Update dependencies (#1851)

69 files changed, 8168 insertions, 2784 deletions

diff --git a/vendor/github.com/SevereCloud/vksdk/v2/.golangci.yml b/vendor/github.com/SevereCloud/vksdk/v2/.golangci.yml
index ff1db59c..66712f60 100644
--- a/vendor/github.com/SevereCloud/vksdk/v2/.golangci.yml
+++ b/vendor/github.com/SevereCloud/vksdk/v2/.golangci.yml
@@ -57,6 +57,8 @@ linters:
     - grouper
     - decorder
     - containedctx
+    # - execinquery # FIXME: panic in 1.46.0
+    - nosprintfhostport
 
 # - wrapcheck # TODO: v3 Fix
 # - testpackage # TODO: Fix testpackage
@@ -87,6 +89,7 @@ linters:
 # - varnamelen
 # - errchkjson
 # - maintidx
+# - nonamedreturns
 
 # depricated
 # - maligned
diff --git a/vendor/github.com/SevereCloud/vksdk/v2/doc.go b/vendor/github.com/SevereCloud/vksdk/v2/doc.go
index 3e3c12a1..aa532f9c 100644
--- a/vendor/github.com/SevereCloud/vksdk/v2/doc.go
+++ b/vendor/github.com/SevereCloud/vksdk/v2/doc.go
@@ -7,6 +7,6 @@ package vksdk
 
 // Module constants.
 const (
-	Version = "2.14.0"
+	Version = "2.14.1"
 	API     = "5.131"
 )
diff --git a/vendor/github.com/SevereCloud/vksdk/v2/object/groups.go b/vendor/github.com/SevereCloud/vksdk/v2/object/groups.go
index 6cabae01..8f1636b4 100644
--- a/vendor/github.com/SevereCloud/vksdk/v2/object/groups.go
+++ b/vendor/github.com/SevereCloud/vksdk/v2/object/groups.go
@@ -210,7 +210,7 @@ type GroupsGroup struct {
 	MainSection     int                  `json:"main_section,omitempty"`
 	OnlineStatus    GroupsOnlineStatus   `json:"online_status,omitempty"` // Status of replies in community messages
 	AgeLimits       int                  `json:"age_limits,omitempty"`    // Information whether age limit
-	BanInfo         GroupsGroupBanInfo   `json:"ban_info,omitempty"`      // User ban info
+	BanInfo         *GroupsGroupBanInfo  `json:"ban_info,omitempty"`      // User ban info
 	Addresses       GroupsAddressesInfo  `json:"addresses,omitempty"`     // Info about addresses in Groups
 	LiveCovers      GroupsLiveCovers     `json:"live_covers,omitempty"`
 	CropPhoto       UsersCropPhoto       `json:"crop_photo,omitempty"`
@@ -963,10 +963,10 @@ type GroupsOnlineStatus struct {
 
 // GroupsOwnerXtrBanInfo struct.
 type GroupsOwnerXtrBanInfo struct {
-	BanInfo GroupsBanInfo `json:"ban_info"`
-	Group   GroupsGroup   `json:"group"`
-	Profile UsersUser     `json:"profile"`
-	Type    string        `json:"type"`
+	BanInfo *GroupsBanInfo `json:"ban_info"`
+	Group   GroupsGroup    `json:"group"`
+	Profile UsersUser      `json:"profile"`
+	Type    string         `json:"type"`
 }
 
 // GroupsSubjectItem struct.
diff --git a/vendor/github.com/SevereCloud/vksdk/v2/object/video.go b/vendor/github.com/SevereCloud/vksdk/v2/object/video.go
index d86143a9..6774fa00 100644
--- a/vendor/github.com/SevereCloud/vksdk/v2/object/video.go
+++ b/vendor/github.com/SevereCloud/vksdk/v2/object/video.go
@@ -31,7 +31,7 @@ type VideoVideo struct {
 	CanLike BaseBoolInt `json:"can_like"`
 
 	// Information whether current user can download the video.
-	CanDownload BaseBoolInt `json:"can_download"`
+	CanDownload int `json:"can_download"`
 
 	// Information whether current user can repost this video.
 	CanRepost         BaseBoolInt       `json:"can_repost"`
diff --git a/vendor/github.com/d5/tengo/v2/compiler.go b/vendor/github.com/d5/tengo/v2/compiler.go
index e4e04303..6591603c 100644
--- a/vendor/github.com/d5/tengo/v2/compiler.go
+++ b/vendor/github.com/d5/tengo/v2/compiler.go
@@ -1220,14 +1220,14 @@ func (c *Compiler) optimizeFunc(node parser.Node) {
 	iterateInstructions(c.scopes[c.scopeIndex].Instructions,
 		func(pos int, opcode parser.Opcode, operands []int) bool {
 			switch {
+			case dsts[pos]:
+				dstIdx++
+				deadCode = false
 			case opcode == parser.OpReturn:
 				if deadCode {
 					return true
 				}
 				deadCode = true
-			case dsts[pos]:
-				dstIdx++
-				deadCode = false
 			case deadCode:
 				return true
 			}
@@ -1242,6 +1242,7 @@ func (c *Compiler) optimizeFunc(node parser.Node) {
 	var appendReturn bool
 	endPos := len(c.scopes[c.scopeIndex].Instructions)
 	newEndPost := len(newInsts)
+
 	iterateInstructions(newInsts,
 		func(pos int, opcode parser.Opcode, operands []int) bool {
 			switch opcode {
diff --git a/vendor/github.com/d5/tengo/v2/parser/parser.go b/vendor/github.com/d5/tengo/v2/parser/parser.go
index fd20423b..eaa9dc93 100644
--- a/vendor/github.com/d5/tengo/v2/parser/parser.go
+++ b/vendor/github.com/d5/tengo/v2/parser/parser.go
@@ -375,7 +375,12 @@ func (p *Parser) parseOperand() Expr {
 	case token.Ident:
 		return p.parseIdent()
 	case token.Int:
-		v, _ := strconv.ParseInt(p.tokenLit, 10, 64)
+		v, err := strconv.ParseInt(p.tokenLit, 0, 64)
+		if err == strconv.ErrRange {
+			p.error(p.pos, "number out of range")
+		} else if err != nil {
+			p.error(p.pos, "invalid integer")
+		}
 		x := &IntLit{
 			Value:    v,
 			ValuePos: p.pos,
@@ -383,8 +388,14 @@ func (p *Parser) parseOperand() Expr {
 		}
 		p.next()
 		return x
+
 	case token.Float:
-		v, _ := strconv.ParseFloat(p.tokenLit, 64)
+		v, err := strconv.ParseFloat(p.tokenLit, 64)
+		if err == strconv.ErrRange {
+			p.error(p.pos, "number out of range")
+		} else if err != nil {
+			p.error(p.pos, "invalid float")
+		}
 		x := &FloatLit{
 			Value:    v,
 			ValuePos: p.pos,
@@ -447,10 +458,11 @@ func (p *Parser) parseOperand() Expr {
 		return p.parseErrorExpr()
 	case token.Immutable: // immutable expression
 		return p.parseImmutableExpr()
+	default:
+		p.errorExpected(p.pos, "operand")
 	}
 
 	pos := p.pos
-	p.errorExpected(pos, "operand")
 	p.advance(stmtStart)
 	return &BadExpr{From: pos, To: p.pos}
 }
diff --git a/vendor/github.com/d5/tengo/v2/parser/scanner.go b/vendor/github.com/d5/tengo/v2/parser/scanner.go
index f1d820a4..5f797706 100644
--- a/vendor/github.com/d5/tengo/v2/parser/scanner.go
+++ b/vendor/github.com/d5/tengo/v2/parser/scanner.go
@@ -93,9 +93,9 @@ func (s *Scanner) Scan() (
 			token.Export, token.True, token.False, token.Undefined:
 			insertSemi = true
 		}
-	case '0' <= ch && ch <= '9':
+	case ('0' <= ch && ch <= '9') || (ch == '.' && '0' <= s.peek() && s.peek() <= '9'):
 		insertSemi = true
-		tok, literal = s.scanNumber(false)
+		tok, literal = s.scanNumber()
 	default:
 		s.next() // always make progress
 
@@ -125,16 +125,11 @@ func (s *Scanner) Scan() (
 		case ':':
 			tok = s.switch2(token.Colon, token.Define)
 		case '.':
-			if '0' <= s.ch && s.ch <= '9' {
-				insertSemi = true
-				tok, literal = s.scanNumber(true)
-			} else {
-				tok = token.Period
-				if s.ch == '.' && s.peek() == '.' {
-					s.next()
-					s.next() // consume last '.'
-					tok = token.Ellipsis
-				}
+			tok = token.Period
+			if s.ch == '.' && s.peek() == '.' {
+				s.next()
+				s.next() // consume last '.'
+				tok = token.Ellipsis
 			}
 		case ',':
 			tok = token.Comma
@@ -379,86 +374,58 @@ func (s *Scanner) scanIdentifier() string {
 	return string(s.src[offs:s.offset])
 }
 
-func (s *Scanner) scanMantissa(base int) {
-	for digitVal(s.ch) < base {
+func (s *Scanner) scanDigits(base int) {
+	for s.ch == '_' || digitVal(s.ch) < base {
 		s.next()
 	}
 }
 
-func (s *Scanner) scanNumber(
-	seenDecimalPoint bool,
-) (tok token.Token, lit string) {
-	// digitVal(s.ch) < 10
+func (s *Scanner) scanNumber() (token.Token, string) {
 	offs := s.offset
-	tok = token.Int
-
-	defer func() {
-		lit = string(s.src[offs:s.offset])
-	}()
-
-	if seenDecimalPoint {
-		offs--
-		tok = token.Float
-		s.scanMantissa(10)
-		goto exponent
-	}
+	tok := token.Int
+	base := 10
 
-	if s.ch == '0' {
-		// int or float
-		offs := s.offset
+	// Determine base
+	switch {
+	case s.ch == '0' && lower(s.peek()) == 'b':
+		base = 2
+		s.next()
+		s.next()
+	case s.ch == '0' && lower(s.peek()) == 'o':
+		base = 8
+		s.next()
+		s.next()
+	case s.ch == '0' && lower(s.peek()) == 'x':
+		base = 16
+		s.next()
 		s.next()
-		if s.ch == 'x' || s.ch == 'X' {
-			// hexadecimal int
-			s.next()
-			s.scanMantissa(16)
-			if s.offset-offs <= 2 {
-				// only scanned "0x" or "0X"
-				s.error(offs, "illegal hexadecimal number")
-			}
-		} else {
-			// octal int or float
-			seenDecimalDigit := false
-			s.scanMantissa(8)
-			if s.ch == '8' || s.ch == '9' {
-				// illegal octal int or float
-				seenDecimalDigit = true
-				s.scanMantissa(10)
-			}
-			if s.ch == '.' || s.ch == 'e' || s.ch == 'E' || s.ch == 'i' {
-				goto fraction
-			}
-			// octal int
-			if seenDecimalDigit {
-				s.error(offs, "illegal octal number")
-			}
-		}
-		return
 	}
 
-	// decimal int or float
-	s.scanMantissa(10)
+	// Scan whole number
+	s.scanDigits(base)
 
-fraction:
-	if s.ch == '.' {
+	// Scan fractional part
+	if s.ch == '.' && (base == 10 || base == 16) {
 		tok = token.Float
 		s.next()
-		s.scanMantissa(10)
+		s.scanDigits(base)
 	}
 
-exponent:
-	if s.ch == 'e' || s.ch == 'E' {
+	// Scan exponent
+	if s.ch == 'e' || s.ch == 'E' || s.ch == 'p' || s.ch == 'P' {
 		tok = token.Float
 		s.next()
 		if s.ch == '-' || s.ch == '+' {
 			s.next()
 		}
-		if digitVal(s.ch) < 10 {
-			s.scanMantissa(10)
-		} else {
-			s.error(offs, "illegal floating-point exponent")
+		offs := s.offset
+		s.scanDigits(10)
+		if offs == s.offset {
+			s.error(offs, "exponent has no digits")
 		}
 	}
-	return
+
+	return tok, string(s.src[offs:s.offset])
 }
 
 func (s *Scanner) scanEscape(quote rune) bool {
@@ -687,3 +654,7 @@ func digitVal(ch rune) int {
 	}
 	return 16 // larger than any legal digit val
 }
+
+func lower(c byte) byte {
+	return c | ('x' - 'X')
+}
diff --git a/vendor/github.com/gomarkdown/markdown/.gitpod.yml b/vendor/github.com/gomarkdown/markdown/.gitpod.yml
new file mode 100644
index 00000000..dde1f1d9
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/.gitpod.yml
@@ -0,0 +1,9 @@
+# This configuration file was automatically generated by Gitpod.
+# Please adjust to your needs (see https://www.gitpod.io/docs/config-gitpod-file)
+# and commit this file to your remote git repository to share the goodness with others.
+
+tasks:
+  - init: go get && go build ./... && go test ./...
+    command: go run
+
+
diff --git a/vendor/github.com/gomarkdown/markdown/parser/block.go b/vendor/github.com/gomarkdown/markdown/parser/block.go
index eeebec73..7c2401f1 100644
--- a/vendor/github.com/gomarkdown/markdown/parser/block.go
+++ b/vendor/github.com/gomarkdown/markdown/parser/block.go
@@ -1675,6 +1675,12 @@ func (p *Parser) paragraph(data []byte) int {
 			return i
 		}
 
+		// if there's a block quote, paragraph is over
+		if p.quotePrefix(current) > 0 {
+			p.renderParagraph(data[:i])
+			return i
+		}
+
 		// if there's a fenced code block, paragraph is over
 		if p.extensions&FencedCode != 0 {
 			if p.fencedCodeBlock(current, false) > 0 {
diff --git a/vendor/github.com/klauspost/compress/.gitignore b/vendor/github.com/klauspost/compress/.gitignore
index b35f8449..d31b3781 100644
--- a/vendor/github.com/klauspost/compress/.gitignore
+++ b/vendor/github.com/klauspost/compress/.gitignore
@@ -23,3 +23,10 @@ _testmain.go
 *.test
 *.prof
 /s2/cmd/_s2sx/sfx-exe
+
+# Linux perf files
+perf.data
+perf.data.old
+
+# gdb history
+.gdb_history
diff --git a/vendor/github.com/klauspost/compress/README.md b/vendor/github.com/klauspost/compress/README.md
index 0e2dc116..5c3c2a25 100644
--- a/vendor/github.com/klauspost/compress/README.md
+++ b/vendor/github.com/klauspost/compress/README.md
@@ -17,6 +17,41 @@ This package provides various compression algorithms.
 

 # changelog

 

+* May 25, 2022 (v1.15.5)

+	* s2: Add concurrent stream decompression https://github.com/klauspost/compress/pull/602

+	* s2: Fix final emit oob read crash on amd64 https://github.com/klauspost/compress/pull/601

+	* huff0: asm implementation of Decompress1X by @WojciechMula https://github.com/klauspost/compress/pull/596

+	* zstd: Use 1 less goroutine for stream decoding https://github.com/klauspost/compress/pull/588

+	* zstd: Copy literal in 16 byte blocks when possible https://github.com/klauspost/compress/pull/592

+	* zstd: Speed up when WithDecoderLowmem(false) https://github.com/klauspost/compress/pull/599

+	* zstd: faster next state update in BMI2 version of decode by @WojciechMula in https://github.com/klauspost/compress/pull/593

+	* huff0: Do not check max size when reading table. https://github.com/klauspost/compress/pull/586

+	* flate: Inplace hashing for level 7-9 by @klauspost in https://github.com/klauspost/compress/pull/590

+

+

+* May 11, 2022 (v1.15.4)

+	* huff0: decompress directly into output by @WojciechMula in [#577](https://github.com/klauspost/compress/pull/577)

+	* inflate: Keep dict on stack [#581](https://github.com/klauspost/compress/pull/581)

+	* zstd: Faster decoding memcopy in asm [#583](https://github.com/klauspost/compress/pull/583)

+	* zstd: Fix ignored crc [#580](https://github.com/klauspost/compress/pull/580)

+

+* May 5, 2022 (v1.15.3)

+	* zstd: Allow to ignore checksum checking by @WojciechMula [#572](https://github.com/klauspost/compress/pull/572)

+	* s2: Fix incorrect seek for io.SeekEnd in [#575](https://github.com/klauspost/compress/pull/575)

+

+* Apr 26, 2022 (v1.15.2)

+	* zstd: Add x86-64 assembly for decompression on streams and blocks. Contributed by [@WojciechMula](https://github.com/WojciechMula). Typically 2x faster.  [#528](https://github.com/klauspost/compress/pull/528) [#531](https://github.com/klauspost/compress/pull/531) [#545](https://github.com/klauspost/compress/pull/545) [#537](https://github.com/klauspost/compress/pull/537)

+	* zstd: Add options to ZipDecompressor and fixes [#539](https://github.com/klauspost/compress/pull/539)

+	* s2: Use sorted search for index [#555](https://github.com/klauspost/compress/pull/555)

+	* Minimum version is Go 1.16, added CI test on 1.18.

+

+* Mar 11, 2022 (v1.15.1)

+	* huff0: Add x86 assembly of Decode4X by @WojciechMula in [#512](https://github.com/klauspost/compress/pull/512)

+	* zstd: Reuse zip decoders in [#514](https://github.com/klauspost/compress/pull/514)

+	* zstd: Detect extra block data and report as corrupted in [#520](https://github.com/klauspost/compress/pull/520)

+	* zstd: Handle zero sized frame content size stricter in [#521](https://github.com/klauspost/compress/pull/521)

+	* zstd: Add stricter block size checks in [#523](https://github.com/klauspost/compress/pull/523)

+

 * Mar 3, 2022 (v1.15.0)

 	* zstd: Refactor decoder by @klauspost in [#498](https://github.com/klauspost/compress/pull/498)

 	* zstd: Add stream encoding without goroutines by @klauspost in [#505](https://github.com/klauspost/compress/pull/505)

@@ -60,6 +95,9 @@ While the release has been extensively tested, it is recommended to testing when
 	* zstd: add arm64 xxhash assembly in [#464](https://github.com/klauspost/compress/pull/464)

 	* Add garbled for binaries for s2 in [#445](https://github.com/klauspost/compress/pull/445)

 

+<details>

+	<summary>See changes to v1.13.x</summary>

+	

 * Aug 30, 2021 (v1.13.5)

 	* gz/zlib/flate: Alias stdlib errors [#425](https://github.com/klauspost/compress/pull/425)

 	* s2: Add block support to commandline tools [#413](https://github.com/klauspost/compress/pull/413)

@@ -88,6 +126,8 @@ While the release has been extensively tested, it is recommended to testing when
 	* Added [gzhttp](https://github.com/klauspost/compress/tree/master/gzhttp#gzip-handler) which allows wrapping HTTP servers and clients with GZIP compressors.

 	* zstd: Detect short invalid signatures [#382](https://github.com/klauspost/compress/pull/382)

 	* zstd: Spawn decoder goroutine only if needed. [#380](https://github.com/klauspost/compress/pull/380)

+</details>

+

 

 <details>

 	<summary>See changes to v1.12.x</summary>

diff --git a/vendor/github.com/klauspost/compress/huff0/autogen.go b/vendor/github.com/klauspost/compress/huff0/autogen.go
deleted file mode 100644
index ff2c69d6..00000000
--- a/vendor/github.com/klauspost/compress/huff0/autogen.go
+++ /dev/null
@@ -1,5 +0,0 @@
-package huff0
-
-//go:generate go run generate.go
-//go:generate asmfmt -w decompress_amd64.s
-//go:generate asmfmt -w decompress_8b_amd64.s
diff --git a/vendor/github.com/klauspost/compress/huff0/bitreader.go b/vendor/github.com/klauspost/compress/huff0/bitreader.go
index 451160ed..504a7be9 100644
--- a/vendor/github.com/klauspost/compress/huff0/bitreader.go
+++ b/vendor/github.com/klauspost/compress/huff0/bitreader.go
@@ -165,11 +165,6 @@ func (b *bitReaderShifted) peekBitsFast(n uint8) uint16 {
 	return uint16(b.value >> ((64 - n) & 63))
 }
 
-// peekTopBits(n) is equvialent to peekBitFast(64 - n)
-func (b *bitReaderShifted) peekTopBits(n uint8) uint16 {
-	return uint16(b.value >> n)
-}
-
 func (b *bitReaderShifted) advance(n uint8) {
 	b.bitsRead += n
 	b.value <<= n & 63
@@ -220,11 +215,6 @@ func (b *bitReaderShifted) fill() {
 	}
 }
 
-// finished returns true if all bits have been read from the bit stream.
-func (b *bitReaderShifted) finished() bool {
-	return b.off == 0 && b.bitsRead >= 64
-}
-
 func (b *bitReaderShifted) remaining() uint {
 	return b.off*8 + uint(64-b.bitsRead)
 }
diff --git a/vendor/github.com/klauspost/compress/huff0/bitwriter.go b/vendor/github.com/klauspost/compress/huff0/bitwriter.go
index 6bce4e87..ec71f7a3 100644
--- a/vendor/github.com/klauspost/compress/huff0/bitwriter.go
+++ b/vendor/github.com/klauspost/compress/huff0/bitwriter.go
@@ -5,8 +5,6 @@
 
 package huff0
 
-import "fmt"
-
 // bitWriter will write bits.
 // First bit will be LSB of the first byte of output.
 type bitWriter struct {
@@ -23,14 +21,6 @@ var bitMask16 = [32]uint16{
 	0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF,
 	0xFFFF, 0xFFFF} /* up to 16 bits */
 
-// addBits16NC will add up to 16 bits.
-// It will not check if there is space for them,
-// so the caller must ensure that it has flushed recently.
-func (b *bitWriter) addBits16NC(value uint16, bits uint8) {
-	b.bitContainer |= uint64(value&bitMask16[bits&31]) << (b.nBits & 63)
-	b.nBits += bits
-}
-
 // addBits16Clean will add up to 16 bits. value may not contain more set bits than indicated.
 // It will not check if there is space for them, so the caller must ensure that it has flushed recently.
 func (b *bitWriter) addBits16Clean(value uint16, bits uint8) {
@@ -70,104 +60,6 @@ func (b *bitWriter) encTwoSymbols(ct cTable, av, bv byte) {
 	b.nBits += encA.nBits + encB.nBits
 }
 
-// addBits16ZeroNC will add up to 16 bits.
-// It will not check if there is space for them,
-// so the caller must ensure that it has flushed recently.
-// This is fastest if bits can be zero.
-func (b *bitWriter) addBits16ZeroNC(value uint16, bits uint8) {
-	if bits == 0 {
-		return
-	}
-	value <<= (16 - bits) & 15
-	value >>= (16 - bits) & 15
-	b.bitContainer |= uint64(value) << (b.nBits & 63)
-	b.nBits += bits
-}
-
-// flush will flush all pending full bytes.
-// There will be at least 56 bits available for writing when this has been called.
-// Using flush32 is faster, but leaves less space for writing.
-func (b *bitWriter) flush() {
-	v := b.nBits >> 3
-	switch v {
-	case 0:
-		return
-	case 1:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-		)
-		b.bitContainer >>= 1 << 3
-	case 2:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-		)
-		b.bitContainer >>= 2 << 3
-	case 3:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-			byte(b.bitContainer>>16),
-		)
-		b.bitContainer >>= 3 << 3
-	case 4:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-			byte(b.bitContainer>>16),
-			byte(b.bitContainer>>24),
-		)
-		b.bitContainer >>= 4 << 3
-	case 5:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-			byte(b.bitContainer>>16),
-			byte(b.bitContainer>>24),
-			byte(b.bitContainer>>32),
-		)
-		b.bitContainer >>= 5 << 3
-	case 6:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-			byte(b.bitContainer>>16),
-			byte(b.bitContainer>>24),
-			byte(b.bitContainer>>32),
-			byte(b.bitContainer>>40),
-		)
-		b.bitContainer >>= 6 << 3
-	case 7:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-			byte(b.bitContainer>>16),
-			byte(b.bitContainer>>24),
-			byte(b.bitContainer>>32),
-			byte(b.bitContainer>>40),
-			byte(b.bitContainer>>48),
-		)
-		b.bitContainer >>= 7 << 3
-	case 8:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-			byte(b.bitContainer>>16),
-			byte(b.bitContainer>>24),
-			byte(b.bitContainer>>32),
-			byte(b.bitContainer>>40),
-			byte(b.bitContainer>>48),
-			byte(b.bitContainer>>56),
-		)
-		b.bitContainer = 0
-		b.nBits = 0
-		return
-	default:
-		panic(fmt.Errorf("bits (%d) > 64", b.nBits))
-	}
-	b.nBits &= 7
-}
-
 // flush32 will flush out, so there are at least 32 bits available for writing.
 func (b *bitWriter) flush32() {
 	if b.nBits < 32 {
@@ -201,10 +93,3 @@ func (b *bitWriter) close() error {
 	b.flushAlign()
 	return nil
 }
-
-// reset and continue writing by appending to out.
-func (b *bitWriter) reset(out []byte) {
-	b.bitContainer = 0
-	b.nBits = 0
-	b.out = out
-}
diff --git a/vendor/github.com/klauspost/compress/huff0/bytereader.go b/vendor/github.com/klauspost/compress/huff0/bytereader.go
index 50bcdf6e..4dcab8d2 100644
--- a/vendor/github.com/klauspost/compress/huff0/bytereader.go
+++ b/vendor/github.com/klauspost/compress/huff0/bytereader.go
@@ -20,11 +20,6 @@ func (b *byteReader) init(in []byte) {
 	b.off = 0
 }
 
-// advance the stream b n bytes.
-func (b *byteReader) advance(n uint) {
-	b.off += int(n)
-}
-
 // Int32 returns a little endian int32 starting at current offset.
 func (b byteReader) Int32() int32 {
 	v3 := int32(b.b[b.off+3])
@@ -43,11 +38,6 @@ func (b byteReader) Uint32() uint32 {
 	return (v3 << 24) | (v2 << 16) | (v1 << 8) | v0
 }
 
-// unread returns the unread portion of the input.
-func (b byteReader) unread() []byte {
-	return b.b[b.off:]
-}
-
 // remain will return the number of bytes remaining.
 func (b byteReader) remain() int {
 	return len(b.b) - b.off
diff --git a/vendor/github.com/klauspost/compress/huff0/compress.go b/vendor/github.com/klauspost/compress/huff0/compress.go
index bc95ac62..4d14542f 100644
--- a/vendor/github.com/klauspost/compress/huff0/compress.go
+++ b/vendor/github.com/klauspost/compress/huff0/compress.go
@@ -404,6 +404,7 @@ func (s *Scratch) canUseTable(c cTable) bool {
 	return true
 }
 
+//lint:ignore U1000 used for debugging
 func (s *Scratch) validateTable(c cTable) bool {
 	if len(c) < int(s.symbolLen) {
 		return false
diff --git a/vendor/github.com/klauspost/compress/huff0/decompress.go b/vendor/github.com/klauspost/compress/huff0/decompress.go
index 04f65299..c0c48bd7 100644
--- a/vendor/github.com/klauspost/compress/huff0/decompress.go
+++ b/vendor/github.com/klauspost/compress/huff0/decompress.go
@@ -11,7 +11,6 @@ import (
 
 type dTable struct {
 	single []dEntrySingle
-	double []dEntryDouble
 }
 
 // single-symbols decoding
@@ -19,13 +18,6 @@ type dEntrySingle struct {
 	entry uint16
 }
 
-// double-symbols decoding
-type dEntryDouble struct {
-	seq   [4]byte
-	nBits uint8
-	len   uint8
-}
-
 // Uses special code for all tables that are < 8 bits.
 const use8BitTables = true
 
@@ -35,7 +27,7 @@ const use8BitTables = true
 // If no Scratch is provided a new one is allocated.
 // The returned Scratch can be used for encoding or decoding input using this table.
 func ReadTable(in []byte, s *Scratch) (s2 *Scratch, remain []byte, err error) {
-	s, err = s.prepare(in)
+	s, err = s.prepare(nil)
 	if err != nil {
 		return s, nil, err
 	}
@@ -236,108 +228,6 @@ func (d *Decoder) buffer() *[4][256]byte {
 	return &[4][256]byte{}
 }
 
-// Decompress1X will decompress a 1X encoded stream.
-// The cap of the output buffer will be the maximum decompressed size.
-// The length of the supplied input must match the end of a block exactly.
-func (d *Decoder) Decompress1X(dst, src []byte) ([]byte, error) {
-	if len(d.dt.single) == 0 {
-		return nil, errors.New("no table loaded")
-	}
-	if use8BitTables && d.actualTableLog <= 8 {
-		return d.decompress1X8Bit(dst, src)
-	}
-	var br bitReaderShifted
-	err := br.init(src)
-	if err != nil {
-		return dst, err
-	}
-	maxDecodedSize := cap(dst)
-	dst = dst[:0]
-
-	// Avoid bounds check by always having full sized table.
-	const tlSize = 1 << tableLogMax
-	const tlMask = tlSize - 1
-	dt := d.dt.single[:tlSize]
-
-	// Use temp table to avoid bound checks/append penalty.
-	bufs := d.buffer()
-	buf := &bufs[0]
-	var off uint8
-
-	for br.off >= 8 {
-		br.fillFast()
-		v := dt[br.peekBitsFast(d.actualTableLog)&tlMask]
-		br.advance(uint8(v.entry))
-		buf[off+0] = uint8(v.entry >> 8)
-
-		v = dt[br.peekBitsFast(d.actualTableLog)&tlMask]
-		br.advance(uint8(v.entry))
-		buf[off+1] = uint8(v.entry >> 8)
-
-		// Refill
-		br.fillFast()
-
-		v = dt[br.peekBitsFast(d.actualTableLog)&tlMask]
-		br.advance(uint8(v.entry))
-		buf[off+2] = uint8(v.entry >> 8)
-
-		v = dt[br.peekBitsFast(d.actualTableLog)&tlMask]
-		br.advance(uint8(v.entry))
-		buf[off+3] = uint8(v.entry >> 8)
-
-		off += 4
-		if off == 0 {
-			if len(dst)+256 > maxDecodedSize {
-				br.close()
-				d.bufs.Put(bufs)
-				return nil, ErrMaxDecodedSizeExceeded
-			}
-			dst = append(dst, buf[:]...)
-		}
-	}
-
-	if len(dst)+int(off) > maxDecodedSize {
-		d.bufs.Put(bufs)
-		br.close()
-		return nil, ErrMaxDecodedSizeExceeded
-	}
-	dst = append(dst, buf[:off]...)
-
-	// br < 8, so uint8 is fine
-	bitsLeft := uint8(br.off)*8 + 64 - br.bitsRead
-	for bitsLeft > 0 {
-		br.fill()
-		if false && br.bitsRead >= 32 {
-			if br.off >= 4 {
-				v := br.in[br.off-4:]
-				v = v[:4]
-				low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
-				br.value = (br.value << 32) | uint64(low)
-				br.bitsRead -= 32
-				br.off -= 4
-			} else {
-				for br.off > 0 {
-					br.value = (br.value << 8) | uint64(br.in[br.off-1])
-					br.bitsRead -= 8
-					br.off--
-				}
-			}
-		}
-		if len(dst) >= maxDecodedSize {
-			d.bufs.Put(bufs)
-			br.close()
-			return nil, ErrMaxDecodedSizeExceeded
-		}
-		v := d.dt.single[br.peekBitsFast(d.actualTableLog)&tlMask]
-		nBits := uint8(v.entry)
-		br.advance(nBits)
-		bitsLeft -= nBits
-		dst = append(dst, uint8(v.entry>>8))
-	}
-	d.bufs.Put(bufs)
-	return dst, br.close()
-}
-
 // decompress1X8Bit will decompress a 1X encoded stream with tablelog <= 8.
 // The cap of the output buffer will be the maximum decompressed size.
 // The length of the supplied input must match the end of a block exactly.
@@ -995,7 +885,6 @@ func (d *Decoder) decompress4X8bitExactly(dst, src []byte) ([]byte, error) {
 
 	const shift = 56
 	const tlSize = 1 << 8
-	const tlMask = tlSize - 1
 	single := d.dt.single[:tlSize]
 
 	// Use temp table to avoid bound checks/append penalty.
diff --git a/vendor/github.com/klauspost/compress/huff0/decompress_8b_amd64.s b/vendor/github.com/klauspost/compress/huff0/decompress_8b_amd64.s
deleted file mode 100644
index 0d6cb1a9..00000000
--- a/vendor/github.com/klauspost/compress/huff0/decompress_8b_amd64.s
+++ /dev/null
@@ -1,488 +0,0 @@
-// +build !appengine
-// +build gc
-// +build !noasm
-
-#include "textflag.h"
-#include "funcdata.h"
-#include "go_asm.h"
-
-#define bufoff      256 // see decompress.go, we're using [4][256]byte table
-
-// func decompress4x_main_loop_x86(pbr0, pbr1, pbr2, pbr3 *bitReaderShifted,
-//	peekBits uint8, buf *byte, tbl *dEntrySingle) (int, bool)
-TEXT ·decompress4x_8b_loop_x86(SB), NOSPLIT, $8
-#define off             R8
-#define buffer          DI
-#define table           SI
-
-#define br_bits_read    R9
-#define br_value        R10
-#define br_offset       R11
-#define peek_bits       R12
-#define exhausted       DX
-
-#define br0             R13
-#define br1             R14
-#define br2             R15
-#define br3             BP
-
-	MOVQ BP, 0(SP)
-
-	XORQ exhausted, exhausted // exhausted = false
-	XORQ off, off             // off = 0
-
-	MOVBQZX peekBits+32(FP), peek_bits
-	MOVQ    buf+40(FP), buffer
-	MOVQ    tbl+48(FP), table
-
-	MOVQ pbr0+0(FP), br0
-	MOVQ pbr1+8(FP), br1
-	MOVQ pbr2+16(FP), br2
-	MOVQ pbr3+24(FP), br3
-
-main_loop:
-
-	// const stream = 0
-	// br0.fillFast()
-	MOVBQZX bitReaderShifted_bitsRead(br0), br_bits_read
-	MOVQ    bitReaderShifted_value(br0), br_value
-	MOVQ    bitReaderShifted_off(br0), br_offset
-
-	// if b.bitsRead >= 32 {
-	CMPQ br_bits_read, $32
-	JB   skip_fill0
-
-	SUBQ $32, br_bits_read // b.bitsRead -= 32
-	SUBQ $4, br_offset     // b.off -= 4
-
-	// v := b.in[b.off-4 : b.off]
-	// v = v[:4]
-	// low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
-	MOVQ bitReaderShifted_in(br0), AX
-	MOVL 0(br_offset)(AX*1), AX       // AX = uint32(b.in[b.off:b.off+4])
-
-	// b.value |= uint64(low) << (b.bitsRead & 63)
-	MOVQ br_bits_read, CX
-	SHLQ CL, AX
-	ORQ  AX, br_value
-
-	// exhausted = exhausted || (br0.off < 4)
-	CMPQ  br_offset, $4
-	SETLT DL
-	ORB   DL, DH
-
-	// }
-skip_fill0:
-
-	// val0 := br0.peekTopBits(peekBits)
-	MOVQ br_value, AX
-	MOVQ peek_bits, CX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v0 := table[val0&mask]
-	MOVW 0(table)(AX*2), AX // AX - v0
-
-	// br0.advance(uint8(v0.entry))
-	MOVB    AH, BL           // BL = uint8(v0.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// val1 := br0.peekTopBits(peekBits)
-	MOVQ peek_bits, CX
-	MOVQ br_value, AX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v1 := table[val1&mask]
-	MOVW 0(table)(AX*2), AX // AX - v1
-
-	// br0.advance(uint8(v1.entry))
-	MOVB    AH, BH           // BH = uint8(v1.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CX, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// these two writes get coalesced
-	// buf[stream][off] = uint8(v0.entry >> 8)
-	// buf[stream][off+1] = uint8(v1.entry >> 8)
-	MOVW BX, 0(buffer)(off*1)
-
-	// SECOND PART:
-	// val2 := br0.peekTopBits(peekBits)
-	MOVQ br_value, AX
-	MOVQ peek_bits, CX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v2 := table[val0&mask]
-	MOVW 0(table)(AX*2), AX // AX - v0
-
-	// br0.advance(uint8(v0.entry))
-	MOVB    AH, BL           // BL = uint8(v0.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// val3 := br0.peekTopBits(peekBits)
-	MOVQ peek_bits, CX
-	MOVQ br_value, AX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v3 := table[val1&mask]
-	MOVW 0(table)(AX*2), AX // AX - v1
-
-	// br0.advance(uint8(v1.entry))
-	MOVB    AH, BH           // BH = uint8(v1.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CX, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// these two writes get coalesced
-	// buf[stream][off+2] = uint8(v2.entry >> 8)
-	// buf[stream][off+3] = uint8(v3.entry >> 8)
-	MOVW BX, 0+2(buffer)(off*1)
-
-	// update the bitrader reader structure
-	MOVB br_bits_read, bitReaderShifted_bitsRead(br0)
-	MOVQ br_value, bitReaderShifted_value(br0)
-	MOVQ br_offset, bitReaderShifted_off(br0)
-
-	// const stream = 1
-	// br1.fillFast()
-	MOVBQZX bitReaderShifted_bitsRead(br1), br_bits_read
-	MOVQ    bitReaderShifted_value(br1), br_value
-	MOVQ    bitReaderShifted_off(br1), br_offset
-
-	// if b.bitsRead >= 32 {
-	CMPQ br_bits_read, $32
-	JB   skip_fill1
-
-	SUBQ $32, br_bits_read // b.bitsRead -= 32
-	SUBQ $4, br_offset     // b.off -= 4
-
-	// v := b.in[b.off-4 : b.off]
-	// v = v[:4]
-	// low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
-	MOVQ bitReaderShifted_in(br1), AX
-	MOVL 0(br_offset)(AX*1), AX       // AX = uint32(b.in[b.off:b.off+4])
-
-	// b.value |= uint64(low) << (b.bitsRead & 63)
-	MOVQ br_bits_read, CX
-	SHLQ CL, AX
-	ORQ  AX, br_value
-
-	// exhausted = exhausted || (br1.off < 4)
-	CMPQ  br_offset, $4
-	SETLT DL
-	ORB   DL, DH
-
-	// }
-skip_fill1:
-
-	// val0 := br1.peekTopBits(peekBits)
-	MOVQ br_value, AX
-	MOVQ peek_bits, CX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v0 := table[val0&mask]
-	MOVW 0(table)(AX*2), AX // AX - v0
-
-	// br1.advance(uint8(v0.entry))
-	MOVB    AH, BL           // BL = uint8(v0.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// val1 := br1.peekTopBits(peekBits)
-	MOVQ peek_bits, CX
-	MOVQ br_value, AX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v1 := table[val1&mask]
-	MOVW 0(table)(AX*2), AX // AX - v1
-
-	// br1.advance(uint8(v1.entry))
-	MOVB    AH, BH           // BH = uint8(v1.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CX, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// these two writes get coalesced
-	// buf[stream][off] = uint8(v0.entry >> 8)
-	// buf[stream][off+1] = uint8(v1.entry >> 8)
-	MOVW BX, 256(buffer)(off*1)
-
-	// SECOND PART:
-	// val2 := br1.peekTopBits(peekBits)
-	MOVQ br_value, AX
-	MOVQ peek_bits, CX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v2 := table[val0&mask]
-	MOVW 0(table)(AX*2), AX // AX - v0
-
-	// br1.advance(uint8(v0.entry))
-	MOVB    AH, BL           // BL = uint8(v0.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// val3 := br1.peekTopBits(peekBits)
-	MOVQ peek_bits, CX
-	MOVQ br_value, AX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v3 := table[val1&mask]
-	MOVW 0(table)(AX*2), AX // AX - v1
-
-	// br1.advance(uint8(v1.entry))
-	MOVB    AH, BH           // BH = uint8(v1.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CX, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// these two writes get coalesced
-	// buf[stream][off+2] = uint8(v2.entry >> 8)
-	// buf[stream][off+3] = uint8(v3.entry >> 8)
-	MOVW BX, 256+2(buffer)(off*1)
-
-	// update the bitrader reader structure
-	MOVB br_bits_read, bitReaderShifted_bitsRead(br1)
-	MOVQ br_value, bitReaderShifted_value(br1)
-	MOVQ br_offset, bitReaderShifted_off(br1)
-
-	// const stream = 2
-	// br2.fillFast()
-	MOVBQZX bitReaderShifted_bitsRead(br2), br_bits_read
-	MOVQ    bitReaderShifted_value(br2), br_value
-	MOVQ    bitReaderShifted_off(br2), br_offset
-
-	// if b.bitsRead >= 32 {
-	CMPQ br_bits_read, $32
-	JB   skip_fill2
-
-	SUBQ $32, br_bits_read // b.bitsRead -= 32
-	SUBQ $4, br_offset     // b.off -= 4
-
-	// v := b.in[b.off-4 : b.off]
-	// v = v[:4]
-	// low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
-	MOVQ bitReaderShifted_in(br2), AX
-	MOVL 0(br_offset)(AX*1), AX       // AX = uint32(b.in[b.off:b.off+4])
-
-	// b.value |= uint64(low) << (b.bitsRead & 63)
-	MOVQ br_bits_read, CX
-	SHLQ CL, AX
-	ORQ  AX, br_value
-
-	// exhausted = exhausted || (br2.off < 4)
-	CMPQ  br_offset, $4
-	SETLT DL
-	ORB   DL, DH
-
-	// }
-skip_fill2:
-
-	// val0 := br2.peekTopBits(peekBits)
-	MOVQ br_value, AX
-	MOVQ peek_bits, CX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v0 := table[val0&mask]
-	MOVW 0(table)(AX*2), AX // AX - v0
-
-	// br2.advance(uint8(v0.entry))
-	MOVB    AH, BL           // BL = uint8(v0.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// val1 := br2.peekTopBits(peekBits)
-	MOVQ peek_bits, CX
-	MOVQ br_value, AX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v1 := table[val1&mask]
-	MOVW 0(table)(AX*2), AX // AX - v1
-
-	// br2.advance(uint8(v1.entry))
-	MOVB    AH, BH           // BH = uint8(v1.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CX, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// these two writes get coalesced
-	// buf[stream][off] = uint8(v0.entry >> 8)
-	// buf[stream][off+1] = uint8(v1.entry >> 8)
-	MOVW BX, 512(buffer)(off*1)
-
-	// SECOND PART:
-	// val2 := br2.peekTopBits(peekBits)
-	MOVQ br_value, AX
-	MOVQ peek_bits, CX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v2 := table[val0&mask]
-	MOVW 0(table)(AX*2), AX // AX - v0
-
-	// br2.advance(uint8(v0.entry))
-	MOVB    AH, BL           // BL = uint8(v0.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// val3 := br2.peekTopBits(peekBits)
-	MOVQ peek_bits, CX
-	MOVQ br_value, AX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v3 := table[val1&mask]
-	MOVW 0(table)(AX*2), AX // AX - v1
-
-	// br2.advance(uint8(v1.entry))
-	MOVB    AH, BH           // BH = uint8(v1.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CX, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// these two writes get coalesced
-	// buf[stream][off+2] = uint8(v2.entry >> 8)
-	// buf[stream][off+3] = uint8(v3.entry >> 8)
-	MOVW BX, 512+2(buffer)(off*1)
-
-	// update the bitrader reader structure
-	MOVB br_bits_read, bitReaderShifted_bitsRead(br2)
-	MOVQ br_value, bitReaderShifted_value(br2)
-	MOVQ br_offset, bitReaderShifted_off(br2)
-
-	// const stream = 3
-	// br3.fillFast()
-	MOVBQZX bitReaderShifted_bitsRead(br3), br_bits_read
-	MOVQ    bitReaderShifted_value(br3), br_value
-	MOVQ    bitReaderShifted_off(br3), br_offset
-
-	// if b.bitsRead >= 32 {
-	CMPQ br_bits_read, $32
-	JB   skip_fill3
-
-	SUBQ $32, br_bits_read // b.bitsRead -= 32
-	SUBQ $4, br_offset     // b.off -= 4
-
-	// v := b.in[b.off-4 : b.off]
-	// v = v[:4]
-	// low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
-	MOVQ bitReaderShifted_in(br3), AX
-	MOVL 0(br_offset)(AX*1), AX       // AX = uint32(b.in[b.off:b.off+4])
-
-	// b.value |= uint64(low) << (b.bitsRead & 63)
-	MOVQ br_bits_read, CX
-	SHLQ CL, AX
-	ORQ  AX, br_value
-
-	// exhausted = exhausted || (br3.off < 4)
-	CMPQ  br_offset, $4
-	SETLT DL
-	ORB   DL, DH
-
-	// }
-skip_fill3:
-
-	// val0 := br3.peekTopBits(peekBits)
-	MOVQ br_value, AX
-	MOVQ peek_bits, CX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v0 := table[val0&mask]
-	MOVW 0(table)(AX*2), AX // AX - v0
-
-	// br3.advance(uint8(v0.entry))
-	MOVB    AH, BL           // BL = uint8(v0.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// val1 := br3.peekTopBits(peekBits)
-	MOVQ peek_bits, CX
-	MOVQ br_value, AX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v1 := table[val1&mask]
-	MOVW 0(table)(AX*2), AX // AX - v1
-
-	// br3.advance(uint8(v1.entry))
-	MOVB    AH, BH           // BH = uint8(v1.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CX, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// these two writes get coalesced
-	// buf[stream][off] = uint8(v0.entry >> 8)
-	// buf[stream][off+1] = uint8(v1.entry >> 8)
-	MOVW BX, 768(buffer)(off*1)
-
-	// SECOND PART:
-	// val2 := br3.peekTopBits(peekBits)
-	MOVQ br_value, AX
-	MOVQ peek_bits, CX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v2 := table[val0&mask]
-	MOVW 0(table)(AX*2), AX // AX - v0
-
-	// br3.advance(uint8(v0.entry))
-	MOVB    AH, BL           // BL = uint8(v0.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// val3 := br3.peekTopBits(peekBits)
-	MOVQ peek_bits, CX
-	MOVQ br_value, AX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-	// v3 := table[val1&mask]
-	MOVW 0(table)(AX*2), AX // AX - v1
-
-	// br3.advance(uint8(v1.entry))
-	MOVB    AH, BH           // BH = uint8(v1.entry >> 8)
-	MOVBQZX AL, CX
-	SHLQ    CX, br_value     // value <<= n
-	ADDQ    CX, br_bits_read // bits_read += n
-
-	// these two writes get coalesced
-	// buf[stream][off+2] = uint8(v2.entry >> 8)
-	// buf[stream][off+3] = uint8(v3.entry >> 8)
-	MOVW BX, 768+2(buffer)(off*1)
-
-	// update the bitrader reader structure
-	MOVB br_bits_read, bitReaderShifted_bitsRead(br3)
-	MOVQ br_value, bitReaderShifted_value(br3)
-	MOVQ br_offset, bitReaderShifted_off(br3)
-
-	ADDQ $4, off // off += 2
-
-	TESTB DH, DH // any br[i].ofs < 4?
-	JNZ   end
-
-	CMPQ off, $bufoff
-	JL   main_loop
-
-end:
-	MOVQ 0(SP), BP
-
-	MOVB off, ret+56(FP)
-	RET
-
-#undef off
-#undef buffer
-#undef table
-
-#undef br_bits_read
-#undef br_value
-#undef br_offset
-#undef peek_bits
-#undef exhausted
-
-#undef br0
-#undef br1
-#undef br2
-#undef br3
diff --git a/vendor/github.com/klauspost/compress/huff0/decompress_8b_amd64.s.in b/vendor/github.com/klauspost/compress/huff0/decompress_8b_amd64.s.in
deleted file mode 100644
index 6d477a2c..00000000
--- a/vendor/github.com/klauspost/compress/huff0/decompress_8b_amd64.s.in
+++ /dev/null
@@ -1,197 +0,0 @@
-// +build !appengine
-// +build gc
-// +build !noasm
-
-#include "textflag.h"
-#include "funcdata.h"
-#include "go_asm.h"
-
-
-#define bufoff      256     // see decompress.go, we're using [4][256]byte table
-
-//func decompress4x_main_loop_x86(pbr0, pbr1, pbr2, pbr3 *bitReaderShifted,
-//	peekBits uint8, buf *byte, tbl *dEntrySingle) (int, bool)
-TEXT ·decompress4x_8b_loop_x86(SB), NOSPLIT, $8
-#define off             R8
-#define buffer          DI
-#define table           SI
-
-#define br_bits_read    R9
-#define br_value        R10
-#define br_offset       R11
-#define peek_bits       R12
-#define exhausted       DX
-
-#define br0             R13
-#define br1             R14
-#define br2             R15
-#define br3             BP
-
-    MOVQ    BP, 0(SP)
-
-    XORQ    exhausted, exhausted    // exhausted = false
-    XORQ    off, off                // off = 0
-
-    MOVBQZX peekBits+32(FP), peek_bits
-    MOVQ    buf+40(FP), buffer
-    MOVQ    tbl+48(FP), table
-
-    MOVQ    pbr0+0(FP), br0
-    MOVQ    pbr1+8(FP), br1
-    MOVQ    pbr2+16(FP), br2
-    MOVQ    pbr3+24(FP), br3
-
-main_loop:
-{{ define "decode_2_values_x86" }}
-    // const stream = {{ var "id" }}
-    // br{{ var "id"}}.fillFast()
-    MOVBQZX bitReaderShifted_bitsRead(br{{ var "id" }}), br_bits_read
-    MOVQ    bitReaderShifted_value(br{{ var "id" }}), br_value
-    MOVQ    bitReaderShifted_off(br{{ var "id" }}), br_offset
-
-	// if b.bitsRead >= 32 {
-    CMPQ    br_bits_read, $32
-    JB      skip_fill{{ var "id" }}
-
-    SUBQ    $32, br_bits_read       // b.bitsRead -= 32
-    SUBQ    $4, br_offset           // b.off -= 4
-
-	// v := b.in[b.off-4 : b.off]
-	// v = v[:4]
-	// low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
-    MOVQ    bitReaderShifted_in(br{{ var "id" }}), AX
-    MOVL    0(br_offset)(AX*1), AX  // AX = uint32(b.in[b.off:b.off+4])
-
-	// b.value |= uint64(low) << (b.bitsRead & 63)
-    MOVQ    br_bits_read, CX
-    SHLQ    CL, AX
-    ORQ     AX, br_value
-
-    // exhausted = exhausted || (br{{ var "id"}}.off < 4)
-    CMPQ    br_offset, $4
-    SETLT   DL
-    ORB     DL, DH
-    // }
-skip_fill{{ var "id" }}:
-
-    // val0 := br{{ var "id"}}.peekTopBits(peekBits)
-    MOVQ    br_value, AX
-    MOVQ    peek_bits, CX
-    SHRQ    CL, AX                  // AX = (value >> peek_bits) & mask
-
-    // v0 := table[val0&mask]
-    MOVW    0(table)(AX*2), AX      // AX - v0
-
-    // br{{ var "id"}}.advance(uint8(v0.entry))
-    MOVB    AH, BL                  // BL = uint8(v0.entry >> 8)
-    MOVBQZX AL, CX
-    SHLQ    CL, br_value            // value <<= n
-    ADDQ    CX, br_bits_read        // bits_read += n
-
-    // val1 := br{{ var "id"}}.peekTopBits(peekBits)
-    MOVQ    peek_bits, CX
-    MOVQ    br_value, AX
-    SHRQ    CL, AX                  // AX = (value >> peek_bits) & mask
-
-    // v1 := table[val1&mask]
-    MOVW    0(table)(AX*2), AX      // AX - v1
-
-    // br{{ var "id"}}.advance(uint8(v1.entry))
-    MOVB    AH, BH                  // BH = uint8(v1.entry >> 8)
-    MOVBQZX AL, CX
-    SHLQ    CX, br_value            // value <<= n
-    ADDQ    CX, br_bits_read        // bits_read += n
-
-
-    // these two writes get coalesced
-    // buf[stream][off] = uint8(v0.entry >> 8)
-    // buf[stream][off+1] = uint8(v1.entry >> 8)
-    MOVW    BX, {{ var "bufofs" }}(buffer)(off*1)
-
-    // SECOND PART:
-    // val2 := br{{ var "id"}}.peekTopBits(peekBits)
-    MOVQ    br_value, AX
-    MOVQ    peek_bits, CX
-    SHRQ    CL, AX                  // AX = (value >> peek_bits) & mask
-
-    // v2 := table[val0&mask]
-    MOVW    0(table)(AX*2), AX      // AX - v0
-
-    // br{{ var "id"}}.advance(uint8(v0.entry))
-    MOVB    AH, BL                  // BL = uint8(v0.entry >> 8)
-    MOVBQZX AL, CX
-    SHLQ    CL, br_value            // value <<= n
-    ADDQ    CX, br_bits_read        // bits_read += n
-
-    // val3 := br{{ var "id"}}.peekTopBits(peekBits)
-    MOVQ    peek_bits, CX
-    MOVQ    br_value, AX
-    SHRQ    CL, AX                  // AX = (value >> peek_bits) & mask
-
-    // v3 := table[val1&mask]
-    MOVW    0(table)(AX*2), AX      // AX - v1
-
-    // br{{ var "id"}}.advance(uint8(v1.entry))
-    MOVB    AH, BH                  // BH = uint8(v1.entry >> 8)
-    MOVBQZX AL, CX
-    SHLQ    CX, br_value            // value <<= n
-    ADDQ    CX, br_bits_read        // bits_read += n
-
-
-    // these two writes get coalesced
-    // buf[stream][off+2] = uint8(v2.entry >> 8)
-    // buf[stream][off+3] = uint8(v3.entry >> 8)
-    MOVW    BX, {{ var "bufofs" }}+2(buffer)(off*1)
-
-    // update the bitrader reader structure
-    MOVB    br_bits_read, bitReaderShifted_bitsRead(br{{ var "id" }})
-    MOVQ    br_value, bitReaderShifted_value(br{{ var "id" }})
-    MOVQ    br_offset, bitReaderShifted_off(br{{ var "id" }})
-{{ end }}
-
-    {{ set "id" "0" }}
-    {{ set "ofs" "0" }}
-    {{ set "bufofs" "0" }} {{/* id * bufoff */}}
-    {{ template "decode_2_values_x86" . }}
-
-    {{ set "id" "1" }}
-    {{ set "ofs" "8" }}
-    {{ set "bufofs" "256" }}
-    {{ template "decode_2_values_x86" . }}
-
-    {{ set "id" "2" }}
-    {{ set "ofs" "16" }}
-    {{ set "bufofs" "512" }}
-    {{ template "decode_2_values_x86" . }}
-
-    {{ set "id" "3" }}
-    {{ set "ofs" "24" }}
-    {{ set "bufofs" "768" }}
-    {{ template "decode_2_values_x86" . }}
-
-    ADDQ    $4, off     // off += 2
-
-    TESTB   DH, DH      // any br[i].ofs < 4?
-    JNZ     end
-
-    CMPQ    off, $bufoff
-    JL      main_loop
-end:
-    MOVQ    0(SP), BP
-
-    MOVB    off, ret+56(FP)
-    RET
-#undef  off
-#undef  buffer
-#undef  table
-
-#undef  br_bits_read
-#undef  br_value
-#undef  br_offset
-#undef  peek_bits
-#undef  exhausted
-
-#undef  br0
-#undef  br1
-#undef  br2
-#undef  br3
diff --git a/vendor/github.com/klauspost/compress/huff0/decompress_amd64.go b/vendor/github.com/klauspost/compress/huff0/decompress_amd64.go
index d47f6644..671e630a 100644
--- a/vendor/github.com/klauspost/compress/huff0/decompress_amd64.go
+++ b/vendor/github.com/klauspost/compress/huff0/decompress_amd64.go
@@ -2,30 +2,43 @@
 // +build amd64,!appengine,!noasm,gc
 
