||||
MOVB R2, 0(R8)
|
||||
MOVB R4, 1(R8)
|
||||
ADD $2, R8, R8
|
||||
ADD $2, R6, R6
|
||||
B memmove
|
||||
|
||||
oneByte:
|
||||
LSLW $2, R4, R4
|
||||
MOVB R4, 0(R8)
|
||||
ADD $1, R8, R8
|
||||
ADD $1, R6, R6
|
||||
|
||||
memmove:
|
||||
MOVD R6, ret+48(FP)
|
||||
|
||||
// copy(dst[i:], lit)
|
||||
//
|
||||
// This means calling runtime·memmove(&dst[i], &lit[0], len(lit)), so we push
|
||||
// R8, R10 and R3 as arguments.
|
||||
MOVD R8, 8(RSP)
|
||||
MOVD R10, 16(RSP)
|
||||
MOVD R3, 24(RSP)
|
||||
CALL runtime·memmove(SB)
|
||||
RET
|
||||
|
||||
// ----------------------------------------------------------------------------
|
||||
|
||||
// func emitCopy(dst []byte, offset, length int) int
|
||||
//
|
||||
// All local variables fit into registers. The register allocation:
|
||||
// - R3 length
|
||||
// - R7 &dst[0]
|
||||
// - R8 &dst[i]
|
||||
// - R11 offset
|
||||
//
|
||||
// The unusual register allocation of local variables, such as R11 for the
|
||||
// offset, matches the allocation used at the call site in encodeBlock, which
|
||||
// makes it easier to manually inline this function.
|
||||
TEXT ·emitCopy(SB), NOSPLIT, $0-48
|
||||
MOVD dst_base+0(FP), R8
|
||||
MOVD R8, R7
|
||||
MOVD offset+24(FP), R11
|
||||
MOVD length+32(FP), R3
|
||||
|
||||
loop0:
|
||||
// for length >= 68 { etc }
|
||||
CMPW $68, R3
|
||||
BLT step1
|
||||
|
||||
// Emit a length 64 copy, encoded as 3 bytes.
|
||||
MOVD $0xfe, R2
|
||||
MOVB R2, 0(R8)
|
||||
MOVW R11, 1(R8)
|
||||
ADD $3, R8, R8
|
||||
SUB $64, R3, R3
|
||||
B loop0
|
||||
|
||||
step1:
|
||||
// if length > 64 { etc }
|
||||
CMP $64, R3
|
||||
BLE step2
|
||||
|
||||
// Emit a length 60 copy, encoded as 3 bytes.
|
||||
MOVD $0xee, R2
|
||||
MOVB R2, 0(R8)
|
||||
MOVW R11, 1(R8)
|
||||
ADD $3, R8, R8
|
||||
SUB $60, R3, R3
|
||||
|
||||
step2:
|
||||
// if length >= 12 || offset >= 2048 { goto step3 }
|
||||
CMP $12, R3
|
||||
BGE step3
|
||||
CMPW $2048, R11
|
||||
BGE step3
|
||||
|
||||
// Emit the remaining copy, encoded as 2 bytes.
|
||||
MOVB R11, 1(R8)
|
||||
LSRW $3, R11, R11
|
||||
AND $0xe0, R11, R11
|
||||
SUB $4, R3, R3
|
||||
LSLW $2, R3
|
||||
AND $0xff, R3, R3
|
||||
ORRW R3, R11, R11
|
||||
ORRW $1, R11, R11
|
||||
MOVB R11, 0(R8)
|
||||
ADD $2, R8, R8
|
||||
|
||||
// Return the number of bytes written.
|
||||
SUB R7, R8, R8
|
||||
MOVD R8, ret+40(FP)
|
||||
RET
|
||||
|
||||
step3:
|
||||
// Emit the remaining copy, encoded as 3 bytes.
|
||||
SUB $1, R3, R3
|
||||
AND $0xff, R3, R3
|
||||
LSLW $2, R3, R3
|
||||
ORRW $2, R3, R3
|
||||
MOVB R3, 0(R8)
|
||||
MOVW R11, 1(R8)
|
||||
ADD $3, R8, R8
|
||||
|
||||
// Return the number of bytes written.
|
||||
SUB R7, R8, R8
|
||||
MOVD R8, ret+40(FP)
|
||||
RET
|
||||
|
||||
// ----------------------------------------------------------------------------
|
||||
|
||||
// func extendMatch(src []byte, i, j int) int
|
||||
//
|
||||
// All local variables fit into registers. The register allocation:
|
||||
// - R6 &src[0]
|
||||
// - R7 &src[j]
|
||||
// - R13 &src[len(src) - 8]
|
||||
// - R14 &src[len(src)]
|
||||
// - R15 &src[i]
|
||||
//
|
||||
// The unusual register allocation of local variables, such as R15 for a source
|
||||
// pointer, matches the allocation used at the call site in encodeBlock, which
|
||||
// makes it easier to manually inline this function.