 // This file contains the specialisation of Decoder.Decompress4X
-// that uses an asm implementation of its main loop.
+// and Decoder.Decompress1X that use an asm implementation of thir main loops.
 package huff0
 
 import (
 	"errors"
 	"fmt"
+
+	"github.com/klauspost/compress/internal/cpuinfo"
 )
 
 // decompress4x_main_loop_x86 is an x86 assembler implementation
 // of Decompress4X when tablelog > 8.
-// go:noescape
-func decompress4x_main_loop_x86(pbr0, pbr1, pbr2, pbr3 *bitReaderShifted,
-	peekBits uint8, buf *byte, tbl *dEntrySingle) uint8
+//go:noescape
+func decompress4x_main_loop_amd64(ctx *decompress4xContext)
 
 // decompress4x_8b_loop_x86 is an x86 assembler implementation
 // of Decompress4X when tablelog <= 8 which decodes 4 entries
 // per loop.
-// go:noescape
-func decompress4x_8b_loop_x86(pbr0, pbr1, pbr2, pbr3 *bitReaderShifted,
-	peekBits uint8, buf *byte, tbl *dEntrySingle) uint8
+//go:noescape
+func decompress4x_8b_main_loop_amd64(ctx *decompress4xContext)
 
 // fallback8BitSize is the size where using Go version is faster.
 const fallback8BitSize = 800
 
+type decompress4xContext struct {
+	pbr0     *bitReaderShifted
+	pbr1     *bitReaderShifted
+	pbr2     *bitReaderShifted
+	pbr3     *bitReaderShifted
+	peekBits uint8
+	out      *byte
+	dstEvery int
+	tbl      *dEntrySingle
+	decoded  int
+	limit    *byte
+}
+
 // Decompress4X will decompress a 4X encoded stream.
 // The length of the supplied input must match the end of a block exactly.
 // The *capacity* of the dst slice must match the destination size of
@@ -42,6 +55,7 @@ func (d *Decoder) Decompress4X(dst, src []byte) ([]byte, error) {
 	if cap(dst) < fallback8BitSize && use8BitTables {
 		return d.decompress4X8bit(dst, src)
 	}
+
 	var br [4]bitReaderShifted
 	// Decode "jump table"
 	start := 6
@@ -71,70 +85,28 @@ func (d *Decoder) Decompress4X(dst, src []byte) ([]byte, error) {
 	const tlMask = tlSize - 1
 	single := d.dt.single[:tlSize]
 
-	// Use temp table to avoid bound checks/append penalty.
-	buf := d.buffer()
-	var off uint8
 	var decoded int
 
-	const debug = false
-
-	// see: bitReaderShifted.peekBitsFast()
-	peekBits := uint8((64 - d.actualTableLog) & 63)
-
-	// Decode 2 values from each decoder/loop.
-	const bufoff = 256
-	for {
-		if br[0].off < 4 || br[1].off < 4 || br[2].off < 4 || br[3].off < 4 {
-			break
+	if len(out) > 4*4 && !(br[0].off < 4 || br[1].off < 4 || br[2].off < 4 || br[3].off < 4) {
+		ctx := decompress4xContext{
+			pbr0:     &br[0],
+			pbr1:     &br[1],
+			pbr2:     &br[2],
+			pbr3:     &br[3],
+			peekBits: uint8((64 - d.actualTableLog) & 63), // see: bitReaderShifted.peekBitsFast()
+			out:      &out[0],
+			dstEvery: dstEvery,
+			tbl:      &single[0],
+			limit:    &out[dstEvery-4], // Always stop decoding when first buffer gets here to avoid writing OOB on last.
 		}
-
 		if use8BitTables {
-			off = decompress4x_8b_loop_x86(&br[0], &br[1], &br[2], &br[3], peekBits, &buf[0][0], &single[0])
+			decompress4x_8b_main_loop_amd64(&ctx)
 		} else {
-			off = decompress4x_main_loop_x86(&br[0], &br[1], &br[2], &br[3], peekBits, &buf[0][0], &single[0])
-		}
-		if debug {
-			fmt.Print("DEBUG: ")
-			fmt.Printf("off=%d,", off)
-			for i := 0; i < 4; i++ {
-				fmt.Printf(" br[%d]={bitsRead=%d, value=%x, off=%d}",
-					i, br[i].bitsRead, br[i].value, br[i].off)
-			}
-			fmt.Println("")
-		}
-
-		if off != 0 {
-			break
+			decompress4x_main_loop_amd64(&ctx)
 		}
 
-		if bufoff > dstEvery {
-			d.bufs.Put(buf)
-			return nil, errors.New("corruption detected: stream overrun 1")
-		}
-		copy(out, buf[0][:])
-		copy(out[dstEvery:], buf[1][:])
-		copy(out[dstEvery*2:], buf[2][:])
-		copy(out[dstEvery*3:], buf[3][:])
-		out = out[bufoff:]
-		decoded += bufoff * 4
-		// There must at least be 3 buffers left.
-		if len(out) < dstEvery*3 {
-			d.bufs.Put(buf)
-			return nil, errors.New("corruption detected: stream overrun 2")
-		}
-	}
-	if off > 0 {
-		ioff := int(off)
-		if len(out) < dstEvery*3+ioff {
-			d.bufs.Put(buf)
-			return nil, errors.New("corruption detected: stream overrun 3")
-		}
-		copy(out, buf[0][:off])
-		copy(out[dstEvery:], buf[1][:off])
-		copy(out[dstEvery*2:], buf[2][:off])
-		copy(out[dstEvery*3:], buf[3][:off])
-		decoded += int(off) * 4
-		out = out[off:]
+		decoded = ctx.decoded
+		out = out[decoded/4:]
 	}
 
 	// Decode remaining.
@@ -150,7 +122,6 @@ func (d *Decoder) Decompress4X(dst, src []byte) ([]byte, error) {
 		for bitsLeft > 0 {
 			br.fill()
 			if offset >= endsAt {
-				d.bufs.Put(buf)
 				return nil, errors.New("corruption detected: stream overrun 4")
 			}
 
@@ -164,7 +135,6 @@ func (d *Decoder) Decompress4X(dst, src []byte) ([]byte, error) {
 			offset++
 		}
 		if offset != endsAt {
-			d.bufs.Put(buf)
 			return nil, fmt.Errorf("corruption detected: short output block %d, end %d != %d", i, offset, endsAt)
 		}
 		decoded += offset - dstEvery*i
@@ -173,9 +143,86 @@ func (d *Decoder) Decompress4X(dst, src []byte) ([]byte, error) {
 			return nil, err
 		}
 	}
-	d.bufs.Put(buf)
 	if dstSize != decoded {
 		return nil, errors.New("corruption detected: short output block")
 	}
 	return dst, nil
 }
+
+// decompress4x_main_loop_x86 is an x86 assembler implementation
+// of Decompress1X when tablelog > 8.
+//go:noescape
+func decompress1x_main_loop_amd64(ctx *decompress1xContext)
+
+// decompress4x_main_loop_x86 is an x86 with BMI2 assembler implementation
+// of Decompress1X when tablelog > 8.
+//go:noescape
+func decompress1x_main_loop_bmi2(ctx *decompress1xContext)
+
+type decompress1xContext struct {
+	pbr      *bitReaderShifted
+	peekBits uint8
+	out      *byte
+	outCap   int
+	tbl      *dEntrySingle
+	decoded  int
+}
+
+// Error reported by asm implementations
+const error_max_decoded_size_exeeded = -1
+
+// Decompress1X will decompress a 1X encoded stream.
+// The cap of the output buffer will be the maximum decompressed size.
+// The length of the supplied input must match the end of a block exactly.
+func (d *Decoder) Decompress1X(dst, src []byte) ([]byte, error) {
+	if len(d.dt.single) == 0 {
+		return nil, errors.New("no table loaded")
+	}
+	var br bitReaderShifted
+	err := br.init(src)
+	if err != nil {
+		return dst, err
+	}
+	maxDecodedSize := cap(dst)
+	dst = dst[:maxDecodedSize]
+
+	const tlSize = 1 << tableLogMax
+	const tlMask = tlSize - 1
+
+	if maxDecodedSize >= 4 {
+		ctx := decompress1xContext{
+			pbr:      &br,
+			out:      &dst[0],
+			outCap:   maxDecodedSize,
+			peekBits: uint8((64 - d.actualTableLog) & 63), // see: bitReaderShifted.peekBitsFast()
+			tbl:      &d.dt.single[0],
+		}
+
+		if cpuinfo.HasBMI2() {
+			decompress1x_main_loop_bmi2(&ctx)
+		} else {
+			decompress1x_main_loop_amd64(&ctx)
+		}
+		if ctx.decoded == error_max_decoded_size_exeeded {
+			return nil, ErrMaxDecodedSizeExceeded
+		}
+
+		dst = dst[:ctx.decoded]
+	}
+
+	// br < 8, so uint8 is fine
+	bitsLeft := uint8(br.off)*8 + 64 - br.bitsRead
+	for bitsLeft > 0 {
+		br.fill()
+		if len(dst) >= maxDecodedSize {
+			br.close()
+			return nil, ErrMaxDecodedSizeExceeded
+		}
+		v := d.dt.single[br.peekBitsFast(d.actualTableLog)&tlMask]
+		nBits := uint8(v.entry)
+		br.advance(nBits)
+		bitsLeft -= nBits
+		dst = append(dst, uint8(v.entry>>8))
+	}
+	return dst, br.close()
+}
diff --git a/vendor/github.com/klauspost/compress/huff0/decompress_amd64.s b/vendor/github.com/klauspost/compress/huff0/decompress_amd64.s
index 2edad3ea..6c65c6e2 100644
--- a/vendor/github.com/klauspost/compress/huff0/decompress_amd64.s
+++ b/vendor/github.com/klauspost/compress/huff0/decompress_amd64.s
@@ -1,506 +1,865 @@
-// +build !appengine
-// +build gc
-// +build !noasm
-
-#include "textflag.h"
-#include "funcdata.h"
-#include "go_asm.h"
-
-#ifdef GOAMD64_v4
-#ifndef GOAMD64_v3
-#define GOAMD64_v3
-#endif
-#endif
-
-#define bufoff      256 // see decompress.go, we're using [4][256]byte table
-
-// func decompress4x_main_loop_x86(pbr0, pbr1, pbr2, pbr3 *bitReaderShifted,
-//	peekBits uint8, buf *byte, tbl *dEntrySingle) (int, bool)
-TEXT ·decompress4x_main_loop_x86(SB), NOSPLIT, $8
-#define off             R8
-#define buffer          DI
-#define table           SI
-
-#define br_bits_read    R9
-#define br_value        R10
-#define br_offset       R11
-#define peek_bits       R12
-#define exhausted       DX
-
-#define br0             R13
-#define br1             R14
-#define br2             R15
-#define br3             BP
-
-	MOVQ BP, 0(SP)
-
-	XORQ exhausted, exhausted // exhausted = false
-	XORQ off, off             // off = 0
-
-	MOVBQZX peekBits+32(FP), peek_bits
-	MOVQ    buf+40(FP), buffer
-	MOVQ    tbl+48(FP), table
-
-	MOVQ pbr0+0(FP), br0
-	MOVQ pbr1+8(FP), br1
-	MOVQ pbr2+16(FP), br2
-	MOVQ pbr3+24(FP), br3
-
+// Code generated by command: go run gen.go -out ../decompress_amd64.s -pkg=huff0. DO NOT EDIT.
+
+//go:build amd64 && !appengine && !noasm && gc
+// +build amd64,!appengine,!noasm,gc
+
+// func decompress4x_main_loop_amd64(ctx *decompress4xContext)
+TEXT ·decompress4x_main_loop_amd64(SB), $8-8
+	XORQ DX, DX
+
+	// Preload values
+	MOVQ    ctx+0(FP), AX
+	MOVBQZX 32(AX), SI
+	MOVQ    40(AX), DI
+	MOVQ    DI, BX
+	MOVQ    72(AX), CX
+	MOVQ    CX, (SP)
+	MOVQ    48(AX), R8
+	MOVQ    56(AX), R9
+	MOVQ    (AX), R10
+	MOVQ    8(AX), R11
+	MOVQ    16(AX), R12
+	MOVQ    24(AX), R13
+
+	// Main loop
 main_loop:
-
-	// const stream = 0
-	// br0.fillFast()
-	MOVBQZX bitReaderShifted_bitsRead(br0), br_bits_read
-	MOVQ    bitReaderShifted_value(br0), br_value
-	MOVQ    bitReaderShifted_off(br0), br_offset
-
-	// We must have at least 2 * max tablelog left
-	CMPQ br_bits_read, $64-22
-	JBE  skip_fill0
-
-	SUBQ $32, br_bits_read // b.bitsRead -= 32
-	SUBQ $4, br_offset     // b.off -= 4
-
-	// v := b.in[b.off-4 : b.off]
-	// v = v[:4]
-	// low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
-	MOVQ bitReaderShifted_in(br0), AX
+	MOVQ  BX, DI
+	CMPQ  DI, (SP)
+	SETGE DL
+
+	// br0.fillFast32()
+	MOVQ    32(R10), R14
+	MOVBQZX 40(R10), R15
+	CMPQ    R15, $0x20
+	JBE     skip_fill0
+	MOVQ    24(R10), AX
+	SUBQ    $0x20, R15
+	SUBQ    $0x04, AX
+	MOVQ    (R10), BP
 
 	// b.value |= uint64(low) << (b.bitsRead & 63)
-#ifdef GOAMD64_v3
-	SHLXQ br_bits_read, 0(br_offset)(AX*1), AX // AX = uint32(b.in[b.off:b.off+4]) << (b.bitsRead & 63)
-
-#else
-	MOVL 0(br_offset)(AX*1), AX // AX = uint32(b.in[b.off:b.off+4])
-	MOVQ br_bits_read, CX
-	SHLQ CL, AX
-
-#endif
-
-	ORQ AX, br_value
+	MOVL (AX)(BP*1), BP
+	MOVQ R15, CX
+	SHLQ CL, BP
+	MOVQ AX, 24(R10)
+	ORQ  BP, R14
 
 	// exhausted = exhausted || (br0.off < 4)
-	CMPQ  br_offset, $4
-	SETLT DL
-	ORB   DL, DH
+	CMPQ  AX, $0x04
+	SETLT AL
+	ORB   AL, DL
 
-	// }
 skip_fill0:
-
 	// val0 := br0.peekTopBits(peekBits)
-#ifdef GOAMD64_v3
-	SHRXQ peek_bits, br_value, AX // AX = (value >> peek_bits) & mask
-
-#else
-	MOVQ br_value, AX
-	MOVQ peek_bits, CX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-#endif
+	MOVQ R14, BP
+	MOVQ SI, CX
+	SHRQ CL, BP
 
 	// v0 := table[val0&mask]
-	MOVW 0(table)(AX*2), AX // AX - v0
-
-	// br0.advance(uint8(v0.entry))
-	MOVB AH, BL // BL = uint8(v0.entry >> 8)
-
-#ifdef GOAMD64_v3
-	MOVBQZX AL, CX
-	SHLXQ   AX, br_value, br_value // value <<= n
-
-#else
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value // value <<= n
+	MOVW (R9)(BP*2), CX
 
-#endif
+	// br0.advance(uint8(v0.entry)
+	MOVB CH, AL
+	SHLQ CL, R14
+	ADDB CL, R15
 
-	ADDQ CX, br_bits_read // bits_read += n
-
-#ifdef GOAMD64_v3
-	SHRXQ peek_bits, br_value, AX // AX = (value >> peek_bits) & mask
-
-#else
 	// val1 := br0.peekTopBits(peekBits)
-	MOVQ peek_bits, CX
-	MOVQ br_value, AX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-#endif
+	MOVQ SI, CX
+	MOVQ R14, BP
+	SHRQ CL, BP
 
 	// v1 := table[val1&mask]
-	MOVW 0(table)(AX*2), AX // AX - v1
+	MOVW (R9)(BP*2), CX
 
 	// br0.advance(uint8(v1.entry))
-	MOVB AH, BH // BH = uint8(v1.entry >> 8)
-
-#ifdef GOAMD64_v3
-	MOVBQZX AL, CX
-	SHLXQ   AX, br_value, br_value // value <<= n
-
-#else
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value // value <<= n
-
-#endif
-
-	ADDQ CX, br_bits_read // bits_read += n
+	MOVB CH, AH
+	SHLQ CL, R14
+	ADDB CL, R15
 
 	// these two writes get coalesced
-	// buf[stream][off] = uint8(v0.entry >> 8)
-	// buf[stream][off+1] = uint8(v1.entry >> 8)
-	MOVW BX, 0(buffer)(off*1)
+	// out[id * dstEvery + 0] = uint8(v0.entry >> 8)
+	// out[id * dstEvery + 1] = uint8(v1.entry >> 8)
+	MOVW AX, (DI)
 
 	// update the bitrader reader structure
-	MOVB br_bits_read, bitReaderShifted_bitsRead(br0)
-	MOVQ br_value, bitReaderShifted_value(br0)
-	MOVQ br_offset, bitReaderShifted_off(br0)
-
-	// const stream = 1
-	// br1.fillFast()
-	MOVBQZX bitReaderShifted_bitsRead(br1), br_bits_read
-	MOVQ    bitReaderShifted_value(br1), br_value
-	MOVQ    bitReaderShifted_off(br1), br_offset
-
-	// We must have at least 2 * max tablelog left
-	CMPQ br_bits_read, $64-22
-	JBE  skip_fill1
-
-	SUBQ $32, br_bits_read // b.bitsRead -= 32
-	SUBQ $4, br_offset     // b.off -= 4
-
-	// v := b.in[b.off-4 : b.off]
-	// v = v[:4]
-	// low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
-	MOVQ bitReaderShifted_in(br1), AX
+	MOVQ R14, 32(R10)
+	MOVB R15, 40(R10)
+	ADDQ R8, DI
+
+	// br1.fillFast32()
+	MOVQ    32(R11), R14
+	MOVBQZX 40(R11), R15
+	CMPQ    R15, $0x20
+	JBE     skip_fill1
+	MOVQ    24(R11), AX
+	SUBQ    $0x20, R15
+	SUBQ    $0x04, AX
+	MOVQ    (R11), BP
 
 	// b.value |= uint64(low) << (b.bitsRead & 63)
-#ifdef GOAMD64_v3
-	SHLXQ br_bits_read, 0(br_offset)(AX*1), AX // AX = uint32(b.in[b.off:b.off+4]) << (b.bitsRead & 63)
-
-#else
-	MOVL 0(br_offset)(AX*1), AX // AX = uint32(b.in[b.off:b.off+4])
-	MOVQ br_bits_read, CX
-	SHLQ CL, AX
-
-#endif
-
-	ORQ AX, br_value
+	MOVL (AX)(BP*1), BP
+	MOVQ R15, CX
+	SHLQ CL, BP
+	MOVQ AX, 24(R11)
+	ORQ  BP, R14
 
 	// exhausted = exhausted || (br1.off < 4)
-	CMPQ  br_offset, $4
-	SETLT DL
-	ORB   DL, DH
+	CMPQ  AX, $0x04
+	SETLT AL
+	ORB   AL, DL
 
-	// }
 skip_fill1:
-
 	// val0 := br1.peekTopBits(peekBits)
-#ifdef GOAMD64_v3
-	SHRXQ peek_bits, br_value, AX // AX = (value >> peek_bits) & mask
-
-#else
-	MOVQ br_value, AX
-	MOVQ peek_bits, CX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-#endif
+	MOVQ R14, BP
+	MOVQ SI, CX
+	SHRQ CL, BP
 
 	// v0 := table[val0&mask]
-	MOVW 0(table)(AX*2), AX // AX - v0
-
-	// br1.advance(uint8(v0.entry))
-	MOVB AH, BL // BL = uint8(v0.entry >> 8)
-
-#ifdef GOAMD64_v3
-	MOVBQZX AL, CX
-	SHLXQ   AX, br_value, br_value // value <<= n
-
-#else
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value // value <<= n
+	MOVW (R9)(BP*2), CX
 
-#endif
+	// br1.advance(uint8(v0.entry)
+	MOVB CH, AL
+	SHLQ CL, R14
+	ADDB CL, R15
 
-	ADDQ CX, br_bits_read // bits_read += n
-
-#ifdef GOAMD64_v3
-	SHRXQ peek_bits, br_value, AX // AX = (value >> peek_bits) & mask
-
-#else
 	// val1 := br1.peekTopBits(peekBits)
-	MOVQ peek_bits, CX
-	MOVQ br_value, AX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-#endif
+	MOVQ SI, CX
+	MOVQ R14, BP
+	SHRQ CL, BP
 
 	// v1 := table[val1&mask]
-	MOVW 0(table)(AX*2), AX // AX - v1
+	MOVW (R9)(BP*2), CX
 
 	// br1.advance(uint8(v1.entry))
-	MOVB AH, BH // BH = uint8(v1.entry >> 8)
-
-#ifdef GOAMD64_v3
-	MOVBQZX AL, CX
-	SHLXQ   AX, br_value, br_value // value <<= n
-
-#else
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value // value <<= n
-
-#endif
-
-	ADDQ CX, br_bits_read // bits_read += n
+	MOVB CH, AH
+	SHLQ CL, R14
+	ADDB CL, R15
 
 	// these two writes get coalesced
-	// buf[stream][off] = uint8(v0.entry >> 8)
-	// buf[stream][off+1] = uint8(v1.entry >> 8)
-	MOVW BX, 256(buffer)(off*1)
+	// out[id * dstEvery + 0] = uint8(v0.entry >> 8)
+	// out[id * dstEvery + 1] = uint8(v1.entry >> 8)
+	MOVW AX, (DI)
 
 	// update the bitrader reader structure
-	MOVB br_bits_read, bitReaderShifted_bitsRead(br1)
-	MOVQ br_value, bitReaderShifted_value(br1)
-	MOVQ br_offset, bitReaderShifted_off(br1)
-
-	// const stream = 2
-	// br2.fillFast()
-	MOVBQZX bitReaderShifted_bitsRead(br2), br_bits_read
-	MOVQ    bitReaderShifted_value(br2), br_value
-	MOVQ    bitReaderShifted_off(br2), br_offset
-
-	// We must have at least 2 * max tablelog left
-	CMPQ br_bits_read, $64-22
-	JBE  skip_fill2
-
-	SUBQ $32, br_bits_read // b.bitsRead -= 32
-	SUBQ $4, br_offset     // b.off -= 4
-
-	// v := b.in[b.off-4 : b.off]
-	// v = v[:4]
-	// low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
-	MOVQ bitReaderShifted_in(br2), AX
+	MOVQ R14, 32(R11)
+	MOVB R15, 40(R11)
+	ADDQ R8, DI
+
+	// br2.fillFast32()
+	MOVQ    32(R12), R14
+	MOVBQZX 40(R12), R15
+	CMPQ    R15, $0x20
+	JBE     skip_fill2
+	MOVQ    24(R12), AX
+	SUBQ    $0x20, R15
+	SUBQ    $0x04, AX
+	MOVQ    (R12), BP
 
 	// b.value |= uint64(low) << (b.bitsRead & 63)
-#ifdef GOAMD64_v3
-	SHLXQ br_bits_read, 0(br_offset)(AX*1), AX // AX = uint32(b.in[b.off:b.off+4]) << (b.bitsRead & 63)
-
-#else
-	MOVL 0(br_offset)(AX*1), AX // AX = uint32(b.in[b.off:b.off+4])
-	MOVQ br_bits_read, CX
-	SHLQ CL, AX
-
-#endif
-
-	ORQ AX, br_value
+	MOVL (AX)(BP*1), BP
+	MOVQ R15, CX
+	SHLQ CL, BP
+	MOVQ AX, 24(R12)
+	ORQ  BP, R14
 
 	// exhausted = exhausted || (br2.off < 4)
-	CMPQ  br_offset, $4
-	SETLT DL
-	ORB   DL, DH
+	CMPQ  AX, $0x04
+	SETLT AL
+	ORB   AL, DL
 
-	// }
 skip_fill2:
-
 	// val0 := br2.peekTopBits(peekBits)
-#ifdef GOAMD64_v3
-	SHRXQ peek_bits, br_value, AX // AX = (value >> peek_bits) & mask
-
-#else
-	MOVQ br_value, AX
-	MOVQ peek_bits, CX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-#endif
+	MOVQ R14, BP
+	MOVQ SI, CX
+	SHRQ CL, BP
 
 	// v0 := table[val0&mask]
-	MOVW 0(table)(AX*2), AX // AX - v0
-
-	// br2.advance(uint8(v0.entry))
-	MOVB AH, BL // BL = uint8(v0.entry >> 8)
-
-#ifdef GOAMD64_v3
-	MOVBQZX AL, CX
-	SHLXQ   AX, br_value, br_value // value <<= n
-
-#else
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value // value <<= n
+	MOVW (R9)(BP*2), CX
 
-#endif
+	// br2.advance(uint8(v0.entry)
+	MOVB CH, AL
+	SHLQ CL, R14
+	ADDB CL, R15
 
-	ADDQ CX, br_bits_read // bits_read += n
-
-#ifdef GOAMD64_v3
-	SHRXQ peek_bits, br_value, AX // AX = (value >> peek_bits) & mask
-
-#else
 	// val1 := br2.peekTopBits(peekBits)
-	MOVQ peek_bits, CX
-	MOVQ br_value, AX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
-
-#endif
+	MOVQ SI, CX
+	MOVQ R14, BP
+	SHRQ CL, BP
 
 	// v1 := table[val1&mask]
-	MOVW 0(table)(AX*2), AX // AX - v1
+	MOVW (R9)(BP*2), CX
 
 	// br2.advance(uint8(v1.entry))
-	MOVB AH, BH // BH = uint8(v1.entry >> 8)
-
-#ifdef GOAMD64_v3
-	MOVBQZX AL, CX
-	SHLXQ   AX, br_value, br_value // value <<= n
-
-#else
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value // value <<= n
-
-#endif
-
-	ADDQ CX, br_bits_read // bits_read += n
+	MOVB CH, AH
+	SHLQ CL, R14
+	ADDB CL, R15
 
 	// these two writes get coalesced
-	// buf[stream][off] = uint8(v0.entry >> 8)
-	// buf[stream][off+1] = uint8(v1.entry >> 8)
-	MOVW BX, 512(buffer)(off*1)
+	// out[id * dstEvery + 0] = uint8(v0.entry >> 8)
+	// out[id * dstEvery + 1] = uint8(v1.entry >> 8)
+	MOVW AX, (DI)
 
 	// update the bitrader reader structure
-	MOVB br_bits_read, bitReaderShifted_bitsRead(br2)
-	MOVQ br_value, bitReaderShifted_value(br2)
-	MOVQ br_offset, bitReaderShifted_off(br2)
-
-	// const stream = 3
-	// br3.fillFast()
-	MOVBQZX bitReaderShifted_bitsRead(br3), br_bits_read
-	MOVQ    bitReaderShifted_value(br3), br_value
-	MOVQ    bitReaderShifted_off(br3), br_offset
-
-	// We must have at least 2 * max tablelog left
-	CMPQ br_bits_read, $64-22
-	JBE  skip_fill3
-
-	SUBQ $32, br_bits_read // b.bitsRead -= 32
-	SUBQ $4, br_offset     // b.off -= 4
-
-	// v := b.in[b.off-4 : b.off]
-	// v = v[:4]
-	// low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
-	MOVQ bitReaderShifted_in(br3), AX
+	MOVQ R14, 32(R12)
+	MOVB R15, 40(R12)
+	ADDQ R8, DI
+
+	// br3.fillFast32()
+	MOVQ    32(R13), R14
+	MOVBQZX 40(R13), R15
+	CMPQ    R15, $0x20
+	JBE     skip_fill3
+	MOVQ    24(R13), AX
+	SUBQ    $0x20, R15
+	SUBQ    $0x04, AX
+	MOVQ    (R13), BP
 
 	// b.value |= uint64(low) << (b.bitsRead & 63)
-#ifdef GOAMD64_v3
-	SHLXQ br_bits_read, 0(br_offset)(AX*1), AX // AX = uint32(b.in[b.off:b.off+4]) << (b.bitsRead & 63)
-
-#else
-	MOVL 0(br_offset)(AX*1), AX // AX = uint32(b.in[b.off:b.off+4])
-	MOVQ br_bits_read, CX
-	SHLQ CL, AX
-
-#endif
-
-	ORQ AX, br_value
+	MOVL (AX)(BP*1), BP
+	MOVQ R15, CX
+	SHLQ CL, BP
+	MOVQ AX, 24(R13)
+	ORQ  BP, R14
 
 	// exhausted = exhausted || (br3.off < 4)
-	CMPQ  br_offset, $4
-	SETLT DL
-	ORB   DL, DH
+	CMPQ  AX, $0x04
+	SETLT AL
+	ORB   AL, DL
 
-	// }
 skip_fill3:
-
 	// val0 := br3.peekTopBits(peekBits)
-#ifdef GOAMD64_v3
-	SHRXQ peek_bits, br_value, AX // AX = (value >> peek_bits) & mask
+	MOVQ R14, BP
+	MOVQ SI, CX
+	SHRQ CL, BP
 
-#else
-	MOVQ br_value, AX
-	MOVQ peek_bits, CX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
+	// v0 := table[val0&mask]
+	MOVW (R9)(BP*2), CX
 
-#endif
+	// br3.advance(uint8(v0.entry)
+	MOVB CH, AL
+	SHLQ CL, R14
+	ADDB CL, R15
 
-	// v0 := table[val0&mask]
-	MOVW 0(table)(AX*2), AX // AX - v0
+	// val1 := br3.peekTopBits(peekBits)
+	MOVQ SI, CX
+	MOVQ R14, BP
+	SHRQ CL, BP
 
-	// br3.advance(uint8(v0.entry))
-	MOVB AH, BL // BL = uint8(v0.entry >> 8)
+	// v1 := table[val1&mask]
+	MOVW (R9)(BP*2), CX
 
-#ifdef GOAMD64_v3
-	MOVBQZX AL, CX
-	SHLXQ   AX, br_value, br_value // value <<= n
+	// br3.advance(uint8(v1.entry))
+	MOVB CH, AH
+	SHLQ CL, R14
+	ADDB CL, R15
 
-#else
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value // value <<= n
+	// these two writes get coalesced
+	// out[id * dstEvery + 0] = uint8(v0.entry >> 8)
+	// out[id * dstEvery + 1] = uint8(v1.entry >> 8)
+	MOVW AX, (DI)
 
-#endif
+	// update the bitrader reader structure
+	MOVQ  R14, 32(R13)
+	MOVB  R15, 40(R13)
+	ADDQ  $0x02, BX
+	TESTB DL, DL
+	JZ    main_loop
+	MOVQ  ctx+0(FP), AX
+	MOVQ  40(AX), CX
+	MOVQ  BX, DX
+	SUBQ  CX, DX
+	SHLQ  $0x02, DX
+	MOVQ  DX, 64(AX)
+	RET
 
-	ADDQ CX, br_bits_read // bits_read += n
+// func decompress4x_8b_main_loop_amd64(ctx *decompress4xContext)
+TEXT ·decompress4x_8b_main_loop_amd64(SB), $16-8
+	XORQ DX, DX
+
+	// Preload values
+	MOVQ    ctx+0(FP), CX
+	MOVBQZX 32(CX), BX
+	MOVQ    40(CX), SI
+	MOVQ    SI, (SP)
+	MOVQ    72(CX), DX
+	MOVQ    DX, 8(SP)
+	MOVQ    48(CX), DI
+	MOVQ    56(CX), R8
+	MOVQ    (CX), R9
+	MOVQ    8(CX), R10
+	MOVQ    16(CX), R11
+	MOVQ    24(CX), R12
+
+	// Main loop
+main_loop:
+	MOVQ  (SP), SI
+	CMPQ  SI, 8(SP)
+	SETGE DL
+
+	// br1000.fillFast32()
+	MOVQ    32(R9), R13
+	MOVBQZX 40(R9), R14
+	CMPQ    R14, $0x20
+	JBE     skip_fill1000
+	MOVQ    24(R9), R15
+	SUBQ    $0x20, R14
+	SUBQ    $0x04, R15
+	MOVQ    (R9), BP
 
-#ifdef GOAMD64_v3
-	SHRXQ peek_bits, br_value, AX // AX = (value >> peek_bits) & mask
+	// b.value |= uint64(low) << (b.bitsRead & 63)
+	MOVL (R15)(BP*1), BP
+	MOVQ R14, CX
+	SHLQ CL, BP
+	MOVQ R15, 24(R9)
+	ORQ  BP, R13
+
+	// exhausted = exhausted || (br1000.off < 4)
+	CMPQ  R15, $0x04
+	SETLT AL
+	ORB   AL, DL
+
+skip_fill1000:
+	// val0 := br0.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
 
-#else
-	// val1 := br3.peekTopBits(peekBits)
-	MOVQ peek_bits, CX
-	MOVQ br_value, AX
-	SHRQ CL, AX        // AX = (value >> peek_bits) & mask
+	// v0 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
 
-#endif
+	// br0.advance(uint8(v0.entry)
+	MOVB CH, AL
+	SHLQ CL, R13
+	ADDB CL, R14
 
-	// v1 := table[val1&mask]
-	MOVW 0(table)(AX*2), AX // AX - v1
+	// val1 := br0.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
+
+	// v1 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
+
+	// br0.advance(uint8(v1.entry)
+	MOVB   CH, AH
+	SHLQ   CL, R13
+	ADDB   CL, R14
+	BSWAPL AX
+
+	// val2 := br0.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
+
+	// v2 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
+
+	// br0.advance(uint8(v2.entry)
+	MOVB CH, AH
+	SHLQ CL, R13
+	ADDB CL, R14
+
+	// val3 := br0.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
+
+	// v3 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
+
+	// br0.advance(uint8(v3.entry)
+	MOVB   CH, AL
+	SHLQ   CL, R13
+	ADDB   CL, R14
+	BSWAPL AX
+
+	// these four writes get coalesced
+	// out[id * dstEvery + 0] = uint8(v0.entry >> 8)
+	// out[id * dstEvery + 1] = uint8(v1.entry >> 8)
+	// out[id * dstEvery + 3] = uint8(v2.entry >> 8)
+	// out[id * dstEvery + 4] = uint8(v3.entry >> 8)
+	MOVL AX, (SI)
+
+	// update the bitreader reader structure
+	MOVQ R13, 32(R9)
+	MOVB R14, 40(R9)
+	ADDQ DI, SI
+
+	// br1001.fillFast32()
+	MOVQ    32(R10), R13
+	MOVBQZX 40(R10), R14
+	CMPQ    R14, $0x20
+	JBE     skip_fill1001
+	MOVQ    24(R10), R15
+	SUBQ    $0x20, R14
+	SUBQ    $0x04, R15
+	MOVQ    (R10), BP
 
-	// br3.advance(uint8(v1.entry))
-	MOVB AH, BH // BH = uint8(v1.entry >> 8)
+	// b.value |= uint64(low) << (b.bitsRead & 63)
+	MOVL (R15)(BP*1), BP
+	MOVQ R14, CX
+	SHLQ CL, BP
+	MOVQ R15, 24(R10)
+	ORQ  BP, R13
+
+	// exhausted = exhausted || (br1001.off < 4)
+	CMPQ  R15, $0x04
+	SETLT AL
+	ORB   AL, DL
+
+skip_fill1001:
+	// val0 := br1.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
 
-#ifdef GOAMD64_v3
-	MOVBQZX AL, CX
-	SHLXQ   AX, br_value, br_value // value <<= n
+	// v0 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
 
-#else
-	MOVBQZX AL, CX
-	SHLQ    CL, br_value // value <<= n
+	// br1.advance(uint8(v0.entry)
+	MOVB CH, AL
+	SHLQ CL, R13
+	ADDB CL, R14
 
-#endif
+	// val1 := br1.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
+
+	// v1 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
+
+	// br1.advance(uint8(v1.entry)
+	MOVB   CH, AH
+	SHLQ   CL, R13
+	ADDB   CL, R14
+	BSWAPL AX
+
+	// val2 := br1.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
+
+	// v2 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
+
+	// br1.advance(uint8(v2.entry)
+	MOVB CH, AH
+	SHLQ CL, R13
+	ADDB CL, R14
+
+	// val3 := br1.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
+
+	// v3 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
+
+	// br1.advance(uint8(v3.entry)
+	MOVB   CH, AL
+	SHLQ   CL, R13
+	ADDB   CL, R14
+	BSWAPL AX
+
+	// these four writes get coalesced
+	// out[id * dstEvery + 0] = uint8(v0.entry >> 8)
+	// out[id * dstEvery + 1] = uint8(v1.entry >> 8)
+	// out[id * dstEvery + 3] = uint8(v2.entry >> 8)
+	// out[id * dstEvery + 4] = uint8(v3.entry >> 8)
+	MOVL AX, (SI)
+
+	// update the bitreader reader structure
+	MOVQ R13, 32(R10)
+	MOVB R14, 40(R10)
+	ADDQ DI, SI
+
+	// br1002.fillFast32()
+	MOVQ    32(R11), R13
+	MOVBQZX 40(R11), R14
+	CMPQ    R14, $0x20
+	JBE     skip_fill1002
+	MOVQ    24(R11), R15
+	SUBQ    $0x20, R14
+	SUBQ    $0x04, R15
+	MOVQ    (R11), BP
 
-	ADDQ CX, br_bits_read // bits_read += n
+	// b.value |= uint64(low) << (b.bitsRead & 63)
+	MOVL (R15)(BP*1), BP
+	MOVQ R14, CX
+	SHLQ CL, BP
+	MOVQ R15, 24(R11)
+	ORQ  BP, R13
+
+	// exhausted = exhausted || (br1002.off < 4)
+	CMPQ  R15, $0x04
+	SETLT AL
+	ORB   AL, DL
+
+skip_fill1002:
+	// val0 := br2.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
 
-	// these two writes get coalesced
-	// buf[stream][off] = uint8(v0.entry >> 8)
-	// buf[stream][off+1] = uint8(v1.entry >> 8)
-	MOVW BX, 768(buffer)(off*1)
+	// v0 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
 
-	// update the bitrader reader structure
-	MOVB br_bits_read, bitReaderShifted_bitsRead(br3)
-	MOVQ br_value, bitReaderShifted_value(br3)
-	MOVQ br_offset, bitReaderShifted_off(br3)
+	// br2.advance(uint8(v0.entry)
+	MOVB CH, AL
+	SHLQ CL, R13
+	ADDB CL, R14
 
-	ADDQ $2, off // off += 2
+	// val1 := br2.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
+
+	// v1 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
+
+	// br2.advance(uint8(v1.entry)
+	MOVB   CH, AH
+	SHLQ   CL, R13
+	ADDB   CL, R14
+	BSWAPL AX
+
+	// val2 := br2.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
+
+	// v2 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
+
+	// br2.advance(uint8(v2.entry)
+	MOVB CH, AH
+	SHLQ CL, R13
+	ADDB CL, R14
+
+	// val3 := br2.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
+
+	// v3 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
+
+	// br2.advance(uint8(v3.entry)
+	MOVB   CH, AL
+	SHLQ   CL, R13
+	ADDB   CL, R14
+	BSWAPL AX
+
+	// these four writes get coalesced
+	// out[id * dstEvery + 0] = uint8(v0.entry >> 8)
+	// out[id * dstEvery + 1] = uint8(v1.entry >> 8)
+	// out[id * dstEvery + 3] = uint8(v2.entry >> 8)
+	// out[id * dstEvery + 4] = uint8(v3.entry >> 8)
+	MOVL AX, (SI)
+
+	// update the bitreader reader structure
+	MOVQ R13, 32(R11)
+	MOVB R14, 40(R11)
+	ADDQ DI, SI
+
+	// br1003.fillFast32()
+	MOVQ    32(R12), R13
+	MOVBQZX 40(R12), R14
+	CMPQ    R14, $0x20
+	JBE     skip_fill1003
+	MOVQ    24(R12), R15
+	SUBQ    $0x20, R14
+	SUBQ    $0x04, R15
+	MOVQ    (R12), BP
 
-	TESTB DH, DH // any br[i].ofs < 4?
-	JNZ   end
+	// b.value |= uint64(low) << (b.bitsRead & 63)
+	MOVL (R15)(BP*1), BP
+	MOVQ R14, CX
+	SHLQ CL, BP
+	MOVQ R15, 24(R12)
+	ORQ  BP, R13
+
+	// exhausted = exhausted || (br1003.off < 4)
+	CMPQ  R15, $0x04
+	SETLT AL
+	ORB   AL, DL
+
+skip_fill1003:
+	// val0 := br3.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
 
-	CMPQ off, $bufoff
-	JL   main_loop
+	// v0 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
 
-end:
-	MOVQ 0(SP), BP
+	// br3.advance(uint8(v0.entry)
+	MOVB CH, AL
+	SHLQ CL, R13
+	ADDB CL, R14
 
-	MOVB off, ret+56(FP)
+	// val1 := br3.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
+
+	// v1 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
+
+	// br3.advance(uint8(v1.entry)
+	MOVB   CH, AH
+	SHLQ   CL, R13
+	ADDB   CL, R14
+	BSWAPL AX
+
+	// val2 := br3.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
+
+	// v2 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
+
+	// br3.advance(uint8(v2.entry)
+	MOVB CH, AH
+	SHLQ CL, R13
+	ADDB CL, R14
+
+	// val3 := br3.peekTopBits(peekBits)
+	MOVQ R13, R15
+	MOVQ BX, CX
+	SHRQ CL, R15
+
+	// v3 := table[val0&mask]
+	MOVW (R8)(R15*2), CX
+
+	// br3.advance(uint8(v3.entry)
+	MOVB   CH, AL
+	SHLQ   CL, R13
+	ADDB   CL, R14
+	BSWAPL AX
+
+	// these four writes get coalesced
+	// out[id * dstEvery + 0] = uint8(v0.entry >> 8)
+	// out[id * dstEvery + 1] = uint8(v1.entry >> 8)
+	// out[id * dstEvery + 3] = uint8(v2.entry >> 8)
+	// out[id * dstEvery + 4] = uint8(v3.entry >> 8)
+	MOVL AX, (SI)
+
+	// update the bitreader reader structure
+	MOVQ  R13, 32(R12)
+	MOVB  R14, 40(R12)
+	ADDQ  $0x04, (SP)
+	TESTB DL, DL
+	JZ    main_loop
+	MOVQ  ctx+0(FP), AX
+	MOVQ  40(AX), CX
+	MOVQ  (SP), DX
+	SUBQ  CX, DX
+	SHLQ  $0x02, DX
+	MOVQ  DX, 64(AX)
 	RET
 
-#undef off
-#undef buffer
-#undef table
+// func decompress1x_main_loop_amd64(ctx *decompress1xContext)
+TEXT ·decompress1x_main_loop_amd64(SB), $0-8
+	MOVQ    ctx+0(FP), CX
+	MOVQ    16(CX), DX
+	MOVQ    24(CX), BX
+	CMPQ    BX, $0x04
+	JB      error_max_decoded_size_exeeded
+	LEAQ    (DX)(BX*1), BX
+	MOVQ    (CX), SI
+	MOVQ    (SI), R8
+	MOVQ    24(SI), R9
+	MOVQ    32(SI), R10
+	MOVBQZX 40(SI), R11
+	MOVQ    32(CX), SI
+	MOVBQZX 8(CX), DI
+	JMP     loop_condition
 
-#undef br_bits_read
-#undef br_value
-#undef br_offset
-#undef peek_bits
-#undef exhausted
+main_loop:
+	// Check if we have room for 4 bytes in the output buffer
+	LEAQ 4(DX), CX
+	CMPQ CX, BX
+	JGE  error_max_decoded_size_exeeded
+
+	// Decode 4 values
+	CMPQ R11, $0x20
+	JL   bitReader_fillFast_1_end
+	SUBQ $0x20, R11
+	SUBQ $0x04, R9
+	MOVL (R8)(R9*1), R12
+	MOVQ R11, CX
+	SHLQ CL, R12
+	ORQ  R12, R10
+
+bitReader_fillFast_1_end:
+	MOVQ    DI, CX
+	MOVQ    R10, R12
+	SHRQ    CL, R12
+	MOVW    (SI)(R12*2), CX
+	MOVB    CH, AL
+	MOVBQZX CL, CX
+	ADDQ    CX, R11
+	SHLQ    CL, R10
+	MOVQ    DI, CX
+	MOVQ    R10, R12
+	SHRQ    CL, R12
+	MOVW    (SI)(R12*2), CX
+	MOVB    CH, AH
+	MOVBQZX CL, CX
+	ADDQ    CX, R11
+	SHLQ    CL, R10
+	BSWAPL  AX
+	CMPQ    R11, $0x20
+	JL      bitReader_fillFast_2_end
+	SUBQ    $0x20, R11
+	SUBQ    $0x04, R9
+	MOVL    (R8)(R9*1), R12
+	MOVQ    R11, CX
+	SHLQ    CL, R12
+	ORQ     R12, R10
+
+bitReader_fillFast_2_end:
+	MOVQ    DI, CX
+	MOVQ    R10, R12
+	SHRQ    CL, R12
+	MOVW    (SI)(R12*2), CX
+	MOVB    CH, AH
+	MOVBQZX CL, CX
+	ADDQ    CX, R11
+	SHLQ    CL, R10
+	MOVQ    DI, CX
+	MOVQ    R10, R12
+	SHRQ    CL, R12
+	MOVW    (SI)(R12*2), CX
+	MOVB    CH, AL
+	MOVBQZX CL, CX
+	ADDQ    CX, R11
+	SHLQ    CL, R10
+	BSWAPL  AX
+
+	// Store the decoded values
+	MOVL AX, (DX)
+	ADDQ $0x04, DX
+
+loop_condition:
+	CMPQ R9, $0x08
+	JGE  main_loop
+
+	// Update ctx structure
+	MOVQ ctx+0(FP), AX
+	MOVQ DX, CX
+	MOVQ 16(AX), DX
+	SUBQ DX, CX
+	MOVQ CX, 40(AX)
+	MOVQ (AX), AX
+	MOVQ R9, 24(AX)
+	MOVQ R10, 32(AX)
+	MOVB R11, 40(AX)
+	RET
 
-#undef br0
-#undef br1
-#undef br2
-#undef br3
+	// Report error
+error_max_decoded_size_exeeded:
+	MOVQ ctx+0(FP), AX
+	MOVQ $-1, CX
+	MOVQ CX, 40(AX)
+	RET
+
+// func decompress1x_main_loop_bmi2(ctx *decompress1xContext)
+// Requires: BMI2
+TEXT ·decompress1x_main_loop_bmi2(SB), $0-8
+	MOVQ    ctx+0(FP), CX
+	MOVQ    16(CX), DX
+	MOVQ    24(CX), BX
+	CMPQ    BX, $0x04
+	JB      error_max_decoded_size_exeeded
+	LEAQ    (DX)(BX*1), BX
+	MOVQ    (CX), SI
+	MOVQ    (SI), R8
+	MOVQ    24(SI), R9
+	MOVQ    32(SI), R10
+	MOVBQZX 40(SI), R11
+	MOVQ    32(CX), SI
+	MOVBQZX 8(CX), DI
+	JMP     loop_condition
+
+main_loop:
+	// Check if we have room for 4 bytes in the output buffer
+	LEAQ 4(DX), CX
+	CMPQ CX, BX
+	JGE  error_max_decoded_size_exeeded
+
+	// Decode 4 values
+	CMPQ  R11, $0x20
+	JL    bitReader_fillFast_1_end
+	SUBQ  $0x20, R11
+	SUBQ  $0x04, R9
+	MOVL  (R8)(R9*1), CX
+	SHLXQ R11, CX, CX
+	ORQ   CX, R10
+
+bitReader_fillFast_1_end:
+	SHRXQ   DI, R10, CX
+	MOVW    (SI)(CX*2), CX
+	MOVB    CH, AL
+	MOVBQZX CL, CX
+	ADDQ    CX, R11
+	SHLXQ   CX, R10, R10
+	SHRXQ   DI, R10, CX
+	MOVW    (SI)(CX*2), CX
+	MOVB    CH, AH
+	MOVBQZX CL, CX
+	ADDQ    CX, R11
+	SHLXQ   CX, R10, R10
+	BSWAPL  AX
+	CMPQ    R11, $0x20
+	JL      bitReader_fillFast_2_end
+	SUBQ    $0x20, R11
+	SUBQ    $0x04, R9
+	MOVL    (R8)(R9*1), CX
+	SHLXQ   R11, CX, CX
+	ORQ     CX, R10
+
+bitReader_fillFast_2_end:
+	SHRXQ   DI, R10, CX
+	MOVW    (SI)(CX*2), CX
+	MOVB    CH, AH
+	MOVBQZX CL, CX
+	ADDQ    CX, R11
+	SHLXQ   CX, R10, R10
+	SHRXQ   DI, R10, CX
+	MOVW    (SI)(CX*2), CX
+	MOVB    CH, AL
+	MOVBQZX CL, CX
+	ADDQ    CX, R11
+	SHLXQ   CX, R10, R10
+	BSWAPL  AX
+
+	// Store the decoded values
+	MOVL AX, (DX)
+	ADDQ $0x04, DX
+
+loop_condition:
+	CMPQ R9, $0x08
+	JGE  main_loop
+
+	// Update ctx structure
+	MOVQ ctx+0(FP), AX
+	MOVQ DX, CX
+	MOVQ 16(AX), DX
+	SUBQ DX, CX
+	MOVQ CX, 40(AX)
+	MOVQ (AX), AX
+	MOVQ R9, 24(AX)
+	MOVQ R10, 32(AX)
+	MOVB R11, 40(AX)
+	RET
+
+	// Report error
+error_max_decoded_size_exeeded:
+	MOVQ ctx+0(FP), AX
+	MOVQ $-1, CX
+	MOVQ CX, 40(AX)
+	RET
diff --git a/vendor/github.com/klauspost/compress/huff0/decompress_amd64.s.in b/vendor/github.com/klauspost/compress/huff0/decompress_amd64.s.in
deleted file mode 100644
index 330d86ae..00000000
--- a/vendor/github.com/klauspost/compress/huff0/decompress_amd64.s.in
+++ /dev/null
@@ -1,195 +0,0 @@
-// +build !appengine
-// +build gc
-// +build !noasm
-
-#include "textflag.h"
-#include "funcdata.h"
-#include "go_asm.h"
-
-#ifdef GOAMD64_v4
-#ifndef GOAMD64_v3
-#define GOAMD64_v3
-#endif
-#endif
-
-#define bufoff      256     // see decompress.go, we're using [4][256]byte table
-
-//func decompress4x_main_loop_x86(pbr0, pbr1, pbr2, pbr3 *bitReaderShifted,
-//	peekBits uint8, buf *byte, tbl *dEntrySingle) (int, bool)
-TEXT ·decompress4x_main_loop_x86(SB), NOSPLIT, $8
-#define off             R8
-#define buffer          DI
-#define table           SI
-
-#define br_bits_read    R9
-#define br_value        R10
-#define br_offset       R11
-#define peek_bits       R12
-#define exhausted       DX
-
-#define br0             R13
-#define br1             R14
-#define br2             R15
-#define br3             BP
-
-    MOVQ    BP, 0(SP)
-
-    XORQ    exhausted, exhausted    // exhausted = false
-    XORQ    off, off                // off = 0
-
-    MOVBQZX peekBits+32(FP), peek_bits
-    MOVQ    buf+40(FP), buffer
-    MOVQ    tbl+48(FP), table
-
-    MOVQ    pbr0+0(FP), br0
-    MOVQ    pbr1+8(FP), br1
-    MOVQ    pbr2+16(FP), br2
-    MOVQ    pbr3+24(FP), br3
-
-main_loop:
-{{ define "decode_2_values_x86" }}
-    // const stream = {{ var "id" }}
-    // br{{ var "id"}}.fillFast()
-    MOVBQZX bitReaderShifted_bitsRead(br{{ var "id" }}), br_bits_read
-    MOVQ    bitReaderShifted_value(br{{ var "id" }}), br_value
-    MOVQ    bitReaderShifted_off(br{{ var "id" }}), br_offset
-
-    // We must have at least 2 * max tablelog left
-    CMPQ    br_bits_read, $64-22
-    JBE     skip_fill{{ var "id" }}
-
-    SUBQ    $32, br_bits_read       // b.bitsRead -= 32
-    SUBQ    $4, br_offset           // b.off -= 4
-
-	// v := b.in[b.off-4 : b.off]
-	// v = v[:4]
-	// low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
-    MOVQ    bitReaderShifted_in(br{{ var "id" }}), AX
-
-	// b.value |= uint64(low) << (b.bitsRead & 63)
-#ifdef GOAMD64_v3
-    SHLXQ   br_bits_read, 0(br_offset)(AX*1), AX // AX = uint32(b.in[b.off:b.off+4]) << (b.bitsRead & 63)
-#else
-    MOVL    0(br_offset)(AX*1), AX  // AX = uint32(b.in[b.off:b.off+4])
-    MOVQ    br_bits_read, CX
-    SHLQ    CL, AX
-#endif
-
-    ORQ     AX, br_value
-
-    // exhausted = exhausted || (br{{ var "id"}}.off < 4)
-    CMPQ    br_offset, $4
-    SETLT   DL
-    ORB     DL, DH
-    // }
-skip_fill{{ var "id" }}:
-
-    // val0 := br{{ var "id"}}.peekTopBits(peekBits)
-#ifdef GOAMD64_v3
-    SHRXQ   peek_bits, br_value, AX // AX = (value >> peek_bits) & mask
-#else
-    MOVQ    br_value, AX
-    MOVQ    peek_bits, CX
-    SHRQ    CL, AX                  // AX = (value >> peek_bits) & mask
-#endif
-
-    // v0 := table[val0&mask]
-    MOVW    0(table)(AX*2), AX      // AX - v0
-
-    // br{{ var "id"}}.advance(uint8(v0.entry))
-    MOVB    AH, BL                  // BL = uint8(v0.entry >> 8)
-
-#ifdef GOAMD64_v3
-    MOVBQZX AL, CX
-    SHLXQ   AX, br_value, br_value // value <<= n
-#else
-    MOVBQZX AL, CX
-    SHLQ    CL, br_value            // value <<= n
-#endif
-
-    ADDQ    CX, br_bits_read        // bits_read += n
-
-
-#ifdef GOAMD64_v3
-    SHRXQ    peek_bits, br_value, AX  // AX = (value >> peek_bits) & mask
-#else
-    // val1 := br{{ var "id"}}.peekTopBits(peekBits)
-    MOVQ    peek_bits, CX
-    MOVQ    br_value, AX
-    SHRQ    CL, AX                  // AX = (value >> peek_bits) & mask
-#endif
-
-    // v1 := table[val1&mask]
-    MOVW    0(table)(AX*2), AX      // AX - v1
-
-    // br{{ var "id"}}.advance(uint8(v1.entry))
-    MOVB    AH, BH                  // BH = uint8(v1.entry >> 8)
-
-#ifdef GOAMD64_v3
-    MOVBQZX AL, CX
-    SHLXQ   AX, br_value, br_value // value <<= n
-#else
-    MOVBQZX AL, CX
-    SHLQ    CL, br_value            // value <<= n
-#endif
-
-    ADDQ    CX, br_bits_read        // bits_read += n
-
-
-    // these two writes get coalesced
-    // buf[stream][off] = uint8(v0.entry >> 8)
-    // buf[stream][off+1] = uint8(v1.entry >> 8)
-    MOVW    BX, {{ var "bufofs" }}(buffer)(off*1)
-
-    // update the bitrader reader structure
-    MOVB    br_bits_read, bitReaderShifted_bitsRead(br{{ var "id" }})
-    MOVQ    br_value, bitReaderShifted_value(br{{ var "id" }})
-    MOVQ    br_offset, bitReaderShifted_off(br{{ var "id" }})
-{{ end }}
-
-    {{ set "id" "0" }}
-    {{ set "ofs" "0" }}
-    {{ set "bufofs" "0" }} {{/* id * bufoff */}}
-    {{ template "decode_2_values_x86" . }}
-
-    {{ set "id" "1" }}
-    {{ set "ofs" "8" }}
-    {{ set "bufofs" "256" }}
-    {{ template "decode_2_values_x86" . }}
-
-    {{ set "id" "2" }}
-    {{ set "ofs" "16" }}
-    {{ set "bufofs" "512" }}
-    {{ template "decode_2_values_x86" . }}
-
-    {{ set "id" "3" }}
-    {{ set "ofs" "24" }}
-    {{ set "bufofs" "768" }}
-    {{ template "decode_2_values_x86" . }}
-
-    ADDQ    $2, off     // off += 2
-
-    TESTB   DH, DH      // any br[i].ofs < 4?
-    JNZ     end
-
-    CMPQ    off, $bufoff
-    JL      main_loop
-end:
-    MOVQ    0(SP), BP
-
-    MOVB    off, ret+56(FP)
-    RET
-#undef  off
-#undef  buffer
-#undef  table
-
-#undef  br_bits_read
-#undef  br_value
-#undef  br_offset
-#undef  peek_bits
-#undef  exhausted
-
-#undef  br0
-#undef  br1
-#undef  br2
-#undef  br3
diff --git a/vendor/github.com/klauspost/compress/huff0/decompress_generic.go b/vendor/github.com/klauspost/compress/huff0/decompress_generic.go
index 126b4d68..4f6f37cb 100644
--- a/vendor/github.com/klauspost/compress/huff0/decompress_generic.go
+++ b/vendor/github.com/klauspost/compress/huff0/decompress_generic.go
@@ -191,3 +191,105 @@ func (d *Decoder) Decompress4X(dst, src []byte) ([]byte, error) {
 	}
 	return dst, nil
 }
+
+// Decompress1X will decompress a 1X encoded stream.
+// The cap of the output buffer will be the maximum decompressed size.
+// The length of the supplied input must match the end of a block exactly.
+func (d *Decoder) Decompress1X(dst, src []byte) ([]byte, error) {
+	if len(d.dt.single) == 0 {
+		return nil, errors.New("no table loaded")
+	}
+	if use8BitTables && d.actualTableLog <= 8 {
+		return d.decompress1X8Bit(dst, src)
+	}
+	var br bitReaderShifted
+	err := br.init(src)
+	if err != nil {
+		return dst, err
+	}
+	maxDecodedSize := cap(dst)
+	dst = dst[:0]
+
+	// Avoid bounds check by always having full sized table.
+	const tlSize = 1 << tableLogMax
+	const tlMask = tlSize - 1
+	dt := d.dt.single[:tlSize]
+
+	// Use temp table to avoid bound checks/append penalty.
+	bufs := d.buffer()
+	buf := &bufs[0]
+	var off uint8
+
+	for br.off >= 8 {
+		br.fillFast()
+		v := dt[br.peekBitsFast(d.actualTableLog)&tlMask]
+		br.advance(uint8(v.entry))
+		buf[off+0] = uint8(v.entry >> 8)
+
+		v = dt[br.peekBitsFast(d.actualTableLog)&tlMask]
+		br.advance(uint8(v.entry))
+		buf[off+1] = uint8(v.entry >> 8)
+
+		// Refill
+		br.fillFast()
+
+		v = dt[br.peekBitsFast(d.actualTableLog)&tlMask]
+		br.advance(uint8(v.entry))
+		buf[off+2] = uint8(v.entry >> 8)
+
+		v = dt[br.peekBitsFast(d.actualTableLog)&tlMask]
+		br.advance(uint8(v.entry))
+		buf[off+3] = uint8(v.entry >> 8)
+
+		off += 4
+		if off == 0 {
+			if len(dst)+256 > maxDecodedSize {
+				br.close()
+				d.bufs.Put(bufs)
+				return nil, ErrMaxDecodedSizeExceeded
+			}
+			dst = append(dst, buf[:]...)
+		}
+	}
+
+	if len(dst)+int(off) > maxDecodedSize {
+		d.bufs.Put(bufs)
+		br.close()
+		return nil, ErrMaxDecodedSizeExceeded
+	}
+	dst = append(dst, buf[:off]...)
+
+	// br < 8, so uint8 is fine
+	bitsLeft := uint8(br.off)*8 + 64 - br.bitsRead
+	for bitsLeft > 0 {
+		br.fill()
+		if false && br.bitsRead >= 32 {
+			if br.off >= 4 {
+				v := br.in[br.off-4:]
+				v = v[:4]
+				low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
+				br.value = (br.value << 32) | uint64(low)
+				br.bitsRead -= 32
+				br.off -= 4
+			} else {
+				for br.off > 0 {
+					br.value = (br.value << 8) | uint64(br.in[br.off-1])
+					br.bitsRead -= 8
+					br.off--
+				}
+			}
+		}
+		if len(dst) >= maxDecodedSize {
+			d.bufs.Put(bufs)
+			br.close()
+			return nil, ErrMaxDecodedSizeExceeded
+		}
+		v := d.dt.single[br.peekBitsFast(d.actualTableLog)&tlMask]
+		nBits := uint8(v.entry)
+		br.advance(nBits)
+		bitsLeft -= nBits
+		dst = append(dst, uint8(v.entry>>8))
+	}
+	d.bufs.Put(bufs)
+	return dst, br.close()
+}
diff --git a/vendor/github.com/klauspost/compress/internal/cpuinfo/cpuinfo.go b/vendor/github.com/klauspost/compress/internal/cpuinfo/cpuinfo.go
new file mode 100644
index 00000000..3954c512
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/internal/cpuinfo/cpuinfo.go
@@ -0,0 +1,34 @@
+// Package cpuinfo gives runtime info about the current CPU.
+//
+// This is a very limited module meant for use internally
+// in this project. For more versatile solution check
+// https://github.com/klauspost/cpuid.
+package cpuinfo
+
+// HasBMI1 checks whether an x86 CPU supports the BMI1 extension.
+func HasBMI1() bool {
+	return hasBMI1
+}
+
+// HasBMI2 checks whether an x86 CPU supports the BMI2 extension.
+func HasBMI2() bool {
+	return hasBMI2
+}
+
+// DisableBMI2 will disable BMI2, for testing purposes.
+// Call returned function to restore previous state.
+func DisableBMI2() func() {
+	old := hasBMI2
+	hasBMI2 = false
+	return func() {
+		hasBMI2 = old
+	}
+}
+
+// HasBMI checks whether an x86 CPU supports both BMI1 and BMI2 extensions.
+func HasBMI() bool {
+	return HasBMI1() && HasBMI2()
+}
+
+var hasBMI1 bool
+var hasBMI2 bool
diff --git a/vendor/github.com/klauspost/compress/internal/cpuinfo/cpuinfo_amd64.go b/vendor/github.com/klauspost/compress/internal/cpuinfo/cpuinfo_amd64.go
new file mode 100644
index 00000000..e802579c
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/internal/cpuinfo/cpuinfo_amd64.go
@@ -0,0 +1,11 @@
+//go:build amd64 && !appengine && !noasm && gc
+// +build amd64,!appengine,!noasm,gc
+
+package cpuinfo
+
+// go:noescape
+func x86extensions() (bmi1, bmi2 bool)
+
+func init() {
+	hasBMI1, hasBMI2 = x86extensions()
+}
diff --git a/vendor/github.com/klauspost/compress/internal/cpuinfo/cpuinfo_amd64.s b/vendor/github.com/klauspost/compress/internal/cpuinfo/cpuinfo_amd64.s
new file mode 100644
index 00000000..4465fbe9
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/internal/cpuinfo/cpuinfo_amd64.s
@@ -0,0 +1,36 @@
+// +build !appengine
+// +build gc
+// +build !noasm
+
+#include "textflag.h"
+#include "funcdata.h"
+#include "go_asm.h"
+
+TEXT ·x86extensions(SB), NOSPLIT, $0
+	// 1. determine max EAX value
+	XORQ AX, AX
+	CPUID
+
+	CMPQ AX, $7
+	JB   unsupported
+
+	// 2. EAX = 7, ECX = 0 --- see Table 3-8 "Information Returned by CPUID Instruction"
+	MOVQ $7, AX
+	MOVQ $0, CX
+	CPUID
+
+	BTQ   $3, BX // bit 3 = BMI1
+	SETCS AL
+
+	BTQ   $8, BX // bit 8 = BMI2
+	SETCS AH
+
+	MOVB AL, bmi1+0(FP)
+	MOVB AH, bmi2+1(FP)
+	RET
+
+unsupported:
+	XORQ AX, AX
+	MOVB AL, bmi1+0(FP)
+	MOVB AL, bmi2+1(FP)
+	RET
diff --git a/vendor/github.com/klauspost/compress/s2/README.md b/vendor/github.com/klauspost/compress/s2/README.md
index 11979345..73c0c462 100644
--- a/vendor/github.com/klauspost/compress/s2/README.md
+++ b/vendor/github.com/klauspost/compress/s2/README.md
@@ -19,6 +19,7 @@ This is important, so you don't have to worry about spending CPU cycles on alrea
 * Adjustable compression (3 levels) 
 * Concurrent stream compression
 * Faster decompression, even for Snappy compatible content
+* Concurrent Snappy/S2 stream decompression
 * Ability to quickly skip forward in compressed stream
 * Random seeking with indexes
 * Compatible with reading Snappy compressed content
@@ -415,6 +416,25 @@ Without assembly decompression is also very fast; single goroutine decompression
 
 Even though S2 typically compresses better than Snappy, decompression speed is always better. 
 
+### Concurrent Stream Decompression
+
+For full stream decompression S2 offers a [DecodeConcurrent](https://pkg.go.dev/github.com/klauspost/compress/s2#Reader.DecodeConcurrent) 
+that will decode a full stream using multiple goroutines.
+
+Example scaling, AMD Ryzen 3950X, 16 cores, decompression using `s2d -bench=3 <input>`, best of 3: 
+
+| Input                                     | `-cpu=1`   | `-cpu=2`   | `-cpu=4`   | `-cpu=8`   | `-cpu=16`   |
+|-------------------------------------------|------------|------------|------------|------------|-------------|
+| enwik10.snappy                            | 1098.6MB/s | 1819.8MB/s | 3625.6MB/s | 6910.6MB/s | 10818.2MB/s |
+| enwik10.s2                                | 1303.5MB/s | 2606.1MB/s | 4847.9MB/s | 8878.4MB/s | 9592.1MB/s  |
+| sofia-air-quality-dataset.tar.snappy      | 1302.0MB/s | 2165.0MB/s | 4244.5MB/s | 8241.0MB/s | 12920.5MB/s |
+| sofia-air-quality-dataset.tar.s2          | 1399.2MB/s | 2463.2MB/s | 5196.5MB/s | 9639.8MB/s | 11439.5MB/s |
+| sofia-air-quality-dataset.tar.s2 (no asm) | 837.5MB/s  | 1652.6MB/s | 3183.6MB/s | 5945.0MB/s | 9620.7MB/s  |
+
+Scaling can be expected to be pretty linear until memory bandwidth is saturated. 
+
+For now the DecodeConcurrent can only be used for full streams without seeking or combining with regular reads.
+
 ## Block compression
 
 
@@ -873,7 +893,7 @@ for each entry {
     }
     
     // Uncompressed uses previous offset and adds EstBlockSize
-    entry[entryNum].UncompressedOffset = entry[entryNum-1].UncompressedOffset + EstBlockSize
+    entry[entryNum].UncompressedOffset = entry[entryNum-1].UncompressedOffset + EstBlockSize + uOff
 }
 