|
||||
TEXT ·extendMatch(SB), NOSPLIT, $0-48
|
||||
MOVD src_base+0(FP), R6
|
||||
MOVD src_len+8(FP), R14
|
||||
MOVD i+24(FP), R15
|
||||
MOVD j+32(FP), R7
|
||||
ADD R6, R14, R14
|
||||
ADD R6, R15, R15
|
||||
ADD R6, R7, R7
|
||||
MOVD R14, R13
|
||||
SUB $8, R13, R13
|
||||
|
||||
cmp8:
|
||||
// As long as we are 8 or more bytes before the end of src, we can load and
|
||||
// compare 8 bytes at a time. If those 8 bytes are equal, repeat.
|
||||
CMP R13, R7
|
||||
BHI cmp1
|
||||
MOVD (R15), R3
|
||||
MOVD (R7), R4
|
||||
CMP R4, R3
|
||||
BNE bsf
|
||||
ADD $8, R15, R15
|
||||
ADD $8, R7, R7
|
||||
B cmp8
|
||||
|
||||
bsf:
|
||||
// If those 8 bytes were not equal, XOR the two 8 byte values, and return
|
||||
// the index of the first byte that differs.
|
||||
// RBIT reverses the bit order, then CLZ counts the leading zeros, the
|
||||
// combination of which finds the least significant bit which is set.
|
||||
// The arm64 architecture is little-endian, and the shift by 3 converts
|
||||
// a bit index to a byte index.
|
||||
EOR R3, R4, R4
|
||||
RBIT R4, R4
|
||||
CLZ R4, R4
|
||||
ADD R4>>3, R7, R7
|
||||
|
||||
// Convert from &src[ret] to ret.
|
||||
SUB R6, R7, R7
|
||||
MOVD R7, ret+40(FP)
|
||||
RET
|
||||
|
||||
cmp1:
|
||||
// In src's tail, compare 1 byte at a time.
|
||||
CMP R7, R14
|
||||
BLS extendMatchEnd
|
||||
MOVB (R15), R3
|
||||
MOVB (R7), R4
|
||||
CMP R4, R3
|
||||
BNE extendMatchEnd
|
||||
ADD $1, R15, R15
|
||||
ADD $1, R7, R7
|
||||
B cmp1
|
||||
|
||||
extendMatchEnd:
|
||||
// Convert from &src[ret] to ret.
|
||||
SUB R6, R7, R7
|
||||
MOVD R7, ret+40(FP)
|
||||
RET
|
||||
|
||||
// ----------------------------------------------------------------------------
|
||||
|
||||
// func encodeBlock(dst, src []byte) (d int)
|
||||
//
|
||||
// All local variables fit into registers, other than "var table". The register
|
||||
// allocation:
|
||||
// - R3 . .
|
||||
// - R4 . .
|
||||
// - R5 64 shift
|
||||
// - R6 72 &src[0], tableSize
|
||||
// - R7 80 &src[s]
|
||||
// - R8 88 &dst[d]
|
||||
// - R9 96 sLimit
|
||||
// - R10 . &src[nextEmit]
|
||||
// - R11 104 prevHash, currHash, nextHash, offset
|
||||
// - R12 112 &src[base], skip
|
||||
// - R13 . &src[nextS], &src[len(src) - 8]
|
||||
// - R14 . len(src), bytesBetweenHashLookups, &src[len(src)], x
|
||||
// - R15 120 candidate
|
||||
// - R16 . hash constant, 0x1e35a7bd
|
||||
// - R17 . &table
|
||||
// - . 128 table
|
||||
//
|
||||
// The second column (64, 72, etc) is the stack offset to spill the registers
|
||||
// when calling other functions. We could pack this slightly tighter, but it's
|
||||
// simpler to have a dedicated spill map independent of the function called.
|
||||
//
|
||||
// "var table [maxTableSize]uint16" takes up 32768 bytes of stack space. An
|
||||
// extra 64 bytes, to call other functions, and an extra 64 bytes, to spill
|
||||
// local variables (registers) during calls gives 32768 + 64 + 64 = 32896.
|
||||
TEXT ·encodeBlock(SB), 0, $32896-56
|
||||
MOVD dst_base+0(FP), R8
|
||||
MOVD src_base+24(FP), R7
|
||||
MOVD src_len+32(FP), R14
|
||||
|
||||
// shift, tableSize := uint32(32-8), 1<<8
|
||||
MOVD $24, R5
|
||||
MOVD $256, R6
|
||||
MOVW $0xa7bd, R16
|
||||
MOVKW $(0x1e35<<16), R16
|
||||
|
||||
calcShift:
|
||||
// for ; tableSize < maxTableSize && tableSize < len(src); tableSize *= 2 {
|
||||
// shift--
|
||||
// }
|
||||
MOVD $16384, R2
|
||||
CMP R2, R6
|
||||
BGE varTable
|
||||
CMP R14, R6
|
||||
BGE varTable
|
||||
SUB $1, R5, R5
|
||||
LSL $1, R6, R6
|
||||
B calcShift
|
||||
|
||||
varTable:
|
||||
// var table [maxTableSize]uint16
|
||||
//
|
||||
// In the asm code, unlike the Go code, we can zero-initialize only the
|
||||
// first tableSize elements. Each uint16 element is 2 bytes and each
|
||||
// iterations writes 64 bytes, so we can do only tableSize/32 writes
|
||||
// instead of the 2048 writes that would zero-initialize all of table's
|
||||
// 32768 bytes. This clear could overrun the first tableSize elements, but
|
||||
// it won't overrun the allocated stack size.
|
||||
ADD $12 |