 
@@ -901,6 +921,14 @@ for each entry {
 }
 ```
 
+To decode from any given uncompressed offset `(wantOffset)`:
+
+* Iterate entries until `entry[n].UncompressedOffset > wantOffset`.
+* Start decoding from `entry[n-1].CompressedOffset`.
+* Discard `entry[n-1].UncompressedOffset - wantOffset` bytes from the decoded stream.
+
+See [using indexes](https://github.com/klauspost/compress/tree/master/s2#using-indexes) for functions that perform the operations with a simpler interface.
+
 # Format Extensions
 
 * Frame [Stream identifier](https://github.com/google/snappy/blob/master/framing_format.txt#L68) changed from `sNaPpY` to `S2sTwO`.
diff --git a/vendor/github.com/klauspost/compress/s2/decode.go b/vendor/github.com/klauspost/compress/s2/decode.go
index 9e7fce88..2aba9e27 100644
--- a/vendor/github.com/klauspost/compress/s2/decode.go
+++ b/vendor/github.com/klauspost/compress/s2/decode.go
@@ -11,6 +11,8 @@ import (
 	"fmt"
 	"io"
 	"io/ioutil"
+	"runtime"
+	"sync"
 )
 
 var (
@@ -169,6 +171,14 @@ func ReaderSkippableCB(id uint8, fn func(r io.Reader) error) ReaderOption {
 	}
 }
 
+// ReaderIgnoreCRC will make the reader skip CRC calculation and checks.
+func ReaderIgnoreCRC() ReaderOption {
+	return func(r *Reader) error {
+		r.ignoreCRC = true
+		return nil
+	}
+}
+
 // Reader is an io.Reader that can read Snappy-compressed bytes.
 type Reader struct {
 	r           io.Reader
@@ -191,18 +201,19 @@ type Reader struct {
 	paramsOK       bool
 	snappyFrame    bool
 	ignoreStreamID bool
+	ignoreCRC      bool
 }
 
 // ensureBufferSize will ensure that the buffer can take at least n bytes.
 // If false is returned the buffer exceeds maximum allowed size.
 func (r *Reader) ensureBufferSize(n int) bool {
-	if len(r.buf) >= n {
-		return true
-	}
 	if n > r.maxBufSize {
 		r.err = ErrCorrupt
 		return false
 	}
+	if cap(r.buf) >= n {
+		return true
+	}
 	// Realloc buffer.
 	r.buf = make([]byte, n)
 	return true
@@ -220,6 +231,7 @@ func (r *Reader) Reset(reader io.Reader) {
 	r.err = nil
 	r.i = 0
 	r.j = 0
+	r.blockStart = 0
 	r.readHeader = r.ignoreStreamID
 }
 
@@ -344,7 +356,7 @@ func (r *Reader) Read(p []byte) (int, error) {
 				r.err = err
 				return 0, r.err
 			}
-			if crc(r.decoded[:n]) != checksum {
+			if !r.ignoreCRC && crc(r.decoded[:n]) != checksum {
 				r.err = ErrCRC
 				return 0, r.err
 			}
@@ -385,7 +397,7 @@ func (r *Reader) Read(p []byte) (int, error) {
 			if !r.readFull(r.decoded[:n], false) {
 				return 0, r.err
 			}
-			if crc(r.decoded[:n]) != checksum {
+			if !r.ignoreCRC && crc(r.decoded[:n]) != checksum {
 				r.err = ErrCRC
 				return 0, r.err
 			}
@@ -435,6 +447,259 @@ func (r *Reader) Read(p []byte) (int, error) {
 	}
 }
 
+// DecodeConcurrent will decode the full stream to w.
+// This function should not be combined with reading, seeking or other operations.
+// Up to 'concurrent' goroutines will be used.
+// If <= 0, runtime.NumCPU will be used.
+// On success the number of bytes decompressed nil and is returned.
+// This is mainly intended for bigger streams.
+func (r *Reader) DecodeConcurrent(w io.Writer, concurrent int) (written int64, err error) {
+	if r.i > 0 || r.j > 0 || r.blockStart > 0 {
+		return 0, errors.New("DecodeConcurrent called after ")
+	}
+	if concurrent <= 0 {
+		concurrent = runtime.NumCPU()
+	}
+
+	// Write to output
+	var errMu sync.Mutex
+	var aErr error
+	setErr := func(e error) (ok bool) {
+		errMu.Lock()
+		defer errMu.Unlock()
+		if e == nil {
+			return aErr == nil
+		}
+		if aErr == nil {
+			aErr = e
+		}
+		return false
+	}
+	hasErr := func() (ok bool) {
+		errMu.Lock()
+		v := aErr != nil
+		errMu.Unlock()
+		return v
+	}
+
+	var aWritten int64
+	toRead := make(chan []byte, concurrent)
+	writtenBlocks := make(chan []byte, concurrent)
+	queue := make(chan chan []byte, concurrent)
+	reUse := make(chan chan []byte, concurrent)
+	for i := 0; i < concurrent; i++ {
+		toRead <- make([]byte, 0, r.maxBufSize)
+		writtenBlocks <- make([]byte, 0, r.maxBufSize)
+		reUse <- make(chan []byte, 1)
+	}
+	// Writer
+	var wg sync.WaitGroup
+	wg.Add(1)
+	go func() {
+		defer wg.Done()
+		for toWrite := range queue {
+			entry := <-toWrite
+			reUse <- toWrite
+			if hasErr() {
+				writtenBlocks <- entry
+				continue
+			}
+			n, err := w.Write(entry)
+			want := len(entry)
+			writtenBlocks <- entry
+			if err != nil {
+				setErr(err)
+				continue
+			}
+			if n != want {
+				setErr(io.ErrShortWrite)
+				continue
+			}
+			aWritten += int64(n)
+		}
+	}()
+
+	// Reader
+	defer func() {
+		close(queue)
+		if r.err != nil {
+			err = r.err
+			setErr(r.err)
+		}
+		wg.Wait()
+		if err == nil {
+			err = aErr
+		}
+		written = aWritten
+	}()
+
+	for !hasErr() {
+		if !r.readFull(r.buf[:4], true) {
+			if r.err == io.EOF {
+				r.err = nil
+			}
+			return 0, r.err
+		}
+		chunkType := r.buf[0]
+		if !r.readHeader {
+			if chunkType != chunkTypeStreamIdentifier {
+				r.err = ErrCorrupt
+				return 0, r.err
+			}
+			r.readHeader = true
+		}
+		chunkLen := int(r.buf[1]) | int(r.buf[2])<<8 | int(r.buf[3])<<16
+
+		// The chunk types are specified at
+		// https://github.com/google/snappy/blob/master/framing_format.txt
+		switch chunkType {
+		case chunkTypeCompressedData:
+			r.blockStart += int64(r.j)
+			// Section 4.2. Compressed data (chunk type 0x00).
+			if chunkLen < checksumSize {
+				r.err = ErrCorrupt
+				return 0, r.err
+			}
+			if chunkLen > r.maxBufSize {
+				r.err = ErrCorrupt
+				return 0, r.err
+			}
+			orgBuf := <-toRead
+			buf := orgBuf[:chunkLen]
+
+			if !r.readFull(buf, false) {
+				return 0, r.err
+			}
+
+			checksum := uint32(buf[0]) | uint32(buf[1])<<8 | uint32(buf[2])<<16 | uint32(buf[3])<<24
+			buf = buf[checksumSize:]
+
+			n, err := DecodedLen(buf)
+			if err != nil {
+				r.err = err
+				return 0, r.err
+			}
+			if r.snappyFrame && n > maxSnappyBlockSize {
+				r.err = ErrCorrupt
+				return 0, r.err
+			}
+
+			if n > r.maxBlock {
+				r.err = ErrCorrupt
+				return 0, r.err
+			}
+			wg.Add(1)
+
+			decoded := <-writtenBlocks
+			entry := <-reUse
+			queue <- entry
+			go func() {
+				defer wg.Done()
+				decoded = decoded[:n]
+				_, err := Decode(decoded, buf)
+				toRead <- orgBuf
+				if err != nil {
+					writtenBlocks <- decoded
+					setErr(err)
+					return
+				}
+				if !r.ignoreCRC && crc(decoded) != checksum {
+					writtenBlocks <- decoded
+					setErr(ErrCRC)
+					return
+				}
+				entry <- decoded
+			}()
+			continue
+
+		case chunkTypeUncompressedData:
+
+			// Section 4.3. Uncompressed data (chunk type 0x01).
+			if chunkLen < checksumSize {
+				r.err = ErrCorrupt
+				return 0, r.err
+			}
+			if chunkLen > r.maxBufSize {
+				r.err = ErrCorrupt
+				return 0, r.err
+			}
+			// Grab write buffer
+			orgBuf := <-writtenBlocks
+			buf := orgBuf[:checksumSize]
+			if !r.readFull(buf, false) {
+				return 0, r.err
+			}
+			checksum := uint32(buf[0]) | uint32(buf[1])<<8 | uint32(buf[2])<<16 | uint32(buf[3])<<24
+			// Read content.
+			n := chunkLen - checksumSize
+
+			if r.snappyFrame && n > maxSnappyBlockSize {
+				r.err = ErrCorrupt
+				return 0, r.err
+			}
+			if n > r.maxBlock {
+				r.err = ErrCorrupt
+				return 0, r.err
+			}
+			// Read uncompressed
+			buf = orgBuf[:n]
+			if !r.readFull(buf, false) {
+				return 0, r.err
+			}
+
+			if !r.ignoreCRC && crc(buf) != checksum {
+				r.err = ErrCRC
+				return 0, r.err
+			}
+			entry := <-reUse
+			queue <- entry
+			entry <- buf
+			continue
+
+		case chunkTypeStreamIdentifier:
+			// Section 4.1. Stream identifier (chunk type 0xff).
+			if chunkLen != len(magicBody) {
+				r.err = ErrCorrupt
+				return 0, r.err
+			}
+			if !r.readFull(r.buf[:len(magicBody)], false) {
+				return 0, r.err
+			}
+			if string(r.buf[:len(magicBody)]) != magicBody {
+				if string(r.buf[:len(magicBody)]) != magicBodySnappy {
+					r.err = ErrCorrupt
+					return 0, r.err
+				} else {
+					r.snappyFrame = true
+				}
+			} else {
+				r.snappyFrame = false
+			}
+			continue
+		}
+
+		if chunkType <= 0x7f {
+			// Section 4.5. Reserved unskippable chunks (chunk types 0x02-0x7f).
+			// fmt.Printf("ERR chunktype: 0x%x\n", chunkType)
+			r.err = ErrUnsupported
+			return 0, r.err
+		}
+		// Section 4.4 Padding (chunk type 0xfe).
+		// Section 4.6. Reserved skippable chunks (chunk types 0x80-0xfd).
+		if chunkLen > maxChunkSize {
+			// fmt.Printf("ERR chunkLen: 0x%x\n", chunkLen)
+			r.err = ErrUnsupported
+			return 0, r.err
+		}
+
+		// fmt.Printf("skippable: ID: 0x%x, len: 0x%x\n", chunkType, chunkLen)
+		if !r.skippable(r.buf, chunkLen, false, chunkType) {
+			return 0, r.err
+		}
+	}
+	return 0, r.err
+}
+
 // Skip will skip n bytes forward in the decompressed output.
 // For larger skips this consumes less CPU and is faster than reading output and discarding it.
 // CRC is not checked on skipped blocks.
@@ -699,8 +964,16 @@ func (r *ReadSeeker) Seek(offset int64, whence int) (int64, error) {
 	case io.SeekCurrent:
 		offset += r.blockStart + int64(r.i)
 	case io.SeekEnd:
-		offset = -offset
+		if offset > 0 {
+			return 0, errors.New("seek after end of file")
+		}
+		offset = r.index.TotalUncompressed + offset
+	}
+
+	if offset < 0 {
+		return 0, errors.New("seek before start of file")
 	}
+
 	c, u, err := r.index.Find(offset)
 	if err != nil {
 		return r.blockStart + int64(r.i), err
@@ -712,10 +985,6 @@ func (r *ReadSeeker) Seek(offset int64, whence int) (int64, error) {
 		return 0, err
 	}
 
-	if offset < 0 {
-		offset = r.index.TotalUncompressed + offset
-	}
-
 	r.i = r.j // Remove rest of current block.
 	if u < offset {
 		// Forward inside block
diff --git a/vendor/github.com/klauspost/compress/s2/encode.go b/vendor/github.com/klauspost/compress/s2/encode.go
index 59f992ca..1aefabf3 100644
--- a/vendor/github.com/klauspost/compress/s2/encode.go
+++ b/vendor/github.com/klauspost/compress/s2/encode.go
@@ -1119,12 +1119,6 @@ func (w *Writer) closeIndex(idx bool) ([]byte, error) {
 			if w.appendIndex {
 				w.written += int64(len(index))
 			}
-			if true {
-				_, err := w.index.Load(index)
-				if err != nil {
-					panic(err)
-				}
-			}
 		}
 
 		if w.pad > 1 {
diff --git a/vendor/github.com/klauspost/compress/s2/encode_best.go b/vendor/github.com/klauspost/compress/s2/encode_best.go
index 44803477..4bc80bc6 100644
--- a/vendor/github.com/klauspost/compress/s2/encode_best.go
+++ b/vendor/github.com/klauspost/compress/s2/encode_best.go
@@ -370,7 +370,7 @@ func encodeBlockBestSnappy(dst, src []byte) (d int) {
 				}
 				offset := m.s - m.offset
 
-				return score - emitCopySize(offset, m.length)
+				return score - emitCopyNoRepeatSize(offset, m.length)
 			}
 
 			matchAt := func(offset, s int, first uint32) match {
@@ -567,6 +567,10 @@ func emitCopySize(offset, length int) int {
 
 	// Offset no more than 2 bytes.
 	if length > 64 {
+		if offset < 2048 {
+			// Emit 8 bytes, then rest as repeats...
+			return 2 + emitRepeatSize(offset, length-8)
+		}
 		// Emit remaining as repeats, at least 4 bytes remain.
 		return 3 + emitRepeatSize(offset, length-60)
 	}
@@ -577,6 +581,28 @@ func emitCopySize(offset, length int) int {
 	return 2
 }
 
+// emitCopyNoRepeatSize returns the size to encode the offset+length
+//
+// It assumes that:
+//	1 <= offset && offset <= math.MaxUint32
+//	4 <= length && length <= 1 << 24
+func emitCopyNoRepeatSize(offset, length int) int {
+	if offset >= 65536 {
+		return 5 + 5*(length/64)
+	}
+
+	// Offset no more than 2 bytes.
+	if length > 64 {
+		// Emit remaining as repeats, at least 4 bytes remain.
+		return 3 + 3*(length/60)
+	}
+	if length >= 12 || offset >= 2048 {
+		return 3
+	}
+	// Emit the remaining copy, encoded as 2 bytes.
+	return 2
+}
+
 // emitRepeatSize returns the number of bytes required to encode a repeat.
 // Length must be at least 4 and < 1<<24
 func emitRepeatSize(offset, length int) int {
diff --git a/vendor/github.com/klauspost/compress/s2/encode_go.go b/vendor/github.com/klauspost/compress/s2/encode_go.go
index 43d43534..94784b82 100644
--- a/vendor/github.com/klauspost/compress/s2/encode_go.go
+++ b/vendor/github.com/klauspost/compress/s2/encode_go.go
@@ -180,14 +180,23 @@ func emitCopy(dst []byte, offset, length int) int {
 
 	// Offset no more than 2 bytes.
 	if length > 64 {
-		// Emit a length 60 copy, encoded as 3 bytes.
-		// Emit remaining as repeat value (minimum 4 bytes).
-		dst[2] = uint8(offset >> 8)
-		dst[1] = uint8(offset)
-		dst[0] = 59<<2 | tagCopy2
-		length -= 60
+		off := 3
+		if offset < 2048 {
+			// emit 8 bytes as tagCopy1, rest as repeats.
+			dst[1] = uint8(offset)
+			dst[0] = uint8(offset>>8)<<5 | uint8(8-4)<<2 | tagCopy1
+			length -= 8
+			off = 2
+		} else {
+			// Emit a length 60 copy, encoded as 3 bytes.
+			// Emit remaining as repeat value (minimum 4 bytes).
+			dst[2] = uint8(offset >> 8)
+			dst[1] = uint8(offset)
+			dst[0] = 59<<2 | tagCopy2
+			length -= 60
+		}
 		// Emit remaining as repeats, at least 4 bytes remain.
-		return 3 + emitRepeat(dst[3:], offset, length)
+		return off + emitRepeat(dst[off:], offset, length)
 	}
 	if length >= 12 || offset >= 2048 {
 		// Emit the remaining copy, encoded as 3 bytes.
diff --git a/vendor/github.com/klauspost/compress/s2/encodeblock_amd64.go b/vendor/github.com/klauspost/compress/s2/encodeblock_amd64.go
index d9312e5b..88f27c09 100644
--- a/vendor/github.com/klauspost/compress/s2/encodeblock_amd64.go
+++ b/vendor/github.com/klauspost/compress/s2/encodeblock_amd64.go
@@ -5,6 +5,8 @@
 
 package s2
 
+func _dummy_()
+
 // encodeBlockAsm encodes a non-empty src to a guaranteed-large-enough dst.
 // Maximum input 4294967295 bytes.
 // It assumes that the varint-encoded length of the decompressed bytes has already been written.
diff --git a/vendor/github.com/klauspost/compress/s2/encodeblock_amd64.s b/vendor/github.com/klauspost/compress/s2/encodeblock_amd64.s
index 729dbf53..36915d94 100644
--- a/vendor/github.com/klauspost/compress/s2/encodeblock_amd64.s
+++ b/vendor/github.com/klauspost/compress/s2/encodeblock_amd64.s
@@ -5,6 +5,15 @@
 
 #include "textflag.h"
 
+// func _dummy_()
+TEXT ·_dummy_(SB), $0
+#ifdef GOAMD64_v4
+#ifndef GOAMD64_v3
+#define GOAMD64_v3
+#endif
+#endif
+	RET
+
 // func encodeBlockAsm(dst []byte, src []byte) int
 // Requires: BMI, SSE2
 TEXT ·encodeBlockAsm(SB), $65560-56
@@ -253,17 +262,6 @@ matchlen_loopback_repeat_extend_encodeBlockAsm:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -476,6 +474,90 @@ four_bytes_remain_repeat_as_copy_encodeBlockAsm:
 two_byte_offset_repeat_as_copy_encodeBlockAsm:
 	CMPL SI, $0x40
 	JLE  two_byte_offset_short_repeat_as_copy_encodeBlockAsm
+	CMPL DI, $0x00000800
+	JAE  long_offset_short_repeat_as_copy_encodeBlockAsm
+	MOVL $0x00000001, R8
+	LEAL 16(R8), R8
+	MOVB DI, 1(AX)
+	MOVL DI, R9
+	SHRL $0x08, R9
+	SHLL $0x05, R9
+	ORL  R9, R8
+	MOVB R8, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, SI
+
+	// emitRepeat
+	LEAL -4(SI), SI
+	JMP  cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b
+
+emit_repeat_again_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b:
+	MOVL SI, R8
+	LEAL -4(SI), SI
+	CMPL R8, $0x08
+	JLE  repeat_two_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b
+	CMPL R8, $0x0c
+	JGE  cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b
+	CMPL DI, $0x00000800
+	JLT  repeat_two_offset_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b
+
+cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b:
+	CMPL SI, $0x00000104
+	JLT  repeat_three_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b
+	CMPL SI, $0x00010100
+	JLT  repeat_four_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b
+	CMPL SI, $0x0100ffff
+	JLT  repeat_five_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b
+	LEAL -16842747(SI), SI
+	MOVW $0x001d, (AX)
+	MOVW $0xfffb, 2(AX)
+	MOVB $0xff, 4(AX)
+	ADDQ $0x05, AX
+	JMP  emit_repeat_again_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b
+
+repeat_five_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b:
+	LEAL -65536(SI), SI
+	MOVL SI, DI
+	MOVW $0x001d, (AX)
+	MOVW SI, 2(AX)
+	SARL $0x10, DI
+	MOVB DI, 4(AX)
+	ADDQ $0x05, AX
+	JMP  repeat_end_emit_encodeBlockAsm
+
+repeat_four_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b:
+	LEAL -256(SI), SI
+	MOVW $0x0019, (AX)
+	MOVW SI, 2(AX)
+	ADDQ $0x04, AX
+	JMP  repeat_end_emit_encodeBlockAsm
+
+repeat_three_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b:
+	LEAL -4(SI), SI
+	MOVW $0x0015, (AX)
+	MOVB SI, 2(AX)
+	ADDQ $0x03, AX
+	JMP  repeat_end_emit_encodeBlockAsm
+
+repeat_two_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b:
+	SHLL $0x02, SI
+	ORL  $0x01, SI
+	MOVW SI, (AX)
+	ADDQ $0x02, AX
+	JMP  repeat_end_emit_encodeBlockAsm
+
+repeat_two_offset_repeat_as_copy_encodeBlockAsm_emit_copy_short_2b:
+	XORQ R8, R8
+	LEAL 1(R8)(SI*4), SI
+	MOVB DI, 1(AX)
+	SARL $0x08, DI
+	SHLL $0x05, DI
+	ORL  DI, SI
+	MOVB SI, (AX)
+	ADDQ $0x02, AX
+	JMP  repeat_end_emit_encodeBlockAsm
+
+long_offset_short_repeat_as_copy_encodeBlockAsm:
 	MOVB $0xee, (AX)
 	MOVW DI, 1(AX)
 	LEAL -60(SI), SI
@@ -791,17 +873,6 @@ matchlen_loopback_match_nolit_encodeBlockAsm:
 #ifdef GOAMD64_v3
 	TZCNTQ R9, R9
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R9, R9
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R9, R9
 
@@ -944,6 +1015,90 @@ four_bytes_remain_match_nolit_encodeBlockAsm:
 two_byte_offset_match_nolit_encodeBlockAsm:
 	CMPL R10, $0x40
 	JLE  two_byte_offset_short_match_nolit_encodeBlockAsm
+	CMPL SI, $0x00000800
+	JAE  long_offset_short_match_nolit_encodeBlockAsm
+	MOVL $0x00000001, DI
+	LEAL 16(DI), DI
+	MOVB SI, 1(AX)
+	MOVL SI, R8
+	SHRL $0x08, R8
+	SHLL $0x05, R8
+	ORL  R8, DI
+	MOVB DI, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, R10
+
+	// emitRepeat
+	LEAL -4(R10), R10
+	JMP  cant_repeat_two_offset_match_nolit_encodeBlockAsm_emit_copy_short_2b
+
+emit_repeat_again_match_nolit_encodeBlockAsm_emit_copy_short_2b:
+	MOVL R10, DI
+	LEAL -4(R10), R10
+	CMPL DI, $0x08
+	JLE  repeat_two_match_nolit_encodeBlockAsm_emit_copy_short_2b
+	CMPL DI, $0x0c
+	JGE  cant_repeat_two_offset_match_nolit_encodeBlockAsm_emit_copy_short_2b
+	CMPL SI, $0x00000800
+	JLT  repeat_two_offset_match_nolit_encodeBlockAsm_emit_copy_short_2b
+
+cant_repeat_two_offset_match_nolit_encodeBlockAsm_emit_copy_short_2b:
+	CMPL R10, $0x00000104
+	JLT  repeat_three_match_nolit_encodeBlockAsm_emit_copy_short_2b
+	CMPL R10, $0x00010100
+	JLT  repeat_four_match_nolit_encodeBlockAsm_emit_copy_short_2b
+	CMPL R10, $0x0100ffff
+	JLT  repeat_five_match_nolit_encodeBlockAsm_emit_copy_short_2b
+	LEAL -16842747(R10), R10
+	MOVW $0x001d, (AX)
+	MOVW $0xfffb, 2(AX)
+	MOVB $0xff, 4(AX)
+	ADDQ $0x05, AX
+	JMP  emit_repeat_again_match_nolit_encodeBlockAsm_emit_copy_short_2b
+
+repeat_five_match_nolit_encodeBlockAsm_emit_copy_short_2b:
+	LEAL -65536(R10), R10
+	MOVL R10, SI
+	MOVW $0x001d, (AX)
+	MOVW R10, 2(AX)
+	SARL $0x10, SI
+	MOVB SI, 4(AX)
+	ADDQ $0x05, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm
+
+repeat_four_match_nolit_encodeBlockAsm_emit_copy_short_2b:
+	LEAL -256(R10), R10
+	MOVW $0x0019, (AX)
+	MOVW R10, 2(AX)
+	ADDQ $0x04, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm
+
+repeat_three_match_nolit_encodeBlockAsm_emit_copy_short_2b:
+	LEAL -4(R10), R10
+	MOVW $0x0015, (AX)
+	MOVB R10, 2(AX)
+	ADDQ $0x03, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm
+
+repeat_two_match_nolit_encodeBlockAsm_emit_copy_short_2b:
+	SHLL $0x02, R10
+	ORL  $0x01, R10
+	MOVW R10, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm
+
+repeat_two_offset_match_nolit_encodeBlockAsm_emit_copy_short_2b:
+	XORQ DI, DI
+	LEAL 1(DI)(R10*4), R10
+	MOVB SI, 1(AX)
+	SARL $0x08, SI
+	SHLL $0x05, SI
+	ORL  SI, R10
+	MOVB R10, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm
+
+long_offset_short_match_nolit_encodeBlockAsm:
 	MOVB $0xee, (AX)
 	MOVW SI, 1(AX)
 	LEAL -60(R10), R10
@@ -1134,17 +1289,36 @@ memmove_emit_remainder_encodeBlockAsm:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeBlockAsm_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeBlockAsm_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeBlockAsm_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeBlockAsm_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeBlockAsm_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeBlockAsm_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeBlockAsm_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeBlockAsm_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeBlockAsm_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm
+
+emit_lit_memmove_emit_remainder_encodeBlockAsm_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm
+
+emit_lit_memmove_emit_remainder_encodeBlockAsm_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm
 
 emit_lit_memmove_emit_remainder_encodeBlockAsm_memmove_move_8through16:
@@ -1466,17 +1640,6 @@ matchlen_loopback_repeat_extend_encodeBlockAsm4MB:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -1667,6 +1830,77 @@ four_bytes_remain_repeat_as_copy_encodeBlockAsm4MB:
 two_byte_offset_repeat_as_copy_encodeBlockAsm4MB:
 	CMPL SI, $0x40
 	JLE  two_byte_offset_short_repeat_as_copy_encodeBlockAsm4MB
+	CMPL DI, $0x00000800
+	JAE  long_offset_short_repeat_as_copy_encodeBlockAsm4MB
+	MOVL $0x00000001, R8
+	LEAL 16(R8), R8
+	MOVB DI, 1(AX)
+	SHRL $0x08, DI
+	SHLL $0x05, DI
+	ORL  DI, R8
+	MOVB R8, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, SI
+
+	// emitRepeat
+	LEAL -4(SI), SI
+	JMP  cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm4MB_emit_copy_short_2b
+	MOVL SI, R8
+	LEAL -4(SI), SI
+	CMPL R8, $0x08
+	JLE  repeat_two_repeat_as_copy_encodeBlockAsm4MB_emit_copy_short_2b
+	CMPL R8, $0x0c
+	JGE  cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm4MB_emit_copy_short_2b
+	CMPL DI, $0x00000800
+	JLT  repeat_two_offset_repeat_as_copy_encodeBlockAsm4MB_emit_copy_short_2b
+
+cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm4MB_emit_copy_short_2b:
+	CMPL SI, $0x00000104
+	JLT  repeat_three_repeat_as_copy_encodeBlockAsm4MB_emit_copy_short_2b
+	CMPL SI, $0x00010100
+	JLT  repeat_four_repeat_as_copy_encodeBlockAsm4MB_emit_copy_short_2b
+	LEAL -65536(SI), SI
+	MOVL SI, DI
+	MOVW $0x001d, (AX)
+	MOVW SI, 2(AX)
+	SARL $0x10, DI
+	MOVB DI, 4(AX)
+	ADDQ $0x05, AX
+	JMP  repeat_end_emit_encodeBlockAsm4MB
+
+repeat_four_repeat_as_copy_encodeBlockAsm4MB_emit_copy_short_2b:
+	LEAL -256(SI), SI
+	MOVW $0x0019, (AX)
+	MOVW SI, 2(AX)
+	ADDQ $0x04, AX
+	JMP  repeat_end_emit_encodeBlockAsm4MB
+
+repeat_three_repeat_as_copy_encodeBlockAsm4MB_emit_copy_short_2b:
+	LEAL -4(SI), SI
+	MOVW $0x0015, (AX)
+	MOVB SI, 2(AX)
+	ADDQ $0x03, AX
+	JMP  repeat_end_emit_encodeBlockAsm4MB
+
+repeat_two_repeat_as_copy_encodeBlockAsm4MB_emit_copy_short_2b:
+	SHLL $0x02, SI
+	ORL  $0x01, SI
+	MOVW SI, (AX)
+	ADDQ $0x02, AX
+	JMP  repeat_end_emit_encodeBlockAsm4MB
+
+repeat_two_offset_repeat_as_copy_encodeBlockAsm4MB_emit_copy_short_2b:
+	XORQ R8, R8
+	LEAL 1(R8)(SI*4), SI
+	MOVB DI, 1(AX)
+	SARL $0x08, DI
+	SHLL $0x05, DI
+	ORL  DI, SI
+	MOVB SI, (AX)
+	ADDQ $0x02, AX
+	JMP  repeat_end_emit_encodeBlockAsm4MB
+
+long_offset_short_repeat_as_copy_encodeBlockAsm4MB:
 	MOVB $0xee, (AX)
 	MOVW DI, 1(AX)
 	LEAL -60(SI), SI
@@ -1963,17 +2197,6 @@ matchlen_loopback_match_nolit_encodeBlockAsm4MB:
 #ifdef GOAMD64_v3
 	TZCNTQ R9, R9
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R9, R9
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R9, R9
 
@@ -2105,6 +2328,77 @@ four_bytes_remain_match_nolit_encodeBlockAsm4MB:
 two_byte_offset_match_nolit_encodeBlockAsm4MB:
 	CMPL R10, $0x40
 	JLE  two_byte_offset_short_match_nolit_encodeBlockAsm4MB
+	CMPL SI, $0x00000800
+	JAE  long_offset_short_match_nolit_encodeBlockAsm4MB
+	MOVL $0x00000001, DI
+	LEAL 16(DI), DI
+	MOVB SI, 1(AX)
+	SHRL $0x08, SI
+	SHLL $0x05, SI
+	ORL  SI, DI
+	MOVB DI, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, R10
+
+	// emitRepeat
+	LEAL -4(R10), R10
+	JMP  cant_repeat_two_offset_match_nolit_encodeBlockAsm4MB_emit_copy_short_2b
+	MOVL R10, DI
+	LEAL -4(R10), R10
+	CMPL DI, $0x08
+	JLE  repeat_two_match_nolit_encodeBlockAsm4MB_emit_copy_short_2b
+	CMPL DI, $0x0c
+	JGE  cant_repeat_two_offset_match_nolit_encodeBlockAsm4MB_emit_copy_short_2b
+	CMPL SI, $0x00000800
+	JLT  repeat_two_offset_match_nolit_encodeBlockAsm4MB_emit_copy_short_2b
+
+cant_repeat_two_offset_match_nolit_encodeBlockAsm4MB_emit_copy_short_2b:
+	CMPL R10, $0x00000104
+	JLT  repeat_three_match_nolit_encodeBlockAsm4MB_emit_copy_short_2b
+	CMPL R10, $0x00010100
+	JLT  repeat_four_match_nolit_encodeBlockAsm4MB_emit_copy_short_2b
+	LEAL -65536(R10), R10
+	MOVL R10, SI
+	MOVW $0x001d, (AX)
+	MOVW R10, 2(AX)
+	SARL $0x10, SI
+	MOVB SI, 4(AX)
+	ADDQ $0x05, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm4MB
+
+repeat_four_match_nolit_encodeBlockAsm4MB_emit_copy_short_2b:
+	LEAL -256(R10), R10
+	MOVW $0x0019, (AX)
+	MOVW R10, 2(AX)
+	ADDQ $0x04, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm4MB
+
+repeat_three_match_nolit_encodeBlockAsm4MB_emit_copy_short_2b:
+	LEAL -4(R10), R10
+	MOVW $0x0015, (AX)
+	MOVB R10, 2(AX)
+	ADDQ $0x03, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm4MB
+
+repeat_two_match_nolit_encodeBlockAsm4MB_emit_copy_short_2b:
+	SHLL $0x02, R10
+	ORL  $0x01, R10
+	MOVW R10, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm4MB
+
+repeat_two_offset_match_nolit_encodeBlockAsm4MB_emit_copy_short_2b:
+	XORQ DI, DI
+	LEAL 1(DI)(R10*4), R10
+	MOVB SI, 1(AX)
+	SARL $0x08, SI
+	SHLL $0x05, SI
+	ORL  SI, R10
+	MOVB R10, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm4MB
+
+long_offset_short_match_nolit_encodeBlockAsm4MB:
 	MOVB $0xee, (AX)
 	MOVW SI, 1(AX)
 	LEAL -60(R10), R10
@@ -2276,17 +2570,36 @@ memmove_emit_remainder_encodeBlockAsm4MB:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeBlockAsm4MB_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeBlockAsm4MB_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeBlockAsm4MB_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeBlockAsm4MB_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeBlockAsm4MB_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeBlockAsm4MB_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeBlockAsm4MB_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeBlockAsm4MB_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeBlockAsm4MB_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm4MB
+
+emit_lit_memmove_emit_remainder_encodeBlockAsm4MB_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm4MB
+
+emit_lit_memmove_emit_remainder_encodeBlockAsm4MB_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm4MB
 
 emit_lit_memmove_emit_remainder_encodeBlockAsm4MB_memmove_move_8through16:
@@ -2597,17 +2910,6 @@ matchlen_loopback_repeat_extend_encodeBlockAsm12B:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -2706,6 +3008,65 @@ repeat_as_copy_encodeBlockAsm12B:
 two_byte_offset_repeat_as_copy_encodeBlockAsm12B:
 	CMPL SI, $0x40
 	JLE  two_byte_offset_short_repeat_as_copy_encodeBlockAsm12B
+	CMPL DI, $0x00000800
+	JAE  long_offset_short_repeat_as_copy_encodeBlockAsm12B
+	MOVL $0x00000001, R8
+	LEAL 16(R8), R8
+	MOVB DI, 1(AX)
+	SHRL $0x08, DI
+	SHLL $0x05, DI
+	ORL  DI, R8
+	MOVB R8, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, SI
+
+	// emitRepeat
+	LEAL -4(SI), SI
+	JMP  cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm12B_emit_copy_short_2b
+	MOVL SI, R8
+	LEAL -4(SI), SI
+	CMPL R8, $0x08
+	JLE  repeat_two_repeat_as_copy_encodeBlockAsm12B_emit_copy_short_2b
+	CMPL R8, $0x0c
+	JGE  cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm12B_emit_copy_short_2b
+	CMPL DI, $0x00000800
+	JLT  repeat_two_offset_repeat_as_copy_encodeBlockAsm12B_emit_copy_short_2b
+
+cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm12B_emit_copy_short_2b:
+	CMPL SI, $0x00000104
+	JLT  repeat_three_repeat_as_copy_encodeBlockAsm12B_emit_copy_short_2b
+	LEAL -256(SI), SI
+	MOVW $0x0019, (AX)
+	MOVW SI, 2(AX)
+	ADDQ $0x04, AX
+	JMP  repeat_end_emit_encodeBlockAsm12B
+
+repeat_three_repeat_as_copy_encodeBlockAsm12B_emit_copy_short_2b:
+	LEAL -4(SI), SI
+	MOVW $0x0015, (AX)
+	MOVB SI, 2(AX)
+	ADDQ $0x03, AX
+	JMP  repeat_end_emit_encodeBlockAsm12B
+
+repeat_two_repeat_as_copy_encodeBlockAsm12B_emit_copy_short_2b:
+	SHLL $0x02, SI
+	ORL  $0x01, SI
+	MOVW SI, (AX)
+	ADDQ $0x02, AX
+	JMP  repeat_end_emit_encodeBlockAsm12B
+
+repeat_two_offset_repeat_as_copy_encodeBlockAsm12B_emit_copy_short_2b:
+	XORQ R8, R8
+	LEAL 1(R8)(SI*4), SI
+	MOVB DI, 1(AX)
+	SARL $0x08, DI
+	SHLL $0x05, DI
+	ORL  DI, SI
+	MOVB SI, (AX)
+	ADDQ $0x02, AX
+	JMP  repeat_end_emit_encodeBlockAsm12B
+
+long_offset_short_repeat_as_copy_encodeBlockAsm12B:
 	MOVB $0xee, (AX)
 	MOVW DI, 1(AX)
 	LEAL -60(SI), SI
@@ -2979,17 +3340,6 @@ matchlen_loopback_match_nolit_encodeBlockAsm12B:
 #ifdef GOAMD64_v3
 	TZCNTQ R9, R9
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R9, R9
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R9, R9
 
@@ -3041,6 +3391,65 @@ match_nolit_end_encodeBlockAsm12B:
 two_byte_offset_match_nolit_encodeBlockAsm12B:
 	CMPL R10, $0x40
 	JLE  two_byte_offset_short_match_nolit_encodeBlockAsm12B
+	CMPL SI, $0x00000800
+	JAE  long_offset_short_match_nolit_encodeBlockAsm12B
+	MOVL $0x00000001, DI
+	LEAL 16(DI), DI
+	MOVB SI, 1(AX)
+	SHRL $0x08, SI
+	SHLL $0x05, SI
+	ORL  SI, DI
+	MOVB DI, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, R10
+
+	// emitRepeat
+	LEAL -4(R10), R10
+	JMP  cant_repeat_two_offset_match_nolit_encodeBlockAsm12B_emit_copy_short_2b
+	MOVL R10, DI
+	LEAL -4(R10), R10
+	CMPL DI, $0x08
+	JLE  repeat_two_match_nolit_encodeBlockAsm12B_emit_copy_short_2b
+	CMPL DI, $0x0c
+	JGE  cant_repeat_two_offset_match_nolit_encodeBlockAsm12B_emit_copy_short_2b
+	CMPL SI, $0x00000800
+	JLT  repeat_two_offset_match_nolit_encodeBlockAsm12B_emit_copy_short_2b
+
+cant_repeat_two_offset_match_nolit_encodeBlockAsm12B_emit_copy_short_2b:
+	CMPL R10, $0x00000104
+	JLT  repeat_three_match_nolit_encodeBlockAsm12B_emit_copy_short_2b
+	LEAL -256(R10), R10
+	MOVW $0x0019, (AX)
+	MOVW R10, 2(AX)
+	ADDQ $0x04, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm12B
+
+repeat_three_match_nolit_encodeBlockAsm12B_emit_copy_short_2b:
+	LEAL -4(R10), R10
+	MOVW $0x0015, (AX)
+	MOVB R10, 2(AX)
+	ADDQ $0x03, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm12B
+
+repeat_two_match_nolit_encodeBlockAsm12B_emit_copy_short_2b:
+	SHLL $0x02, R10
+	ORL  $0x01, R10
+	MOVW R10, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm12B
+
+repeat_two_offset_match_nolit_encodeBlockAsm12B_emit_copy_short_2b:
+	XORQ DI, DI
+	LEAL 1(DI)(R10*4), R10
+	MOVB SI, 1(AX)
+	SARL $0x08, SI
+	SHLL $0x05, SI
+	ORL  SI, R10
+	MOVB R10, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm12B
+
+long_offset_short_match_nolit_encodeBlockAsm12B:
 	MOVB $0xee, (AX)
 	MOVW SI, 1(AX)
 	LEAL -60(R10), R10
@@ -3189,17 +3598,36 @@ memmove_emit_remainder_encodeBlockAsm12B:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeBlockAsm12B_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeBlockAsm12B_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeBlockAsm12B_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeBlockAsm12B_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeBlockAsm12B_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeBlockAsm12B_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeBlockAsm12B_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeBlockAsm12B_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeBlockAsm12B_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm12B
+
+emit_lit_memmove_emit_remainder_encodeBlockAsm12B_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm12B
+
+emit_lit_memmove_emit_remainder_encodeBlockAsm12B_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm12B
 
 emit_lit_memmove_emit_remainder_encodeBlockAsm12B_memmove_move_8through16:
@@ -3510,17 +3938,6 @@ matchlen_loopback_repeat_extend_encodeBlockAsm10B:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -3619,6 +4036,65 @@ repeat_as_copy_encodeBlockAsm10B:
 two_byte_offset_repeat_as_copy_encodeBlockAsm10B:
 	CMPL SI, $0x40
 	JLE  two_byte_offset_short_repeat_as_copy_encodeBlockAsm10B
+	CMPL DI, $0x00000800
+	JAE  long_offset_short_repeat_as_copy_encodeBlockAsm10B
+	MOVL $0x00000001, R8
+	LEAL 16(R8), R8
+	MOVB DI, 1(AX)
+	SHRL $0x08, DI
+	SHLL $0x05, DI
+	ORL  DI, R8
+	MOVB R8, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, SI
+
+	// emitRepeat
+	LEAL -4(SI), SI
+	JMP  cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm10B_emit_copy_short_2b
+	MOVL SI, R8
+	LEAL -4(SI), SI
+	CMPL R8, $0x08
+	JLE  repeat_two_repeat_as_copy_encodeBlockAsm10B_emit_copy_short_2b
+	CMPL R8, $0x0c
+	JGE  cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm10B_emit_copy_short_2b
+	CMPL DI, $0x00000800
+	JLT  repeat_two_offset_repeat_as_copy_encodeBlockAsm10B_emit_copy_short_2b
+
+cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm10B_emit_copy_short_2b:
+	CMPL SI, $0x00000104
+	JLT  repeat_three_repeat_as_copy_encodeBlockAsm10B_emit_copy_short_2b
+	LEAL -256(SI), SI
+	MOVW $0x0019, (AX)
+	MOVW SI, 2(AX)
+	ADDQ $0x04, AX
+	JMP  repeat_end_emit_encodeBlockAsm10B
+
+repeat_three_repeat_as_copy_encodeBlockAsm10B_emit_copy_short_2b:
+	LEAL -4(SI), SI
+	MOVW $0x0015, (AX)
+	MOVB SI, 2(AX)
+	ADDQ $0x03, AX
+	JMP  repeat_end_emit_encodeBlockAsm10B
+
+repeat_two_repeat_as_copy_encodeBlockAsm10B_emit_copy_short_2b:
+	SHLL $0x02, SI
+	ORL  $0x01, SI
+	MOVW SI, (AX)
+	ADDQ $0x02, AX
+	JMP  repeat_end_emit_encodeBlockAsm10B
+
+repeat_two_offset_repeat_as_copy_encodeBlockAsm10B_emit_copy_short_2b:
+	XORQ R8, R8
+	LEAL 1(R8)(SI*4), SI
+	MOVB DI, 1(AX)
+	SARL $0x08, DI
+	SHLL $0x05, DI
+	ORL  DI, SI
+	MOVB SI, (AX)
+	ADDQ $0x02, AX
+	JMP  repeat_end_emit_encodeBlockAsm10B
+
+long_offset_short_repeat_as_copy_encodeBlockAsm10B:
 	MOVB $0xee, (AX)
 	MOVW DI, 1(AX)
 	LEAL -60(SI), SI
@@ -3892,17 +4368,6 @@ matchlen_loopback_match_nolit_encodeBlockAsm10B:
 #ifdef GOAMD64_v3
 	TZCNTQ R9, R9
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R9, R9
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R9, R9
 
@@ -3954,6 +4419,65 @@ match_nolit_end_encodeBlockAsm10B:
 two_byte_offset_match_nolit_encodeBlockAsm10B:
 	CMPL R10, $0x40
 	JLE  two_byte_offset_short_match_nolit_encodeBlockAsm10B
+	CMPL SI, $0x00000800
+	JAE  long_offset_short_match_nolit_encodeBlockAsm10B
+	MOVL $0x00000001, DI
+	LEAL 16(DI), DI
+	MOVB SI, 1(AX)
+	SHRL $0x08, SI
+	SHLL $0x05, SI
+	ORL  SI, DI
+	MOVB DI, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, R10
+
+	// emitRepeat
+	LEAL -4(R10), R10
+	JMP  cant_repeat_two_offset_match_nolit_encodeBlockAsm10B_emit_copy_short_2b
+	MOVL R10, DI
+	LEAL -4(R10), R10
+	CMPL DI, $0x08
+	JLE  repeat_two_match_nolit_encodeBlockAsm10B_emit_copy_short_2b
+	CMPL DI, $0x0c
+	JGE  cant_repeat_two_offset_match_nolit_encodeBlockAsm10B_emit_copy_short_2b
+	CMPL SI, $0x00000800
+	JLT  repeat_two_offset_match_nolit_encodeBlockAsm10B_emit_copy_short_2b
+
+cant_repeat_two_offset_match_nolit_encodeBlockAsm10B_emit_copy_short_2b:
+	CMPL R10, $0x00000104
+	JLT  repeat_three_match_nolit_encodeBlockAsm10B_emit_copy_short_2b
+	LEAL -256(R10), R10
+	MOVW $0x0019, (AX)
+	MOVW R10, 2(AX)
+	ADDQ $0x04, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm10B
+
+repeat_three_match_nolit_encodeBlockAsm10B_emit_copy_short_2b:
+	LEAL -4(R10), R10
+	MOVW $0x0015, (AX)
+	MOVB R10, 2(AX)
+	ADDQ $0x03, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm10B
+
+repeat_two_match_nolit_encodeBlockAsm10B_emit_copy_short_2b:
+	SHLL $0x02, R10
+	ORL  $0x01, R10
+	MOVW R10, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm10B
+
+repeat_two_offset_match_nolit_encodeBlockAsm10B_emit_copy_short_2b:
+	XORQ DI, DI
+	LEAL 1(DI)(R10*4), R10
+	MOVB SI, 1(AX)
+	SARL $0x08, SI
+	SHLL $0x05, SI
+	ORL  SI, R10
+	MOVB R10, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm10B
+
+long_offset_short_match_nolit_encodeBlockAsm10B:
 	MOVB $0xee, (AX)
 	MOVW SI, 1(AX)
 	LEAL -60(R10), R10
@@ -4102,17 +4626,36 @@ memmove_emit_remainder_encodeBlockAsm10B:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeBlockAsm10B_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeBlockAsm10B_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeBlockAsm10B_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeBlockAsm10B_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeBlockAsm10B_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeBlockAsm10B_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeBlockAsm10B_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeBlockAsm10B_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeBlockAsm10B_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm10B
+
+emit_lit_memmove_emit_remainder_encodeBlockAsm10B_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm10B
+
+emit_lit_memmove_emit_remainder_encodeBlockAsm10B_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm10B
 
 emit_lit_memmove_emit_remainder_encodeBlockAsm10B_memmove_move_8through16:
@@ -4423,17 +4966,6 @@ matchlen_loopback_repeat_extend_encodeBlockAsm8B:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -4528,6 +5060,61 @@ repeat_as_copy_encodeBlockAsm8B:
 two_byte_offset_repeat_as_copy_encodeBlockAsm8B:
 	CMPL SI, $0x40
 	JLE  two_byte_offset_short_repeat_as_copy_encodeBlockAsm8B
+	CMPL DI, $0x00000800
+	JAE  long_offset_short_repeat_as_copy_encodeBlockAsm8B
+	MOVL $0x00000001, R8
+	LEAL 16(R8), R8
+	MOVB DI, 1(AX)
+	SHRL $0x08, DI
+	SHLL $0x05, DI
+	ORL  DI, R8
+	MOVB R8, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, SI
+
+	// emitRepeat
+	LEAL -4(SI), SI
+	JMP  cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm8B_emit_copy_short_2b
+	MOVL SI, DI
+	LEAL -4(SI), SI
+	CMPL DI, $0x08
+	JLE  repeat_two_repeat_as_copy_encodeBlockAsm8B_emit_copy_short_2b
+	CMPL DI, $0x0c
+	JGE  cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm8B_emit_copy_short_2b
+
+cant_repeat_two_offset_repeat_as_copy_encodeBlockAsm8B_emit_copy_short_2b:
+	CMPL SI, $0x00000104
+	JLT  repeat_three_repeat_as_copy_encodeBlockAsm8B_emit_copy_short_2b
+	LEAL -256(SI), SI
+	MOVW $0x0019, (AX)
+	MOVW SI, 2(AX)
+	ADDQ $0x04, AX
+	JMP  repeat_end_emit_encodeBlockAsm8B
+
+repeat_three_repeat_as_copy_encodeBlockAsm8B_emit_copy_short_2b:
+	LEAL -4(SI), SI
+	MOVW $0x0015, (AX)
+	MOVB SI, 2(AX)
+	ADDQ $0x03, AX
+	JMP  repeat_end_emit_encodeBlockAsm8B
+
+repeat_two_repeat_as_copy_encodeBlockAsm8B_emit_copy_short_2b:
+	SHLL $0x02, SI
+	ORL  $0x01, SI
+	MOVW SI, (AX)
+	ADDQ $0x02, AX
+	JMP  repeat_end_emit_encodeBlockAsm8B
+	XORQ R8, R8
+	LEAL 1(R8)(SI*4), SI
+	MOVB DI, 1(AX)
+	SARL $0x08, DI
+	SHLL $0x05, DI
+	ORL  DI, SI
+	MOVB SI, (AX)
+	ADDQ $0x02, AX
+	JMP  repeat_end_emit_encodeBlockAsm8B
+
+long_offset_short_repeat_as_copy_encodeBlockAsm8B:
 	MOVB $0xee, (AX)
 	MOVW DI, 1(AX)
 	LEAL -60(SI), SI
@@ -4795,17 +5382,6 @@ matchlen_loopback_match_nolit_encodeBlockAsm8B:
 #ifdef GOAMD64_v3
 	TZCNTQ R9, R9
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R9, R9
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R9, R9
 
@@ -4857,6 +5433,61 @@ match_nolit_end_encodeBlockAsm8B:
 two_byte_offset_match_nolit_encodeBlockAsm8B:
 	CMPL R10, $0x40
 	JLE  two_byte_offset_short_match_nolit_encodeBlockAsm8B
+	CMPL SI, $0x00000800
+	JAE  long_offset_short_match_nolit_encodeBlockAsm8B
+	MOVL $0x00000001, DI
+	LEAL 16(DI), DI
+	MOVB SI, 1(AX)
+	SHRL $0x08, SI
+	SHLL $0x05, SI
+	ORL  SI, DI
+	MOVB DI, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, R10
+
+	// emitRepeat
+	LEAL -4(R10), R10
+	JMP  cant_repeat_two_offset_match_nolit_encodeBlockAsm8B_emit_copy_short_2b
+	MOVL R10, SI
+	LEAL -4(R10), R10
+	CMPL SI, $0x08
+	JLE  repeat_two_match_nolit_encodeBlockAsm8B_emit_copy_short_2b
+	CMPL SI, $0x0c
+	JGE  cant_repeat_two_offset_match_nolit_encodeBlockAsm8B_emit_copy_short_2b
+
+cant_repeat_two_offset_match_nolit_encodeBlockAsm8B_emit_copy_short_2b:
+	CMPL R10, $0x00000104
+	JLT  repeat_three_match_nolit_encodeBlockAsm8B_emit_copy_short_2b
+	LEAL -256(R10), R10
+	MOVW $0x0019, (AX)
+	MOVW R10, 2(AX)
+	ADDQ $0x04, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm8B
+
+repeat_three_match_nolit_encodeBlockAsm8B_emit_copy_short_2b:
+	LEAL -4(R10), R10
+	MOVW $0x0015, (AX)
+	MOVB R10, 2(AX)
+	ADDQ $0x03, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm8B
+
+repeat_two_match_nolit_encodeBlockAsm8B_emit_copy_short_2b:
+	SHLL $0x02, R10
+	ORL  $0x01, R10
+	MOVW R10, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm8B
+	XORQ DI, DI
+	LEAL 1(DI)(R10*4), R10
+	MOVB SI, 1(AX)
+	SARL $0x08, SI
+	SHLL $0x05, SI
+	ORL  SI, R10
+	MOVB R10, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBlockAsm8B
+
+long_offset_short_match_nolit_encodeBlockAsm8B:
 	MOVB $0xee, (AX)
 	MOVW SI, 1(AX)
 	LEAL -60(R10), R10
@@ -4999,17 +5630,36 @@ memmove_emit_remainder_encodeBlockAsm8B:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeBlockAsm8B_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeBlockAsm8B_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeBlockAsm8B_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeBlockAsm8B_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeBlockAsm8B_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeBlockAsm8B_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeBlockAsm8B_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeBlockAsm8B_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeBlockAsm8B_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm8B
+
+emit_lit_memmove_emit_remainder_encodeBlockAsm8B_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm8B
+
+emit_lit_memmove_emit_remainder_encodeBlockAsm8B_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeBlockAsm8B
 
 emit_lit_memmove_emit_remainder_encodeBlockAsm8B_memmove_move_8through16:
@@ -5225,17 +5875,6 @@ matchlen_loopback_match_nolit_encodeBetterBlockAsm:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -5541,6 +6180,90 @@ four_bytes_remain_match_nolit_encodeBetterBlockAsm:
 two_byte_offset_match_nolit_encodeBetterBlockAsm:
 	CMPL R12, $0x40
 	JLE  two_byte_offset_short_match_nolit_encodeBetterBlockAsm
+	CMPL R8, $0x00000800
+	JAE  long_offset_short_match_nolit_encodeBetterBlockAsm
+	MOVL $0x00000001, SI
+	LEAL 16(SI), SI
+	MOVB R8, 1(AX)
+	MOVL R8, R9
+	SHRL $0x08, R9
+	SHLL $0x05, R9
+	ORL  R9, SI
+	MOVB SI, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, R12
+
+	// emitRepeat
+	LEAL -4(R12), R12
+	JMP  cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b
+
+emit_repeat_again_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b:
+	MOVL R12, SI
+	LEAL -4(R12), R12
+	CMPL SI, $0x08
+	JLE  repeat_two_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b
+	CMPL SI, $0x0c
+	JGE  cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b
+	CMPL R8, $0x00000800
+	JLT  repeat_two_offset_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b
+
+cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b:
+	CMPL R12, $0x00000104
+	JLT  repeat_three_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b
+	CMPL R12, $0x00010100
+	JLT  repeat_four_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b
+	CMPL R12, $0x0100ffff
+	JLT  repeat_five_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b
+	LEAL -16842747(R12), R12
+	MOVW $0x001d, (AX)
+	MOVW $0xfffb, 2(AX)
+	MOVB $0xff, 4(AX)
+	ADDQ $0x05, AX
+	JMP  emit_repeat_again_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b
+
+repeat_five_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b:
+	LEAL -65536(R12), R12
+	MOVL R12, R8
+	MOVW $0x001d, (AX)
+	MOVW R12, 2(AX)
+	SARL $0x10, R8
+	MOVB R8, 4(AX)
+	ADDQ $0x05, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm
+
+repeat_four_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b:
+	LEAL -256(R12), R12
+	MOVW $0x0019, (AX)
+	MOVW R12, 2(AX)
+	ADDQ $0x04, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm
+
+repeat_three_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b:
+	LEAL -4(R12), R12
+	MOVW $0x0015, (AX)
+	MOVB R12, 2(AX)
+	ADDQ $0x03, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm
+
+repeat_two_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b:
+	SHLL $0x02, R12
+	ORL  $0x01, R12
+	MOVW R12, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm
+
+repeat_two_offset_match_nolit_encodeBetterBlockAsm_emit_copy_short_2b:
+	XORQ SI, SI
+	LEAL 1(SI)(R12*4), R12
+	MOVB R8, 1(AX)
+	SARL $0x08, R8
+	SHLL $0x05, R8
+	ORL  R8, R12
+	MOVB R12, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm
+
+long_offset_short_match_nolit_encodeBetterBlockAsm:
 	MOVB $0xee, (AX)
 	MOVW R8, 1(AX)
 	LEAL -60(R12), R12
@@ -5979,8 +6702,9 @@ memmove_emit_remainder_encodeBetterBlockAsm:
 	MOVL SI, BX
 
 	// genMemMoveShort
-	CMPQ BX, $0x04
-	JLE  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm_memmove_move_4
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm_memmove_move_3
 	CMPQ BX, $0x08
 	JB   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm_memmove_move_4through7
 	CMPQ BX, $0x10
@@ -5989,9 +6713,18 @@ memmove_emit_remainder_encodeBetterBlockAsm:
 	JBE  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeBetterBlockAsm_memmove_move_4:
-	MOVL (CX), SI
-	MOVL SI, (AX)
+emit_lit_memmove_emit_remainder_encodeBetterBlockAsm_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeBetterBlockAsm
+
+emit_lit_memmove_emit_remainder_encodeBetterBlockAsm_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
 	JMP  memmove_end_copy_emit_remainder_encodeBetterBlockAsm
 
 emit_lit_memmove_emit_remainder_encodeBetterBlockAsm_memmove_move_4through7:
@@ -6214,17 +6947,6 @@ matchlen_loopback_match_nolit_encodeBetterBlockAsm4MB:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -6511,6 +7233,77 @@ four_bytes_remain_match_nolit_encodeBetterBlockAsm4MB:
 two_byte_offset_match_nolit_encodeBetterBlockAsm4MB:
 	CMPL R12, $0x40
 	JLE  two_byte_offset_short_match_nolit_encodeBetterBlockAsm4MB
+	CMPL R8, $0x00000800
+	JAE  long_offset_short_match_nolit_encodeBetterBlockAsm4MB
+	MOVL $0x00000001, SI
+	LEAL 16(SI), SI
+	MOVB R8, 1(AX)
+	SHRL $0x08, R8
+	SHLL $0x05, R8
+	ORL  R8, SI
+	MOVB SI, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, R12
+
+	// emitRepeat
+	LEAL -4(R12), R12
+	JMP  cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm4MB_emit_copy_short_2b
+	MOVL R12, SI
+	LEAL -4(R12), R12
+	CMPL SI, $0x08
+	JLE  repeat_two_match_nolit_encodeBetterBlockAsm4MB_emit_copy_short_2b
+	CMPL SI, $0x0c
+	JGE  cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm4MB_emit_copy_short_2b
+	CMPL R8, $0x00000800
+	JLT  repeat_two_offset_match_nolit_encodeBetterBlockAsm4MB_emit_copy_short_2b
+
+cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm4MB_emit_copy_short_2b:
+	CMPL R12, $0x00000104
+	JLT  repeat_three_match_nolit_encodeBetterBlockAsm4MB_emit_copy_short_2b
+	CMPL R12, $0x00010100
+	JLT  repeat_four_match_nolit_encodeBetterBlockAsm4MB_emit_copy_short_2b
+	LEAL -65536(R12), R12
+	MOVL R12, R8
+	MOVW $0x001d, (AX)
+	MOVW R12, 2(AX)
+	SARL $0x10, R8
+	MOVB R8, 4(AX)
+	ADDQ $0x05, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm4MB
+
+repeat_four_match_nolit_encodeBetterBlockAsm4MB_emit_copy_short_2b:
+	LEAL -256(R12), R12
+	MOVW $0x0019, (AX)
+	MOVW R12, 2(AX)
+	ADDQ $0x04, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm4MB
+
+repeat_three_match_nolit_encodeBetterBlockAsm4MB_emit_copy_short_2b:
+	LEAL -4(R12), R12
+	MOVW $0x0015, (AX)
+	MOVB R12, 2(AX)
+	ADDQ $0x03, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm4MB
+
+repeat_two_match_nolit_encodeBetterBlockAsm4MB_emit_copy_short_2b:
+	SHLL $0x02, R12
+	ORL  $0x01, R12
+	MOVW R12, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm4MB
+
+repeat_two_offset_match_nolit_encodeBetterBlockAsm4MB_emit_copy_short_2b:
+	XORQ SI, SI
+	LEAL 1(SI)(R12*4), R12
+	MOVB R8, 1(AX)
+	SARL $0x08, R8
+	SHLL $0x05, R8
+	ORL  R8, R12
+	MOVB R12, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm4MB
+
+long_offset_short_match_nolit_encodeBetterBlockAsm4MB:
 	MOVB $0xee, (AX)
 	MOVW R8, 1(AX)
 	LEAL -60(R12), R12
@@ -6911,8 +7704,9 @@ memmove_emit_remainder_encodeBetterBlockAsm4MB:
 	MOVL SI, BX
 
 	// genMemMoveShort
-	CMPQ BX, $0x04
-	JLE  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm4MB_memmove_move_4
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm4MB_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm4MB_memmove_move_3
 	CMPQ BX, $0x08
 	JB   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm4MB_memmove_move_4through7
 	CMPQ BX, $0x10
@@ -6921,9 +7715,18 @@ memmove_emit_remainder_encodeBetterBlockAsm4MB:
 	JBE  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm4MB_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm4MB_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeBetterBlockAsm4MB_memmove_move_4:
-	MOVL (CX), SI
-	MOVL SI, (AX)
+emit_lit_memmove_emit_remainder_encodeBetterBlockAsm4MB_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeBetterBlockAsm4MB
+
+emit_lit_memmove_emit_remainder_encodeBetterBlockAsm4MB_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
 	JMP  memmove_end_copy_emit_remainder_encodeBetterBlockAsm4MB
 
 emit_lit_memmove_emit_remainder_encodeBetterBlockAsm4MB_memmove_move_4through7:
@@ -7138,17 +7941,6 @@ matchlen_loopback_match_nolit_encodeBetterBlockAsm12B:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -7335,6 +8127,65 @@ emit_literal_done_match_emit_encodeBetterBlockAsm12B:
 two_byte_offset_match_nolit_encodeBetterBlockAsm12B:
 	CMPL R12, $0x40
 	JLE  two_byte_offset_short_match_nolit_encodeBetterBlockAsm12B
+	CMPL R8, $0x00000800
+	JAE  long_offset_short_match_nolit_encodeBetterBlockAsm12B
+	MOVL $0x00000001, SI
+	LEAL 16(SI), SI
+	MOVB R8, 1(AX)
+	SHRL $0x08, R8
+	SHLL $0x05, R8
+	ORL  R8, SI
+	MOVB SI, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, R12
+
+	// emitRepeat
+	LEAL -4(R12), R12
+	JMP  cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm12B_emit_copy_short_2b
+	MOVL R12, SI
+	LEAL -4(R12), R12
+	CMPL SI, $0x08
+	JLE  repeat_two_match_nolit_encodeBetterBlockAsm12B_emit_copy_short_2b
+	CMPL SI, $0x0c
+	JGE  cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm12B_emit_copy_short_2b
+	CMPL R8, $0x00000800
+	JLT  repeat_two_offset_match_nolit_encodeBetterBlockAsm12B_emit_copy_short_2b
+
+cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm12B_emit_copy_short_2b:
+	CMPL R12, $0x00000104
+	JLT  repeat_three_match_nolit_encodeBetterBlockAsm12B_emit_copy_short_2b
+	LEAL -256(R12), R12
+	MOVW $0x0019, (AX)
+	MOVW R12, 2(AX)
+	ADDQ $0x04, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm12B
+
+repeat_three_match_nolit_encodeBetterBlockAsm12B_emit_copy_short_2b:
+	LEAL -4(R12), R12
+	MOVW $0x0015, (AX)
+	MOVB R12, 2(AX)
+	ADDQ $0x03, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm12B
+
+repeat_two_match_nolit_encodeBetterBlockAsm12B_emit_copy_short_2b:
+	SHLL $0x02, R12
+	ORL  $0x01, R12
+	MOVW R12, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm12B
+
+repeat_two_offset_match_nolit_encodeBetterBlockAsm12B_emit_copy_short_2b:
+	XORQ SI, SI
+	LEAL 1(SI)(R12*4), R12
+	MOVB R8, 1(AX)
+	SARL $0x08, R8
+	SHLL $0x05, R8
+	ORL  R8, R12
+	MOVB R12, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm12B
+
+long_offset_short_match_nolit_encodeBetterBlockAsm12B:
 	MOVB $0xee, (AX)
 	MOVW R8, 1(AX)
 	LEAL -60(R12), R12
@@ -7689,8 +8540,9 @@ memmove_emit_remainder_encodeBetterBlockAsm12B:
 	MOVL SI, BX
 
 	// genMemMoveShort
-	CMPQ BX, $0x04
-	JLE  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm12B_memmove_move_4
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm12B_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm12B_memmove_move_3
 	CMPQ BX, $0x08
 	JB   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm12B_memmove_move_4through7
 	CMPQ BX, $0x10
@@ -7699,9 +8551,18 @@ memmove_emit_remainder_encodeBetterBlockAsm12B:
 	JBE  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm12B_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm12B_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeBetterBlockAsm12B_memmove_move_4:
-	MOVL (CX), SI
-	MOVL SI, (AX)
+emit_lit_memmove_emit_remainder_encodeBetterBlockAsm12B_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeBetterBlockAsm12B
+
+emit_lit_memmove_emit_remainder_encodeBetterBlockAsm12B_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
 	JMP  memmove_end_copy_emit_remainder_encodeBetterBlockAsm12B
 
 emit_lit_memmove_emit_remainder_encodeBetterBlockAsm12B_memmove_move_4through7:
@@ -7916,17 +8777,6 @@ matchlen_loopback_match_nolit_encodeBetterBlockAsm10B:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -8113,6 +8963,65 @@ emit_literal_done_match_emit_encodeBetterBlockAsm10B:
 two_byte_offset_match_nolit_encodeBetterBlockAsm10B:
 	CMPL R12, $0x40
 	JLE  two_byte_offset_short_match_nolit_encodeBetterBlockAsm10B
+	CMPL R8, $0x00000800
+	JAE  long_offset_short_match_nolit_encodeBetterBlockAsm10B
+	MOVL $0x00000001, SI
+	LEAL 16(SI), SI
+	MOVB R8, 1(AX)
+	SHRL $0x08, R8
+	SHLL $0x05, R8
+	ORL  R8, SI
+	MOVB SI, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, R12
+
+	// emitRepeat
+	LEAL -4(R12), R12
+	JMP  cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm10B_emit_copy_short_2b
+	MOVL R12, SI
+	LEAL -4(R12), R12
+	CMPL SI, $0x08
+	JLE  repeat_two_match_nolit_encodeBetterBlockAsm10B_emit_copy_short_2b
+	CMPL SI, $0x0c
+	JGE  cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm10B_emit_copy_short_2b
+	CMPL R8, $0x00000800
+	JLT  repeat_two_offset_match_nolit_encodeBetterBlockAsm10B_emit_copy_short_2b
+
+cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm10B_emit_copy_short_2b:
+	CMPL R12, $0x00000104
+	JLT  repeat_three_match_nolit_encodeBetterBlockAsm10B_emit_copy_short_2b
+	LEAL -256(R12), R12
+	MOVW $0x0019, (AX)
+	MOVW R12, 2(AX)
+	ADDQ $0x04, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm10B
+
+repeat_three_match_nolit_encodeBetterBlockAsm10B_emit_copy_short_2b:
+	LEAL -4(R12), R12
+	MOVW $0x0015, (AX)
+	MOVB R12, 2(AX)
+	ADDQ $0x03, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm10B
+
+repeat_two_match_nolit_encodeBetterBlockAsm10B_emit_copy_short_2b:
+	SHLL $0x02, R12
+	ORL  $0x01, R12
+	MOVW R12, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm10B
+
+repeat_two_offset_match_nolit_encodeBetterBlockAsm10B_emit_copy_short_2b:
+	XORQ SI, SI
+	LEAL 1(SI)(R12*4), R12
+	MOVB R8, 1(AX)
+	SARL $0x08, R8
+	SHLL $0x05, R8
+	ORL  R8, R12
+	MOVB R12, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm10B
+
+long_offset_short_match_nolit_encodeBetterBlockAsm10B:
 	MOVB $0xee, (AX)
 	MOVW R8, 1(AX)
 	LEAL -60(R12), R12
@@ -8467,8 +9376,9 @@ memmove_emit_remainder_encodeBetterBlockAsm10B:
 	MOVL SI, BX
 
 	// genMemMoveShort
-	CMPQ BX, $0x04
-	JLE  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm10B_memmove_move_4
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm10B_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm10B_memmove_move_3
 	CMPQ BX, $0x08
 	JB   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm10B_memmove_move_4through7
 	CMPQ BX, $0x10
@@ -8477,9 +9387,18 @@ memmove_emit_remainder_encodeBetterBlockAsm10B:
 	JBE  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm10B_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm10B_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeBetterBlockAsm10B_memmove_move_4:
-	MOVL (CX), SI
-	MOVL SI, (AX)
+emit_lit_memmove_emit_remainder_encodeBetterBlockAsm10B_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeBetterBlockAsm10B
+
+emit_lit_memmove_emit_remainder_encodeBetterBlockAsm10B_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
 	JMP  memmove_end_copy_emit_remainder_encodeBetterBlockAsm10B
 
 emit_lit_memmove_emit_remainder_encodeBetterBlockAsm10B_memmove_move_4through7:
@@ -8694,17 +9613,6 @@ matchlen_loopback_match_nolit_encodeBetterBlockAsm8B:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -8891,6 +9799,61 @@ emit_literal_done_match_emit_encodeBetterBlockAsm8B:
 two_byte_offset_match_nolit_encodeBetterBlockAsm8B:
 	CMPL R12, $0x40
 	JLE  two_byte_offset_short_match_nolit_encodeBetterBlockAsm8B
+	CMPL R8, $0x00000800
+	JAE  long_offset_short_match_nolit_encodeBetterBlockAsm8B
+	MOVL $0x00000001, SI
+	LEAL 16(SI), SI
+	MOVB R8, 1(AX)
+	SHRL $0x08, R8
+	SHLL $0x05, R8
+	ORL  R8, SI
+	MOVB SI, (AX)
+	ADDQ $0x02, AX
+	SUBL $0x08, R12
+
+	// emitRepeat
+	LEAL -4(R12), R12
+	JMP  cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm8B_emit_copy_short_2b
+	MOVL R12, SI
+	LEAL -4(R12), R12
+	CMPL SI, $0x08
+	JLE  repeat_two_match_nolit_encodeBetterBlockAsm8B_emit_copy_short_2b
+	CMPL SI, $0x0c
+	JGE  cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm8B_emit_copy_short_2b
+
+cant_repeat_two_offset_match_nolit_encodeBetterBlockAsm8B_emit_copy_short_2b:
+	CMPL R12, $0x00000104
+	JLT  repeat_three_match_nolit_encodeBetterBlockAsm8B_emit_copy_short_2b
+	LEAL -256(R12), R12
+	MOVW $0x0019, (AX)
+	MOVW R12, 2(AX)
+	ADDQ $0x04, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm8B
+
+repeat_three_match_nolit_encodeBetterBlockAsm8B_emit_copy_short_2b:
+	LEAL -4(R12), R12
+	MOVW $0x0015, (AX)
+	MOVB R12, 2(AX)
+	ADDQ $0x03, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm8B
+
+repeat_two_match_nolit_encodeBetterBlockAsm8B_emit_copy_short_2b:
+	SHLL $0x02, R12
+	ORL  $0x01, R12
+	MOVW R12, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm8B
+	XORQ SI, SI
+	LEAL 1(SI)(R12*4), R12
+	MOVB R8, 1(AX)
+	SARL $0x08, R8
+	SHLL $0x05, R8
+	ORL  R8, R12
+	MOVB R12, (AX)
+	ADDQ $0x02, AX
+	JMP  match_nolit_emitcopy_end_encodeBetterBlockAsm8B
+
+long_offset_short_match_nolit_encodeBetterBlockAsm8B:
 	MOVB $0xee, (AX)
 	MOVW R8, 1(AX)
 	LEAL -60(R12), R12
@@ -9235,8 +10198,9 @@ memmove_emit_remainder_encodeBetterBlockAsm8B:
 	MOVL SI, BX
 
 	// genMemMoveShort
-	CMPQ BX, $0x04
-	JLE  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm8B_memmove_move_4
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm8B_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm8B_memmove_move_3
 	CMPQ BX, $0x08
 	JB   emit_lit_memmove_emit_remainder_encodeBetterBlockAsm8B_memmove_move_4through7
 	CMPQ BX, $0x10
@@ -9245,9 +10209,18 @@ memmove_emit_remainder_encodeBetterBlockAsm8B:
 	JBE  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm8B_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeBetterBlockAsm8B_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeBetterBlockAsm8B_memmove_move_4:
-	MOVL (CX), SI
-	MOVL SI, (AX)
+emit_lit_memmove_emit_remainder_encodeBetterBlockAsm8B_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeBetterBlockAsm8B
+
+emit_lit_memmove_emit_remainder_encodeBetterBlockAsm8B_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
 	JMP  memmove_end_copy_emit_remainder_encodeBetterBlockAsm8B
 
 emit_lit_memmove_emit_remainder_encodeBetterBlockAsm8B_memmove_move_4through7:
@@ -9584,17 +10557,6 @@ matchlen_loopback_repeat_extend_encodeSnappyBlockAsm:
 #ifdef GOAMD64_v3
 	TZCNTQ R10, R10
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R10, R10
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R10, R10
 
@@ -9918,17 +10880,6 @@ matchlen_loopback_match_nolit_encodeSnappyBlockAsm:
 #ifdef GOAMD64_v3
 	TZCNTQ R9, R9
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R9, R9
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R9, R9
 
@@ -10127,17 +11078,36 @@ memmove_emit_remainder_encodeSnappyBlockAsm:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm
+
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm
+
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm
 
 emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm_memmove_move_8through16:
@@ -10448,17 +11418,6 @@ matchlen_loopback_repeat_extend_encodeSnappyBlockAsm64K:
 #ifdef GOAMD64_v3
 	TZCNTQ R10, R10
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R10, R10
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R10, R10
 
@@ -10739,17 +11698,6 @@ matchlen_loopback_match_nolit_encodeSnappyBlockAsm64K:
 #ifdef GOAMD64_v3
 	TZCNTQ R9, R9
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R9, R9
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R9, R9
 
@@ -10905,17 +11853,36 @@ memmove_emit_remainder_encodeSnappyBlockAsm64K:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm64K_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm64K_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm64K_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm64K_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm64K_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm64K_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm64K_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm64K_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm64K_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm64K
+
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm64K_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm64K
+
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm64K_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm64K
 
 emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm64K_memmove_move_8through16:
@@ -11226,17 +12193,6 @@ matchlen_loopback_repeat_extend_encodeSnappyBlockAsm12B:
 #ifdef GOAMD64_v3
 	TZCNTQ R10, R10
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R10, R10
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R10, R10
 
@@ -11517,17 +12473,6 @@ matchlen_loopback_match_nolit_encodeSnappyBlockAsm12B:
 #ifdef GOAMD64_v3
 	TZCNTQ R9, R9
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R9, R9
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R9, R9
 
@@ -11683,17 +12628,36 @@ memmove_emit_remainder_encodeSnappyBlockAsm12B:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm12B_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm12B_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm12B_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm12B_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm12B_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm12B_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm12B_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm12B_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm12B_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm12B
+
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm12B_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm12B
+
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm12B_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm12B
 
 emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm12B_memmove_move_8through16:
@@ -12004,17 +12968,6 @@ matchlen_loopback_repeat_extend_encodeSnappyBlockAsm10B:
 #ifdef GOAMD64_v3
 	TZCNTQ R10, R10
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R10, R10
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R10, R10
 
@@ -12295,17 +13248,6 @@ matchlen_loopback_match_nolit_encodeSnappyBlockAsm10B:
 #ifdef GOAMD64_v3
 	TZCNTQ R9, R9
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R9, R9
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R9, R9
 
@@ -12461,17 +13403,36 @@ memmove_emit_remainder_encodeSnappyBlockAsm10B:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm10B_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm10B_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm10B_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm10B_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm10B_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm10B_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm10B_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm10B_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm10B_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm10B
+
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm10B_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm10B
+
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm10B_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm10B
 
 emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm10B_memmove_move_8through16:
@@ -12782,17 +13743,6 @@ matchlen_loopback_repeat_extend_encodeSnappyBlockAsm8B:
 #ifdef GOAMD64_v3
 	TZCNTQ R10, R10
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R10, R10
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R10, R10
 
@@ -13071,17 +14021,6 @@ matchlen_loopback_match_nolit_encodeSnappyBlockAsm8B:
 #ifdef GOAMD64_v3
 	TZCNTQ R9, R9
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R9, R9
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R9, R9
 
@@ -13235,17 +14174,36 @@ memmove_emit_remainder_encodeSnappyBlockAsm8B:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm8B_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm8B_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm8B_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm8B_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm8B_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm8B_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm8B_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm8B_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm8B_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm8B
+
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm8B_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm8B
+
+emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm8B_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeSnappyBlockAsm8B
 
 emit_lit_memmove_emit_remainder_encodeSnappyBlockAsm8B_memmove_move_8through16:
@@ -13461,17 +14419,6 @@ matchlen_loopback_match_nolit_encodeSnappyBetterBlockAsm:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -13850,17 +14797,36 @@ memmove_emit_remainder_encodeSnappyBetterBlockAsm:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm
+
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm
+
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm
 
 emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm_memmove_move_8through16:
@@ -14068,17 +15034,6 @@ matchlen_loopback_match_nolit_encodeSnappyBetterBlockAsm64K:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -14386,17 +15341,36 @@ memmove_emit_remainder_encodeSnappyBetterBlockAsm64K:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm64K_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm64K_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm64K_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm64K_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm64K_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm64K_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm64K_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm64K_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm64K_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm64K
+
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm64K_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm64K
+
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm64K_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm64K
 
 emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm64K_memmove_move_8through16:
@@ -14604,17 +15578,6 @@ matchlen_loopback_match_nolit_encodeSnappyBetterBlockAsm12B:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -14922,17 +15885,36 @@ memmove_emit_remainder_encodeSnappyBetterBlockAsm12B:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm12B_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm12B_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm12B_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm12B_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm12B_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm12B_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm12B_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm12B_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm12B_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm12B
+
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm12B_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm12B
+
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm12B_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm12B
 
 emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm12B_memmove_move_8through16:
@@ -15140,17 +16122,6 @@ matchlen_loopback_match_nolit_encodeSnappyBetterBlockAsm10B:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -15458,17 +16429,36 @@ memmove_emit_remainder_encodeSnappyBetterBlockAsm10B:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm10B_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm10B_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm10B_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm10B_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm10B_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm10B_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm10B_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm10B_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm10B_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm10B
+
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm10B_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm10B
+
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm10B_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm10B
 
 emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm10B_memmove_move_8through16:
@@ -15676,17 +16666,6 @@ matchlen_loopback_match_nolit_encodeSnappyBetterBlockAsm8B:
 #ifdef GOAMD64_v3
 	TZCNTQ R11, R11
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ R11, R11
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ R11, R11
 
@@ -15992,17 +16971,36 @@ memmove_emit_remainder_encodeSnappyBetterBlockAsm8B:
 	MOVL SI, BX
 
 	// genMemMoveShort
+	CMPQ BX, $0x03
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm8B_memmove_move_1or2
+	JE   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm8B_memmove_move_3
 	CMPQ BX, $0x08
-	JLE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm8B_memmove_move_8
+	JB   emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm8B_memmove_move_4through7
 	CMPQ BX, $0x10
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm8B_memmove_move_8through16
 	CMPQ BX, $0x20
 	JBE  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm8B_memmove_move_17through32
 	JMP  emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm8B_memmove_move_33through64
 
-emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm8B_memmove_move_8:
-	MOVQ (CX), SI
-	MOVQ SI, (AX)
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm8B_memmove_move_1or2:
+	MOVB (CX), SI
+	MOVB -1(CX)(BX*1), CL
+	MOVB SI, (AX)
+	MOVB CL, -1(AX)(BX*1)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm8B
+
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm8B_memmove_move_3:
+	MOVW (CX), SI
+	MOVB 2(CX), CL
+	MOVW SI, (AX)
+	MOVB CL, 2(AX)
+	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm8B
+
+emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm8B_memmove_move_4through7:
+	MOVL (CX), SI
+	MOVL -4(CX)(BX*1), CX
+	MOVL SI, (AX)
+	MOVL CX, -4(AX)(BX*1)
 	JMP  memmove_end_copy_emit_remainder_encodeSnappyBetterBlockAsm8B
 
 emit_lit_memmove_emit_remainder_encodeSnappyBetterBlockAsm8B_memmove_move_8through16:
@@ -16443,6 +17441,97 @@ four_bytes_remain_standalone:
 two_byte_offset_standalone:
 	CMPL DX, $0x40
 	JLE  two_byte_offset_short_standalone
+	CMPL CX, $0x00000800
+	JAE  long_offset_short_standalone
+	MOVL $0x00000001, SI
+	LEAL 16(SI), SI
+	MOVB CL, 1(AX)
+	MOVL CX, DI
+	SHRL $0x08, DI
+	SHLL $0x05, DI
+	ORL  DI, SI
+	MOVB SI, (AX)
+	ADDQ $0x02, BX
+	ADDQ $0x02, AX
+	SUBL $0x08, DX
+
+	// emitRepeat
+	LEAL -4(DX), DX
+	JMP  cant_repeat_two_offset_standalone_emit_copy_short_2b
+
+emit_repeat_again_standalone_emit_copy_short_2b:
+	MOVL DX, SI
+	LEAL -4(DX), DX
+	CMPL SI, $0x08
+	JLE  repeat_two_standalone_emit_copy_short_2b
+	CMPL SI, $0x0c
+	JGE  cant_repeat_two_offset_standalone_emit_copy_short_2b
+	CMPL CX, $0x00000800
+	JLT  repeat_two_offset_standalone_emit_copy_short_2b
+
+cant_repeat_two_offset_standalone_emit_copy_short_2b:
+	CMPL DX, $0x00000104
+	JLT  repeat_three_standalone_emit_copy_short_2b
+	CMPL DX, $0x00010100
+	JLT  repeat_four_standalone_emit_copy_short_2b
+	CMPL DX, $0x0100ffff
+	JLT  repeat_five_standalone_emit_copy_short_2b
+	LEAL -16842747(DX), DX
+	MOVW $0x001d, (AX)
+	MOVW $0xfffb, 2(AX)
+	MOVB $0xff, 4(AX)
+	ADDQ $0x05, AX
+	ADDQ $0x05, BX
+	JMP  emit_repeat_again_standalone_emit_copy_short_2b
+
+repeat_five_standalone_emit_copy_short_2b:
+	LEAL -65536(DX), DX
+	MOVL DX, CX
+	MOVW $0x001d, (AX)
+	MOVW DX, 2(AX)
+	SARL $0x10, CX
+	MOVB CL, 4(AX)
+	ADDQ $0x05, BX
+	ADDQ $0x05, AX
+	JMP  gen_emit_copy_end
+
+repeat_four_standalone_emit_copy_short_2b:
+	LEAL -256(DX), DX
+	MOVW $0x0019, (AX)
+	MOVW DX, 2(AX)
+	ADDQ $0x04, BX
+	ADDQ $0x04, AX
+	JMP  gen_emit_copy_end
+
+repeat_three_standalone_emit_copy_short_2b:
+	LEAL -4(DX), DX
+	MOVW $0x0015, (AX)
+	MOVB DL, 2(AX)
+	ADDQ $0x03, BX
+	ADDQ $0x03, AX
+	JMP  gen_emit_copy_end
+
+repeat_two_standalone_emit_copy_short_2b:
+	SHLL $0x02, DX
+	ORL  $0x01, DX
+	MOVW DX, (AX)
+	ADDQ $0x02, BX
+	ADDQ $0x02, AX
+	JMP  gen_emit_copy_end
+
+repeat_two_offset_standalone_emit_copy_short_2b:
+	XORQ SI, SI
+	LEAL 1(SI)(DX*4), DX
+	MOVB CL, 1(AX)
+	SARL $0x08, CX
+	SHLL $0x05, CX
+	ORL  CX, DX
+	MOVB DL, (AX)
+	ADDQ $0x02, BX
+	ADDQ $0x02, AX
+	JMP  gen_emit_copy_end
+
+long_offset_short_standalone:
 	MOVB $0xee, (AX)
 	MOVW CX, 1(AX)
 	LEAL -60(DX), DX
@@ -16644,17 +17733,6 @@ matchlen_loopback_standalone:
 #ifdef GOAMD64_v3
 	TZCNTQ BX, BX
 
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef GOAMD64_v4
-	TZCNTQ BX, BX
-
-#define TZCNTQ_EMITTED 1
-#endif
-
-#ifdef TZCNTQ_EMITTED
-#undef TZCNTQ_EMITTED
 #else
 	BSFQ BX, BX
 
diff --git a/vendor/github.com/klauspost/compress/s2/index.go b/vendor/github.com/klauspost/compress/s2/index.go
index fd857682..7b24a006 100644
--- a/vendor/github.com/klauspost/compress/s2/index.go
+++ b/vendor/github.com/klauspost/compress/s2/index.go
@@ -10,6 +10,7 @@ import (
 	"encoding/json"
 	"fmt"
 	"io"
+	"sort"
 )
 
 const (
@@ -100,6 +101,15 @@ func (i *Index) Find(offset int64) (compressedOff, uncompressedOff int64, err er
 	if offset > i.TotalUncompressed {
 		return 0, 0, io.ErrUnexpectedEOF
 	}
+	if len(i.info) > 200 {
+		n := sort.Search(len(i.info), func(n int) bool {
+			return i.info[n].uncompressedOffset > offset
+		})
+		if n == 0 {
+			n = 1
+		}
+		return i.info[n-1].compressedOffset, i.info[n-1].uncompressedOffset, nil
+	}
 	for _, info := range i.info {
 		if info.uncompressedOffset > offset {
 			break
diff --git a/vendor/github.com/klauspost/compress/zstd/README.md b/vendor/github.com/klauspost/compress/zstd/README.md
index e3445ac1..beb7fa87 100644
--- a/vendor/github.com/klauspost/compress/zstd/README.md
+++ b/vendor/github.com/klauspost/compress/zstd/README.md
@@ -386,47 +386,31 @@ In practice this means that concurrency is often limited to utilizing about 3 co
   
 ### Benchmarks
 
-These are some examples of performance compared to [datadog cgo library](https://github.com/DataDog/zstd).
-
 The first two are streaming decodes and the last are smaller inputs. 
- 
+
+Running on AMD Ryzen 9 3950X 16-Core Processor. AMD64 assembly used.
+
 ```
-BenchmarkDecoderSilesia-8                          3     385000067 ns/op     550.51 MB/s        5498 B/op          8 allocs/op
-BenchmarkDecoderSilesiaCgo-8                       6     197666567 ns/op    1072.25 MB/s      270672 B/op          8 allocs/op
-
-BenchmarkDecoderEnwik9-8                           1    2027001600 ns/op     493.34 MB/s       10496 B/op         18 allocs/op
-BenchmarkDecoderEnwik9Cgo-8                        2     979499200 ns/op    1020.93 MB/s      270672 B/op          8 allocs/op
-
-Concurrent performance:
-
-BenchmarkDecoder_DecodeAllParallel/kppkn.gtb.zst-16                28915         42469 ns/op    4340.07 MB/s         114 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallel/geo.protodata.zst-16           116505          9965 ns/op    11900.16 MB/s         16 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallel/plrabn12.txt.zst-16              8952        134272 ns/op    3588.70 MB/s         915 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallel/lcet10.txt.zst-16               11820        102538 ns/op    4161.90 MB/s         594 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallel/asyoulik.txt.zst-16             34782         34184 ns/op    3661.88 MB/s          60 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallel/alice29.txt.zst-16              27712         43447 ns/op    3500.58 MB/s          99 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallel/html_x_4.zst-16                 62826         18750 ns/op    21845.10 MB/s        104 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallel/paper-100k.pdf.zst-16          631545          1794 ns/op    57078.74 MB/s          2 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallel/fireworks.jpeg.zst-16         1690140           712 ns/op    172938.13 MB/s         1 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallel/urls.10K.zst-16                 10432        113593 ns/op    6180.73 MB/s        1143 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallel/html.zst-16                    113206         10671 ns/op    9596.27 MB/s          15 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallel/comp-data.bin.zst-16          1530615           779 ns/op    5229.49 MB/s           0 B/op          0 allocs/op
-
-BenchmarkDecoder_DecodeAllParallelCgo/kppkn.gtb.zst-16             65217         16192 ns/op    11383.34 MB/s         46 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallelCgo/geo.protodata.zst-16        292671          4039 ns/op    29363.19 MB/s          6 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallelCgo/plrabn12.txt.zst-16          26314         46021 ns/op    10470.43 MB/s        293 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallelCgo/lcet10.txt.zst-16            33897         34900 ns/op    12227.96 MB/s        205 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallelCgo/asyoulik.txt.zst-16         104348         11433 ns/op    10949.01 MB/s         20 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallelCgo/alice29.txt.zst-16           75949         15510 ns/op    9805.60 MB/s          32 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallelCgo/html_x_4.zst-16             173910          6756 ns/op    60624.29 MB/s         37 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallelCgo/paper-100k.pdf.zst-16       923076          1339 ns/op    76474.87 MB/s          1 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallelCgo/fireworks.jpeg.zst-16       922920          1351 ns/op    91102.57 MB/s          2 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallelCgo/urls.10K.zst-16              27649         43618 ns/op    16096.19 MB/s        407 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallelCgo/html.zst-16                 279073          4160 ns/op    24614.18 MB/s          6 B/op          0 allocs/op
-BenchmarkDecoder_DecodeAllParallelCgo/comp-data.bin.zst-16        749938          1579 ns/op    2581.71 MB/s           0 B/op          0 allocs/op
+BenchmarkDecoderSilesia-32    	                   5	 206878840 ns/op	1024.50 MB/s	   49808 B/op	      43 allocs/op
+BenchmarkDecoderEnwik9-32                          1	1271809000 ns/op	 786.28 MB/s	   72048 B/op	      52 allocs/op
+
+Concurrent blocks, performance:
+
+BenchmarkDecoder_DecodeAllParallel/kppkn.gtb.zst-32         	   67356	     17857 ns/op	10321.96 MB/s	        22.48 pct	     102 B/op	       0 allocs/op
+BenchmarkDecoder_DecodeAllParallel/geo.protodata.zst-32     	  266656	      4421 ns/op	26823.21 MB/s	        11.89 pct	      19 B/op	       0 allocs/op
+BenchmarkDecoder_DecodeAllParallel/plrabn12.txt.zst-32      	   20992	     56842 ns/op	8477.17 MB/s	        39.90 pct	     754 B/op	       0 allocs/op
+BenchmarkDecoder_DecodeAllParallel/lcet10.txt.zst-32        	   27456	     43932 ns/op	9714.01 MB/s	        33.27 pct	     524 B/op	       0 allocs/op
+BenchmarkDecoder_DecodeAllParallel/asyoulik.txt.zst-32      	   78432	     15047 ns/op	8319.15 MB/s	        40.34 pct	      66 B/op	       0 allocs/op
+BenchmarkDecoder_DecodeAllParallel/alice29.txt.zst-32       	   65800	     18436 ns/op	8249.63 MB/s	        37.75 pct	      88 B/op	       0 allocs/op
+BenchmarkDecoder_DecodeAllParallel/html_x_4.zst-32          	  102993	     11523 ns/op	35546.09 MB/s	         3.637 pct	     143 B/op	       0 allocs/op
+BenchmarkDecoder_DecodeAllParallel/paper-100k.pdf.zst-32    	 1000000	      1070 ns/op	95720.98 MB/s	        80.53 pct	       3 B/op	       0 allocs/op
+BenchmarkDecoder_DecodeAllParallel/fireworks.jpeg.zst-32    	  749802	      1752 ns/op	70272.35 MB/s	       100.0 pct	       5 B/op	       0 allocs/op
+BenchmarkDecoder_DecodeAllParallel/urls.10K.zst-32          	   22640	     52934 ns/op	13263.37 MB/s	        26.25 pct	    1014 B/op	       0 allocs/op
+BenchmarkDecoder_DecodeAllParallel/html.zst-32              	  226412	      5232 ns/op	19572.27 MB/s	        14.49 pct	      20 B/op	       0 allocs/op
+BenchmarkDecoder_DecodeAllParallel/comp-data.bin.zst-32     	  923041	      1276 ns/op	3194.71 MB/s	        31.26 pct	       0 B/op	       0 allocs/op
 ```
 
-This reflects the performance around May 2020, but this may be out of date.
+This reflects the performance around May 2022, but this may be out of date.
 
 ## Zstd inside ZIP files
 
diff --git a/vendor/github.com/klauspost/compress/zstd/bitreader.go b/vendor/github.com/klauspost/compress/zstd/bitreader.go
index d7cd15ba..97299d49 100644
--- a/vendor/github.com/klauspost/compress/zstd/bitreader.go
+++ b/vendor/github.com/klauspost/compress/zstd/bitreader.go
@@ -63,13 +63,6 @@ func (b *bitReader) get32BitsFast(n uint8) uint32 {
 	return v
 }
 
-func (b *bitReader) get16BitsFast(n uint8) uint16 {
-	const regMask = 64 - 1
-	v := uint16((b.value << (b.bitsRead & regMask)) >> ((regMask + 1 - n) & regMask))
-	b.bitsRead += n
-	return v
-}
-
 // fillFast() will make sure at least 32 bits are available.
 // There must be at least 4 bytes available.
 func (b *bitReader) fillFast() {
diff --git a/vendor/github.com/klauspost/compress/zstd/bitwriter.go b/vendor/github.com/klauspost/compress/zstd/bitwriter.go
index b3661828..78b3c61b 100644
--- a/vendor/github.com/klauspost/compress/zstd/bitwriter.go
+++ b/vendor/github.com/klauspost/compress/zstd/bitwriter.go
@@ -5,8 +5,6 @@
 
 package zstd
 
-import "fmt"
-
 // bitWriter will write bits.
 // First bit will be LSB of the first byte of output.
 type bitWriter struct {
@@ -73,80 +71,6 @@ func (b *bitWriter) addBits16Clean(value uint16, bits uint8) {
 	b.nBits += bits
 }
 
-// flush will flush all pending full bytes.
-// There will be at least 56 bits available for writing when this has been called.
-// Using flush32 is faster, but leaves less space for writing.
-func (b *bitWriter) flush() {
-	v := b.nBits >> 3
-	switch v {
-	case 0:
-	case 1:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-		)
-	case 2:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-		)
-	case 3:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-			byte(b.bitContainer>>16),
-		)
-	case 4:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-			byte(b.bitContainer>>16),
-			byte(b.bitContainer>>24),
-		)
-	case 5:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-			byte(b.bitContainer>>16),
-			byte(b.bitContainer>>24),
-			byte(b.bitContainer>>32),
-		)
-	case 6:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-			byte(b.bitContainer>>16),
-			byte(b.bitContainer>>24),
-			byte(b.bitContainer>>32),
-			byte(b.bitContainer>>40),
-		)
-	case 7:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-			byte(b.bitContainer>>16),
-			byte(b.bitContainer>>24),
-			byte(b.bitContainer>>32),
-			byte(b.bitContainer>>40),
-			byte(b.bitContainer>>48),
-		)
-	case 8:
-		b.out = append(b.out,
-			byte(b.bitContainer),
-			byte(b.bitContainer>>8),
-			byte(b.bitContainer>>16),
-			byte(b.bitContainer>>24),
-			byte(b.bitContainer>>32),
-			byte(b.bitContainer>>40),
-			byte(b.bitContainer>>48),
-			byte(b.bitContainer>>56),
-		)
-	default:
-		panic(fmt.Errorf("bits (%d) > 64", b.nBits))
-	}
-	b.bitContainer >>= v << 3
-	b.nBits &= 7
-}
-
 // flush32 will flush out, so there are at least 32 bits available for writing.
 func (b *bitWriter) flush32() {
 	if b.nBits < 32 {
diff --git a/vendor/github.com/klauspost/compress/zstd/blockdec.go b/vendor/github.com/klauspost/compress/zstd/blockdec.go
index 7d567a54..7eed729b 100644
--- a/vendor/github.com/klauspost/compress/zstd/blockdec.go
+++ b/vendor/github.com/klauspost/compress/zstd/blockdec.go
@@ -5,9 +5,14 @@
 package zstd
 
 import (
+	"bytes"
+	"encoding/binary"
 	"errors"
 	"fmt"
 	"io"
+	"io/ioutil"
+	"os"
+	"path/filepath"
 	"sync"
 
 	"github.com/klauspost/compress/huff0"
@@ -38,14 +43,14 @@ const (
 	// maxCompressedBlockSize is the biggest allowed compressed block size (128KB)
 	maxCompressedBlockSize = 128 << 10
 
+	compressedBlockOverAlloc    = 16
+	maxCompressedBlockSizeAlloc = 128<<10 + compressedBlockOverAlloc
+
 	// Maximum possible block size (all Raw+Uncompressed).
 	maxBlockSize = (1 << 21) - 1
 
-	// https://github.com/facebook/zstd/blob/dev/doc/zstd_compression_format.md#literals_section_header
-	maxCompressedLiteralSize = 1 << 18
-	maxRLELiteralSize        = 1 << 20
-	maxMatchLen              = 131074
-	maxSequences             = 0x7f00 + 0xffff
+	maxMatchLen  = 131074
+	maxSequences = 0x7f00 + 0xffff
 
 	// We support slightly less than the reference decoder to be able to
 	// use ints on 32 bit archs.
@@ -97,7 +102,6 @@ type blockDec struct {
 
 	// Block is RLE, this is the size.
 	RLESize uint32
-	tmp     [4]byte
 
 	Type blockType
 
@@ -136,7 +140,7 @@ func (b *blockDec) reset(br byteBuffer, windowSize uint64) error {
 	b.Type = blockType((bh >> 1) & 3)
 	// find size.
 	cSize := int(bh >> 3)
-	maxSize := maxBlockSize
+	maxSize := maxCompressedBlockSizeAlloc
 	switch b.Type {
 	case blockTypeReserved:
 		return ErrReservedBlockType
@@ -157,9 +161,9 @@ func (b *blockDec) reset(br byteBuffer, windowSize uint64) error {
 			println("Data size on stream:", cSize)
 		}
 		b.RLESize = 0
-		maxSize = maxCompressedBlockSize
+		maxSize = maxCompressedBlockSizeAlloc
 		if windowSize < maxCompressedBlockSize && b.lowMem {
-			maxSize = int(windowSize)
+			maxSize = int(windowSize) + compressedBlockOverAlloc
 		}
 		if cSize > maxCompressedBlockSize || uint64(cSize) > b.WindowSize {
 			if debugDecoder {
@@ -190,9 +194,9 @@ func (b *blockDec) reset(br byteBuffer, windowSize uint64) error {
 	// Read block data.
 	if cap(b.dataStorage) < cSize {
 		if b.lowMem || cSize > maxCompressedBlockSize {
-			b.dataStorage = make([]byte, 0, cSize)
+			b.dataStorage = make([]byte, 0, cSize+compressedBlockOverAlloc)
 		} else {
-			b.dataStorage = make([]byte, 0, maxCompressedBlockSize)
+			b.dataStorage = make([]byte, 0, maxCompressedBlockSizeAlloc)
 		}
 	}
 	if cap(b.dst) <= maxSize {
@@ -360,14 +364,9 @@ func (b *blockDec) decodeLiterals(in []byte, hist *history) (remain []byte, err
 		}
 		if cap(b.literalBuf) < litRegenSize {
 			if b.lowMem {
-				b.literalBuf = make([]byte, litRegenSize)
+				b.literalBuf = make([]byte, litRegenSize, litRegenSize+compressedBlockOverAlloc)
 			} else {
-				if litRegenSize > maxCompressedLiteralSize {
-					// Exceptional
-					b.literalBuf = make([]byte, litRegenSize)
-				} else {
-					b.literalBuf = make([]byte, litRegenSize, maxCompressedLiteralSize)
-				}
+				b.literalBuf = make([]byte, litRegenSize, maxCompressedBlockSize+compressedBlockOverAlloc)
 			}
 		}
 		literals = b.literalBuf[:litRegenSize]
@@ -397,14 +396,14 @@ func (b *blockDec) decodeLiterals(in []byte, hist *history) (remain []byte, err
 		// Ensure we have space to store it.
 		if cap(b.literalBuf) < litRegenSize {
 			if b.lowMem {
-				b.literalBuf = make([]byte, 0, litRegenSize)
+				b.literalBuf = make([]byte, 0, litRegenSize+compressedBlockOverAlloc)
 			} else {
-				b.literalBuf = make([]byte, 0, maxCompressedLiteralSize)
+				b.literalBuf = make([]byte, 0, maxCompressedBlockSize+compressedBlockOverAlloc)
 			}
 		}
 		var err error
 		// Use our out buffer.
-		huff.MaxDecodedSize = maxCompressedBlockSize
+		huff.MaxDecodedSize = litRegenSize
 		if fourStreams {
 			literals, err = huff.Decoder().Decompress4X(b.literalBuf[:0:litRegenSize], literals)
 		} else {
@@ -429,9 +428,9 @@ func (b *blockDec) decodeLiterals(in []byte, hist *history) (remain []byte, err
 		// Ensure we have space to store it.
 		if cap(b.literalBuf) < litRegenSize {
 			if b.lowMem {
-				b.literalBuf = make([]byte, 0, litRegenSize)
+				b.literalBuf = make([]byte, 0, litRegenSize+compressedBlockOverAlloc)
 			} else {
-				b.literalBuf = make([]byte, 0, maxCompressedBlockSize)
+				b.literalBuf = make([]byte, 0, maxCompressedBlockSize+compressedBlockOverAlloc)
 			}
 		}
 		huff := hist.huffTree
@@ -448,7 +447,7 @@ func (b *blockDec) decodeLiterals(in []byte, hist *history) (remain []byte, err
 			return in, err
 		}
 		hist.huffTree = huff
-		huff.MaxDecodedSize = maxCompressedBlockSize
+		huff.MaxDecodedSize = litRegenSize
 		// Use our out buffer.
 		if fourStreams {
 			literals, err = huff.Decoder().Decompress4X(b.literalBuf[:0:litRegenSize], literals)
@@ -463,6 +462,8 @@ func (b *blockDec) decodeLiterals(in []byte, hist *history) (remain []byte, err
 		if len(literals) != litRegenSize {
 			return in, fmt.Errorf("literal output size mismatch want %d, got %d", litRegenSize, len(literals))
 		}
+		// Re-cap to get extra size.
+		literals = b.literalBuf[:len(literals)]
 		if debugDecoder {
 			printf("Decompressed %d literals into %d bytes\n", litCompSize, litRegenSize)
 		}
@@ -486,10 +487,15 @@ func (b *blockDec) decodeCompressed(hist *history) error {
 		b.dst = append(b.dst, hist.decoders.literals...)
 		return nil
 	}
-	err = hist.decoders.decodeSync(hist)
+	before := len(hist.decoders.out)
+	err = hist.decoders.decodeSync(hist.b[hist.ignoreBuffer:])
 	if err != nil {
 		return err
 	}
+	if hist.decoders.maxSyncLen > 0 {
+		hist.decoders.maxSyncLen += uint64(before)
+		hist.decoders.maxSyncLen -= uint64(len(hist.decoders.out))
+	}
 	b.dst = hist.decoders.out
 	hist.recentOffsets = hist.decoders.prevOffset
 	return nil
@@ -632,6 +638,22 @@ func (b *blockDec) prepareSequences(in []byte, hist *history) (err error) {
 		println("initializing sequences:", err)
 		return err
 	}
+	// Extract blocks...
+	if false && hist.dict == nil {
+		fatalErr := func(err error) {
+			if err != nil {
+				panic(err)
+			}
+		}
+		fn := fmt.Sprintf("n-%d-lits-%d-prev-%d-%d-%d-win-%d.blk", hist.decoders.nSeqs, len(hist.decoders.literals), hist.recentOffsets[0], hist.recentOffsets[1], hist.recentOffsets[2], hist.windowSize)
+		var buf bytes.Buffer
+		fatalErr(binary.Write(&buf, binary.LittleEndian, hist.decoders.litLengths.fse))
+		fatalErr(binary.Write(&buf, binary.LittleEndian, hist.decoders.matchLengths.fse))
+		fatalErr(binary.Write(&buf, binary.LittleEndian, hist.decoders.offsets.fse))
+		buf.Write(in)
+		ioutil.WriteFile(filepath.Join("testdata", "seqs", fn), buf.Bytes(), os.ModePerm)
+	}
+
 	return nil
 }
 
@@ -650,6 +672,7 @@ func (b *blockDec) decodeSequences(hist *history) error {
 	}
 	hist.decoders.windowSize = hist.windowSize
 	hist.decoders.prevOffset = hist.recentOffsets
+
 	err := hist.decoders.decode(b.sequence)
 	hist.recentOffsets = hist.decoders.prevOffset
 	return err
diff --git a/vendor/github.com/klauspost/compress/zstd/bytebuf.go b/vendor/github.com/klauspost/compress/zstd/bytebuf.go
index b80191e4..4493baa7 100644
--- a/vendor/github.com/klauspost/compress/zstd/bytebuf.go
+++ b/vendor/github.com/klauspost/compress/zstd/bytebuf.go
@@ -52,10 +52,6 @@ func (b *byteBuf) readBig(n int, dst []byte) ([]byte, error) {
 	return r, nil
 }
 
-func (b *byteBuf) remain() []byte {
-	return *b
-}
-
 func (b *byteBuf) readByte() (byte, error) {
 	bb := *b
 	if len(bb) < 1 {
diff --git a/vendor/github.com/klauspost/compress/zstd/bytereader.go b/vendor/github.com/klauspost/compress/zstd/bytereader.go
index 2c4fca17..0e59a242 100644
--- a/vendor/github.com/klauspost/compress/zstd/bytereader.go
+++ b/vendor/github.com/klauspost/compress/zstd/bytereader.go
@@ -13,12 +13,6 @@ type byteReader struct {
 	off int
 }
 
-// init will initialize the reader and set the input.
-func (b *byteReader) init(in []byte) {
-	b.b = in
-	b.off = 0
-}
-
 // advance the stream b n bytes.
 func (b *byteReader) advance(n uint) {
 	b.off += int(n)
diff --git a/vendor/github.com/klauspost/compress/zstd/decoder.go b/vendor/github.com/klauspost/compress/zstd/decoder.go
index 9fcdaac1..286c8f9d 100644
--- a/vendor/github.com/klauspost/compress/zstd/decoder.go
+++ b/vendor/github.com/klauspost/compress/zstd/decoder.go
@@ -347,18 +347,20 @@ func (d *Decoder) DecodeAll(input, dst []byte) ([]byte, error) {
 			}
 			frame.history.setDict(&dict)
 		}
-
-		if frame.FrameContentSize != fcsUnknown && frame.FrameContentSize > d.o.maxDecodedSize-uint64(len(dst)) {
-			return dst, ErrDecoderSizeExceeded
+		if frame.WindowSize > d.o.maxWindowSize {
+			return dst, ErrWindowSizeExceeded
 		}
-		if frame.FrameContentSize < 1<<30 {
-			// Never preallocate more than 1 GB up front.
+		if frame.FrameContentSize != fcsUnknown {
+			if frame.FrameContentSize > d.o.maxDecodedSize-uint64(len(dst)) {
+				return dst, ErrDecoderSizeExceeded
+			}
 			if cap(dst)-len(dst) < int(frame.FrameContentSize) {
-				dst2 := make([]byte, len(dst), len(dst)+int(frame.FrameContentSize))
+				dst2 := make([]byte, len(dst), len(dst)+int(frame.FrameContentSize)+compressedBlockOverAlloc)
 				copy(dst2, dst)
 				dst = dst2
 			}
 		}
+
 		if cap(dst) == 0 {
 			// Allocate len(input) * 2 by default if nothing is provided
 			// and we didn't get frame content size.
@@ -437,7 +439,7 @@ func (d *Decoder) nextBlock(blocking bool) (ok bool) {
 		println("got", len(d.current.b), "bytes, error:", d.current.err, "data crc:", tmp)
 	}
 
-	if len(next.b) > 0 {
+	if !d.o.ignoreChecksum && len(next.b) > 0 {
 		n, err := d.current.crc.Write(next.b)
 		if err == nil {
 			if n != len(next.b) {
@@ -449,7 +451,7 @@ func (d *Decoder) nextBlock(blocking bool) (ok bool) {
 		got := d.current.crc.Sum64()
 		var tmp [4]byte
 		binary.LittleEndian.PutUint32(tmp[:], uint32(got))
-		if !bytes.Equal(tmp[:], next.d.checkCRC) && !ignoreCRC {
+		if !d.o.ignoreChecksum && !bytes.Equal(tmp[:], next.d.checkCRC) {
 			if debugDecoder {
 				println("CRC Check Failed:", tmp[:], " (got) !=", next.d.checkCRC, "(on stream)")
 			}
@@ -533,9 +535,15 @@ func (d *Decoder) nextBlockSync() (ok bool) {
 
 		// Update/Check CRC
 		if d.frame.HasCheckSum {
-			d.frame.crc.Write(d.current.b)
+			if !d.o.ignoreChecksum {
+				d.frame.crc.Write(d.current.b)
+			}
 			if d.current.d.Last {
-				d.current.err = d.frame.checkCRC()
+				if !d.o.ignoreChecksum {
+					d.current.err = d.frame.checkCRC()
+				} else {
+					d.current.err = d.frame.consumeCRC()
+				}
 				if d.current.err != nil {
 					println("CRC error:", d.current.err)
 					return false
@@ -629,60 +637,18 @@ func (d *Decoder) startSyncDecoder(r io.Reader) error {
 
 // Create Decoder:
 // ASYNC:
-// Spawn 4 go routines.
-// 0: Read frames and decode blocks.
-// 1: Decode block and literals. Receives hufftree and seqdecs, returns seqdecs and huff tree.
-// 2: Wait for recentOffsets if needed. Decode sequences, send recentOffsets.
-// 3: Wait for stream history, execute sequences, send stream history.
+// Spawn 3 go routines.
+// 0: Read frames and decode block literals.
+// 1: Decode sequences.
+// 2: Execute sequences, send to output.
 func (d *Decoder) startStreamDecoder(ctx context.Context, r io.Reader, output chan decodeOutput) {
 	defer d.streamWg.Done()
 	br := readerWrapper{r: r}
 
-	var seqPrepare = make(chan *blockDec, d.o.concurrent)
 	var seqDecode = make(chan *blockDec, d.o.concurrent)
 	var seqExecute = make(chan *blockDec, d.o.concurrent)
 
-	// Async 1: Prepare blocks...
-	go func() {
-		var hist history
-		var hasErr bool
-		for block := range seqPrepare {
-			if hasErr {
-				if block != nil {
-					seqDecode <- block
-				}
-				continue
-			}
-			if block.async.newHist != nil {
-				if debugDecoder {
-					println("Async 1: new history")
-				}
-				hist.reset()
-				if block.async.newHist.dict != nil {
-					hist.setDict(block.async.newHist.dict)
-				}
-			}
-			if block.err != nil || block.Type != blockTypeCompressed {
-				hasErr = block.err != nil
-				seqDecode <- block
-				continue
-			}
-
-			remain, err := block.decodeLiterals(block.data, &hist)
-			block.err = err
-			hasErr = block.err != nil
-			if err == nil {
-				block.async.literals = hist.decoders.literals
-				block.async.seqData = remain
-			} else if debugDecoder {
-				println("decodeLiterals error:", err)
-			}
-			seqDecode <- block
-		}
-		close(seqDecode)
-	}()
-
-	// Async 2: Decode sequences...
+	// Async 1: Decode sequences...
 	go func() {
 		var hist history
 		var hasErr bool
@@ -696,7 +662,7 @@ func (d *Decoder) startStreamDecoder(ctx context.Context, r io.Reader, output ch
 			}
 			if block.async.newHist != nil {
 				if debugDecoder {
-					println("Async 2: new history, recent:", block.async.newHist.recentOffsets)
+					println("Async 1: new history, recent:", block.async.newHist.recentOffsets)
 				}
 				hist.decoders = block.async.newHist.decoders
 				hist.recentOffsets = block.async.newHist.recentOffsets
@@ -750,7 +716,7 @@ func (d *Decoder) startStreamDecoder(ctx context.Context, r io.Reader, output ch
 			}
 			if block.async.newHist != nil {
 				if debugDecoder {
-					println("Async 3: new history")
+					println("Async 2: new history")
 				}
 				hist.windowSize = block.async.newHist.windowSize
 				hist.allocFrameBuffer = block.async.newHist.allocFrameBuffer
@@ -837,6 +803,33 @@ func (d *Decoder) startStreamDecoder(ctx context.Context, r io.Reader, output ch
 
 decodeStream:
 	for {
+		var hist history
+		var hasErr bool
+
+		decodeBlock := func(block *blockDec) {
+			if hasErr {
+				if block != nil {
+					seqDecode <- block
+				}
+				return
+			}
+			if block.err != nil || block.Type != blockTypeCompressed {
+				hasErr = block.err != nil
+				seqDecode <- block
+				return
+			}
+
+			remain, err := block.decodeLiterals(block.data, &hist)
+			block.err = err
+			hasErr = block.err != nil
+			if err == nil {
+				block.async.literals = hist.decoders.literals
+				block.async.seqData = remain
+			} else if debugDecoder {
+				println("decodeLiterals error:", err)
+			}
+			seqDecode <- block
+		}
 		frame := d.frame
 		if debugDecoder {
 			println("New frame...")
@@ -863,7 +856,7 @@ decodeStream:
 			case <-ctx.Done():
 			case dec := <-d.decoders:
 				dec.sendErr(err)
-				seqPrepare <- dec
+				decodeBlock(dec)
 			}
 			break decodeStream
 		}
@@ -883,6 +876,10 @@ decodeStream:
 				if debugDecoder {
 					println("Alloc History:", h.allocFrameBuffer)
 				}
+				hist.reset()
+				if h.dict != nil {
+					hist.setDict(h.dict)
+				}
 				dec.async.newHist = &h
 				dec.async.fcs = frame.FrameContentSize
 				historySent = true
@@ -909,7 +906,7 @@ decodeStream:
 			}
 			err = dec.err
 			last := dec.Last
-			seqPrepare <- dec
+			decodeBlock(dec)
 			if err != nil {
 				break decodeStream
 			}
@@ -918,7 +915,7 @@ decodeStream:
 			}
 		}
 	}
-	close(seqPrepare)
+	close(seqDecode)
 	wg.Wait()
 	d.frame.history.b = frameHistCache
 }
diff --git a/vendor/github.com/klauspost/compress/zstd/decoder_options.go b/vendor/github.com/klauspost/compress/zstd/decoder_options.go
index fd05c9bb..c70e6fa0 100644
--- a/vendor/github.com/klauspost/compress/zstd/decoder_options.go
+++ b/vendor/github.com/klauspost/compress/zstd/decoder_options.go
@@ -19,6 +19,7 @@ type decoderOptions struct {
 	maxDecodedSize uint64
 	maxWindowSize  uint64
 	dicts          []dict
+	ignoreChecksum bool
 }
 
 func (o *decoderOptions) setDefault() {
@@ -31,7 +32,7 @@ func (o *decoderOptions) setDefault() {
 	if o.concurrent > 4 {
 		o.concurrent = 4
 	}
-	o.maxDecodedSize = 1 << 63
+	o.maxDecodedSize = 64 << 30
 }
 
 // WithDecoderLowmem will set whether to use a lower amount of memory,
@@ -66,7 +67,7 @@ func WithDecoderConcurrency(n int) DOption {
 // WithDecoderMaxMemory allows to set a maximum decoded size for in-memory
 // non-streaming operations or maximum window size for streaming operations.
 // This can be used to control memory usage of potentially hostile content.
-// Maximum and default is 1 << 63 bytes.
+// Maximum is 1 << 63 bytes. Default is 64GiB.
 func WithDecoderMaxMemory(n uint64) DOption {
 	return func(o *decoderOptions) error {
 		if n == 0 {
@@ -112,3 +113,11 @@ func WithDecoderMaxWindow(size uint64) DOption {
 		return nil
 	}
 }
+
+// IgnoreChecksum allows to forcibly ignore checksum checking.
+func IgnoreChecksum(b bool) DOption {
+	return func(o *decoderOptions) error {
+		o.ignoreChecksum = b
+		return nil
+	}
+}
diff --git a/vendor/github.com/klauspost/compress/zstd/enc_better.go b/vendor/github.com/klauspost/compress/zstd/enc_better.go
index 602c05ee..c769f694 100644
--- a/vendor/github.com/klauspost/compress/zstd/enc_better.go
+++ b/vendor/github.com/klauspost/compress/zstd/enc_better.go
@@ -156,8 +156,8 @@ encodeLoop:
 				panic("offset0 was 0")
 			}
 
-			nextHashS := hashLen(cv, betterShortTableBits, betterShortLen)
 			nextHashL := hashLen(cv, betterLongTableBits, betterLongLen)
+			nextHashS := hashLen(cv, betterShortTableBits, betterShortLen)
 			candidateL := e.longTable[nextHashL]
 			candidateS := e.table[nextHashS]
 
@@ -518,8 +518,8 @@ encodeLoop:
 			}
 
 			// Store this, since we have it.
-			nextHashS := hashLen(cv, betterShortTableBits, betterShortLen)
 			nextHashL := hashLen(cv, betterLongTableBits, betterLongLen)
+			nextHashS := hashLen(cv, betterShortTableBits, betterShortLen)
 
 			// We have at least 4 byte match.
 			// No need to check backwards. We come straight from a match
@@ -674,8 +674,8 @@ encodeLoop:
 				panic("offset0 was 0")
 			}
 
-			nextHashS := hashLen(cv, betterShortTableBits, betterShortLen)
 			nextHashL := hashLen(cv, betterLongTableBits, betterLongLen)
+			nextHashS := hashLen(cv, betterShortTableBits, betterShortLen)
 			candidateL := e.longTable[nextHashL]
 			candidateS := e.table[nextHashS]
 
@@ -1047,8 +1047,8 @@ encodeLoop:
 			}
 
 			// Store this, since we have it.
-			nextHashS := hashLen(cv, betterShortTableBits, betterShortLen)
 			nextHashL := hashLen(cv, betterLongTableBits, betterLongLen)
+			nextHashS := hashLen(cv, betterShortTableBits, betterShortLen)
 
 			// We have at least 4 byte match.
 			// No need to check backwards. We come straight from a match
diff --git a/vendor/github.com/klauspost/compress/zstd/enc_dfast.go b/vendor/github.com/klauspost/compress/zstd/enc_dfast.go
index d6b31042..7ff0c64f 100644
--- a/vendor/github.com/klauspost/compress/zstd/enc_dfast.go
+++ b/vendor/github.com/klauspost/compress/zstd/enc_dfast.go
@@ -127,8 +127,8 @@ encodeLoop:
 				panic("offset0 was 0")
 			}
 
-			nextHashS := hashLen(cv, dFastShortTableBits, dFastShortLen)
 			nextHashL := hashLen(cv, dFastLongTableBits, dFastLongLen)
+			nextHashS := hashLen(cv, dFastShortTableBits, dFastShortLen)
 			candidateL := e.longTable[nextHashL]
 			candidateS := e.table[nextHashS]
 
@@ -439,8 +439,8 @@ encodeLoop:
 		var t int32
 		for {
 
-			nextHashS := hashLen(cv, dFastShortTableBits, dFastShortLen)
 			nextHashL := hashLen(cv, dFastLongTableBits, dFastLongLen)
+			nextHashS := hashLen(cv, dFastShortTableBits, dFastShortLen)
 			candidateL := e.longTable[nextHashL]
 			candidateS := e.table[nextHashS]
 
@@ -785,8 +785,8 @@ encodeLoop:
 				panic("offset0 was 0")
 			}
 
-			nextHashS := hashLen(cv, dFastShortTableBits, dFastShortLen)
 			nextHashL := hashLen(cv, dFastLongTableBits, dFastLongLen)
+			nextHashS := hashLen(cv, dFastShortTableBits, dFastShortLen)
 			candidateL := e.longTable[nextHashL]
 			candidateS := e.table[nextHashS]
 
@@ -969,7 +969,7 @@ encodeLoop:
 		te0 := tableEntry{offset: index0 + e.cur, val: uint32(cv0)}
 		te1 := tableEntry{offset: index1 + e.cur, val: uint32(cv1)}
 		longHash1 := hashLen(cv0, dFastLongTableBits, dFastLongLen)
-		longHash2 := hashLen(cv0, dFastLongTableBits, dFastLongLen)
+		longHash2 := hashLen(cv1, dFastLongTableBits, dFastLongLen)
 		e.longTable[longHash1] = te0
 		e.longTable[longHash2] = te1
 		e.markLongShardDirty(longHash1)
@@ -1002,8 +1002,8 @@ encodeLoop:
 			}
 
 			// Store this, since we have it.
-			nextHashS := hashLen(cv, dFastShortTableBits, dFastShortLen)
 			nextHashL := hashLen(cv, dFastLongTableBits, dFastLongLen)
+			nextHashS := hashLen(cv, dFastShortTableBits, dFastShortLen)
 
 			// We have at least 4 byte match.
 			// No need to check backwards. We come straight from a match
diff --git a/vendor/github.com/klauspost/compress/zstd/encoder.go b/vendor/github.com/klauspost/compress/zstd/encoder.go
index dcc987a7..e6b1d01c 100644
--- a/vendor/github.com/klauspost/compress/zstd/encoder.go
+++ b/vendor/github.com/klauspost/compress/zstd/encoder.go
@@ -551,7 +551,7 @@ func (e *Encoder) EncodeAll(src, dst []byte) []byte {
 	}
 
 	// If we can do everything in one block, prefer that.
-	if len(src) <= maxCompressedBlockSize {
+	if len(src) <= e.o.blockSize {
 		enc.Reset(e.o.dict, true)
 		// Slightly faster with no history and everything in one block.
 		if e.o.crc {
diff --git a/vendor/github.com/klauspost/compress/zstd/framedec.go b/vendor/github.com/klauspost/compress/zstd/framedec.go
index 11089d22..fa0a633f 100644
--- a/vendor/github.com/klauspost/compress/zstd/framedec.go
+++ b/vendor/github.com/klauspost/compress/zstd/framedec.go
@@ -253,10 +253,11 @@ func (d *frameDec) reset(br byteBuffer) error {
 		return ErrWindowSizeTooSmall
 	}
 	d.history.windowSize = int(d.WindowSize)
-	if d.o.lowMem && d.history.windowSize < maxBlockSize {
+	if !d.o.lowMem || d.history.windowSize < maxBlockSize {
+		// Alloc 2x window size if not low-mem, or very small window size.
 		d.history.allocFrameBuffer = d.history.windowSize * 2
-		// TODO: Maybe use FrameContent size
 	} else {
+		// Alloc with one additional block
 		d.history.allocFrameBuffer = d.history.windowSize + maxBlockSize
 	}
 
@@ -290,13 +291,6 @@ func (d *frameDec) checkCRC() error {
 	if !d.HasCheckSum {
 		return nil
 	}
-	var tmp [4]byte
-	got := d.crc.Sum64()
-	// Flip to match file order.
-	tmp[0] = byte(got >> 0)
-	tmp[1] = byte(got >> 8)
-	tmp[2] = byte(got >> 16)
-	tmp[3] = byte(got >> 24)
 
 	// We can overwrite upper tmp now
 	want, err := d.rawInput.readSmall(4)
@@ -305,7 +299,19 @@ func (d *frameDec) checkCRC() error {
 		return err
 	}
 
-	if !bytes.Equal(tmp[:], want) && !ignoreCRC {
+	if d.o.ignoreChecksum {
+		return nil
+	}
+
+	var tmp [4]byte
+	got := d.crc.Sum64()
+	// Flip to match file order.
+	tmp[0] = byte(got >> 0)
+	tmp[1] = byte(got >> 8)
+	tmp[2] = byte(got >> 16)
+	tmp[3] = byte(got >> 24)
+
+	if !bytes.Equal(tmp[:], want) {
 		if debugDecoder {
 			println("CRC Check Failed:", tmp[:], "!=", want)
 		}
@@ -317,6 +323,19 @@ func (d *frameDec) checkCRC() error {
 	return nil
 }
 
+// consumeCRC reads the checksum data if the frame has one.
+func (d *frameDec) consumeCRC() error {
+	if d.HasCheckSum {
+		_, err := d.rawInput.readSmall(4)
+		if err != nil {
+			println("CRC missing?", err)
+			return err
+		}
+	}
+
+	return nil
+}
+
 // runDecoder will create a sync decoder that will decode a block of data.
 func (d *frameDec) runDecoder(dst []byte, dec *blockDec) ([]byte, error) {
 	saved := d.history.b
@@ -326,6 +345,19 @@ func (d *frameDec) runDecoder(dst []byte, dec *blockDec) ([]byte, error) {
 	d.history.ignoreBuffer = len(dst)
 	// Store input length, so we only check new data.
 	crcStart := len(dst)
+	d.history.decoders.maxSyncLen = 0
+	if d.FrameContentSize != fcsUnknown {
+		d.history.decoders.maxSyncLen = d.FrameContentSize + uint64(len(dst))
+		if d.history.decoders.maxSyncLen > d.o.maxDecodedSize {
+			return dst, ErrDecoderSizeExceeded
+		}
+		if uint64(cap(dst)) < d.history.decoders.maxSyncLen {
+			// Alloc for output
+			dst2 := make([]byte, len(dst), d.history.decoders.maxSyncLen+compressedBlockOverAlloc)
+			copy(dst2, dst)
+			dst = dst2
+		}
+	}
 	var err error
 	for {
 		err = dec.reset(d.rawInput, d.WindowSize)
@@ -360,13 +392,17 @@ func (d *frameDec) runDecoder(dst []byte, dec *blockDec) ([]byte, error) {
 		if d.FrameContentSize != fcsUnknown && uint64(len(d.history.b)-crcStart) != d.FrameContentSize {
 			err = ErrFrameSizeMismatch
 		} else if d.HasCheckSum {
-			var n int
-			n, err = d.crc.Write(dst[crcStart:])
-			if err == nil {
-				if n != len(dst)-crcStart {
-					err = io.ErrShortWrite
-				} else {
-					err = d.checkCRC()
+			if d.o.ignoreChecksum {
+				err = d.consumeCRC()
+			} else {
+				var n int
+				n, err = d.crc.Write(dst[crcStart:])
+				if err == nil {
+					if n != len(dst)-crcStart {
+						err = io.ErrShortWrite
+					} else {
+						err = d.checkCRC()
+					}
 				}
 			}
 		}
diff --git a/vendor/github.com/klauspost/compress/zstd/fse_decoder.go b/vendor/github.com/klauspost/compress/zstd/fse_decoder.go
index bb3d4fd6..23333b96 100644
--- a/vendor/github.com/klauspost/compress/zstd/fse_decoder.go
+++ b/vendor/github.com/klauspost/compress/zstd/fse_decoder.go
@@ -5,8 +5,10 @@
 package zstd
 
 import (
+	"encoding/binary"
 	"errors"
 	"fmt"
+	"io"
 )
 
 const (
@@ -182,6 +184,29 @@ func (s *fseDecoder) readNCount(b *byteReader, maxSymbol uint16) error {
 	return s.buildDtable()
 }
 
+func (s *fseDecoder) mustReadFrom(r io.Reader) {
+	fatalErr := func(err error) {
+		if err != nil {
+			panic(err)
+		}
+	}
+	// 	dt             [maxTablesize]decSymbol // Decompression table.
+	//	symbolLen      uint16                  // Length of active part of the symbol table.
+	//	actualTableLog uint8                   // Selected tablelog.
+	//	maxBits        uint8                   // Maximum number of additional bits
+	//	// used for table creation to avoid allocations.
+	//	stateTable [256]uint16
+	//	norm       [maxSymbolValue + 1]int16
+	//	preDefined bool
+	fatalErr(binary.Read(r, binary.LittleEndian, &s.dt))
+	fatalErr(binary.Read(r, binary.LittleEndian, &s.symbolLen))
+	fatalErr(binary.Read(r, binary.LittleEndian, &s.actualTableLog))
+	fatalErr(binary.Read(r, binary.LittleEndian, &s.maxBits))
+	fatalErr(binary.Read(r, binary.LittleEndian, &s.stateTable))
+	fatalErr(binary.Read(r, binary.LittleEndian, &s.norm))
+	fatalErr(binary.Read(r, binary.LittleEndian, &s.preDefined))
+}
+
 // decSymbol contains information about a state entry,
 // Including the state offset base, the output symbol and
 // the number of bits to read for the low part of the destination state.
@@ -204,18 +229,10 @@ func (d decSymbol) newState() uint16 {
 	return uint16(d >> 16)
 }
 
-func (d decSymbol) baseline() uint32 {
-	return uint32(d >> 32)
-}
-
 func (d decSymbol) baselineInt() int {
 	return int(d >> 32)
 }
 
-func (d *decSymbol) set(nbits, addBits uint8, newState uint16, baseline uint32) {
-	*d = decSymbol(nbits) | (decSymbol(addBits) << 8) | (decSymbol(newState) << 16) | (decSymbol(baseline) << 32)
-}
-
 func (d *decSymbol) setNBits(nBits uint8) {
 	const mask = 0xffffffffffffff00
 	*d = (*d & mask) | decSymbol(nBits)
@@ -231,11 +248,6 @@ func (d *decSymbol) setNewState(state uint16) {
 	*d = (*d & mask) | decSymbol(state)<<16
 }
 
-func (d *decSymbol) setBaseline(baseline uint32) {
-	const mask = 0xffffffff
-	*d = (*d & mask) | decSymbol(baseline)<<32
-}
-
 func (d *decSymbol) setExt(addBits uint8, baseline uint32) {
 	const mask = 0xffff00ff
 	*d = (*d & mask) | (decSymbol(addBits) << 8) | (decSymbol(baseline) << 32)
@@ -352,34 +364,7 @@ func (s *fseState) init(br *bitReader, tableLog uint8, dt []decSymbol) {
 	s.state = dt[br.getBits(tableLog)]
 }
 
-// next returns the current symbol and sets the next state.
-// At least tablelog bits must be available in the bit reader.
-func (s *fseState) next(br *bitReader) {
-	lowBits := uint16(br.getBits(s.state.nbBits()))
-	s.state = s.dt[s.state.newState()+lowBits]
-}
-
-// finished returns true if all bits have been read from the bitstream
-// and the next state would require reading bits from the input.
-func (s *fseState) finished(br *bitReader) bool {
-	return br.finished() && s.state.nbBits() > 0
-}
-
-// final returns the current state symbol without decoding the next.
-func (s *fseState) final() (int, uint8) {
-	return s.state.baselineInt(), s.state.addBits()
-}
-
 // final returns the current state symbol without decoding the next.
 func (s decSymbol) final() (int, uint8) {
 	return s.baselineInt(), s.addBits()
 }
-
-// nextFast returns the next symbol and sets the next state.
-// This can only be used if no symbols are 0 bits.
-// At least tablelog bits must be available in the bit reader.
-func (s *fseState) nextFast(br *bitReader) (uint32, uint8) {
-	lowBits := br.get16BitsFast(s.state.nbBits())
-	s.state = s.dt[s.state.newState()+lowBits]
-	return s.state.baseline(), s.state.addBits()
-}
diff --git a/vendor/github.com/klauspost/compress/zstd/fse_encoder.go b/vendor/github.com/klauspost/compress/zstd/fse_encoder.go
index 5442061b..ab26326a 100644
--- a/vendor/github.com/klauspost/compress/zstd/fse_encoder.go
+++ b/vendor/github.com/klauspost/compress/zstd/fse_encoder.go
@@ -76,21 +76,6 @@ func (s *fseEncoder) HistogramFinished(maxSymbol uint8, maxCount int) {
 	s.clearCount = maxCount != 0
 }
 
-// prepare will prepare and allocate scratch tables used for both compression and decompression.
-func (s *fseEncoder) prepare() (*fseEncoder, error) {
-	if s == nil {
-		s = &fseEncoder{}
-	}
-	s.useRLE = false
-	if s.clearCount && s.maxCount == 0 {
-		for i := range s.count {
-			s.count[i] = 0
-		}
-		s.clearCount = false
-	}
-	return s, nil
-}
-
 // allocCtable will allocate tables needed for compression.
 // If existing tables a re big enough, they are simply re-used.
 func (s *fseEncoder) allocCtable() {
@@ -709,14 +694,6 @@ func (c *cState) init(bw *bitWriter, ct *cTable, first symbolTransform) {
 	c.state = c.stateTable[lu]
 }
 
-// encode the output symbol provided and write it to the bitstream.
-func (c *cState) encode(symbolTT symbolTransform) {
-	nbBitsOut := (uint32(c.state) + symbolTT.deltaNbBits) >> 16
-	dstState := int32(c.state>>(nbBitsOut&15)) + int32(symbolTT.deltaFindState)
-	c.bw.addBits16NC(c.state, uint8(nbBitsOut))
-	c.state = c.stateTable[dstState]
-}
-
 // flush will write the tablelog to the output and flush the remaining full bytes.
 func (c *cState) flush(tableLog uint8) {
 	c.bw.flush32()
diff --git a/vendor/github.com/klauspost/compress/zstd/fuzz.go b/vendor/github.com/klauspost/compress/zstd/fuzz.go
deleted file mode 100644
index 7f2210e0..00000000
--- a/vendor/github.com/klauspost/compress/zstd/fuzz.go
+++ /dev/null
@@ -1,11 +0,0 @@
-//go:build ignorecrc
-// +build ignorecrc
-
-// Copyright 2019+ Klaus Post. All rights reserved.
-// License information can be found in the LICENSE file.
-// Based on work by Yann Collet, released under BSD License.
-
-package zstd
-
-// ignoreCRC can be used for fuzz testing to ignore CRC values...
-const ignoreCRC = true
diff --git a/vendor/github.com/klauspost/compress/zstd/fuzz_none.go b/vendor/github.com/klauspost/compress/zstd/fuzz_none.go
deleted file mode 100644
index 6811c68a..00000000
--- a/vendor/github.com/klauspost/compress/zstd/fuzz_none.go
+++ /dev/null
@@ -1,11 +0,0 @@
-//go:build !ignorecrc
-// +build !ignorecrc
-
-// Copyright 2019+ Klaus Post. All rights reserved.
-// License information can be found in the LICENSE file.
-// Based on work by Yann Collet, released under BSD License.
-
-package zstd
-
-// ignoreCRC can be used for fuzz testing to ignore CRC values...
-const ignoreCRC = false
diff --git a/vendor/github.com/klauspost/compress/zstd/hash.go b/vendor/github.com/klauspost/compress/zstd/hash.go
index cf33f29a..5d73c21e 100644
--- a/vendor/github.com/klauspost/compress/zstd/hash.go
+++ b/vendor/github.com/klauspost/compress/zstd/hash.go
@@ -33,9 +33,3 @@ func hashLen(u uint64, length, mls uint8) uint32 {
 		return (uint32(u) * prime4bytes) >> (32 - length)
 	}
 }
-
-// hash3 returns the hash of the lower 3 bytes of u to fit in a hash table with h bits.
-// Preferably h should be a constant and should always be <32.
-func hash3(u uint32, h uint8) uint32 {
-	return ((u << (32 - 24)) * prime3bytes) >> ((32 - h) & 31)
-}
diff --git a/vendor/github.com/klauspost/compress/zstd/seqdec.go b/vendor/github.com/klauspost/compress/zstd/seqdec.go
index 819f1461..df044720 100644
--- a/vendor/github.com/klauspost/compress/zstd/seqdec.go
+++ b/vendor/github.com/klauspost/compress/zstd/seqdec.go
@@ -73,6 +73,7 @@ type sequenceDecs struct {
 	seqSize      int
 	windowSize   int
 	maxBits      uint8
+	maxSyncLen   uint64
 }
 
 // initialize all 3 decoders from the stream input.
@@ -98,153 +99,13 @@ func (s *sequenceDecs) initialize(br *bitReader, hist *history, out []byte) erro
 	return nil
 }
 
-// decode sequences from the stream with the provided history.
-func (s *sequenceDecs) decode(seqs []seqVals) error {
-	br := s.br
-
-	// Grab full sizes tables, to avoid bounds checks.
-	llTable, mlTable, ofTable := s.litLengths.fse.dt[:maxTablesize], s.matchLengths.fse.dt[:maxTablesize], s.offsets.fse.dt[:maxTablesize]
-	llState, mlState, ofState := s.litLengths.state.state, s.matchLengths.state.state, s.offsets.state.state
-	s.seqSize = 0
-	litRemain := len(s.literals)
-	maxBlockSize := maxCompressedBlockSize
-	if s.windowSize < maxBlockSize {
-		maxBlockSize = s.windowSize
-	}
-	for i := range seqs {
-		var ll, mo, ml int
-		if br.off > 4+((maxOffsetBits+16+16)>>3) {
-			// inlined function:
-			// ll, mo, ml = s.nextFast(br, llState, mlState, ofState)
-
-			// Final will not read from stream.
-			var llB, mlB, moB uint8
-			ll, llB = llState.final()
-			ml, mlB = mlState.final()
-			mo, moB = ofState.final()
-
-			// extra bits are stored in reverse order.
-			br.fillFast()
-			mo += br.getBits(moB)
-			if s.maxBits > 32 {
-				br.fillFast()
-			}
-			ml += br.getBits(mlB)
-			ll += br.getBits(llB)
-
-			if moB > 1 {
-				s.prevOffset[2] = s.prevOffset[1]
-				s.prevOffset[1] = s.prevOffset[0]
-				s.prevOffset[0] = mo
-			} else {
-				// mo = s.adjustOffset(mo, ll, moB)
-				// Inlined for rather big speedup
-				if ll == 0 {
-					// There is an exception though, when current sequence's literals_length = 0.
-					// In this case, repeated offsets are shifted by one, so an offset_value of 1 means Repeated_Offset2,
-					// an offset_value of 2 means Repeated_Offset3, and an offset_value of 3 means Repeated_Offset1 - 1_byte.
-					mo++
-				}
-
-				if mo == 0 {
-					mo = s.prevOffset[0]
-				} else {
-					var temp int
-					if mo == 3 {
-						temp = s.prevOffset[0] - 1
-					} else {
-						temp = s.prevOffset[mo]
-					}
-
-					if temp == 0 {
-						// 0 is not valid; input is corrupted; force offset to 1
-						println("WARNING: temp was 0")
-						temp = 1
-					}
-
-					if mo != 1 {
-						s.prevOffset[2] = s.prevOffset[1]
-					}
-					s.prevOffset[1] = s.prevOffset[0]
-					s.prevOffset[0] = temp
-					mo = temp
-				}
-			}
-			br.fillFast()
-		} else {
-			if br.overread() {
-				if debugDecoder {
-					printf("reading sequence %d, exceeded available data\n", i)
-				}
-				return io.ErrUnexpectedEOF
-			}
-			ll, mo, ml = s.next(br, llState, mlState, ofState)
-			br.fill()
-		}
-
-		if debugSequences {
-			println("Seq", i, "Litlen:", ll, "mo:", mo, "(abs) ml:", ml)
-		}
-		// Evaluate.
-		// We might be doing this async, so do it early.
-		if mo == 0 && ml > 0 {
-			return fmt.Errorf("zero matchoff and matchlen (%d) > 0", ml)
-		}
-		if ml > maxMatchLen {
-			return fmt.Errorf("match len (%d) bigger than max allowed length", ml)
-		}
-		s.seqSize += ll + ml
-		if s.seqSize > maxBlockSize {
-			return fmt.Errorf("output (%d) bigger than max block size (%d)", s.seqSize, maxBlockSize)
-		}
-		litRemain -= ll
-		if litRemain < 0 {
-			return fmt.Errorf("unexpected literal count, want %d bytes, but only %d is available", ll, litRemain+ll)
-		}
-		seqs[i] = seqVals{
-			ll: ll,
-			ml: ml,
-			mo: mo,
-		}
-		if i == len(seqs)-1 {
-			// This is the last sequence, so we shouldn't update state.
-			break
-		}
-
-		// Manually inlined, ~ 5-20% faster
-		// Update all 3 states at once. Approx 20% faster.
-		nBits := llState.nbBits() + mlState.nbBits() + ofState.nbBits()
-		if nBits == 0 {
-			llState = llTable[llState.newState()&maxTableMask]
-			mlState = mlTable[mlState.newState()&maxTableMask]
-			ofState = ofTable[ofState.newState()&maxTableMask]
-		} else {
-			bits := br.get32BitsFast(nBits)
-			lowBits := uint16(bits >> ((ofState.nbBits() + mlState.nbBits()) & 31))
-			llState = llTable[(llState.newState()+lowBits)&maxTableMask]
-
-			lowBits = uint16(bits >> (ofState.nbBits() & 31))
-			lowBits &= bitMask[mlState.nbBits()&15]
-			mlState = mlTable[(mlState.newState()+lowBits)&maxTableMask]
-
-			lowBits = uint16(bits) & bitMask[ofState.nbBits()&15]
-			ofState = ofTable[(ofState.newState()+lowBits)&maxTableMask]
-		}
-	}
-	s.seqSize += litRemain
-	if s.seqSize > maxBlockSize {
-		return fmt.Errorf("output (%d) bigger than max block size (%d)", s.seqSize, maxBlockSize)
-	}
-	err := br.close()
-	if err != nil {
-		printf("Closing sequences: %v, %+v\n", err, *br)
-	}
-	return err
-}
-
 // execute will execute the decoded sequence with the provided history.
 // The sequence must be evaluated before being sent.
 func (s *sequenceDecs) execute(seqs []seqVals, hist []byte) error {
+	if len(s.dict) == 0 {
+		return s.executeSimple(seqs, hist)
+	}
+
 	// Ensure we have enough output size...
 	if len(s.out)+s.seqSize > cap(s.out) {
 		addBytes := s.seqSize + len(s.out)
@@ -327,6 +188,7 @@ func (s *sequenceDecs) execute(seqs []seqVals, hist []byte) error {
 			}
 		}
 	}
+
 	// Add final literals
 	copy(out[t:], s.literals)
 	if debugDecoder {
@@ -341,14 +203,18 @@ func (s *sequenceDecs) execute(seqs []seqVals, hist []byte) error {
 }
 
 // decode sequences from the stream with the provided history.
-func (s *sequenceDecs) decodeSync(history *history) error {
+func (s *sequenceDecs) decodeSync(hist []byte) error {
+	supported, err := s.decodeSyncSimple(hist)
+	if supported {
+		return err
+	}
+
 	br := s.br
 	seqs := s.nSeqs
 	startSize := len(s.out)
 	// Grab full sizes tables, to avoid bounds checks.
 	llTable, mlTable, ofTable := s.litLengths.fse.dt[:maxTablesize], s.matchLengths.fse.dt[:maxTablesize], s.offsets.fse.dt[:maxTablesize]
 	llState, mlState, ofState := s.litLengths.state.state, s.matchLengths.state.state, s.offsets.state.state
-	hist := history.b[history.ignoreBuffer:]
 	out := s.out
 	maxBlockSize := maxCompressedBlockSize
 	if s.windowSize < maxBlockSize {
@@ -433,7 +299,7 @@ func (s *sequenceDecs) decodeSync(history *history) error {
 		}
 		size := ll + ml + len(out)
 		if size-startSize > maxBlockSize {
-			return fmt.Errorf("output (%d) bigger than max block size (%d)", size, maxBlockSize)
+			return fmt.Errorf("output (%d) bigger than max block size (%d)", size-startSize, maxBlockSize)
 		}
 		if size > cap(out) {
 			// Not enough size, which can happen under high volume block streaming conditions
@@ -463,13 +329,13 @@ func (s *sequenceDecs) decodeSync(history *history) error {
 
 		if mo > len(out)+len(hist) || mo > s.windowSize {
 			if len(s.dict) == 0 {
-				return fmt.Errorf("match offset (%d) bigger than current history (%d)", mo, len(out)+len(hist))
+				return fmt.Errorf("match offset (%d) bigger than current history (%d)", mo, len(out)+len(hist)-startSize)
 			}
 
 			// we may be in dictionary.
 			dictO := len(s.dict) - (mo - (len(out) + len(hist)))
 			if dictO < 0 || dictO >= len(s.dict) {
-				return fmt.Errorf("match offset (%d) bigger than current history (%d)", mo, len(out)+len(hist))
+				return fmt.Errorf("match offset (%d) bigger than current history (%d)", mo, len(out)+len(hist)-startSize)
 			}
 			end := dictO + ml
 			if end > len(s.dict) {
@@ -530,6 +396,7 @@ func (s *sequenceDecs) decodeSync(history *history) error {
 			ofState = ofTable[ofState.newState()&maxTableMask]
 		} else {
 			bits := br.get32BitsFast(nBits)
+
 			lowBits := uint16(bits >> ((ofState.nbBits() + mlState.nbBits()) & 31))
 			llState = llTable[(llState.newState()+lowBits)&maxTableMask]
 
@@ -543,8 +410,8 @@ func (s *sequenceDecs) decodeSync(history *history) error {
 	}
 
 	// Check if space for literals
-	if len(s.literals)+len(s.out)-startSize > maxBlockSize {
-		return fmt.Errorf("output (%d) bigger than max block size (%d)", len(s.out), maxBlockSize)
+	if size := len(s.literals) + len(s.out) - startSize; size > maxBlockSize {
+		return fmt.Errorf("output (%d) bigger than max block size (%d)", size, maxBlockSize)
 	}
 
 	// Add final literals
@@ -552,16 +419,6 @@ func (s *sequenceDecs) decodeSync(history *history) error {
 	return br.close()
 }
 
-// update states, at least 27 bits must be available.
-func (s *sequenceDecs) update(br *bitReader) {
-	// Max 8 bits
-	s.litLengths.state.next(br)
-	// Max 9 bits
-	s.matchLengths.state.next(br)
-	// Max 8 bits
-	s.offsets.state.next(br)
-}
-
 var bitMask [16]uint16
 
 func init() {
@@ -570,87 +427,6 @@ func init() {
 	}
 }
 
-// update states, at least 27 bits must be available.
-func (s *sequenceDecs) updateAlt(br *bitReader) {
-	// Update all 3 states at once. Approx 20% faster.
-	a, b, c := s.litLengths.state.state, s.matchLengths.state.state, s.offsets.state.state
-
-	nBits := a.nbBits() + b.nbBits() + c.nbBits()
-	if nBits == 0 {
-		s.litLengths.state.state = s.litLengths.state.dt[a.newState()]
-		s.matchLengths.state.state = s.matchLengths.state.dt[b.newState()]
-		s.offsets.state.state = s.offsets.state.dt[c.newState()]
-		return
-	}
-	bits := br.get32BitsFast(nBits)
-	lowBits := uint16(bits >> ((c.nbBits() + b.nbBits()) & 31))
-	s.litLengths.state.state = s.litLengths.state.dt[a.newState()+lowBits]
-
-	lowBits = uint16(bits >> (c.nbBits() & 31))
-	lowBits &= bitMask[b.nbBits()&15]
-	s.matchLengths.state.state = s.matchLengths.state.dt[b.newState()+lowBits]
-
-	lowBits = uint16(bits) & bitMask[c.nbBits()&15]
-	s.offsets.state.state = s.offsets.state.dt[c.newState()+lowBits]
-}
-
-// nextFast will return new states when there are at least 4 unused bytes left on the stream when done.
-func (s *sequenceDecs) nextFast(br *bitReader, llState, mlState, ofState decSymbol) (ll, mo, ml int) {
-	// Final will not read from stream.
-	ll, llB := llState.final()
-	ml, mlB := mlState.final()
-	mo, moB := ofState.final()
-
-	// extra bits are stored in reverse order.
-	br.fillFast()
-	mo += br.getBits(moB)
-	if s.maxBits > 32 {
-		br.fillFast()
-	}
-	ml += br.getBits(mlB)
-	ll += br.getBits(llB)
-
-	if moB > 1 {
-		s.prevOffset[2] = s.prevOffset[1]
-		s.prevOffset[1] = s.prevOffset[0]
-		s.prevOffset[0] = mo
-		return
-	}
-	// mo = s.adjustOffset(mo, ll, moB)
-	// Inlined for rather big speedup
-	if ll == 0 {
-		// There is an exception though, when current sequence's literals_length = 0.
-		// In this case, repeated offsets are shifted by one, so an offset_value of 1 means Repeated_Offset2,
-		// an offset_value of 2 means Repeated_Offset3, and an offset_value of 3 means Repeated_Offset1 - 1_byte.
-		mo++
-	}
-
-	if mo == 0 {
-		mo = s.prevOffset[0]
-		return
-	}
-	var temp int
-	if mo == 3 {
-		temp = s.prevOffset[0] - 1
-	} else {
-		temp = s.prevOffset[mo]
-	}
-
-	if temp == 0 {
-		// 0 is not valid; input is corrupted; force offset to 1
-		println("temp was 0")
-		temp = 1
-	}
-
-	if mo != 1 {
-		s.prevOffset[2] = s.prevOffset[1]
-	}
-	s.prevOffset[1] = s.prevOffset[0]
-	s.prevOffset[0] = temp
-	mo = temp
-	return
-}
-
 func (s *sequenceDecs) next(br *bitReader, llState, mlState, ofState decSymbol) (ll, mo, ml int) {
 	// Final will not read from stream.
 	ll, llB := llState.final()
diff --git a/vendor/github.com/klauspost/compress/zstd/seqdec_amd64.go b/vendor/github.com/klauspost/compress/zstd/seqdec_amd64.go
new file mode 100644
index 00000000..847b322a
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/zstd/seqdec_amd64.go
@@ -0,0 +1,362 @@
+//go:build amd64 && !appengine && !noasm && gc
+// +build amd64,!appengine,!noasm,gc
+
+package zstd
+
+import (
+	"fmt"
+
+	"github.com/klauspost/compress/internal/cpuinfo"
+)
+
+type decodeSyncAsmContext struct {
+	llTable     []decSymbol
+	mlTable     []decSymbol
+	ofTable     []decSymbol
+	llState     uint64
+	mlState     uint64
+	ofState     uint64
+	iteration   int
+	litRemain   int
+	out         []byte
+	outPosition int
+	literals    []byte
+	litPosition int
+	history     []byte
+	windowSize  int
+	ll          int // set on error (not for all errors, please refer to _generate/gen.go)
+	ml          int // set on error (not for all errors, please refer to _generate/gen.go)
+	mo          int // set on error (not for all errors, please refer to _generate/gen.go)
+}
+
+// sequenceDecs_decodeSync_amd64 implements the main loop of sequenceDecs.decodeSync in x86 asm.
+//
+// Please refer to seqdec_generic.go for the reference implementation.
+//go:noescape
+func sequenceDecs_decodeSync_amd64(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int
+
+// sequenceDecs_decodeSync_bmi2 implements the main loop of sequenceDecs.decodeSync in x86 asm with BMI2 extensions.
+//go:noescape
+func sequenceDecs_decodeSync_bmi2(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int
+
+// sequenceDecs_decodeSync_safe_amd64 does the same as above, but does not write more than output buffer.
+//go:noescape
+func sequenceDecs_decodeSync_safe_amd64(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int
+
+// sequenceDecs_decodeSync_safe_bmi2 does the same as above, but does not write more than output buffer.
+//go:noescape
+func sequenceDecs_decodeSync_safe_bmi2(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int
+
+// decode sequences from the stream with the provided history but without a dictionary.
+func (s *sequenceDecs) decodeSyncSimple(hist []byte) (bool, error) {
+	if len(s.dict) > 0 {
+		return false, nil
+	}
+	if s.maxSyncLen == 0 && cap(s.out)-len(s.out) < maxCompressedBlockSize {
+		return false, nil
+	}
+	useSafe := false
+	if s.maxSyncLen == 0 && cap(s.out)-len(s.out) < maxCompressedBlockSizeAlloc {
+		useSafe = true
+	}
+	if s.maxSyncLen > 0 && cap(s.out)-len(s.out)-compressedBlockOverAlloc < int(s.maxSyncLen) {
+		useSafe = true
+	}
+	if cap(s.literals) < len(s.literals)+compressedBlockOverAlloc {
+		useSafe = true
+	}
+
+	br := s.br
+
+	maxBlockSize := maxCompressedBlockSize
+	if s.windowSize < maxBlockSize {
+		maxBlockSize = s.windowSize
+	}
+
+	ctx := decodeSyncAsmContext{
+		llTable:     s.litLengths.fse.dt[:maxTablesize],
+		mlTable:     s.matchLengths.fse.dt[:maxTablesize],
+		ofTable:     s.offsets.fse.dt[:maxTablesize],
+		llState:     uint64(s.litLengths.state.state),
+		mlState:     uint64(s.matchLengths.state.state),
+		ofState:     uint64(s.offsets.state.state),
+		iteration:   s.nSeqs - 1,
+		litRemain:   len(s.literals),
+		out:         s.out,
+		outPosition: len(s.out),
+		literals:    s.literals,
+		windowSize:  s.windowSize,
+		history:     hist,
+	}
+
+	s.seqSize = 0
+	startSize := len(s.out)
+
+	var errCode int
+	if cpuinfo.HasBMI2() {
+		if useSafe {
+			errCode = sequenceDecs_decodeSync_safe_bmi2(s, br, &ctx)
+		} else {
+			errCode = sequenceDecs_decodeSync_bmi2(s, br, &ctx)
+		}
+	} else {
+		if useSafe {
+			errCode = sequenceDecs_decodeSync_safe_amd64(s, br, &ctx)
+		} else {
+			errCode = sequenceDecs_decodeSync_amd64(s, br, &ctx)
+		}
+	}
+	switch errCode {
+	case noError:
+		break
+
+	case errorMatchLenOfsMismatch:
+		return true, fmt.Errorf("zero matchoff and matchlen (%d) > 0", ctx.ml)
+
+	case errorMatchLenTooBig:
+		return true, fmt.Errorf("match len (%d) bigger than max allowed length", ctx.ml)
+
+	case errorMatchOffTooBig:
+		return true, fmt.Errorf("match offset (%d) bigger than current history (%d)",
+			ctx.mo, ctx.outPosition+len(hist)-startSize)
+
+	case errorNotEnoughLiterals:
+		return true, fmt.Errorf("unexpected literal count, want %d bytes, but only %d is available",
+			ctx.ll, ctx.litRemain+ctx.ll)
+
+	case errorNotEnoughSpace:
+		size := ctx.outPosition + ctx.ll + ctx.ml
+		if debugDecoder {
+			println("msl:", s.maxSyncLen, "cap", cap(s.out), "bef:", startSize, "sz:", size-startSize, "mbs:", maxBlockSize, "outsz:", cap(s.out)-startSize)
+		}
+		return true, fmt.Errorf("output (%d) bigger than max block size (%d)", size-startSize, maxBlockSize)
+
+	default:
+		return true, fmt.Errorf("sequenceDecs_decode returned erronous code %d", errCode)
+	}
+
+	s.seqSize += ctx.litRemain
+	if s.seqSize > maxBlockSize {
+		return true, fmt.Errorf("output (%d) bigger than max block size (%d)", s.seqSize, maxBlockSize)
+	}
+	err := br.close()
+	if err != nil {
+		printf("Closing sequences: %v, %+v\n", err, *br)
+		return true, err
+	}
+
+	s.literals = s.literals[ctx.litPosition:]
+	t := ctx.outPosition
+	s.out = s.out[:t]
+
+	// Add final literals
+	s.out = append(s.out, s.literals...)
+	if debugDecoder {
+		t += len(s.literals)
+		if t != len(s.out) {
+			panic(fmt.Errorf("length mismatch, want %d, got %d", len(s.out), t))
+		}
+	}
+
+	return true, nil
+}
+
+// --------------------------------------------------------------------------------
+
+type decodeAsmContext struct {
+	llTable   []decSymbol
+	mlTable   []decSymbol
+	ofTable   []decSymbol
+	llState   uint64
+	mlState   uint64
+	ofState   uint64
+	iteration int
+	seqs      []seqVals
+	litRemain int
+}
+
+const noError = 0
+
+// error reported when mo == 0 && ml > 0
+const errorMatchLenOfsMismatch = 1
+
+// error reported when ml > maxMatchLen
+const errorMatchLenTooBig = 2
+
+// error reported when mo > available history or mo > s.windowSize
+const errorMatchOffTooBig = 3
+
+// error reported when the sum of literal lengths exeeceds the literal buffer size
+const errorNotEnoughLiterals = 4
+
+// error reported when capacity of `out` is too small
+const errorNotEnoughSpace = 5
+
+// sequenceDecs_decode implements the main loop of sequenceDecs in x86 asm.
+//
+// Please refer to seqdec_generic.go for the reference implementation.
+//go:noescape
+func sequenceDecs_decode_amd64(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int
+
+// sequenceDecs_decode implements the main loop of sequenceDecs in x86 asm.
+//
+// Please refer to seqdec_generic.go for the reference implementation.
+//go:noescape
+func sequenceDecs_decode_56_amd64(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int
+
+// sequenceDecs_decode implements the main loop of sequenceDecs in x86 asm with BMI2 extensions.
+//go:noescape
+func sequenceDecs_decode_bmi2(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int
+
+// sequenceDecs_decode implements the main loop of sequenceDecs in x86 asm with BMI2 extensions.
+//go:noescape
+func sequenceDecs_decode_56_bmi2(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int
+
+// decode sequences from the stream without the provided history.
+func (s *sequenceDecs) decode(seqs []seqVals) error {
+	br := s.br
+
+	maxBlockSize := maxCompressedBlockSize
+	if s.windowSize < maxBlockSize {
+		maxBlockSize = s.windowSize
+	}
+
+	ctx := decodeAsmContext{
+		llTable:   s.litLengths.fse.dt[:maxTablesize],
+		mlTable:   s.matchLengths.fse.dt[:maxTablesize],
+		ofTable:   s.offsets.fse.dt[:maxTablesize],
+		llState:   uint64(s.litLengths.state.state),
+		mlState:   uint64(s.matchLengths.state.state),
+		ofState:   uint64(s.offsets.state.state),
+		seqs:      seqs,
+		iteration: len(seqs) - 1,
+		litRemain: len(s.literals),
+	}
+
+	s.seqSize = 0
+	lte56bits := s.maxBits+s.offsets.fse.actualTableLog+s.matchLengths.fse.actualTableLog+s.litLengths.fse.actualTableLog <= 56
+	var errCode int
+	if cpuinfo.HasBMI2() {
+		if lte56bits {
+			errCode = sequenceDecs_decode_56_bmi2(s, br, &ctx)
+		} else {
+			errCode = sequenceDecs_decode_bmi2(s, br, &ctx)
+		}
+	} else {
+		if lte56bits {
+			errCode = sequenceDecs_decode_56_amd64(s, br, &ctx)
+		} else {
+			errCode = sequenceDecs_decode_amd64(s, br, &ctx)
+		}
+	}
+	if errCode != 0 {
+		i := len(seqs) - ctx.iteration - 1
+		switch errCode {
+		case errorMatchLenOfsMismatch:
+			ml := ctx.seqs[i].ml
+			return fmt.Errorf("zero matchoff and matchlen (%d) > 0", ml)
+
+		case errorMatchLenTooBig:
+			ml := ctx.seqs[i].ml
+			return fmt.Errorf("match len (%d) bigger than max allowed length", ml)
+
+		case errorNotEnoughLiterals:
+			ll := ctx.seqs[i].ll
+			return fmt.Errorf("unexpected literal count, want %d bytes, but only %d is available", ll, ctx.litRemain+ll)
+		}
+
+		return fmt.Errorf("sequenceDecs_decode_amd64 returned erronous code %d", errCode)
+	}
+
+	if ctx.litRemain < 0 {
+		return fmt.Errorf("literal count is too big: total available %d, total requested %d",
+			len(s.literals), len(s.literals)-ctx.litRemain)
+	}
+
+	s.seqSize += ctx.litRemain
+	if s.seqSize > maxBlockSize {
+		return fmt.Errorf("output (%d) bigger than max block size (%d)", s.seqSize, maxBlockSize)
+	}
+	err := br.close()
+	if err != nil {
+		printf("Closing sequences: %v, %+v\n", err, *br)
+	}
+	return err
+}
+
+// --------------------------------------------------------------------------------
+
+type executeAsmContext struct {
+	seqs        []seqVals
+	seqIndex    int
+	out         []byte
+	history     []byte
+	literals    []byte
+	outPosition int
+	litPosition int
+	windowSize  int
+}
+
+// sequenceDecs_executeSimple_amd64 implements the main loop of sequenceDecs.executeSimple in x86 asm.
+//
+// Returns false if a match offset is too big.
+//
+// Please refer to seqdec_generic.go for the reference implementation.
+//go:noescape
+func sequenceDecs_executeSimple_amd64(ctx *executeAsmContext) bool
+
+// Same as above, but with safe memcopies
+//go:noescape
+func sequenceDecs_executeSimple_safe_amd64(ctx *executeAsmContext) bool
+
+// executeSimple handles cases when dictionary is not used.
+func (s *sequenceDecs) executeSimple(seqs []seqVals, hist []byte) error {
+	// Ensure we have enough output size...
+	if len(s.out)+s.seqSize+compressedBlockOverAlloc > cap(s.out) {
+		addBytes := s.seqSize + len(s.out) + compressedBlockOverAlloc
+		s.out = append(s.out, make([]byte, addBytes)...)
+		s.out = s.out[:len(s.out)-addBytes]
+	}
+
+	if debugDecoder {
+		printf("Execute %d seqs with literals: %d into %d bytes\n", len(seqs), len(s.literals), s.seqSize)
+	}
+
+	var t = len(s.out)
+	out := s.out[:t+s.seqSize]
+
+	ctx := executeAsmContext{
+		seqs:        seqs,
+		seqIndex:    0,
+		out:         out,
+		history:     hist,
+		outPosition: t,
+		litPosition: 0,
+		literals:    s.literals,
+		windowSize:  s.windowSize,
+	}
+	var ok bool
+	if cap(s.literals) < len(s.literals)+compressedBlockOverAlloc {
+		ok = sequenceDecs_executeSimple_safe_amd64(&ctx)
+	} else {
+		ok = sequenceDecs_executeSimple_amd64(&ctx)
+	}
+	if !ok {
+		return fmt.Errorf("match offset (%d) bigger than current history (%d)",
+			seqs[ctx.seqIndex].mo, ctx.outPosition+len(hist))
+	}
+	s.literals = s.literals[ctx.litPosition:]
+	t = ctx.outPosition
+
+	// Add final literals
+	copy(out[t:], s.literals)
+	if debugDecoder {
+		t += len(s.literals)
+		if t != len(out) {
+			panic(fmt.Errorf("length mismatch, want %d, got %d, ss: %d", len(out), t, s.seqSize))
+		}
+	}
+	s.out = out
+
+	return nil
+}
diff --git a/vendor/github.com/klauspost/compress/zstd/seqdec_amd64.s b/vendor/github.com/klauspost/compress/zstd/seqdec_amd64.s
new file mode 100644
index 00000000..212c6cac
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/zstd/seqdec_amd64.s
@@ -0,0 +1,3689 @@
+// Code generated by command: go run gen.go -out ../seqdec_amd64.s -pkg=zstd. DO NOT EDIT.
+
+//go:build !appengine && !noasm && gc && !noasm
+// +build !appengine,!noasm,gc,!noasm
+
+// func sequenceDecs_decode_amd64(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int
+// Requires: CMOV
+TEXT ·sequenceDecs_decode_amd64(SB), $8-32
+	MOVQ    br+8(FP), AX
+	MOVQ    32(AX), DX
+	MOVBQZX 40(AX), BX
+	MOVQ    24(AX), SI
+	MOVQ    (AX), AX
+	ADDQ    SI, AX
+	MOVQ    AX, (SP)
+	MOVQ    ctx+16(FP), AX
+	MOVQ    72(AX), DI
+	MOVQ    80(AX), R8
+	MOVQ    88(AX), R9
+	MOVQ    104(AX), R10
+	MOVQ    s+0(FP), AX
+	MOVQ    144(AX), R11
+	MOVQ    152(AX), R12
+	MOVQ    160(AX), R13
+
+sequenceDecs_decode_amd64_main_loop:
+	MOVQ (SP), R14
+
+	// Fill bitreader to have enough for the offset and match length.
+	CMPQ SI, $0x08
+	JL   sequenceDecs_decode_amd64_fill_byte_by_byte
+	MOVQ BX, AX
+	SHRQ $0x03, AX
+	SUBQ AX, R14
+	MOVQ (R14), DX
+	SUBQ AX, SI
+	ANDQ $0x07, BX
+	JMP  sequenceDecs_decode_amd64_fill_end
+
+sequenceDecs_decode_amd64_fill_byte_by_byte:
+	CMPQ    SI, $0x00
+	JLE     sequenceDecs_decode_amd64_fill_end
+	CMPQ    BX, $0x07
+	JLE     sequenceDecs_decode_amd64_fill_end
+	SHLQ    $0x08, DX
+	SUBQ    $0x01, R14
+	SUBQ    $0x01, SI
+	SUBQ    $0x08, BX
+	MOVBQZX (R14), AX
+	ORQ     AX, DX
+	JMP     sequenceDecs_decode_amd64_fill_byte_by_byte
+
+sequenceDecs_decode_amd64_fill_end:
+	// Update offset
+	MOVQ    R9, AX
+	MOVQ    BX, CX
+	MOVQ    DX, R15
+	SHLQ    CL, R15
+	MOVB    AH, CL
+	ADDQ    CX, BX
+	NEGL    CX
+	SHRQ    CL, R15
+	SHRQ    $0x20, AX
+	TESTQ   CX, CX
+	CMOVQEQ CX, R15
+	ADDQ    R15, AX
+	MOVQ    AX, 16(R10)
+
+	// Update match length
+	MOVQ    R8, AX
+	MOVQ    BX, CX
+	MOVQ    DX, R15
+	SHLQ    CL, R15
+	MOVB    AH, CL
+	ADDQ    CX, BX
+	NEGL    CX
+	SHRQ    CL, R15
+	SHRQ    $0x20, AX
+	TESTQ   CX, CX
+	CMOVQEQ CX, R15
+	ADDQ    R15, AX
+	MOVQ    AX, 8(R10)
+
+	// Fill bitreader to have enough for the remaining
+	CMPQ SI, $0x08
+	JL   sequenceDecs_decode_amd64_fill_2_byte_by_byte
+	MOVQ BX, AX
+	SHRQ $0x03, AX
+	SUBQ AX, R14
+	MOVQ (R14), DX
+	SUBQ AX, SI
+	ANDQ $0x07, BX
+	JMP  sequenceDecs_decode_amd64_fill_2_end
+
+sequenceDecs_decode_amd64_fill_2_byte_by_byte:
+	CMPQ    SI, $0x00
+	JLE     sequenceDecs_decode_amd64_fill_2_end
+	CMPQ    BX, $0x07
+	JLE     sequenceDecs_decode_amd64_fill_2_end
+	SHLQ    $0x08, DX
+	SUBQ    $0x01, R14
+	SUBQ    $0x01, SI
+	SUBQ    $0x08, BX
+	MOVBQZX (R14), AX
+	ORQ     AX, DX
+	JMP     sequenceDecs_decode_amd64_fill_2_byte_by_byte
+
+sequenceDecs_decode_amd64_fill_2_end:
+	// Update literal length
+	MOVQ    DI, AX
+	MOVQ    BX, CX
+	MOVQ    DX, R15
+	SHLQ    CL, R15
+	MOVB    AH, CL
+	ADDQ    CX, BX
+	NEGL    CX
+	SHRQ    CL, R15
+	SHRQ    $0x20, AX
+	TESTQ   CX, CX
+	CMOVQEQ CX, R15
+	ADDQ    R15, AX
+	MOVQ    AX, (R10)
+
+	// Fill bitreader for state updates
+	MOVQ    R14, (SP)
+	MOVQ    R9, AX
+	SHRQ    $0x08, AX
+	MOVBQZX AL, AX
+	MOVQ    ctx+16(FP), CX
+	CMPQ    96(CX), $0x00
+	JZ      sequenceDecs_decode_amd64_skip_update
+
+	// Update Literal Length State
+	MOVBQZX DI, R14
+	SHRQ    $0x10, DI
+	MOVWQZX DI, DI
+	CMPQ    R14, $0x00
+	JZ      sequenceDecs_decode_amd64_llState_updateState_skip_zero
+	MOVQ    BX, CX
+	ADDQ    R14, BX
+	MOVQ    DX, R15
+	SHLQ    CL, R15
+	MOVQ    R14, CX
+	NEGQ    CX
+	SHRQ    CL, R15
+	ADDQ    R15, DI
+
+sequenceDecs_decode_amd64_llState_updateState_skip_zero:
+	// Load ctx.llTable
+	MOVQ ctx+16(FP), CX
+	MOVQ (CX), CX
+	MOVQ (CX)(DI*8), DI
+
+	// Update Match Length State
+	MOVBQZX R8, R14
+	SHRQ    $0x10, R8
+	MOVWQZX R8, R8
+	CMPQ    R14, $0x00
+	JZ      sequenceDecs_decode_amd64_mlState_updateState_skip_zero
+	MOVQ    BX, CX
+	ADDQ    R14, BX
+	MOVQ    DX, R15
+	SHLQ    CL, R15
+	MOVQ    R14, CX
+	NEGQ    CX
+	SHRQ    CL, R15
+	ADDQ    R15, R8
+
+sequenceDecs_decode_amd64_mlState_updateState_skip_zero:
+	// Load ctx.mlTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 24(CX), CX
+	MOVQ (CX)(R8*8), R8
+
+	// Update Offset State
+	MOVBQZX R9, R14
+	SHRQ    $0x10, R9
+	MOVWQZX R9, R9
+	CMPQ    R14, $0x00
+	JZ      sequenceDecs_decode_amd64_ofState_updateState_skip_zero
+	MOVQ    BX, CX
+	ADDQ    R14, BX
+	MOVQ    DX, R15
+	SHLQ    CL, R15
+	MOVQ    R14, CX
+	NEGQ    CX
+	SHRQ    CL, R15
+	ADDQ    R15, R9
+
+sequenceDecs_decode_amd64_ofState_updateState_skip_zero:
+	// Load ctx.ofTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 48(CX), CX
+	MOVQ (CX)(R9*8), R9
+
+sequenceDecs_decode_amd64_skip_update:
+	// Adjust offset
+	MOVQ 16(R10), CX
+	CMPQ AX, $0x01
+	JBE  sequenceDecs_decode_amd64_adjust_offsetB_1_or_0
+	MOVQ R12, R13
+	MOVQ R11, R12
+	MOVQ CX, R11
+	JMP  sequenceDecs_decode_amd64_adjust_end
+
+sequenceDecs_decode_amd64_adjust_offsetB_1_or_0:
+	CMPQ (R10), $0x00000000
+	JNE  sequenceDecs_decode_amd64_adjust_offset_maybezero
+	INCQ CX
+	JMP  sequenceDecs_decode_amd64_adjust_offset_nonzero
+
+sequenceDecs_decode_amd64_adjust_offset_maybezero:
+	TESTQ CX, CX
+	JNZ   sequenceDecs_decode_amd64_adjust_offset_nonzero
+	MOVQ  R11, CX
+	JMP   sequenceDecs_decode_amd64_adjust_end
+
+sequenceDecs_decode_amd64_adjust_offset_nonzero:
+	CMPQ CX, $0x01
+	JB   sequenceDecs_decode_amd64_adjust_zero
+	JEQ  sequenceDecs_decode_amd64_adjust_one
+	CMPQ CX, $0x02
+	JA   sequenceDecs_decode_amd64_adjust_three
+	JMP  sequenceDecs_decode_amd64_adjust_two
+
+sequenceDecs_decode_amd64_adjust_zero:
+	MOVQ R11, AX
+	JMP  sequenceDecs_decode_amd64_adjust_test_temp_valid
+
+sequenceDecs_decode_amd64_adjust_one:
+	MOVQ R12, AX
+	JMP  sequenceDecs_decode_amd64_adjust_test_temp_valid
+
+sequenceDecs_decode_amd64_adjust_two:
+	MOVQ R13, AX
+	JMP  sequenceDecs_decode_amd64_adjust_test_temp_valid
+
+sequenceDecs_decode_amd64_adjust_three:
+	LEAQ -1(R11), AX
+
+sequenceDecs_decode_amd64_adjust_test_temp_valid:
+	TESTQ AX, AX
+	JNZ   sequenceDecs_decode_amd64_adjust_temp_valid
+	MOVQ  $0x00000001, AX
+
+sequenceDecs_decode_amd64_adjust_temp_valid:
+	CMPQ    CX, $0x01
+	CMOVQNE R12, R13
+	MOVQ    R11, R12
+	MOVQ    AX, R11
+	MOVQ    AX, CX
+
+sequenceDecs_decode_amd64_adjust_end:
+	MOVQ CX, 16(R10)
+
+	// Check values
+	MOVQ  8(R10), AX
+	MOVQ  (R10), R14
+	LEAQ  (AX)(R14*1), R15
+	MOVQ  s+0(FP), BP
+	ADDQ  R15, 256(BP)
+	MOVQ  ctx+16(FP), R15
+	SUBQ  R14, 128(R15)
+	JS    error_not_enough_literals
+	CMPQ  AX, $0x00020002
+	JA    sequenceDecs_decode_amd64_error_match_len_too_big
+	TESTQ CX, CX
+	JNZ   sequenceDecs_decode_amd64_match_len_ofs_ok
+	TESTQ AX, AX
+	JNZ   sequenceDecs_decode_amd64_error_match_len_ofs_mismatch
+
+sequenceDecs_decode_amd64_match_len_ofs_ok:
+	ADDQ $0x18, R10
+	MOVQ ctx+16(FP), AX
+	DECQ 96(AX)
+	JNS  sequenceDecs_decode_amd64_main_loop
+	MOVQ s+0(FP), AX
+	MOVQ R11, 144(AX)
+	MOVQ R12, 152(AX)
+	MOVQ R13, 160(AX)
+	MOVQ br+8(FP), AX
+	MOVQ DX, 32(AX)
+	MOVB BL, 40(AX)
+	MOVQ SI, 24(AX)
+
+	// Return success
+	MOVQ $0x00000000, ret+24(FP)
+	RET
+
+	// Return with match length error
+sequenceDecs_decode_amd64_error_match_len_ofs_mismatch:
+	MOVQ $0x00000001, ret+24(FP)
+	RET
+
+	// Return with match too long error
+sequenceDecs_decode_amd64_error_match_len_too_big:
+	MOVQ $0x00000002, ret+24(FP)
+	RET
+
+	// Return with match offset too long error
+	MOVQ $0x00000003, ret+24(FP)
+	RET
+
+	// Return with not enough literals error
+error_not_enough_literals:
+	MOVQ $0x00000004, ret+24(FP)
+	RET
+
+	// Return with not enough output space error
+	MOVQ $0x00000005, ret+24(FP)
+	RET
+
+// func sequenceDecs_decode_56_amd64(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int
+// Requires: CMOV
+TEXT ·sequenceDecs_decode_56_amd64(SB), $8-32
+	MOVQ    br+8(FP), AX
+	MOVQ    32(AX), DX
+	MOVBQZX 40(AX), BX
+	MOVQ    24(AX), SI
+	MOVQ    (AX), AX
+	ADDQ    SI, AX
+	MOVQ    AX, (SP)
+	MOVQ    ctx+16(FP), AX
+	MOVQ    72(AX), DI
+	MOVQ    80(AX), R8
+	MOVQ    88(AX), R9
+	MOVQ    104(AX), R10
+	MOVQ    s+0(FP), AX
+	MOVQ    144(AX), R11
+	MOVQ    152(AX), R12
+	MOVQ    160(AX), R13
+
+sequenceDecs_decode_56_amd64_main_loop:
+	MOVQ (SP), R14
+
+	// Fill bitreader to have enough for the offset and match length.
+	CMPQ SI, $0x08
+	JL   sequenceDecs_decode_56_amd64_fill_byte_by_byte
+	MOVQ BX, AX
+	SHRQ $0x03, AX
+	SUBQ AX, R14
+	MOVQ (R14), DX
+	SUBQ AX, SI
+	ANDQ $0x07, BX
+	JMP  sequenceDecs_decode_56_amd64_fill_end
+
+sequenceDecs_decode_56_amd64_fill_byte_by_byte:
+	CMPQ    SI, $0x00
+	JLE     sequenceDecs_decode_56_amd64_fill_end
+	CMPQ    BX, $0x07
+	JLE     sequenceDecs_decode_56_amd64_fill_end
+	SHLQ    $0x08, DX
+	SUBQ    $0x01, R14
+	SUBQ    $0x01, SI
+	SUBQ    $0x08, BX
+	MOVBQZX (R14), AX
+	ORQ     AX, DX
+	JMP     sequenceDecs_decode_56_amd64_fill_byte_by_byte
+
+sequenceDecs_decode_56_amd64_fill_end:
+	// Update offset
+	MOVQ    R9, AX
+	MOVQ    BX, CX
+	MOVQ    DX, R15
+	SHLQ    CL, R15
+	MOVB    AH, CL
+	ADDQ    CX, BX
+	NEGL    CX
+	SHRQ    CL, R15
+	SHRQ    $0x20, AX
+	TESTQ   CX, CX
+	CMOVQEQ CX, R15
+	ADDQ    R15, AX
+	MOVQ    AX, 16(R10)
+
+	// Update match length
+	MOVQ    R8, AX
+	MOVQ    BX, CX
+	MOVQ    DX, R15
+	SHLQ    CL, R15
+	MOVB    AH, CL
+	ADDQ    CX, BX
+	NEGL    CX
+	SHRQ    CL, R15
+	SHRQ    $0x20, AX
+	TESTQ   CX, CX
+	CMOVQEQ CX, R15
+	ADDQ    R15, AX
+	MOVQ    AX, 8(R10)
+
+	// Update literal length
+	MOVQ    DI, AX
+	MOVQ    BX, CX
+	MOVQ    DX, R15
+	SHLQ    CL, R15
+	MOVB    AH, CL
+	ADDQ    CX, BX
+	NEGL    CX
+	SHRQ    CL, R15
+	SHRQ    $0x20, AX
+	TESTQ   CX, CX
+	CMOVQEQ CX, R15
+	ADDQ    R15, AX
+	MOVQ    AX, (R10)
+
+	// Fill bitreader for state updates
+	MOVQ    R14, (SP)
+	MOVQ    R9, AX
+	SHRQ    $0x08, AX
+	MOVBQZX AL, AX
+	MOVQ    ctx+16(FP), CX
+	CMPQ    96(CX), $0x00
+	JZ      sequenceDecs_decode_56_amd64_skip_update
+
+	// Update Literal Length State
+	MOVBQZX DI, R14
+	SHRQ    $0x10, DI
+	MOVWQZX DI, DI
+	CMPQ    R14, $0x00
+	JZ      sequenceDecs_decode_56_amd64_llState_updateState_skip_zero
+	MOVQ    BX, CX
+	ADDQ    R14, BX
+	MOVQ    DX, R15
+	SHLQ    CL, R15
+	MOVQ    R14, CX
+	NEGQ    CX
+	SHRQ    CL, R15
+	ADDQ    R15, DI
+
+sequenceDecs_decode_56_amd64_llState_updateState_skip_zero:
+	// Load ctx.llTable
+	MOVQ ctx+16(FP), CX
+	MOVQ (CX), CX
+	MOVQ (CX)(DI*8), DI
+
+	// Update Match Length State
+	MOVBQZX R8, R14
+	SHRQ    $0x10, R8
+	MOVWQZX R8, R8
+	CMPQ    R14, $0x00
+	JZ      sequenceDecs_decode_56_amd64_mlState_updateState_skip_zero
+	MOVQ    BX, CX
+	ADDQ    R14, BX
+	MOVQ    DX, R15
+	SHLQ    CL, R15
+	MOVQ    R14, CX
+	NEGQ    CX
+	SHRQ    CL, R15
+	ADDQ    R15, R8
+
+sequenceDecs_decode_56_amd64_mlState_updateState_skip_zero:
+	// Load ctx.mlTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 24(CX), CX
+	MOVQ (CX)(R8*8), R8
+
+	// Update Offset State
+	MOVBQZX R9, R14
+	SHRQ    $0x10, R9
+	MOVWQZX R9, R9
+	CMPQ    R14, $0x00
+	JZ      sequenceDecs_decode_56_amd64_ofState_updateState_skip_zero
+	MOVQ    BX, CX
+	ADDQ    R14, BX
+	MOVQ    DX, R15
+	SHLQ    CL, R15
+	MOVQ    R14, CX
+	NEGQ    CX
+	SHRQ    CL, R15
+	ADDQ    R15, R9
+
+sequenceDecs_decode_56_amd64_ofState_updateState_skip_zero:
+	// Load ctx.ofTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 48(CX), CX
+	MOVQ (CX)(R9*8), R9
+
+sequenceDecs_decode_56_amd64_skip_update:
+	// Adjust offset
+	MOVQ 16(R10), CX
+	CMPQ AX, $0x01
+	JBE  sequenceDecs_decode_56_amd64_adjust_offsetB_1_or_0
+	MOVQ R12, R13
+	MOVQ R11, R12
+	MOVQ CX, R11
+	JMP  sequenceDecs_decode_56_amd64_adjust_end
+
+sequenceDecs_decode_56_amd64_adjust_offsetB_1_or_0:
+	CMPQ (R10), $0x00000000
+	JNE  sequenceDecs_decode_56_amd64_adjust_offset_maybezero
+	INCQ CX
+	JMP  sequenceDecs_decode_56_amd64_adjust_offset_nonzero
+
+sequenceDecs_decode_56_amd64_adjust_offset_maybezero:
+	TESTQ CX, CX
+	JNZ   sequenceDecs_decode_56_amd64_adjust_offset_nonzero
+	MOVQ  R11, CX
+	JMP   sequenceDecs_decode_56_amd64_adjust_end
+
+sequenceDecs_decode_56_amd64_adjust_offset_nonzero:
+	CMPQ CX, $0x01
+	JB   sequenceDecs_decode_56_amd64_adjust_zero
+	JEQ  sequenceDecs_decode_56_amd64_adjust_one
+	CMPQ CX, $0x02
+	JA   sequenceDecs_decode_56_amd64_adjust_three
+	JMP  sequenceDecs_decode_56_amd64_adjust_two
+
+sequenceDecs_decode_56_amd64_adjust_zero:
+	MOVQ R11, AX
+	JMP  sequenceDecs_decode_56_amd64_adjust_test_temp_valid
+
+sequenceDecs_decode_56_amd64_adjust_one:
+	MOVQ R12, AX
+	JMP  sequenceDecs_decode_56_amd64_adjust_test_temp_valid
+
+sequenceDecs_decode_56_amd64_adjust_two:
+	MOVQ R13, AX
+	JMP  sequenceDecs_decode_56_amd64_adjust_test_temp_valid
+
+sequenceDecs_decode_56_amd64_adjust_three:
+	LEAQ -1(R11), AX
+
+sequenceDecs_decode_56_amd64_adjust_test_temp_valid:
+	TESTQ AX, AX
+	JNZ   sequenceDecs_decode_56_amd64_adjust_temp_valid
+	MOVQ  $0x00000001, AX
+
+sequenceDecs_decode_56_amd64_adjust_temp_valid:
+	CMPQ    CX, $0x01
+	CMOVQNE R12, R13
+	MOVQ    R11, R12
+	MOVQ    AX, R11
+	MOVQ    AX, CX
+
+sequenceDecs_decode_56_amd64_adjust_end:
+	MOVQ CX, 16(R10)
+
+	// Check values
+	MOVQ  8(R10), AX
+	MOVQ  (R10), R14
+	LEAQ  (AX)(R14*1), R15
+	MOVQ  s+0(FP), BP
+	ADDQ  R15, 256(BP)
+	MOVQ  ctx+16(FP), R15
+	SUBQ  R14, 128(R15)
+	JS    error_not_enough_literals
+	CMPQ  AX, $0x00020002
+	JA    sequenceDecs_decode_56_amd64_error_match_len_too_big
+	TESTQ CX, CX
+	JNZ   sequenceDecs_decode_56_amd64_match_len_ofs_ok
+	TESTQ AX, AX
+	JNZ   sequenceDecs_decode_56_amd64_error_match_len_ofs_mismatch
+
+sequenceDecs_decode_56_amd64_match_len_ofs_ok:
+	ADDQ $0x18, R10
+	MOVQ ctx+16(FP), AX
+	DECQ 96(AX)
+	JNS  sequenceDecs_decode_56_amd64_main_loop
+	MOVQ s+0(FP), AX
+	MOVQ R11, 144(AX)
+	MOVQ R12, 152(AX)
+	MOVQ R13, 160(AX)
+	MOVQ br+8(FP), AX
+	MOVQ DX, 32(AX)
+	MOVB BL, 40(AX)
+	MOVQ SI, 24(AX)
+
+	// Return success
+	MOVQ $0x00000000, ret+24(FP)
+	RET
+
+	// Return with match length error
+sequenceDecs_decode_56_amd64_error_match_len_ofs_mismatch:
+	MOVQ $0x00000001, ret+24(FP)
+	RET
+
+	// Return with match too long error
+sequenceDecs_decode_56_amd64_error_match_len_too_big:
+	MOVQ $0x00000002, ret+24(FP)
+	RET
+
+	// Return with match offset too long error
+	MOVQ $0x00000003, ret+24(FP)
+	RET
+
+	// Return with not enough literals error
+error_not_enough_literals:
+	MOVQ $0x00000004, ret+24(FP)
+	RET
+
+	// Return with not enough output space error
+	MOVQ $0x00000005, ret+24(FP)
+	RET
+
+// func sequenceDecs_decode_bmi2(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int
+// Requires: BMI, BMI2, CMOV
+TEXT ·sequenceDecs_decode_bmi2(SB), $8-32
+	MOVQ    br+8(FP), CX
+	MOVQ    32(CX), AX
+	MOVBQZX 40(CX), DX
+	MOVQ    24(CX), BX
+	MOVQ    (CX), CX
+	ADDQ    BX, CX
+	MOVQ    CX, (SP)
+	MOVQ    ctx+16(FP), CX
+	MOVQ    72(CX), SI
+	MOVQ    80(CX), DI
+	MOVQ    88(CX), R8
+	MOVQ    104(CX), R9
+	MOVQ    s+0(FP), CX
+	MOVQ    144(CX), R10
+	MOVQ    152(CX), R11
+	MOVQ    160(CX), R12
+
+sequenceDecs_decode_bmi2_main_loop:
+	MOVQ (SP), R13
+
+	// Fill bitreader to have enough for the offset and match length.
+	CMPQ BX, $0x08
+	JL   sequenceDecs_decode_bmi2_fill_byte_by_byte
+	MOVQ DX, CX
+	SHRQ $0x03, CX
+	SUBQ CX, R13
+	MOVQ (R13), AX
+	SUBQ CX, BX
+	ANDQ $0x07, DX
+	JMP  sequenceDecs_decode_bmi2_fill_end
+
+sequenceDecs_decode_bmi2_fill_byte_by_byte:
+	CMPQ    BX, $0x00
+	JLE     sequenceDecs_decode_bmi2_fill_end
+	CMPQ    DX, $0x07
+	JLE     sequenceDecs_decode_bmi2_fill_end
+	SHLQ    $0x08, AX
+	SUBQ    $0x01, R13
+	SUBQ    $0x01, BX
+	SUBQ    $0x08, DX
+	MOVBQZX (R13), CX
+	ORQ     CX, AX
+	JMP     sequenceDecs_decode_bmi2_fill_byte_by_byte
+
+sequenceDecs_decode_bmi2_fill_end:
+	// Update offset
+	MOVQ   $0x00000808, CX
+	BEXTRQ CX, R8, R14
+	MOVQ   AX, R15
+	LEAQ   (DX)(R14*1), CX
+	ROLQ   CL, R15
+	BZHIQ  R14, R15, R15
+	MOVQ   CX, DX
+	MOVQ   R8, CX
+	SHRQ   $0x20, CX
+	ADDQ   R15, CX
+	MOVQ   CX, 16(R9)
+
+	// Update match length
+	MOVQ   $0x00000808, CX
+	BEXTRQ CX, DI, R14
+	MOVQ   AX, R15
+	LEAQ   (DX)(R14*1), CX
+	ROLQ   CL, R15
+	BZHIQ  R14, R15, R15
+	MOVQ   CX, DX
+	MOVQ   DI, CX
+	SHRQ   $0x20, CX
+	ADDQ   R15, CX
+	MOVQ   CX, 8(R9)
+
+	// Fill bitreader to have enough for the remaining
+	CMPQ BX, $0x08
+	JL   sequenceDecs_decode_bmi2_fill_2_byte_by_byte
+	MOVQ DX, CX
+	SHRQ $0x03, CX
+	SUBQ CX, R13
+	MOVQ (R13), AX
+	SUBQ CX, BX
+	ANDQ $0x07, DX
+	JMP  sequenceDecs_decode_bmi2_fill_2_end
+
+sequenceDecs_decode_bmi2_fill_2_byte_by_byte:
+	CMPQ    BX, $0x00
+	JLE     sequenceDecs_decode_bmi2_fill_2_end
+	CMPQ    DX, $0x07
+	JLE     sequenceDecs_decode_bmi2_fill_2_end
+	SHLQ    $0x08, AX
+	SUBQ    $0x01, R13
+	SUBQ    $0x01, BX
+	SUBQ    $0x08, DX
+	MOVBQZX (R13), CX
+	ORQ     CX, AX
+	JMP     sequenceDecs_decode_bmi2_fill_2_byte_by_byte
+
+sequenceDecs_decode_bmi2_fill_2_end:
+	// Update literal length
+	MOVQ   $0x00000808, CX
+	BEXTRQ CX, SI, R14
+	MOVQ   AX, R15
+	LEAQ   (DX)(R14*1), CX
+	ROLQ   CL, R15
+	BZHIQ  R14, R15, R15
+	MOVQ   CX, DX
+	MOVQ   SI, CX
+	SHRQ   $0x20, CX
+	ADDQ   R15, CX
+	MOVQ   CX, (R9)
+
+	// Fill bitreader for state updates
+	MOVQ    R13, (SP)
+	MOVQ    $0x00000808, CX
+	BEXTRQ  CX, R8, R13
+	MOVQ    ctx+16(FP), CX
+	CMPQ    96(CX), $0x00
+	JZ      sequenceDecs_decode_bmi2_skip_update
+	LEAQ    (SI)(DI*1), R14
+	ADDQ    R8, R14
+	MOVBQZX R14, R14
+	LEAQ    (DX)(R14*1), CX
+	MOVQ    AX, R15
+	MOVQ    CX, DX
+	ROLQ    CL, R15
+	BZHIQ   R14, R15, R15
+
+	// Update Offset State
+	BZHIQ  R8, R15, CX
+	SHRXQ  R8, R15, R15
+	MOVQ   $0x00001010, R14
+	BEXTRQ R14, R8, R8
+	ADDQ   CX, R8
+
+	// Load ctx.ofTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 48(CX), CX
+	MOVQ (CX)(R8*8), R8
+
+	// Update Match Length State
+	BZHIQ  DI, R15, CX
+	SHRXQ  DI, R15, R15
+	MOVQ   $0x00001010, R14
+	BEXTRQ R14, DI, DI
+	ADDQ   CX, DI
+
+	// Load ctx.mlTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 24(CX), CX
+	MOVQ (CX)(DI*8), DI
+
+	// Update Literal Length State
+	BZHIQ  SI, R15, CX
+	MOVQ   $0x00001010, R14
+	BEXTRQ R14, SI, SI
+	ADDQ   CX, SI
+
+	// Load ctx.llTable
+	MOVQ ctx+16(FP), CX
+	MOVQ (CX), CX
+	MOVQ (CX)(SI*8), SI
+
+sequenceDecs_decode_bmi2_skip_update:
+	// Adjust offset
+	MOVQ 16(R9), CX
+	CMPQ R13, $0x01
+	JBE  sequenceDecs_decode_bmi2_adjust_offsetB_1_or_0
+	MOVQ R11, R12
+	MOVQ R10, R11
+	MOVQ CX, R10
+	JMP  sequenceDecs_decode_bmi2_adjust_end
+
+sequenceDecs_decode_bmi2_adjust_offsetB_1_or_0:
+	CMPQ (R9), $0x00000000
+	JNE  sequenceDecs_decode_bmi2_adjust_offset_maybezero
+	INCQ CX
+	JMP  sequenceDecs_decode_bmi2_adjust_offset_nonzero
+
+sequenceDecs_decode_bmi2_adjust_offset_maybezero:
+	TESTQ CX, CX
+	JNZ   sequenceDecs_decode_bmi2_adjust_offset_nonzero
+	MOVQ  R10, CX
+	JMP   sequenceDecs_decode_bmi2_adjust_end
+
+sequenceDecs_decode_bmi2_adjust_offset_nonzero:
+	CMPQ CX, $0x01
+	JB   sequenceDecs_decode_bmi2_adjust_zero
+	JEQ  sequenceDecs_decode_bmi2_adjust_one
+	CMPQ CX, $0x02
+	JA   sequenceDecs_decode_bmi2_adjust_three
+	JMP  sequenceDecs_decode_bmi2_adjust_two
+
+sequenceDecs_decode_bmi2_adjust_zero:
+	MOVQ R10, R13
+	JMP  sequenceDecs_decode_bmi2_adjust_test_temp_valid
+
+sequenceDecs_decode_bmi2_adjust_one:
+	MOVQ R11, R13
+	JMP  sequenceDecs_decode_bmi2_adjust_test_temp_valid
+
+sequenceDecs_decode_bmi2_adjust_two:
+	MOVQ R12, R13
+	JMP  sequenceDecs_decode_bmi2_adjust_test_temp_valid
+
+sequenceDecs_decode_bmi2_adjust_three:
+	LEAQ -1(R10), R13
+
+sequenceDecs_decode_bmi2_adjust_test_temp_valid:
+	TESTQ R13, R13
+	JNZ   sequenceDecs_decode_bmi2_adjust_temp_valid
+	MOVQ  $0x00000001, R13
+
+sequenceDecs_decode_bmi2_adjust_temp_valid:
+	CMPQ    CX, $0x01
+	CMOVQNE R11, R12
+	MOVQ    R10, R11
+	MOVQ    R13, R10
+	MOVQ    R13, CX
+
+sequenceDecs_decode_bmi2_adjust_end:
+	MOVQ CX, 16(R9)
+
+	// Check values
+	MOVQ  8(R9), R13
+	MOVQ  (R9), R14
+	LEAQ  (R13)(R14*1), R15
+	MOVQ  s+0(FP), BP
+	ADDQ  R15, 256(BP)
+	MOVQ  ctx+16(FP), R15
+	SUBQ  R14, 128(R15)
+	JS    error_not_enough_literals
+	CMPQ  R13, $0x00020002
+	JA    sequenceDecs_decode_bmi2_error_match_len_too_big
+	TESTQ CX, CX
+	JNZ   sequenceDecs_decode_bmi2_match_len_ofs_ok
+	TESTQ R13, R13
+	JNZ   sequenceDecs_decode_bmi2_error_match_len_ofs_mismatch
+
+sequenceDecs_decode_bmi2_match_len_ofs_ok:
+	ADDQ $0x18, R9
+	MOVQ ctx+16(FP), CX
+	DECQ 96(CX)
+	JNS  sequenceDecs_decode_bmi2_main_loop
+	MOVQ s+0(FP), CX
+	MOVQ R10, 144(CX)
+	MOVQ R11, 152(CX)
+	MOVQ R12, 160(CX)
+	MOVQ br+8(FP), CX
+	MOVQ AX, 32(CX)
+	MOVB DL, 40(CX)
+	MOVQ BX, 24(CX)
+
+	// Return success
+	MOVQ $0x00000000, ret+24(FP)
+	RET
+
+	// Return with match length error
+sequenceDecs_decode_bmi2_error_match_len_ofs_mismatch:
+	MOVQ $0x00000001, ret+24(FP)
+	RET
+
+	// Return with match too long error
+sequenceDecs_decode_bmi2_error_match_len_too_big:
+	MOVQ $0x00000002, ret+24(FP)
+	RET
+
+	// Return with match offset too long error
+	MOVQ $0x00000003, ret+24(FP)
+	RET
+
+	// Return with not enough literals error
+error_not_enough_literals:
+	MOVQ $0x00000004, ret+24(FP)
+	RET
+
+	// Return with not enough output space error
+	MOVQ $0x00000005, ret+24(FP)
+	RET
+
+// func sequenceDecs_decode_56_bmi2(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int
+// Requires: BMI, BMI2, CMOV
+TEXT ·sequenceDecs_decode_56_bmi2(SB), $8-32
+	MOVQ    br+8(FP), CX
+	MOVQ    32(CX), AX
+	MOVBQZX 40(CX), DX
+	MOVQ    24(CX), BX
+	MOVQ    (CX), CX
+	ADDQ    BX, CX
+	MOVQ    CX, (SP)
+	MOVQ    ctx+16(FP), CX
+	MOVQ    72(CX), SI
+	MOVQ    80(CX), DI
+	MOVQ    88(CX), R8
+	MOVQ    104(CX), R9
+	MOVQ    s+0(FP), CX
+	MOVQ    144(CX), R10
+	MOVQ    152(CX), R11
+	MOVQ    160(CX), R12
+
+sequenceDecs_decode_56_bmi2_main_loop:
+	MOVQ (SP), R13
+
+	// Fill bitreader to have enough for the offset and match length.
+	CMPQ BX, $0x08
+	JL   sequenceDecs_decode_56_bmi2_fill_byte_by_byte
+	MOVQ DX, CX
+	SHRQ $0x03, CX
+	SUBQ CX, R13
+	MOVQ (R13), AX
+	SUBQ CX, BX
+	ANDQ $0x07, DX
+	JMP  sequenceDecs_decode_56_bmi2_fill_end
+
+sequenceDecs_decode_56_bmi2_fill_byte_by_byte:
+	CMPQ    BX, $0x00
+	JLE     sequenceDecs_decode_56_bmi2_fill_end
+	CMPQ    DX, $0x07
+	JLE     sequenceDecs_decode_56_bmi2_fill_end
+	SHLQ    $0x08, AX
+	SUBQ    $0x01, R13
+	SUBQ    $0x01, BX
+	SUBQ    $0x08, DX
+	MOVBQZX (R13), CX
+	ORQ     CX, AX
+	JMP     sequenceDecs_decode_56_bmi2_fill_byte_by_byte
+
+sequenceDecs_decode_56_bmi2_fill_end:
+	// Update offset
+	MOVQ   $0x00000808, CX
+	BEXTRQ CX, R8, R14
+	MOVQ   AX, R15
+	LEAQ   (DX)(R14*1), CX
+	ROLQ   CL, R15
+	BZHIQ  R14, R15, R15
+	MOVQ   CX, DX
+	MOVQ   R8, CX
+	SHRQ   $0x20, CX
+	ADDQ   R15, CX
+	MOVQ   CX, 16(R9)
+
+	// Update match length
+	MOVQ   $0x00000808, CX
+	BEXTRQ CX, DI, R14
+	MOVQ   AX, R15
+	LEAQ   (DX)(R14*1), CX
+	ROLQ   CL, R15
+	BZHIQ  R14, R15, R15
+	MOVQ   CX, DX
+	MOVQ   DI, CX
+	SHRQ   $0x20, CX
+	ADDQ   R15, CX
+	MOVQ   CX, 8(R9)
+
+	// Update literal length
+	MOVQ   $0x00000808, CX
+	BEXTRQ CX, SI, R14
+	MOVQ   AX, R15
+	LEAQ   (DX)(R14*1), CX
+	ROLQ   CL, R15
+	BZHIQ  R14, R15, R15
+	MOVQ   CX, DX
+	MOVQ   SI, CX
+	SHRQ   $0x20, CX
+	ADDQ   R15, CX
+	MOVQ   CX, (R9)
+
+	// Fill bitreader for state updates
+	MOVQ    R13, (SP)
+	MOVQ    $0x00000808, CX
+	BEXTRQ  CX, R8, R13
+	MOVQ    ctx+16(FP), CX
+	CMPQ    96(CX), $0x00
+	JZ      sequenceDecs_decode_56_bmi2_skip_update
+	LEAQ    (SI)(DI*1), R14
+	ADDQ    R8, R14
+	MOVBQZX R14, R14
+	LEAQ    (DX)(R14*1), CX
+	MOVQ    AX, R15
+	MOVQ    CX, DX
+	ROLQ    CL, R15
+	BZHIQ   R14, R15, R15
+
+	// Update Offset State
+	BZHIQ  R8, R15, CX
+	SHRXQ  R8, R15, R15
+	MOVQ   $0x00001010, R14
+	BEXTRQ R14, R8, R8
+	ADDQ   CX, R8
+
+	// Load ctx.ofTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 48(CX), CX
+	MOVQ (CX)(R8*8), R8
+
+	// Update Match Length State
+	BZHIQ  DI, R15, CX
+	SHRXQ  DI, R15, R15
+	MOVQ   $0x00001010, R14
+	BEXTRQ R14, DI, DI
+	ADDQ   CX, DI
+
+	// Load ctx.mlTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 24(CX), CX
+	MOVQ (CX)(DI*8), DI
+
+	// Update Literal Length State
+	BZHIQ  SI, R15, CX
+	MOVQ   $0x00001010, R14
+	BEXTRQ R14, SI, SI
+	ADDQ   CX, SI
+
+	// Load ctx.llTable
+	MOVQ ctx+16(FP), CX
+	MOVQ (CX), CX
+	MOVQ (CX)(SI*8), SI
+
+sequenceDecs_decode_56_bmi2_skip_update:
+	// Adjust offset
+	MOVQ 16(R9), CX
+	CMPQ R13, $0x01
+	JBE  sequenceDecs_decode_56_bmi2_adjust_offsetB_1_or_0
+	MOVQ R11, R12
+	MOVQ R10, R11
+	MOVQ CX, R10
+	JMP  sequenceDecs_decode_56_bmi2_adjust_end
+
+sequenceDecs_decode_56_bmi2_adjust_offsetB_1_or_0:
+	CMPQ (R9), $0x00000000
+	JNE  sequenceDecs_decode_56_bmi2_adjust_offset_maybezero
+	INCQ CX
+	JMP  sequenceDecs_decode_56_bmi2_adjust_offset_nonzero
+
+sequenceDecs_decode_56_bmi2_adjust_offset_maybezero:
+	TESTQ CX, CX
+	JNZ   sequenceDecs_decode_56_bmi2_adjust_offset_nonzero
+	MOVQ  R10, CX
+	JMP   sequenceDecs_decode_56_bmi2_adjust_end
+
+sequenceDecs_decode_56_bmi2_adjust_offset_nonzero:
+	CMPQ CX, $0x01
+	JB   sequenceDecs_decode_56_bmi2_adjust_zero
+	JEQ  sequenceDecs_decode_56_bmi2_adjust_one
+	CMPQ CX, $0x02
+	JA   sequenceDecs_decode_56_bmi2_adjust_three
+	JMP  sequenceDecs_decode_56_bmi2_adjust_two
+
+sequenceDecs_decode_56_bmi2_adjust_zero:
+	MOVQ R10, R13
+	JMP  sequenceDecs_decode_56_bmi2_adjust_test_temp_valid
+
+sequenceDecs_decode_56_bmi2_adjust_one:
+	MOVQ R11, R13
+	JMP  sequenceDecs_decode_56_bmi2_adjust_test_temp_valid
+
+sequenceDecs_decode_56_bmi2_adjust_two:
+	MOVQ R12, R13
+	JMP  sequenceDecs_decode_56_bmi2_adjust_test_temp_valid
+
+sequenceDecs_decode_56_bmi2_adjust_three:
+	LEAQ -1(R10), R13
+
+sequenceDecs_decode_56_bmi2_adjust_test_temp_valid:
+	TESTQ R13, R13
+	JNZ   sequenceDecs_decode_56_bmi2_adjust_temp_valid
+	MOVQ  $0x00000001, R13
+
+sequenceDecs_decode_56_bmi2_adjust_temp_valid:
+	CMPQ    CX, $0x01
+	CMOVQNE R11, R12
+	MOVQ    R10, R11
+	MOVQ    R13, R10
+	MOVQ    R13, CX
+
+sequenceDecs_decode_56_bmi2_adjust_end:
+	MOVQ CX, 16(R9)
+
+	// Check values
+	MOVQ  8(R9), R13
+	MOVQ  (R9), R14
+	LEAQ  (R13)(R14*1), R15
+	MOVQ  s+0(FP), BP
+	ADDQ  R15, 256(BP)
+	MOVQ  ctx+16(FP), R15
+	SUBQ  R14, 128(R15)
+	JS    error_not_enough_literals
+	CMPQ  R13, $0x00020002
+	JA    sequenceDecs_decode_56_bmi2_error_match_len_too_big
+	TESTQ CX, CX
+	JNZ   sequenceDecs_decode_56_bmi2_match_len_ofs_ok
+	TESTQ R13, R13
+	JNZ   sequenceDecs_decode_56_bmi2_error_match_len_ofs_mismatch
+
+sequenceDecs_decode_56_bmi2_match_len_ofs_ok:
+	ADDQ $0x18, R9
+	MOVQ ctx+16(FP), CX
+	DECQ 96(CX)
+	JNS  sequenceDecs_decode_56_bmi2_main_loop
+	MOVQ s+0(FP), CX
+	MOVQ R10, 144(CX)
+	MOVQ R11, 152(CX)
+	MOVQ R12, 160(CX)
+	MOVQ br+8(FP), CX
+	MOVQ AX, 32(CX)
+	MOVB DL, 40(CX)
+	MOVQ BX, 24(CX)
+
+	// Return success
+	MOVQ $0x00000000, ret+24(FP)
+	RET
+
+	// Return with match length error
+sequenceDecs_decode_56_bmi2_error_match_len_ofs_mismatch:
+	MOVQ $0x00000001, ret+24(FP)
+	RET
+
+	// Return with match too long error
+sequenceDecs_decode_56_bmi2_error_match_len_too_big:
+	MOVQ $0x00000002, ret+24(FP)
+	RET
+
+	// Return with match offset too long error
+	MOVQ $0x00000003, ret+24(FP)
+	RET
+
+	// Return with not enough literals error
+error_not_enough_literals:
+	MOVQ $0x00000004, ret+24(FP)
+	RET
+
+	// Return with not enough output space error
+	MOVQ $0x00000005, ret+24(FP)
+	RET
+
+// func sequenceDecs_executeSimple_amd64(ctx *executeAsmContext) bool
+// Requires: SSE
+TEXT ·sequenceDecs_executeSimple_amd64(SB), $8-9
+	MOVQ  ctx+0(FP), R10
+	MOVQ  8(R10), CX
+	TESTQ CX, CX
+	JZ    empty_seqs
+	MOVQ  (R10), AX
+	MOVQ  24(R10), DX
+	MOVQ  32(R10), BX
+	MOVQ  80(R10), SI
+	MOVQ  104(R10), DI
+	MOVQ  120(R10), R8
+	MOVQ  56(R10), R9
+	MOVQ  64(R10), R10
+	ADDQ  R10, R9
+
+	// seqsBase += 24 * seqIndex
+	LEAQ (DX)(DX*2), R11
+	SHLQ $0x03, R11
+	ADDQ R11, AX
+
+	// outBase += outPosition
+	ADDQ DI, BX
+
+main_loop:
+	MOVQ (AX), R11
+	MOVQ 16(AX), R12
+	MOVQ 8(AX), R13
+
+	// Copy literals
+	TESTQ R11, R11
+	JZ    check_offset
+	XORQ  R14, R14
+
+copy_1:
+	MOVUPS (SI)(R14*1), X0
+	MOVUPS X0, (BX)(R14*1)
+	ADDQ   $0x10, R14
+	CMPQ   R14, R11
+	JB     copy_1
+	ADDQ   R11, SI
+	ADDQ   R11, BX
+	ADDQ   R11, DI
+
+	// Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize)
+check_offset:
+	LEAQ (DI)(R10*1), R11
+	CMPQ R12, R11
+	JG   error_match_off_too_big
+	CMPQ R12, R8
+	JG   error_match_off_too_big
+
+	// Copy match from history
+	MOVQ  R12, R11
+	SUBQ  DI, R11
+	JLS   copy_match
+	MOVQ  R9, R14
+	SUBQ  R11, R14
+	CMPQ  R13, R11
+	JGE   copy_all_from_history
+	XORQ  R11, R11
+	TESTQ $0x00000001, R13
+	JZ    copy_4_word
+	MOVB  (R14)(R11*1), R12
+	MOVB  R12, (BX)(R11*1)
+	ADDQ  $0x01, R11
+
+copy_4_word:
+	TESTQ $0x00000002, R13
+	JZ    copy_4_dword
+	MOVW  (R14)(R11*1), R12
+	MOVW  R12, (BX)(R11*1)
+	ADDQ  $0x02, R11
+
+copy_4_dword:
+	TESTQ $0x00000004, R13
+	JZ    copy_4_qword
+	MOVL  (R14)(R11*1), R12
+	MOVL  R12, (BX)(R11*1)
+	ADDQ  $0x04, R11
+
+copy_4_qword:
+	TESTQ $0x00000008, R13
+	JZ    copy_4_test
+	MOVQ  (R14)(R11*1), R12
+	MOVQ  R12, (BX)(R11*1)
+	ADDQ  $0x08, R11
+	JMP   copy_4_test
+
+copy_4:
+	MOVUPS (R14)(R11*1), X0
+	MOVUPS X0, (BX)(R11*1)
+	ADDQ   $0x10, R11
+
+copy_4_test:
+	CMPQ R11, R13
+	JB   copy_4
+	ADDQ R13, DI
+	ADDQ R13, BX
+	ADDQ $0x18, AX
+	INCQ DX
+	CMPQ DX, CX
+	JB   main_loop
+	JMP  loop_finished
+
+copy_all_from_history:
+	XORQ  R15, R15
+	TESTQ $0x00000001, R11
+	JZ    copy_5_word
+	MOVB  (R14)(R15*1), BP
+	MOVB  BP, (BX)(R15*1)
+	ADDQ  $0x01, R15
+
+copy_5_word:
+	TESTQ $0x00000002, R11
+	JZ    copy_5_dword
+	MOVW  (R14)(R15*1), BP
+	MOVW  BP, (BX)(R15*1)
+	ADDQ  $0x02, R15
+
+copy_5_dword:
+	TESTQ $0x00000004, R11
+	JZ    copy_5_qword
+	MOVL  (R14)(R15*1), BP
+	MOVL  BP, (BX)(R15*1)
+	ADDQ  $0x04, R15
+
+copy_5_qword:
+	TESTQ $0x00000008, R11
+	JZ    copy_5_test
+	MOVQ  (R14)(R15*1), BP
+	MOVQ  BP, (BX)(R15*1)
+	ADDQ  $0x08, R15
+	JMP   copy_5_test
+
+copy_5:
+	MOVUPS (R14)(R15*1), X0
+	MOVUPS X0, (BX)(R15*1)
+	ADDQ   $0x10, R15
+
+copy_5_test:
+	CMPQ R15, R11
+	JB   copy_5
+	ADDQ R11, BX
+	ADDQ R11, DI
+	SUBQ R11, R13
+
+	// Copy match from the current buffer
+copy_match:
+	TESTQ R13, R13
+	JZ    handle_loop
+	MOVQ  BX, R11
+	SUBQ  R12, R11
+
+	// ml <= mo
+	CMPQ R13, R12
+	JA   copy_overlapping_match
+
+	// Copy non-overlapping match
+	ADDQ R13, DI
+	MOVQ BX, R12
+	ADDQ R13, BX
+
+copy_2:
+	MOVUPS (R11), X0
+	MOVUPS X0, (R12)
+	ADDQ   $0x10, R11
+	ADDQ   $0x10, R12
+	SUBQ   $0x10, R13
+	JHI    copy_2
+	JMP    handle_loop
+
+	// Copy overlapping match
+copy_overlapping_match:
+	ADDQ R13, DI
+
+copy_slow_3:
+	MOVB (R11), R12
+	MOVB R12, (BX)
+	INCQ R11
+	INCQ BX
+	DECQ R13
+	JNZ  copy_slow_3
+
+handle_loop:
+	ADDQ $0x18, AX
+	INCQ DX
+	CMPQ DX, CX
+	JB   main_loop
+
+loop_finished:
+	// Return value
+	MOVB $0x01, ret+8(FP)
+
+	// Update the context
+	MOVQ ctx+0(FP), AX
+	MOVQ DX, 24(AX)
+	MOVQ DI, 104(AX)
+	MOVQ 80(AX), CX
+	SUBQ CX, SI
+	MOVQ SI, 112(AX)
+	RET
+
+error_match_off_too_big:
+	// Return value
+	MOVB $0x00, ret+8(FP)
+
+	// Update the context
+	MOVQ ctx+0(FP), AX
+	MOVQ DX, 24(AX)
+	MOVQ DI, 104(AX)
+	MOVQ 80(AX), CX
+	SUBQ CX, SI
+	MOVQ SI, 112(AX)
+	RET
+
+empty_seqs:
+	// Return value
+	MOVB $0x01, ret+8(FP)
+	RET
+
+// func sequenceDecs_executeSimple_safe_amd64(ctx *executeAsmContext) bool
+// Requires: SSE
+TEXT ·sequenceDecs_executeSimple_safe_amd64(SB), $8-9
+	MOVQ  ctx+0(FP), R10
+	MOVQ  8(R10), CX
+	TESTQ CX, CX
+	JZ    empty_seqs
+	MOVQ  (R10), AX
+	MOVQ  24(R10), DX
+	MOVQ  32(R10), BX
+	MOVQ  80(R10), SI
+	MOVQ  104(R10), DI
+	MOVQ  120(R10), R8
+	MOVQ  56(R10), R9
+	MOVQ  64(R10), R10
+	ADDQ  R10, R9
+
+	// seqsBase += 24 * seqIndex
+	LEAQ (DX)(DX*2), R11
+	SHLQ $0x03, R11
+	ADDQ R11, AX
+
+	// outBase += outPosition
+	ADDQ DI, BX
+
+main_loop:
+	MOVQ (AX), R11
+	MOVQ 16(AX), R12
+	MOVQ 8(AX), R13
+
+	// Copy literals
+	TESTQ R11, R11
+	JZ    check_offset
+	XORQ  R14, R14
+	TESTQ $0x00000001, R11
+	JZ    copy_1_word
+	MOVB  (SI)(R14*1), R15
+	MOVB  R15, (BX)(R14*1)
+	ADDQ  $0x01, R14
+
+copy_1_word:
+	TESTQ $0x00000002, R11
+	JZ    copy_1_dword
+	MOVW  (SI)(R14*1), R15
+	MOVW  R15, (BX)(R14*1)
+	ADDQ  $0x02, R14
+
+copy_1_dword:
+	TESTQ $0x00000004, R11
+	JZ    copy_1_qword
+	MOVL  (SI)(R14*1), R15
+	MOVL  R15, (BX)(R14*1)
+	ADDQ  $0x04, R14
+
+copy_1_qword:
+	TESTQ $0x00000008, R11
+	JZ    copy_1_test
+	MOVQ  (SI)(R14*1), R15
+	MOVQ  R15, (BX)(R14*1)
+	ADDQ  $0x08, R14
+	JMP   copy_1_test
+
+copy_1:
+	MOVUPS (SI)(R14*1), X0
+	MOVUPS X0, (BX)(R14*1)
+	ADDQ   $0x10, R14
+
+copy_1_test:
+	CMPQ R14, R11
+	JB   copy_1
+	ADDQ R11, SI
+	ADDQ R11, BX
+	ADDQ R11, DI
+
+	// Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize)
+check_offset:
+	LEAQ (DI)(R10*1), R11
+	CMPQ R12, R11
+	JG   error_match_off_too_big
+	CMPQ R12, R8
+	JG   error_match_off_too_big
+
+	// Copy match from history
+	MOVQ  R12, R11
+	SUBQ  DI, R11
+	JLS   copy_match
+	MOVQ  R9, R14
+	SUBQ  R11, R14
+	CMPQ  R13, R11
+	JGE   copy_all_from_history
+	XORQ  R11, R11
+	TESTQ $0x00000001, R13
+	JZ    copy_4_word
+	MOVB  (R14)(R11*1), R12
+	MOVB  R12, (BX)(R11*1)
+	ADDQ  $0x01, R11
+
+copy_4_word:
+	TESTQ $0x00000002, R13
+	JZ    copy_4_dword
+	MOVW  (R14)(R11*1), R12
+	MOVW  R12, (BX)(R11*1)
+	ADDQ  $0x02, R11
+
+copy_4_dword:
+	TESTQ $0x00000004, R13
+	JZ    copy_4_qword
+	MOVL  (R14)(R11*1), R12
+	MOVL  R12, (BX)(R11*1)
+	ADDQ  $0x04, R11
+
+copy_4_qword:
+	TESTQ $0x00000008, R13
+	JZ    copy_4_test
+	MOVQ  (R14)(R11*1), R12
+	MOVQ  R12, (BX)(R11*1)
+	ADDQ  $0x08, R11
+	JMP   copy_4_test
+
+copy_4:
+	MOVUPS (R14)(R11*1), X0
+	MOVUPS X0, (BX)(R11*1)
+	ADDQ   $0x10, R11
+
+copy_4_test:
+	CMPQ R11, R13
+	JB   copy_4
+	ADDQ R13, DI
+	ADDQ R13, BX
+	ADDQ $0x18, AX
+	INCQ DX
+	CMPQ DX, CX
+	JB   main_loop
+	JMP  loop_finished
+
+copy_all_from_history:
+	XORQ  R15, R15
+	TESTQ $0x00000001, R11
+	JZ    copy_5_word
+	MOVB  (R14)(R15*1), BP
+	MOVB  BP, (BX)(R15*1)
+	ADDQ  $0x01, R15
+
+copy_5_word:
+	TESTQ $0x00000002, R11
+	JZ    copy_5_dword
+	MOVW  (R14)(R15*1), BP
+	MOVW  BP, (BX)(R15*1)
+	ADDQ  $0x02, R15
+
+copy_5_dword:
+	TESTQ $0x00000004, R11
+	JZ    copy_5_qword
+	MOVL  (R14)(R15*1), BP
+	MOVL  BP, (BX)(R15*1)
+	ADDQ  $0x04, R15
+
+copy_5_qword:
+	TESTQ $0x00000008, R11
+	JZ    copy_5_test
+	MOVQ  (R14)(R15*1), BP
+	MOVQ  BP, (BX)(R15*1)
+	ADDQ  $0x08, R15
+	JMP   copy_5_test
+
+copy_5:
+	MOVUPS (R14)(R15*1), X0
+	MOVUPS X0, (BX)(R15*1)
+	ADDQ   $0x10, R15
+
+copy_5_test:
+	CMPQ R15, R11
+	JB   copy_5
+	ADDQ R11, BX
+	ADDQ R11, DI
+	SUBQ R11, R13
+
+	// Copy match from the current buffer
+copy_match:
+	TESTQ R13, R13
+	JZ    handle_loop
+	MOVQ  BX, R11
+	SUBQ  R12, R11
+
+	// ml <= mo
+	CMPQ R13, R12
+	JA   copy_overlapping_match
+
+	// Copy non-overlapping match
+	ADDQ  R13, DI
+	XORQ  R12, R12
+	TESTQ $0x00000001, R13
+	JZ    copy_2_word
+	MOVB  (R11)(R12*1), R14
+	MOVB  R14, (BX)(R12*1)
+	ADDQ  $0x01, R12
+
+copy_2_word:
+	TESTQ $0x00000002, R13
+	JZ    copy_2_dword
+	MOVW  (R11)(R12*1), R14
+	MOVW  R14, (BX)(R12*1)
+	ADDQ  $0x02, R12
+
+copy_2_dword:
+	TESTQ $0x00000004, R13
+	JZ    copy_2_qword
+	MOVL  (R11)(R12*1), R14
+	MOVL  R14, (BX)(R12*1)
+	ADDQ  $0x04, R12
+
+copy_2_qword:
+	TESTQ $0x00000008, R13
+	JZ    copy_2_test
+	MOVQ  (R11)(R12*1), R14
+	MOVQ  R14, (BX)(R12*1)
+	ADDQ  $0x08, R12
+	JMP   copy_2_test
+
+copy_2:
+	MOVUPS (R11)(R12*1), X0
+	MOVUPS X0, (BX)(R12*1)
+	ADDQ   $0x10, R12
+
+copy_2_test:
+	CMPQ R12, R13
+	JB   copy_2
+	ADDQ R13, BX
+	JMP  handle_loop
+
+	// Copy overlapping match
+copy_overlapping_match:
+	ADDQ R13, DI
+
+copy_slow_3:
+	MOVB (R11), R12
+	MOVB R12, (BX)
+	INCQ R11
+	INCQ BX
+	DECQ R13
+	JNZ  copy_slow_3
+
+handle_loop:
+	ADDQ $0x18, AX
+	INCQ DX
+	CMPQ DX, CX
+	JB   main_loop
+
+loop_finished:
+	// Return value
+	MOVB $0x01, ret+8(FP)
+
+	// Update the context
+	MOVQ ctx+0(FP), AX
+	MOVQ DX, 24(AX)
+	MOVQ DI, 104(AX)
+	MOVQ 80(AX), CX
+	SUBQ CX, SI
+	MOVQ SI, 112(AX)
+	RET
+
+error_match_off_too_big:
+	// Return value
+	MOVB $0x00, ret+8(FP)
+
+	// Update the context
+	MOVQ ctx+0(FP), AX
+	MOVQ DX, 24(AX)
+	MOVQ DI, 104(AX)
+	MOVQ 80(AX), CX
+	SUBQ CX, SI
+	MOVQ SI, 112(AX)
+	RET
+
+empty_seqs:
+	// Return value
+	MOVB $0x01, ret+8(FP)
+	RET
+
+// func sequenceDecs_decodeSync_amd64(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int
+// Requires: CMOV, SSE
+TEXT ·sequenceDecs_decodeSync_amd64(SB), $64-32
+	MOVQ    br+8(FP), AX
+	MOVQ    32(AX), DX
+	MOVBQZX 40(AX), BX
+	MOVQ    24(AX), SI
+	MOVQ    (AX), AX
+	ADDQ    SI, AX
+	MOVQ    AX, (SP)
+	MOVQ    ctx+16(FP), AX
+	MOVQ    72(AX), DI
+	MOVQ    80(AX), R8
+	MOVQ    88(AX), R9
+	MOVQ    112(AX), R10
+	MOVQ    128(AX), CX
+	MOVQ    CX, 32(SP)
+	MOVQ    144(AX), R11
+	MOVQ    136(AX), R12
+	MOVQ    200(AX), CX
+	MOVQ    CX, 56(SP)
+	MOVQ    176(AX), CX
+	MOVQ    CX, 48(SP)
+	MOVQ    184(AX), AX
+	MOVQ    AX, 40(SP)
+	MOVQ    40(SP), AX
+	ADDQ    AX, 48(SP)
+
+	// Calculate poiter to s.out[cap(s.out)] (a past-end pointer)
+	ADDQ R10, 32(SP)
+
+	// outBase += outPosition
+	ADDQ R12, R10
+
+sequenceDecs_decodeSync_amd64_main_loop:
+	MOVQ (SP), R13
+
+	// Fill bitreader to have enough for the offset and match length.
+	CMPQ SI, $0x08
+	JL   sequenceDecs_decodeSync_amd64_fill_byte_by_byte
+	MOVQ BX, AX
+	SHRQ $0x03, AX
+	SUBQ AX, R13
+	MOVQ (R13), DX
+	SUBQ AX, SI
+	ANDQ $0x07, BX
+	JMP  sequenceDecs_decodeSync_amd64_fill_end
+
+sequenceDecs_decodeSync_amd64_fill_byte_by_byte:
+	CMPQ    SI, $0x00
+	JLE     sequenceDecs_decodeSync_amd64_fill_end
+	CMPQ    BX, $0x07
+	JLE     sequenceDecs_decodeSync_amd64_fill_end
+	SHLQ    $0x08, DX
+	SUBQ    $0x01, R13
+	SUBQ    $0x01, SI
+	SUBQ    $0x08, BX
+	MOVBQZX (R13), AX
+	ORQ     AX, DX
+	JMP     sequenceDecs_decodeSync_amd64_fill_byte_by_byte
+
+sequenceDecs_decodeSync_amd64_fill_end:
+	// Update offset
+	MOVQ    R9, AX
+	MOVQ    BX, CX
+	MOVQ    DX, R14
+	SHLQ    CL, R14
+	MOVB    AH, CL
+	ADDQ    CX, BX
+	NEGL    CX
+	SHRQ    CL, R14
+	SHRQ    $0x20, AX
+	TESTQ   CX, CX
+	CMOVQEQ CX, R14
+	ADDQ    R14, AX
+	MOVQ    AX, 8(SP)
+
+	// Update match length
+	MOVQ    R8, AX
+	MOVQ    BX, CX
+	MOVQ    DX, R14
+	SHLQ    CL, R14
+	MOVB    AH, CL
+	ADDQ    CX, BX
+	NEGL    CX
+	SHRQ    CL, R14
+	SHRQ    $0x20, AX
+	TESTQ   CX, CX
+	CMOVQEQ CX, R14
+	ADDQ    R14, AX
+	MOVQ    AX, 16(SP)
+
+	// Fill bitreader to have enough for the remaining
+	CMPQ SI, $0x08
+	JL   sequenceDecs_decodeSync_amd64_fill_2_byte_by_byte
+	MOVQ BX, AX
+	SHRQ $0x03, AX
+	SUBQ AX, R13
+	MOVQ (R13), DX
+	SUBQ AX, SI
+	ANDQ $0x07, BX
+	JMP  sequenceDecs_decodeSync_amd64_fill_2_end
+
+sequenceDecs_decodeSync_amd64_fill_2_byte_by_byte:
+	CMPQ    SI, $0x00
+	JLE     sequenceDecs_decodeSync_amd64_fill_2_end
+	CMPQ    BX, $0x07
+	JLE     sequenceDecs_decodeSync_amd64_fill_2_end
+	SHLQ    $0x08, DX
+	SUBQ    $0x01, R13
+	SUBQ    $0x01, SI
+	SUBQ    $0x08, BX
+	MOVBQZX (R13), AX
+	ORQ     AX, DX
+	JMP     sequenceDecs_decodeSync_amd64_fill_2_byte_by_byte
+
+sequenceDecs_decodeSync_amd64_fill_2_end:
+	// Update literal length
+	MOVQ    DI, AX
+	MOVQ    BX, CX
+	MOVQ    DX, R14
+	SHLQ    CL, R14
+	MOVB    AH, CL
+	ADDQ    CX, BX
+	NEGL    CX
+	SHRQ    CL, R14
+	SHRQ    $0x20, AX
+	TESTQ   CX, CX
+	CMOVQEQ CX, R14
+	ADDQ    R14, AX
+	MOVQ    AX, 24(SP)
+
+	// Fill bitreader for state updates
+	MOVQ    R13, (SP)
+	MOVQ    R9, AX
+	SHRQ    $0x08, AX
+	MOVBQZX AL, AX
+	MOVQ    ctx+16(FP), CX
+	CMPQ    96(CX), $0x00
+	JZ      sequenceDecs_decodeSync_amd64_skip_update
+
+	// Update Literal Length State
+	MOVBQZX DI, R13
+	SHRQ    $0x10, DI
+	MOVWQZX DI, DI
+	CMPQ    R13, $0x00
+	JZ      sequenceDecs_decodeSync_amd64_llState_updateState_skip_zero
+	MOVQ    BX, CX
+	ADDQ    R13, BX
+	MOVQ    DX, R14
+	SHLQ    CL, R14
+	MOVQ    R13, CX
+	NEGQ    CX
+	SHRQ    CL, R14
+	ADDQ    R14, DI
+
+sequenceDecs_decodeSync_amd64_llState_updateState_skip_zero:
+	// Load ctx.llTable
+	MOVQ ctx+16(FP), CX
+	MOVQ (CX), CX
+	MOVQ (CX)(DI*8), DI
+
+	// Update Match Length State
+	MOVBQZX R8, R13
+	SHRQ    $0x10, R8
+	MOVWQZX R8, R8
+	CMPQ    R13, $0x00
+	JZ      sequenceDecs_decodeSync_amd64_mlState_updateState_skip_zero
+	MOVQ    BX, CX
+	ADDQ    R13, BX
+	MOVQ    DX, R14
+	SHLQ    CL, R14
+	MOVQ    R13, CX
+	NEGQ    CX
+	SHRQ    CL, R14
+	ADDQ    R14, R8
+
+sequenceDecs_decodeSync_amd64_mlState_updateState_skip_zero:
+	// Load ctx.mlTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 24(CX), CX
+	MOVQ (CX)(R8*8), R8
+
+	// Update Offset State
+	MOVBQZX R9, R13
+	SHRQ    $0x10, R9
+	MOVWQZX R9, R9
+	CMPQ    R13, $0x00
+	JZ      sequenceDecs_decodeSync_amd64_ofState_updateState_skip_zero
+	MOVQ    BX, CX
+	ADDQ    R13, BX
+	MOVQ    DX, R14
+	SHLQ    CL, R14
+	MOVQ    R13, CX
+	NEGQ    CX
+	SHRQ    CL, R14
+	ADDQ    R14, R9
+
+sequenceDecs_decodeSync_amd64_ofState_updateState_skip_zero:
+	// Load ctx.ofTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 48(CX), CX
+	MOVQ (CX)(R9*8), R9
+
+sequenceDecs_decodeSync_amd64_skip_update:
+	// Adjust offset
+	MOVQ   s+0(FP), CX
+	MOVQ   8(SP), R13
+	CMPQ   AX, $0x01
+	JBE    sequenceDecs_decodeSync_amd64_adjust_offsetB_1_or_0
+	MOVUPS 144(CX), X0
+	MOVQ   R13, 144(CX)
+	MOVUPS X0, 152(CX)
+	JMP    sequenceDecs_decodeSync_amd64_adjust_end
+
+sequenceDecs_decodeSync_amd64_adjust_offsetB_1_or_0:
+	CMPQ 24(SP), $0x00000000
+	JNE  sequenceDecs_decodeSync_amd64_adjust_offset_maybezero
+	INCQ R13
+	JMP  sequenceDecs_decodeSync_amd64_adjust_offset_nonzero
+
+sequenceDecs_decodeSync_amd64_adjust_offset_maybezero:
+	TESTQ R13, R13
+	JNZ   sequenceDecs_decodeSync_amd64_adjust_offset_nonzero
+	MOVQ  144(CX), R13
+	JMP   sequenceDecs_decodeSync_amd64_adjust_end
+
+sequenceDecs_decodeSync_amd64_adjust_offset_nonzero:
+	MOVQ    R13, AX
+	XORQ    R14, R14
+	MOVQ    $-1, R15
+	CMPQ    R13, $0x03
+	CMOVQEQ R14, AX
+	CMOVQEQ R15, R14
+	LEAQ    144(CX), R15
+	ADDQ    (R15)(AX*8), R14
+	JNZ     sequenceDecs_decodeSync_amd64_adjust_temp_valid
+	MOVQ    $0x00000001, R14
+
+sequenceDecs_decodeSync_amd64_adjust_temp_valid:
+	CMPQ R13, $0x01
+	JZ   sequenceDecs_decodeSync_amd64_adjust_skip
+	MOVQ 152(CX), AX
+	MOVQ AX, 160(CX)
+
+sequenceDecs_decodeSync_amd64_adjust_skip:
+	MOVQ 144(CX), AX
+	MOVQ AX, 152(CX)
+	MOVQ R14, 144(CX)
+	MOVQ R14, R13
+
+sequenceDecs_decodeSync_amd64_adjust_end:
+	MOVQ R13, 8(SP)
+
+	// Check values
+	MOVQ  16(SP), AX
+	MOVQ  24(SP), CX
+	LEAQ  (AX)(CX*1), R14
+	MOVQ  s+0(FP), R15
+	ADDQ  R14, 256(R15)
+	MOVQ  ctx+16(FP), R14
+	SUBQ  CX, 104(R14)
+	JS    error_not_enough_literals
+	CMPQ  AX, $0x00020002
+	JA    sequenceDecs_decodeSync_amd64_error_match_len_too_big
+	TESTQ R13, R13
+	JNZ   sequenceDecs_decodeSync_amd64_match_len_ofs_ok
+	TESTQ AX, AX
+	JNZ   sequenceDecs_decodeSync_amd64_error_match_len_ofs_mismatch
+
+sequenceDecs_decodeSync_amd64_match_len_ofs_ok:
+	MOVQ 24(SP), AX
+	MOVQ 8(SP), CX
+	MOVQ 16(SP), R13
+
+	// Check if we have enough space in s.out
+	LEAQ (AX)(R13*1), R14
+	ADDQ R10, R14
+	CMPQ R14, 32(SP)
+	JA   error_not_enough_space
+
+	// Copy literals
+	TESTQ AX, AX
+	JZ    check_offset
+	XORQ  R14, R14
+
+copy_1:
+	MOVUPS (R11)(R14*1), X0
+	MOVUPS X0, (R10)(R14*1)
+	ADDQ   $0x10, R14
+	CMPQ   R14, AX
+	JB     copy_1
+	ADDQ   AX, R11
+	ADDQ   AX, R10
+	ADDQ   AX, R12
+
+	// Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize)
+check_offset:
+	MOVQ R12, AX
+	ADDQ 40(SP), AX
+	CMPQ CX, AX
+	JG   error_match_off_too_big
+	CMPQ CX, 56(SP)
+	JG   error_match_off_too_big
+
+	// Copy match from history
+	MOVQ  CX, AX
+	SUBQ  R12, AX
+	JLS   copy_match
+	MOVQ  48(SP), R14
+	SUBQ  AX, R14
+	CMPQ  R13, AX
+	JGE   copy_all_from_history
+	XORQ  AX, AX
+	TESTQ $0x00000001, R13
+	JZ    copy_4_word
+	MOVB  (R14)(AX*1), CL
+	MOVB  CL, (R10)(AX*1)
+	ADDQ  $0x01, AX
+
+copy_4_word:
+	TESTQ $0x00000002, R13
+	JZ    copy_4_dword
+	MOVW  (R14)(AX*1), CX
+	MOVW  CX, (R10)(AX*1)
+	ADDQ  $0x02, AX
+
+copy_4_dword:
+	TESTQ $0x00000004, R13
+	JZ    copy_4_qword
+	MOVL  (R14)(AX*1), CX
+	MOVL  CX, (R10)(AX*1)
+	ADDQ  $0x04, AX
+
+copy_4_qword:
+	TESTQ $0x00000008, R13
+	JZ    copy_4_test
+	MOVQ  (R14)(AX*1), CX
+	MOVQ  CX, (R10)(AX*1)
+	ADDQ  $0x08, AX
+	JMP   copy_4_test
+
+copy_4:
+	MOVUPS (R14)(AX*1), X0
+	MOVUPS X0, (R10)(AX*1)
+	ADDQ   $0x10, AX
+
+copy_4_test:
+	CMPQ AX, R13
+	JB   copy_4
+	ADDQ R13, R12
+	ADDQ R13, R10
+	JMP  handle_loop
+	JMP loop_finished
+
+copy_all_from_history:
+	XORQ  R15, R15
+	TESTQ $0x00000001, AX
+	JZ    copy_5_word
+	MOVB  (R14)(R15*1), BP
+	MOVB  BP, (R10)(R15*1)
+	ADDQ  $0x01, R15
+
+copy_5_word:
+	TESTQ $0x00000002, AX
+	JZ    copy_5_dword
+	MOVW  (R14)(R15*1), BP
+	MOVW  BP, (R10)(R15*1)
+	ADDQ  $0x02, R15
+
+copy_5_dword:
+	TESTQ $0x00000004, AX
+	JZ    copy_5_qword
+	MOVL  (R14)(R15*1), BP
+	MOVL  BP, (R10)(R15*1)
+	ADDQ  $0x04, R15
+
+copy_5_qword:
+	TESTQ $0x00000008, AX
+	JZ    copy_5_test
+	MOVQ  (R14)(R15*1), BP
+	MOVQ  BP, (R10)(R15*1)
+	ADDQ  $0x08, R15
+	JMP   copy_5_test
+
+copy_5:
+	MOVUPS (R14)(R15*1), X0
+	MOVUPS X0, (R10)(R15*1)
+	ADDQ   $0x10, R15
+
+copy_5_test:
+	CMPQ R15, AX
+	JB   copy_5
+	ADDQ AX, R10
+	ADDQ AX, R12
+	SUBQ AX, R13
+
+	// Copy match from the current buffer
+copy_match:
+	TESTQ R13, R13
+	JZ    handle_loop
+	MOVQ  R10, AX
+	SUBQ  CX, AX
+
+	// ml <= mo
+	CMPQ R13, CX
+	JA   copy_overlapping_match
+
+	// Copy non-overlapping match
+	ADDQ R13, R12
+	MOVQ R10, CX
+	ADDQ R13, R10
+
+copy_2:
+	MOVUPS (AX), X0
+	MOVUPS X0, (CX)
+	ADDQ   $0x10, AX
+	ADDQ   $0x10, CX
+	SUBQ   $0x10, R13
+	JHI    copy_2
+	JMP    handle_loop
+
+	// Copy overlapping match
+copy_overlapping_match:
+	ADDQ R13, R12
+
+copy_slow_3:
+	MOVB (AX), CL
+	MOVB CL, (R10)
+	INCQ AX
+	INCQ R10
+	DECQ R13
+	JNZ  copy_slow_3
+
+handle_loop:
+	MOVQ ctx+16(FP), AX
+	DECQ 96(AX)
+	JNS  sequenceDecs_decodeSync_amd64_main_loop
+
+loop_finished:
+	MOVQ br+8(FP), AX
+	MOVQ DX, 32(AX)
+	MOVB BL, 40(AX)
+	MOVQ SI, 24(AX)
+
+	// Update the context
+	MOVQ ctx+16(FP), AX
+	MOVQ R12, 136(AX)
+	MOVQ 144(AX), CX
+	SUBQ CX, R11
+	MOVQ R11, 168(AX)
+
+	// Return success
+	MOVQ $0x00000000, ret+24(FP)
+	RET
+
+	// Return with match length error
+sequenceDecs_decodeSync_amd64_error_match_len_ofs_mismatch:
+	MOVQ 16(SP), AX
+	MOVQ ctx+16(FP), CX
+	MOVQ AX, 216(CX)
+	MOVQ $0x00000001, ret+24(FP)
+	RET
+
+	// Return with match too long error
+sequenceDecs_decodeSync_amd64_error_match_len_too_big:
+	MOVQ ctx+16(FP), AX
+	MOVQ 16(SP), CX
+	MOVQ CX, 216(AX)
+	MOVQ $0x00000002, ret+24(FP)
+	RET
+
+	// Return with match offset too long error
+error_match_off_too_big:
+	MOVQ ctx+16(FP), AX
+	MOVQ 8(SP), CX
+	MOVQ CX, 224(AX)
+	MOVQ R12, 136(AX)
+	MOVQ $0x00000003, ret+24(FP)
+	RET
+
+	// Return with not enough literals error
+error_not_enough_literals:
+	MOVQ ctx+16(FP), AX
+	MOVQ 24(SP), CX
+	MOVQ CX, 208(AX)
+	MOVQ $0x00000004, ret+24(FP)
+	RET
+
+	// Return with not enough output space error
+error_not_enough_space:
+	MOVQ ctx+16(FP), AX
+	MOVQ 24(SP), CX
+	MOVQ CX, 208(AX)
+	MOVQ 16(SP), CX
+	MOVQ CX, 216(AX)
+	MOVQ R12, 136(AX)
+	MOVQ $0x00000005, ret+24(FP)
+	RET
+
+// func sequenceDecs_decodeSync_bmi2(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int
+// Requires: BMI, BMI2, CMOV, SSE
+TEXT ·sequenceDecs_decodeSync_bmi2(SB), $64-32
+	MOVQ    br+8(FP), CX
+	MOVQ    32(CX), AX
+	MOVBQZX 40(CX), DX
+	MOVQ    24(CX), BX
+	MOVQ    (CX), CX
+	ADDQ    BX, CX
+	MOVQ    CX, (SP)
+	MOVQ    ctx+16(FP), CX
+	MOVQ    72(CX), SI
+	MOVQ    80(CX), DI
+	MOVQ    88(CX), R8
+	MOVQ    112(CX), R9
+	MOVQ    128(CX), R10
+	MOVQ    R10, 32(SP)
+	MOVQ    144(CX), R10
+	MOVQ    136(CX), R11
+	MOVQ    200(CX), R12
+	MOVQ    R12, 56(SP)
+	MOVQ    176(CX), R12
+	MOVQ    R12, 48(SP)
+	MOVQ    184(CX), CX
+	MOVQ    CX, 40(SP)
+	MOVQ    40(SP), CX
+	ADDQ    CX, 48(SP)
+
+	// Calculate poiter to s.out[cap(s.out)] (a past-end pointer)
+	ADDQ R9, 32(SP)
+
+	// outBase += outPosition
+	ADDQ R11, R9
+
+sequenceDecs_decodeSync_bmi2_main_loop:
+	MOVQ (SP), R12
+
+	// Fill bitreader to have enough for the offset and match length.
+	CMPQ BX, $0x08
+	JL   sequenceDecs_decodeSync_bmi2_fill_byte_by_byte
+	MOVQ DX, CX
+	SHRQ $0x03, CX
+	SUBQ CX, R12
+	MOVQ (R12), AX
+	SUBQ CX, BX
+	ANDQ $0x07, DX
+	JMP  sequenceDecs_decodeSync_bmi2_fill_end
+
+sequenceDecs_decodeSync_bmi2_fill_byte_by_byte:
+	CMPQ    BX, $0x00
+	JLE     sequenceDecs_decodeSync_bmi2_fill_end
+	CMPQ    DX, $0x07
+	JLE     sequenceDecs_decodeSync_bmi2_fill_end
+	SHLQ    $0x08, AX
+	SUBQ    $0x01, R12
+	SUBQ    $0x01, BX
+	SUBQ    $0x08, DX
+	MOVBQZX (R12), CX
+	ORQ     CX, AX
+	JMP     sequenceDecs_decodeSync_bmi2_fill_byte_by_byte
+
+sequenceDecs_decodeSync_bmi2_fill_end:
+	// Update offset
+	MOVQ   $0x00000808, CX
+	BEXTRQ CX, R8, R13
+	MOVQ   AX, R14
+	LEAQ   (DX)(R13*1), CX
+	ROLQ   CL, R14
+	BZHIQ  R13, R14, R14
+	MOVQ   CX, DX
+	MOVQ   R8, CX
+	SHRQ   $0x20, CX
+	ADDQ   R14, CX
+	MOVQ   CX, 8(SP)
+
+	// Update match length
+	MOVQ   $0x00000808, CX
+	BEXTRQ CX, DI, R13
+	MOVQ   AX, R14
+	LEAQ   (DX)(R13*1), CX
+	ROLQ   CL, R14
+	BZHIQ  R13, R14, R14
+	MOVQ   CX, DX
+	MOVQ   DI, CX
+	SHRQ   $0x20, CX
+	ADDQ   R14, CX
+	MOVQ   CX, 16(SP)
+
+	// Fill bitreader to have enough for the remaining
+	CMPQ BX, $0x08
+	JL   sequenceDecs_decodeSync_bmi2_fill_2_byte_by_byte
+	MOVQ DX, CX
+	SHRQ $0x03, CX
+	SUBQ CX, R12
+	MOVQ (R12), AX
+	SUBQ CX, BX
+	ANDQ $0x07, DX
+	JMP  sequenceDecs_decodeSync_bmi2_fill_2_end
+
+sequenceDecs_decodeSync_bmi2_fill_2_byte_by_byte:
+	CMPQ    BX, $0x00
+	JLE     sequenceDecs_decodeSync_bmi2_fill_2_end
+	CMPQ    DX, $0x07
+	JLE     sequenceDecs_decodeSync_bmi2_fill_2_end
+	SHLQ    $0x08, AX
+	SUBQ    $0x01, R12
+	SUBQ    $0x01, BX
+	SUBQ    $0x08, DX
+	MOVBQZX (R12), CX
+	ORQ     CX, AX
+	JMP     sequenceDecs_decodeSync_bmi2_fill_2_byte_by_byte
+
+sequenceDecs_decodeSync_bmi2_fill_2_end:
+	// Update literal length
+	MOVQ   $0x00000808, CX
+	BEXTRQ CX, SI, R13
+	MOVQ   AX, R14
+	LEAQ   (DX)(R13*1), CX
+	ROLQ   CL, R14
+	BZHIQ  R13, R14, R14
+	MOVQ   CX, DX
+	MOVQ   SI, CX
+	SHRQ   $0x20, CX
+	ADDQ   R14, CX
+	MOVQ   CX, 24(SP)
+
+	// Fill bitreader for state updates
+	MOVQ    R12, (SP)
+	MOVQ    $0x00000808, CX
+	BEXTRQ  CX, R8, R12
+	MOVQ    ctx+16(FP), CX
+	CMPQ    96(CX), $0x00
+	JZ      sequenceDecs_decodeSync_bmi2_skip_update
+	LEAQ    (SI)(DI*1), R13
+	ADDQ    R8, R13
+	MOVBQZX R13, R13
+	LEAQ    (DX)(R13*1), CX
+	MOVQ    AX, R14
+	MOVQ    CX, DX
+	ROLQ    CL, R14
+	BZHIQ   R13, R14, R14
+
+	// Update Offset State
+	BZHIQ  R8, R14, CX
+	SHRXQ  R8, R14, R14
+	MOVQ   $0x00001010, R13
+	BEXTRQ R13, R8, R8
+	ADDQ   CX, R8
+
+	// Load ctx.ofTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 48(CX), CX
+	MOVQ (CX)(R8*8), R8
+
+	// Update Match Length State
+	BZHIQ  DI, R14, CX
+	SHRXQ  DI, R14, R14
+	MOVQ   $0x00001010, R13
+	BEXTRQ R13, DI, DI
+	ADDQ   CX, DI
+
+	// Load ctx.mlTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 24(CX), CX
+	MOVQ (CX)(DI*8), DI
+
+	// Update Literal Length State
+	BZHIQ  SI, R14, CX
+	MOVQ   $0x00001010, R13
+	BEXTRQ R13, SI, SI
+	ADDQ   CX, SI
+
+	// Load ctx.llTable
+	MOVQ ctx+16(FP), CX
+	MOVQ (CX), CX
+	MOVQ (CX)(SI*8), SI
+
+sequenceDecs_decodeSync_bmi2_skip_update:
+	// Adjust offset
+	MOVQ   s+0(FP), CX
+	MOVQ   8(SP), R13
+	CMPQ   R12, $0x01
+	JBE    sequenceDecs_decodeSync_bmi2_adjust_offsetB_1_or_0
+	MOVUPS 144(CX), X0
+	MOVQ   R13, 144(CX)
+	MOVUPS X0, 152(CX)
+	JMP    sequenceDecs_decodeSync_bmi2_adjust_end
+
+sequenceDecs_decodeSync_bmi2_adjust_offsetB_1_or_0:
+	CMPQ 24(SP), $0x00000000
+	JNE  sequenceDecs_decodeSync_bmi2_adjust_offset_maybezero
+	INCQ R13
+	JMP  sequenceDecs_decodeSync_bmi2_adjust_offset_nonzero
+
+sequenceDecs_decodeSync_bmi2_adjust_offset_maybezero:
+	TESTQ R13, R13
+	JNZ   sequenceDecs_decodeSync_bmi2_adjust_offset_nonzero
+	MOVQ  144(CX), R13
+	JMP   sequenceDecs_decodeSync_bmi2_adjust_end
+
+sequenceDecs_decodeSync_bmi2_adjust_offset_nonzero:
+	MOVQ    R13, R12
+	XORQ    R14, R14
+	MOVQ    $-1, R15
+	CMPQ    R13, $0x03
+	CMOVQEQ R14, R12
+	CMOVQEQ R15, R14
+	LEAQ    144(CX), R15
+	ADDQ    (R15)(R12*8), R14
+	JNZ     sequenceDecs_decodeSync_bmi2_adjust_temp_valid
+	MOVQ    $0x00000001, R14
+
+sequenceDecs_decodeSync_bmi2_adjust_temp_valid:
+	CMPQ R13, $0x01
+	JZ   sequenceDecs_decodeSync_bmi2_adjust_skip
+	MOVQ 152(CX), R12
+	MOVQ R12, 160(CX)
+
+sequenceDecs_decodeSync_bmi2_adjust_skip:
+	MOVQ 144(CX), R12
+	MOVQ R12, 152(CX)
+	MOVQ R14, 144(CX)
+	MOVQ R14, R13
+
+sequenceDecs_decodeSync_bmi2_adjust_end:
+	MOVQ R13, 8(SP)
+
+	// Check values
+	MOVQ  16(SP), CX
+	MOVQ  24(SP), R12
+	LEAQ  (CX)(R12*1), R14
+	MOVQ  s+0(FP), R15
+	ADDQ  R14, 256(R15)
+	MOVQ  ctx+16(FP), R14
+	SUBQ  R12, 104(R14)
+	JS    error_not_enough_literals
+	CMPQ  CX, $0x00020002
+	JA    sequenceDecs_decodeSync_bmi2_error_match_len_too_big
+	TESTQ R13, R13
+	JNZ   sequenceDecs_decodeSync_bmi2_match_len_ofs_ok
+	TESTQ CX, CX
+	JNZ   sequenceDecs_decodeSync_bmi2_error_match_len_ofs_mismatch
+
+sequenceDecs_decodeSync_bmi2_match_len_ofs_ok:
+	MOVQ 24(SP), CX
+	MOVQ 8(SP), R12
+	MOVQ 16(SP), R13
+
+	// Check if we have enough space in s.out
+	LEAQ (CX)(R13*1), R14
+	ADDQ R9, R14
+	CMPQ R14, 32(SP)
+	JA   error_not_enough_space
+
+	// Copy literals
+	TESTQ CX, CX
+	JZ    check_offset
+	XORQ  R14, R14
+
+copy_1:
+	MOVUPS (R10)(R14*1), X0
+	MOVUPS X0, (R9)(R14*1)
+	ADDQ   $0x10, R14
+	CMPQ   R14, CX
+	JB     copy_1
+	ADDQ   CX, R10
+	ADDQ   CX, R9
+	ADDQ   CX, R11
+
+	// Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize)
+check_offset:
+	MOVQ R11, CX
+	ADDQ 40(SP), CX
+	CMPQ R12, CX
+	JG   error_match_off_too_big
+	CMPQ R12, 56(SP)
+	JG   error_match_off_too_big
+
+	// Copy match from history
+	MOVQ  R12, CX
+	SUBQ  R11, CX
+	JLS   copy_match
+	MOVQ  48(SP), R14
+	SUBQ  CX, R14
+	CMPQ  R13, CX
+	JGE   copy_all_from_history
+	XORQ  CX, CX
+	TESTQ $0x00000001, R13
+	JZ    copy_4_word
+	MOVB  (R14)(CX*1), R12
+	MOVB  R12, (R9)(CX*1)
+	ADDQ  $0x01, CX
+
+copy_4_word:
+	TESTQ $0x00000002, R13
+	JZ    copy_4_dword
+	MOVW  (R14)(CX*1), R12
+	MOVW  R12, (R9)(CX*1)
+	ADDQ  $0x02, CX
+
+copy_4_dword:
+	TESTQ $0x00000004, R13
+	JZ    copy_4_qword
+	MOVL  (R14)(CX*1), R12
+	MOVL  R12, (R9)(CX*1)
+	ADDQ  $0x04, CX
+
+copy_4_qword:
+	TESTQ $0x00000008, R13
+	JZ    copy_4_test
+	MOVQ  (R14)(CX*1), R12
+	MOVQ  R12, (R9)(CX*1)
+	ADDQ  $0x08, CX
+	JMP   copy_4_test
+
+copy_4:
+	MOVUPS (R14)(CX*1), X0
+	MOVUPS X0, (R9)(CX*1)
+	ADDQ   $0x10, CX
+
+copy_4_test:
+	CMPQ CX, R13
+	JB   copy_4
+	ADDQ R13, R11
+	ADDQ R13, R9
+	JMP  handle_loop
+	JMP loop_finished
+
+copy_all_from_history:
+	XORQ  R15, R15
+	TESTQ $0x00000001, CX
+	JZ    copy_5_word
+	MOVB  (R14)(R15*1), BP
+	MOVB  BP, (R9)(R15*1)
+	ADDQ  $0x01, R15
+
+copy_5_word:
+	TESTQ $0x00000002, CX
+	JZ    copy_5_dword
+	MOVW  (R14)(R15*1), BP
+	MOVW  BP, (R9)(R15*1)
+	ADDQ  $0x02, R15
+
+copy_5_dword:
+	TESTQ $0x00000004, CX
+	JZ    copy_5_qword
+	MOVL  (R14)(R15*1), BP
+	MOVL  BP, (R9)(R15*1)
+	ADDQ  $0x04, R15
+
+copy_5_qword:
+	TESTQ $0x00000008, CX
+	JZ    copy_5_test
+	MOVQ  (R14)(R15*1), BP
+	MOVQ  BP, (R9)(R15*1)
+	ADDQ  $0x08, R15
+	JMP   copy_5_test
+
+copy_5:
+	MOVUPS (R14)(R15*1), X0
+	MOVUPS X0, (R9)(R15*1)
+	ADDQ   $0x10, R15
+
+copy_5_test:
+	CMPQ R15, CX
+	JB   copy_5
+	ADDQ CX, R9
+	ADDQ CX, R11
+	SUBQ CX, R13
+
+	// Copy match from the current buffer
+copy_match:
+	TESTQ R13, R13
+	JZ    handle_loop
+	MOVQ  R9, CX
+	SUBQ  R12, CX
+
+	// ml <= mo
+	CMPQ R13, R12
+	JA   copy_overlapping_match
+
+	// Copy non-overlapping match
+	ADDQ R13, R11
+	MOVQ R9, R12
+	ADDQ R13, R9
+
+copy_2:
+	MOVUPS (CX), X0
+	MOVUPS X0, (R12)
+	ADDQ   $0x10, CX
+	ADDQ   $0x10, R12
+	SUBQ   $0x10, R13
+	JHI    copy_2
+	JMP    handle_loop
+
+	// Copy overlapping match
+copy_overlapping_match:
+	ADDQ R13, R11
+
+copy_slow_3:
+	MOVB (CX), R12
+	MOVB R12, (R9)
+	INCQ CX
+	INCQ R9
+	DECQ R13
+	JNZ  copy_slow_3
+
+handle_loop:
+	MOVQ ctx+16(FP), CX
+	DECQ 96(CX)
+	JNS  sequenceDecs_decodeSync_bmi2_main_loop
+
+loop_finished:
+	MOVQ br+8(FP), CX
+	MOVQ AX, 32(CX)
+	MOVB DL, 40(CX)
+	MOVQ BX, 24(CX)
+
+	// Update the context
+	MOVQ ctx+16(FP), AX
+	MOVQ R11, 136(AX)
+	MOVQ 144(AX), CX
+	SUBQ CX, R10
+	MOVQ R10, 168(AX)
+
+	// Return success
+	MOVQ $0x00000000, ret+24(FP)
+	RET
+
+	// Return with match length error
+sequenceDecs_decodeSync_bmi2_error_match_len_ofs_mismatch:
+	MOVQ 16(SP), AX
+	MOVQ ctx+16(FP), CX
+	MOVQ AX, 216(CX)
+	MOVQ $0x00000001, ret+24(FP)
+	RET
+
+	// Return with match too long error
+sequenceDecs_decodeSync_bmi2_error_match_len_too_big:
+	MOVQ ctx+16(FP), AX
+	MOVQ 16(SP), CX
+	MOVQ CX, 216(AX)
+	MOVQ $0x00000002, ret+24(FP)
+	RET
+
+	// Return with match offset too long error
+error_match_off_too_big:
+	MOVQ ctx+16(FP), AX
+	MOVQ 8(SP), CX
+	MOVQ CX, 224(AX)
+	MOVQ R11, 136(AX)
+	MOVQ $0x00000003, ret+24(FP)
+	RET
+
+	// Return with not enough literals error
+error_not_enough_literals:
+	MOVQ ctx+16(FP), AX
+	MOVQ 24(SP), CX
+	MOVQ CX, 208(AX)
+	MOVQ $0x00000004, ret+24(FP)
+	RET
+
+	// Return with not enough output space error
+error_not_enough_space:
+	MOVQ ctx+16(FP), AX
+	MOVQ 24(SP), CX
+	MOVQ CX, 208(AX)
+	MOVQ 16(SP), CX
+	MOVQ CX, 216(AX)
+	MOVQ R11, 136(AX)
+	MOVQ $0x00000005, ret+24(FP)
+	RET
+
+// func sequenceDecs_decodeSync_safe_amd64(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int
+// Requires: CMOV, SSE
+TEXT ·sequenceDecs_decodeSync_safe_amd64(SB), $64-32
+	MOVQ    br+8(FP), AX
+	MOVQ    32(AX), DX
+	MOVBQZX 40(AX), BX
+	MOVQ    24(AX), SI
+	MOVQ    (AX), AX
+	ADDQ    SI, AX
+	MOVQ    AX, (SP)
+	MOVQ    ctx+16(FP), AX
+	MOVQ    72(AX), DI
+	MOVQ    80(AX), R8
+	MOVQ    88(AX), R9
+	MOVQ    112(AX), R10
+	MOVQ    128(AX), CX
+	MOVQ    CX, 32(SP)
+	MOVQ    144(AX), R11
+	MOVQ    136(AX), R12
+	MOVQ    200(AX), CX
+	MOVQ    CX, 56(SP)
+	MOVQ    176(AX), CX
+	MOVQ    CX, 48(SP)
+	MOVQ    184(AX), AX
+	MOVQ    AX, 40(SP)
+	MOVQ    40(SP), AX
+	ADDQ    AX, 48(SP)
+
+	// Calculate poiter to s.out[cap(s.out)] (a past-end pointer)
+	ADDQ R10, 32(SP)
+
+	// outBase += outPosition
+	ADDQ R12, R10
+
+sequenceDecs_decodeSync_safe_amd64_main_loop:
+	MOVQ (SP), R13
+
+	// Fill bitreader to have enough for the offset and match length.
+	CMPQ SI, $0x08
+	JL   sequenceDecs_decodeSync_safe_amd64_fill_byte_by_byte
+	MOVQ BX, AX
+	SHRQ $0x03, AX
+	SUBQ AX, R13
+	MOVQ (R13), DX
+	SUBQ AX, SI
+	ANDQ $0x07, BX
+	JMP  sequenceDecs_decodeSync_safe_amd64_fill_end
+
+sequenceDecs_decodeSync_safe_amd64_fill_byte_by_byte:
+	CMPQ    SI, $0x00
+	JLE     sequenceDecs_decodeSync_safe_amd64_fill_end
+	CMPQ    BX, $0x07
+	JLE     sequenceDecs_decodeSync_safe_amd64_fill_end
+	SHLQ    $0x08, DX
+	SUBQ    $0x01, R13
+	SUBQ    $0x01, SI
+	SUBQ    $0x08, BX
+	MOVBQZX (R13), AX
+	ORQ     AX, DX
+	JMP     sequenceDecs_decodeSync_safe_amd64_fill_byte_by_byte
+
+sequenceDecs_decodeSync_safe_amd64_fill_end:
+	// Update offset
+	MOVQ    R9, AX
+	MOVQ    BX, CX
+	MOVQ    DX, R14
+	SHLQ    CL, R14
+	MOVB    AH, CL
+	ADDQ    CX, BX
+	NEGL    CX
+	SHRQ    CL, R14
+	SHRQ    $0x20, AX
+	TESTQ   CX, CX
+	CMOVQEQ CX, R14
+	ADDQ    R14, AX
+	MOVQ    AX, 8(SP)
+
+	// Update match length
+	MOVQ    R8, AX
+	MOVQ    BX, CX
+	MOVQ    DX, R14
+	SHLQ    CL, R14
+	MOVB    AH, CL
+	ADDQ    CX, BX
+	NEGL    CX
+	SHRQ    CL, R14
+	SHRQ    $0x20, AX
+	TESTQ   CX, CX
+	CMOVQEQ CX, R14
+	ADDQ    R14, AX
+	MOVQ    AX, 16(SP)
+
+	// Fill bitreader to have enough for the remaining
+	CMPQ SI, $0x08
+	JL   sequenceDecs_decodeSync_safe_amd64_fill_2_byte_by_byte
+	MOVQ BX, AX
+	SHRQ $0x03, AX
+	SUBQ AX, R13
+	MOVQ (R13), DX
+	SUBQ AX, SI
+	ANDQ $0x07, BX
+	JMP  sequenceDecs_decodeSync_safe_amd64_fill_2_end
+
+sequenceDecs_decodeSync_safe_amd64_fill_2_byte_by_byte:
+	CMPQ    SI, $0x00
+	JLE     sequenceDecs_decodeSync_safe_amd64_fill_2_end
+	CMPQ    BX, $0x07
+	JLE     sequenceDecs_decodeSync_safe_amd64_fill_2_end
+	SHLQ    $0x08, DX
+	SUBQ    $0x01, R13
+	SUBQ    $0x01, SI
+	SUBQ    $0x08, BX
+	MOVBQZX (R13), AX
+	ORQ     AX, DX
+	JMP     sequenceDecs_decodeSync_safe_amd64_fill_2_byte_by_byte
+
+sequenceDecs_decodeSync_safe_amd64_fill_2_end:
+	// Update literal length
+	MOVQ    DI, AX
+	MOVQ    BX, CX
+	MOVQ    DX, R14
+	SHLQ    CL, R14
+	MOVB    AH, CL
+	ADDQ    CX, BX
+	NEGL    CX
+	SHRQ    CL, R14
+	SHRQ    $0x20, AX
+	TESTQ   CX, CX
+	CMOVQEQ CX, R14
+	ADDQ    R14, AX
+	MOVQ    AX, 24(SP)
+
+	// Fill bitreader for state updates
+	MOVQ    R13, (SP)
+	MOVQ    R9, AX
+	SHRQ    $0x08, AX
+	MOVBQZX AL, AX
+	MOVQ    ctx+16(FP), CX
+	CMPQ    96(CX), $0x00
+	JZ      sequenceDecs_decodeSync_safe_amd64_skip_update
+
+	// Update Literal Length State
+	MOVBQZX DI, R13
+	SHRQ    $0x10, DI
+	MOVWQZX DI, DI
+	CMPQ    R13, $0x00
+	JZ      sequenceDecs_decodeSync_safe_amd64_llState_updateState_skip_zero
+	MOVQ    BX, CX
+	ADDQ    R13, BX
+	MOVQ    DX, R14
+	SHLQ    CL, R14
+	MOVQ    R13, CX
+	NEGQ    CX
+	SHRQ    CL, R14
+	ADDQ    R14, DI
+
+sequenceDecs_decodeSync_safe_amd64_llState_updateState_skip_zero:
+	// Load ctx.llTable
+	MOVQ ctx+16(FP), CX
+	MOVQ (CX), CX
+	MOVQ (CX)(DI*8), DI
+
+	// Update Match Length State
+	MOVBQZX R8, R13
+	SHRQ    $0x10, R8
+	MOVWQZX R8, R8
+	CMPQ    R13, $0x00
+	JZ      sequenceDecs_decodeSync_safe_amd64_mlState_updateState_skip_zero
+	MOVQ    BX, CX
+	ADDQ    R13, BX
+	MOVQ    DX, R14
+	SHLQ    CL, R14
+	MOVQ    R13, CX
+	NEGQ    CX
+	SHRQ    CL, R14
+	ADDQ    R14, R8
+
+sequenceDecs_decodeSync_safe_amd64_mlState_updateState_skip_zero:
+	// Load ctx.mlTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 24(CX), CX
+	MOVQ (CX)(R8*8), R8
+
+	// Update Offset State
+	MOVBQZX R9, R13
+	SHRQ    $0x10, R9
+	MOVWQZX R9, R9
+	CMPQ    R13, $0x00
+	JZ      sequenceDecs_decodeSync_safe_amd64_ofState_updateState_skip_zero
+	MOVQ    BX, CX
+	ADDQ    R13, BX
+	MOVQ    DX, R14
+	SHLQ    CL, R14
+	MOVQ    R13, CX
+	NEGQ    CX
+	SHRQ    CL, R14
+	ADDQ    R14, R9
+
+sequenceDecs_decodeSync_safe_amd64_ofState_updateState_skip_zero:
+	// Load ctx.ofTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 48(CX), CX
+	MOVQ (CX)(R9*8), R9
+
+sequenceDecs_decodeSync_safe_amd64_skip_update:
+	// Adjust offset
+	MOVQ   s+0(FP), CX
+	MOVQ   8(SP), R13
+	CMPQ   AX, $0x01
+	JBE    sequenceDecs_decodeSync_safe_amd64_adjust_offsetB_1_or_0
+	MOVUPS 144(CX), X0
+	MOVQ   R13, 144(CX)
+	MOVUPS X0, 152(CX)
+	JMP    sequenceDecs_decodeSync_safe_amd64_adjust_end
+
+sequenceDecs_decodeSync_safe_amd64_adjust_offsetB_1_or_0:
+	CMPQ 24(SP), $0x00000000
+	JNE  sequenceDecs_decodeSync_safe_amd64_adjust_offset_maybezero
+	INCQ R13
+	JMP  sequenceDecs_decodeSync_safe_amd64_adjust_offset_nonzero
+
+sequenceDecs_decodeSync_safe_amd64_adjust_offset_maybezero:
+	TESTQ R13, R13
+	JNZ   sequenceDecs_decodeSync_safe_amd64_adjust_offset_nonzero
+	MOVQ  144(CX), R13
+	JMP   sequenceDecs_decodeSync_safe_amd64_adjust_end
+
+sequenceDecs_decodeSync_safe_amd64_adjust_offset_nonzero:
+	MOVQ    R13, AX
+	XORQ    R14, R14
+	MOVQ    $-1, R15
+	CMPQ    R13, $0x03
+	CMOVQEQ R14, AX
+	CMOVQEQ R15, R14
+	LEAQ    144(CX), R15
+	ADDQ    (R15)(AX*8), R14
+	JNZ     sequenceDecs_decodeSync_safe_amd64_adjust_temp_valid
+	MOVQ    $0x00000001, R14
+
+sequenceDecs_decodeSync_safe_amd64_adjust_temp_valid:
+	CMPQ R13, $0x01
+	JZ   sequenceDecs_decodeSync_safe_amd64_adjust_skip
+	MOVQ 152(CX), AX
+	MOVQ AX, 160(CX)
+
+sequenceDecs_decodeSync_safe_amd64_adjust_skip:
+	MOVQ 144(CX), AX
+	MOVQ AX, 152(CX)
+	MOVQ R14, 144(CX)
+	MOVQ R14, R13
+
+sequenceDecs_decodeSync_safe_amd64_adjust_end:
+	MOVQ R13, 8(SP)
+
+	// Check values
+	MOVQ  16(SP), AX
+	MOVQ  24(SP), CX
+	LEAQ  (AX)(CX*1), R14
+	MOVQ  s+0(FP), R15
+	ADDQ  R14, 256(R15)
+	MOVQ  ctx+16(FP), R14
+	SUBQ  CX, 104(R14)
+	JS    error_not_enough_literals
+	CMPQ  AX, $0x00020002
+	JA    sequenceDecs_decodeSync_safe_amd64_error_match_len_too_big
+	TESTQ R13, R13
+	JNZ   sequenceDecs_decodeSync_safe_amd64_match_len_ofs_ok
+	TESTQ AX, AX
+	JNZ   sequenceDecs_decodeSync_safe_amd64_error_match_len_ofs_mismatch
+
+sequenceDecs_decodeSync_safe_amd64_match_len_ofs_ok:
+	MOVQ 24(SP), AX
+	MOVQ 8(SP), CX
+	MOVQ 16(SP), R13
+
+	// Check if we have enough space in s.out
+	LEAQ (AX)(R13*1), R14
+	ADDQ R10, R14
+	CMPQ R14, 32(SP)
+	JA   error_not_enough_space
+
+	// Copy literals
+	TESTQ AX, AX
+	JZ    check_offset
+	XORQ  R14, R14
+	TESTQ $0x00000001, AX
+	JZ    copy_1_word
+	MOVB  (R11)(R14*1), R15
+	MOVB  R15, (R10)(R14*1)
+	ADDQ  $0x01, R14
+
+copy_1_word:
+	TESTQ $0x00000002, AX
+	JZ    copy_1_dword
+	MOVW  (R11)(R14*1), R15
+	MOVW  R15, (R10)(R14*1)
+	ADDQ  $0x02, R14
+
+copy_1_dword:
+	TESTQ $0x00000004, AX
+	JZ    copy_1_qword
+	MOVL  (R11)(R14*1), R15
+	MOVL  R15, (R10)(R14*1)
+	ADDQ  $0x04, R14
+
+copy_1_qword:
+	TESTQ $0x00000008, AX
+	JZ    copy_1_test
+	MOVQ  (R11)(R14*1), R15
+	MOVQ  R15, (R10)(R14*1)
+	ADDQ  $0x08, R14
+	JMP   copy_1_test
+
+copy_1:
+	MOVUPS (R11)(R14*1), X0
+	MOVUPS X0, (R10)(R14*1)
+	ADDQ   $0x10, R14
+
+copy_1_test:
+	CMPQ R14, AX
+	JB   copy_1
+	ADDQ AX, R11
+	ADDQ AX, R10
+	ADDQ AX, R12
+
+	// Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize)
+check_offset:
+	MOVQ R12, AX
+	ADDQ 40(SP), AX
+	CMPQ CX, AX
+	JG   error_match_off_too_big
+	CMPQ CX, 56(SP)
+	JG   error_match_off_too_big
+
+	// Copy match from history
+	MOVQ  CX, AX
+	SUBQ  R12, AX
+	JLS   copy_match
+	MOVQ  48(SP), R14
+	SUBQ  AX, R14
+	CMPQ  R13, AX
+	JGE   copy_all_from_history
+	XORQ  AX, AX
+	TESTQ $0x00000001, R13
+	JZ    copy_4_word
+	MOVB  (R14)(AX*1), CL
+	MOVB  CL, (R10)(AX*1)
+	ADDQ  $0x01, AX
+
+copy_4_word:
+	TESTQ $0x00000002, R13
+	JZ    copy_4_dword
+	MOVW  (R14)(AX*1), CX
+	MOVW  CX, (R10)(AX*1)
+	ADDQ  $0x02, AX
+
+copy_4_dword:
+	TESTQ $0x00000004, R13
+	JZ    copy_4_qword
+	MOVL  (R14)(AX*1), CX
+	MOVL  CX, (R10)(AX*1)
+	ADDQ  $0x04, AX
+
+copy_4_qword:
+	TESTQ $0x00000008, R13
+	JZ    copy_4_test
+	MOVQ  (R14)(AX*1), CX
+	MOVQ  CX, (R10)(AX*1)
+	ADDQ  $0x08, AX
+	JMP   copy_4_test
+
+copy_4:
+	MOVUPS (R14)(AX*1), X0
+	MOVUPS X0, (R10)(AX*1)
+	ADDQ   $0x10, AX
+
+copy_4_test:
+	CMPQ AX, R13
+	JB   copy_4
+	ADDQ R13, R12
+	ADDQ R13, R10
+	JMP  handle_loop
+	JMP loop_finished
+
+copy_all_from_history:
+	XORQ  R15, R15
+	TESTQ $0x00000001, AX
+	JZ    copy_5_word
+	MOVB  (R14)(R15*1), BP
+	MOVB  BP, (R10)(R15*1)
+	ADDQ  $0x01, R15
+
+copy_5_word:
+	TESTQ $0x00000002, AX
+	JZ    copy_5_dword
+	MOVW  (R14)(R15*1), BP
+	MOVW  BP, (R10)(R15*1)
+	ADDQ  $0x02, R15
+
+copy_5_dword:
+	TESTQ $0x00000004, AX
+	JZ    copy_5_qword
+	MOVL  (R14)(R15*1), BP
+	MOVL  BP, (R10)(R15*1)
+	ADDQ  $0x04, R15
+
+copy_5_qword:
+	TESTQ $0x00000008, AX
+	JZ    copy_5_test
+	MOVQ  (R14)(R15*1), BP
+	MOVQ  BP, (R10)(R15*1)
+	ADDQ  $0x08, R15
+	JMP   copy_5_test
+
+copy_5:
+	MOVUPS (R14)(R15*1), X0
+	MOVUPS X0, (R10)(R15*1)
+	ADDQ   $0x10, R15
+
+copy_5_test:
+	CMPQ R15, AX
+	JB   copy_5
+	ADDQ AX, R10
+	ADDQ AX, R12
+	SUBQ AX, R13
+
+	// Copy match from the current buffer
+copy_match:
+	TESTQ R13, R13
+	JZ    handle_loop
+	MOVQ  R10, AX
+	SUBQ  CX, AX
+
+	// ml <= mo
+	CMPQ R13, CX
+	JA   copy_overlapping_match
+
+	// Copy non-overlapping match
+	ADDQ  R13, R12
+	XORQ  CX, CX
+	TESTQ $0x00000001, R13
+	JZ    copy_2_word
+	MOVB  (AX)(CX*1), R14
+	MOVB  R14, (R10)(CX*1)
+	ADDQ  $0x01, CX
+
+copy_2_word:
+	TESTQ $0x00000002, R13
+	JZ    copy_2_dword
+	MOVW  (AX)(CX*1), R14
+	MOVW  R14, (R10)(CX*1)
+	ADDQ  $0x02, CX
+
+copy_2_dword:
+	TESTQ $0x00000004, R13
+	JZ    copy_2_qword
+	MOVL  (AX)(CX*1), R14
+	MOVL  R14, (R10)(CX*1)
+	ADDQ  $0x04, CX
+
+copy_2_qword:
+	TESTQ $0x00000008, R13
+	JZ    copy_2_test
+	MOVQ  (AX)(CX*1), R14
+	MOVQ  R14, (R10)(CX*1)
+	ADDQ  $0x08, CX
+	JMP   copy_2_test
+
+copy_2:
+	MOVUPS (AX)(CX*1), X0
+	MOVUPS X0, (R10)(CX*1)
+	ADDQ   $0x10, CX
+
+copy_2_test:
+	CMPQ CX, R13
+	JB   copy_2
+	ADDQ R13, R10
+	JMP  handle_loop
+
+	// Copy overlapping match
+copy_overlapping_match:
+	ADDQ R13, R12
+
+copy_slow_3:
+	MOVB (AX), CL
+	MOVB CL, (R10)
+	INCQ AX
+	INCQ R10
+	DECQ R13
+	JNZ  copy_slow_3
+
+handle_loop:
+	MOVQ ctx+16(FP), AX
+	DECQ 96(AX)
+	JNS  sequenceDecs_decodeSync_safe_amd64_main_loop
+
+loop_finished:
+	MOVQ br+8(FP), AX
+	MOVQ DX, 32(AX)
+	MOVB BL, 40(AX)
+	MOVQ SI, 24(AX)
+
+	// Update the context
+	MOVQ ctx+16(FP), AX
+	MOVQ R12, 136(AX)
+	MOVQ 144(AX), CX
+	SUBQ CX, R11
+	MOVQ R11, 168(AX)
+
+	// Return success
+	MOVQ $0x00000000, ret+24(FP)
+	RET
+
+	// Return with match length error
+sequenceDecs_decodeSync_safe_amd64_error_match_len_ofs_mismatch:
+	MOVQ 16(SP), AX
+	MOVQ ctx+16(FP), CX
+	MOVQ AX, 216(CX)
+	MOVQ $0x00000001, ret+24(FP)
+	RET
+
+	// Return with match too long error
+sequenceDecs_decodeSync_safe_amd64_error_match_len_too_big:
+	MOVQ ctx+16(FP), AX
+	MOVQ 16(SP), CX
+	MOVQ CX, 216(AX)
+	MOVQ $0x00000002, ret+24(FP)
+	RET
+
+	// Return with match offset too long error
+error_match_off_too_big:
+	MOVQ ctx+16(FP), AX
+	MOVQ 8(SP), CX
+	MOVQ CX, 224(AX)
+	MOVQ R12, 136(AX)
+	MOVQ $0x00000003, ret+24(FP)
+	RET
+
+	// Return with not enough literals error
+error_not_enough_literals:
+	MOVQ ctx+16(FP), AX
+	MOVQ 24(SP), CX
+	MOVQ CX, 208(AX)
+	MOVQ $0x00000004, ret+24(FP)
+	RET
+
+	// Return with not enough output space error
+error_not_enough_space:
+	MOVQ ctx+16(FP), AX
+	MOVQ 24(SP), CX
+	MOVQ CX, 208(AX)
+	MOVQ 16(SP), CX
+	MOVQ CX, 216(AX)
+	MOVQ R12, 136(AX)
+	MOVQ $0x00000005, ret+24(FP)
+	RET
+
+// func sequenceDecs_decodeSync_safe_bmi2(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int
+// Requires: BMI, BMI2, CMOV, SSE
+TEXT ·sequenceDecs_decodeSync_safe_bmi2(SB), $64-32
+	MOVQ    br+8(FP), CX
+	MOVQ    32(CX), AX
+	MOVBQZX 40(CX), DX
+	MOVQ    24(CX), BX
+	MOVQ    (CX), CX
+	ADDQ    BX, CX
+	MOVQ    CX, (SP)
+	MOVQ    ctx+16(FP), CX
+	MOVQ    72(CX), SI
+	MOVQ    80(CX), DI
+	MOVQ    88(CX), R8
+	MOVQ    112(CX), R9
+	MOVQ    128(CX), R10
+	MOVQ    R10, 32(SP)
+	MOVQ    144(CX), R10
+	MOVQ    136(CX), R11
+	MOVQ    200(CX), R12
+	MOVQ    R12, 56(SP)
+	MOVQ    176(CX), R12
+	MOVQ    R12, 48(SP)
+	MOVQ    184(CX), CX
+	MOVQ    CX, 40(SP)
+	MOVQ    40(SP), CX
+	ADDQ    CX, 48(SP)
+
+	// Calculate poiter to s.out[cap(s.out)] (a past-end pointer)
+	ADDQ R9, 32(SP)
+
+	// outBase += outPosition
+	ADDQ R11, R9
+
+sequenceDecs_decodeSync_safe_bmi2_main_loop:
+	MOVQ (SP), R12
+
+	// Fill bitreader to have enough for the offset and match length.
+	CMPQ BX, $0x08
+	JL   sequenceDecs_decodeSync_safe_bmi2_fill_byte_by_byte
+	MOVQ DX, CX
+	SHRQ $0x03, CX
+	SUBQ CX, R12
+	MOVQ (R12), AX
+	SUBQ CX, BX
+	ANDQ $0x07, DX
+	JMP  sequenceDecs_decodeSync_safe_bmi2_fill_end
+
+sequenceDecs_decodeSync_safe_bmi2_fill_byte_by_byte:
+	CMPQ    BX, $0x00
+	JLE     sequenceDecs_decodeSync_safe_bmi2_fill_end
+	CMPQ    DX, $0x07
+	JLE     sequenceDecs_decodeSync_safe_bmi2_fill_end
+	SHLQ    $0x08, AX
+	SUBQ    $0x01, R12
+	SUBQ    $0x01, BX
+	SUBQ    $0x08, DX
+	MOVBQZX (R12), CX
+	ORQ     CX, AX
+	JMP     sequenceDecs_decodeSync_safe_bmi2_fill_byte_by_byte
+
+sequenceDecs_decodeSync_safe_bmi2_fill_end:
+	// Update offset
+	MOVQ   $0x00000808, CX
+	BEXTRQ CX, R8, R13
+	MOVQ   AX, R14
+	LEAQ   (DX)(R13*1), CX
+	ROLQ   CL, R14
+	BZHIQ  R13, R14, R14
+	MOVQ   CX, DX
+	MOVQ   R8, CX
+	SHRQ   $0x20, CX
+	ADDQ   R14, CX
+	MOVQ   CX, 8(SP)
+
+	// Update match length
+	MOVQ   $0x00000808, CX
+	BEXTRQ CX, DI, R13
+	MOVQ   AX, R14
+	LEAQ   (DX)(R13*1), CX
+	ROLQ   CL, R14
+	BZHIQ  R13, R14, R14
+	MOVQ   CX, DX
+	MOVQ   DI, CX
+	SHRQ   $0x20, CX
+	ADDQ   R14, CX
+	MOVQ   CX, 16(SP)
+
+	// Fill bitreader to have enough for the remaining
+	CMPQ BX, $0x08
+	JL   sequenceDecs_decodeSync_safe_bmi2_fill_2_byte_by_byte
+	MOVQ DX, CX
+	SHRQ $0x03, CX
+	SUBQ CX, R12
+	MOVQ (R12), AX
+	SUBQ CX, BX
+	ANDQ $0x07, DX
+	JMP  sequenceDecs_decodeSync_safe_bmi2_fill_2_end
+
+sequenceDecs_decodeSync_safe_bmi2_fill_2_byte_by_byte:
+	CMPQ    BX, $0x00
+	JLE     sequenceDecs_decodeSync_safe_bmi2_fill_2_end
+	CMPQ    DX, $0x07
+	JLE     sequenceDecs_decodeSync_safe_bmi2_fill_2_end
+	SHLQ    $0x08, AX
+	SUBQ    $0x01, R12
+	SUBQ    $0x01, BX
+	SUBQ    $0x08, DX
+	MOVBQZX (R12), CX
+	ORQ     CX, AX
+	JMP     sequenceDecs_decodeSync_safe_bmi2_fill_2_byte_by_byte
+
+sequenceDecs_decodeSync_safe_bmi2_fill_2_end:
+	// Update literal length
+	MOVQ   $0x00000808, CX
+	BEXTRQ CX, SI, R13
+	MOVQ   AX, R14
+	LEAQ   (DX)(R13*1), CX
+	ROLQ   CL, R14
+	BZHIQ  R13, R14, R14
+	MOVQ   CX, DX
+	MOVQ   SI, CX
+	SHRQ   $0x20, CX
+	ADDQ   R14, CX
+	MOVQ   CX, 24(SP)
+
+	// Fill bitreader for state updates
+	MOVQ    R12, (SP)
+	MOVQ    $0x00000808, CX
+	BEXTRQ  CX, R8, R12
+	MOVQ    ctx+16(FP), CX
+	CMPQ    96(CX), $0x00
+	JZ      sequenceDecs_decodeSync_safe_bmi2_skip_update
+	LEAQ    (SI)(DI*1), R13
+	ADDQ    R8, R13
+	MOVBQZX R13, R13
+	LEAQ    (DX)(R13*1), CX
+	MOVQ    AX, R14
+	MOVQ    CX, DX
+	ROLQ    CL, R14
+	BZHIQ   R13, R14, R14
+
+	// Update Offset State
+	BZHIQ  R8, R14, CX
+	SHRXQ  R8, R14, R14
+	MOVQ   $0x00001010, R13
+	BEXTRQ R13, R8, R8
+	ADDQ   CX, R8
+
+	// Load ctx.ofTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 48(CX), CX
+	MOVQ (CX)(R8*8), R8
+
+	// Update Match Length State
+	BZHIQ  DI, R14, CX
+	SHRXQ  DI, R14, R14
+	MOVQ   $0x00001010, R13
+	BEXTRQ R13, DI, DI
+	ADDQ   CX, DI
+
+	// Load ctx.mlTable
+	MOVQ ctx+16(FP), CX
+	MOVQ 24(CX), CX
+	MOVQ (CX)(DI*8), DI
+
+	// Update Literal Length State
+	BZHIQ  SI, R14, CX
+	MOVQ   $0x00001010, R13
+	BEXTRQ R13, SI, SI
+	ADDQ   CX, SI
+
+	// Load ctx.llTable
+	MOVQ ctx+16(FP), CX
+	MOVQ (CX), CX
+	MOVQ (CX)(SI*8), SI
+
+sequenceDecs_decodeSync_safe_bmi2_skip_update:
+	// Adjust offset
+	MOVQ   s+0(FP), CX
+	MOVQ   8(SP), R13
+	CMPQ   R12, $0x01
+	JBE    sequenceDecs_decodeSync_safe_bmi2_adjust_offsetB_1_or_0
+	MOVUPS 144(CX), X0
+	MOVQ   R13, 144(CX)
+	MOVUPS X0, 152(CX)
+	JMP    sequenceDecs_decodeSync_safe_bmi2_adjust_end
+
+sequenceDecs_decodeSync_safe_bmi2_adjust_offsetB_1_or_0:
+	CMPQ 24(SP), $0x00000000
+	JNE  sequenceDecs_decodeSync_safe_bmi2_adjust_offset_maybezero
+	INCQ R13
+	JMP  sequenceDecs_decodeSync_safe_bmi2_adjust_offset_nonzero
+
+sequenceDecs_decodeSync_safe_bmi2_adjust_offset_maybezero:
+	TESTQ R13, R13
+	JNZ   sequenceDecs_decodeSync_safe_bmi2_adjust_offset_nonzero
+	MOVQ  144(CX), R13
+	JMP   sequenceDecs_decodeSync_safe_bmi2_adjust_end
+
+sequenceDecs_decodeSync_safe_bmi2_adjust_offset_nonzero:
+	MOVQ    R13, R12
+	XORQ    R14, R14
+	MOVQ    $-1, R15
+	CMPQ    R13, $0x03
+	CMOVQEQ R14, R12
+	CMOVQEQ R15, R14
+	LEAQ    144(CX), R15
+	ADDQ    (R15)(R12*8), R14
+	JNZ     sequenceDecs_decodeSync_safe_bmi2_adjust_temp_valid
+	MOVQ    $0x00000001, R14
+
+sequenceDecs_decodeSync_safe_bmi2_adjust_temp_valid:
+	CMPQ R13, $0x01
+	JZ   sequenceDecs_decodeSync_safe_bmi2_adjust_skip
+	MOVQ 152(CX), R12
+	MOVQ R12, 160(CX)
+
+sequenceDecs_decodeSync_safe_bmi2_adjust_skip:
+	MOVQ 144(CX), R12
+	MOVQ R12, 152(CX)
+	MOVQ R14, 144(CX)
+	MOVQ R14, R13
+
+sequenceDecs_decodeSync_safe_bmi2_adjust_end:
+	MOVQ R13, 8(SP)
+
+	// Check values
+	MOVQ  16(SP), CX
+	MOVQ  24(SP), R12
+	LEAQ  (CX)(R12*1), R14
+	MOVQ  s+0(FP), R15
+	ADDQ  R14, 256(R15)
+	MOVQ  ctx+16(FP), R14
+	SUBQ  R12, 104(R14)
+	JS    error_not_enough_literals
+	CMPQ  CX, $0x00020002
+	JA    sequenceDecs_decodeSync_safe_bmi2_error_match_len_too_big
+	TESTQ R13, R13
+	JNZ   sequenceDecs_decodeSync_safe_bmi2_match_len_ofs_ok
+	TESTQ CX, CX
+	JNZ   sequenceDecs_decodeSync_safe_bmi2_error_match_len_ofs_mismatch
+
+sequenceDecs_decodeSync_safe_bmi2_match_len_ofs_ok:
+	MOVQ 24(SP), CX
+	MOVQ 8(SP), R12
+	MOVQ 16(SP), R13
+
+	// Check if we have enough space in s.out
+	LEAQ (CX)(R13*1), R14
+	ADDQ R9, R14
+	CMPQ R14, 32(SP)
+	JA   error_not_enough_space
+
+	// Copy literals
+	TESTQ CX, CX
+	JZ    check_offset
+	XORQ  R14, R14
+	TESTQ $0x00000001, CX
+	JZ    copy_1_word
+	MOVB  (R10)(R14*1), R15
+	MOVB  R15, (R9)(R14*1)
+	ADDQ  $0x01, R14
+
+copy_1_word:
+	TESTQ $0x00000002, CX
+	JZ    copy_1_dword
+	MOVW  (R10)(R14*1), R15
+	MOVW  R15, (R9)(R14*1)
+	ADDQ  $0x02, R14
+
+copy_1_dword:
+	TESTQ $0x00000004, CX
+	JZ    copy_1_qword
+	MOVL  (R10)(R14*1), R15
+	MOVL  R15, (R9)(R14*1)
+	ADDQ  $0x04, R14
+
+copy_1_qword:
+	TESTQ $0x00000008, CX
+	JZ    copy_1_test
+	MOVQ  (R10)(R14*1), R15
+	MOVQ  R15, (R9)(R14*1)
+	ADDQ  $0x08, R14
+	JMP   copy_1_test
+
+copy_1:
+	MOVUPS (R10)(R14*1), X0
+	MOVUPS X0, (R9)(R14*1)
+	ADDQ   $0x10, R14
+
+copy_1_test:
+	CMPQ R14, CX
+	JB   copy_1
+	ADDQ CX, R10
+	ADDQ CX, R9
+	ADDQ CX, R11
+
+	// Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize)
+check_offset:
+	MOVQ R11, CX
+	ADDQ 40(SP), CX
+	CMPQ R12, CX
+	JG   error_match_off_too_big
+	CMPQ R12, 56(SP)
+	JG   error_match_off_too_big
+
+	// Copy match from history
+	MOVQ  R12, CX
+	SUBQ  R11, CX
+	JLS   copy_match
+	MOVQ  48(SP), R14
+	SUBQ  CX, R14
+	CMPQ  R13, CX
+	JGE   copy_all_from_history
+	XORQ  CX, CX
+	TESTQ $0x00000001, R13
+	JZ    copy_4_word
+	MOVB  (R14)(CX*1), R12
+	MOVB  R12, (R9)(CX*1)
+	ADDQ  $0x01, CX
+
+copy_4_word:
+	TESTQ $0x00000002, R13
+	JZ    copy_4_dword
+	MOVW  (R14)(CX*1), R12
+	MOVW  R12, (R9)(CX*1)
+	ADDQ  $0x02, CX
+
+copy_4_dword:
+	TESTQ $0x00000004, R13
+	JZ    copy_4_qword
+	MOVL  (R14)(CX*1), R12
+	MOVL  R12, (R9)(CX*1)
+	ADDQ  $0x04, CX
+
+copy_4_qword:
+	TESTQ $0x00000008, R13
+	JZ    copy_4_test
+	MOVQ  (R14)(CX*1), R12
+	MOVQ  R12, (R9)(CX*1)
+	ADDQ  $0x08, CX
+	JMP   copy_4_test
+
+copy_4:
+	MOVUPS (R14)(CX*1), X0
+	MOVUPS X0, (R9)(CX*1)
+	ADDQ   $0x10, CX
+
+copy_4_test:
+	CMPQ CX, R13
+	JB   copy_4
+	ADDQ R13, R11
+	ADDQ R13, R9
+	JMP  handle_loop
+	JMP loop_finished
+
+copy_all_from_history:
+	XORQ  R15, R15
+	TESTQ $0x00000001, CX
+	JZ    copy_5_word
+	MOVB  (R14)(R15*1), BP
+	MOVB  BP, (R9)(R15*1)
+	ADDQ  $0x01, R15
+
+copy_5_word:
+	TESTQ $0x00000002, CX
+	JZ    copy_5_dword
+	MOVW  (R14)(R15*1), BP
+	MOVW  BP, (R9)(R15*1)
+	ADDQ  $0x02, R15
+
+copy_5_dword:
+	TESTQ $0x00000004, CX
+	JZ    copy_5_qword
+	MOVL  (R14)(R15*1), BP
+	MOVL  BP, (R9)(R15*1)
+	ADDQ  $0x04, R15
+
+copy_5_qword:
+	TESTQ $0x00000008, CX
+	JZ    copy_5_test
+	MOVQ  (R14)(R15*1), BP
+	MOVQ  BP, (R9)(R15*1)
+	ADDQ  $0x08, R15
+	JMP   copy_5_test
+
+copy_5:
+	MOVUPS (R14)(R15*1), X0
+	MOVUPS X0, (R9)(R15*1)
+	ADDQ   $0x10, R15
+
+copy_5_test:
+	CMPQ R15, CX
+	JB   copy_5
+	ADDQ CX, R9
+	ADDQ CX, R11
+	SUBQ CX, R13
+
+	// Copy match from the current buffer
+copy_match:
+	TESTQ R13, R13
+	JZ    handle_loop
+	MOVQ  R9, CX
+	SUBQ  R12, CX
+
+	// ml <= mo
+	CMPQ R13, R12
+	JA   copy_overlapping_match
+
+	// Copy non-overlapping match
+	ADDQ  R13, R11
+	XORQ  R12, R12
+	TESTQ $0x00000001, R13
+	JZ    copy_2_word
+	MOVB  (CX)(R12*1), R14
+	MOVB  R14, (R9)(R12*1)
+	ADDQ  $0x01, R12
+
+copy_2_word:
+	TESTQ $0x00000002, R13
+	JZ    copy_2_dword
+	MOVW  (CX)(R12*1), R14
+	MOVW  R14, (R9)(R12*1)
+	ADDQ  $0x02, R12
+
+copy_2_dword:
+	TESTQ $0x00000004, R13
+	JZ    copy_2_qword
+	MOVL  (CX)(R12*1), R14
+	MOVL  R14, (R9)(R12*1)
+	ADDQ  $0x04, R12
+
+copy_2_qword:
+	TESTQ $0x00000008, R13
+	JZ    copy_2_test
+	MOVQ  (CX)(R12*1), R14
+	MOVQ  R14, (R9)(R12*1)
+	ADDQ  $0x08, R12
+	JMP   copy_2_test
+
+copy_2:
+	MOVUPS (CX)(R12*1), X0
+	MOVUPS X0, (R9)(R12*1)
+	ADDQ   $0x10, R12
+
+copy_2_test:
+	CMPQ R12, R13
+	JB   copy_2
+	ADDQ R13, R9
+	JMP  handle_loop
+
+	// Copy overlapping match
+copy_overlapping_match:
+	ADDQ R13, R11
+
+copy_slow_3:
+	MOVB (CX), R12
+	MOVB R12, (R9)
+	INCQ CX
+	INCQ R9
+	DECQ R13
+	JNZ  copy_slow_3
+
+handle_loop:
+	MOVQ ctx+16(FP), CX
+	DECQ 96(CX)
+	JNS  sequenceDecs_decodeSync_safe_bmi2_main_loop
+
+loop_finished:
+	MOVQ br+8(FP), CX
+	MOVQ AX, 32(CX)
+	MOVB DL, 40(CX)
+	MOVQ BX, 24(CX)
+
+	// Update the context
+	MOVQ ctx+16(FP), AX
+	MOVQ R11, 136(AX)
+	MOVQ 144(AX), CX
+	SUBQ CX, R10
+	MOVQ R10, 168(AX)
+
+	// Return success
+	MOVQ $0x00000000, ret+24(FP)
+	RET
+
+	// Return with match length error
+sequenceDecs_decodeSync_safe_bmi2_error_match_len_ofs_mismatch:
+	MOVQ 16(SP), AX
+	MOVQ ctx+16(FP), CX
+	MOVQ AX, 216(CX)
+	MOVQ $0x00000001, ret+24(FP)
+	RET
+
+	// Return with match too long error
+sequenceDecs_decodeSync_safe_bmi2_error_match_len_too_big:
+	MOVQ ctx+16(FP), AX
+	MOVQ 16(SP), CX
+	MOVQ CX, 216(AX)
+	MOVQ $0x00000002, ret+24(FP)
+	RET
+
+	// Return with match offset too long error
+error_match_off_too_big:
+	MOVQ ctx+16(FP), AX
+	MOVQ 8(SP), CX
+	MOVQ CX, 224(AX)
+	MOVQ R11, 136(AX)
+	MOVQ $0x00000003, ret+24(FP)
+	RET
+
+	// Return with not enough literals error
+error_not_enough_literals:
+	MOVQ ctx+16(FP), AX
+	MOVQ 24(SP), CX
+	MOVQ CX, 208(AX)
+	MOVQ $0x00000004, ret+24(FP)
+	RET
+
+	// Return with not enough output space error
+error_not_enough_space:
+	MOVQ ctx+16(FP), AX
+	MOVQ 24(SP), CX
+	MOVQ CX, 208(AX)
+	MOVQ 16(SP), CX
+	MOVQ CX, 216(AX)
+	MOVQ R11, 136(AX)
+	MOVQ $0x00000005, ret+24(FP)
+	RET
diff --git a/vendor/github.com/klauspost/compress/zstd/seqdec_generic.go b/vendor/github.com/klauspost/compress/zstd/seqdec_generic.go
new file mode 100644
index 00000000..c3452bc3
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/zstd/seqdec_generic.go
@@ -0,0 +1,237 @@
+//go:build !amd64 || appengine || !gc || noasm
+// +build !amd64 appengine !gc noasm
+
+package zstd
+
+import (
+	"fmt"
+	"io"
+)
+
+// decode sequences from the stream with the provided history but without dictionary.
+func (s *sequenceDecs) decodeSyncSimple(hist []byte) (bool, error) {
+	return false, nil
+}
+
+// decode sequences from the stream without the provided history.
+func (s *sequenceDecs) decode(seqs []seqVals) error {
+	br := s.br
+
+	// Grab full sizes tables, to avoid bounds checks.
+	llTable, mlTable, ofTable := s.litLengths.fse.dt[:maxTablesize], s.matchLengths.fse.dt[:maxTablesize], s.offsets.fse.dt[:maxTablesize]
+	llState, mlState, ofState := s.litLengths.state.state, s.matchLengths.state.state, s.offsets.state.state
+	s.seqSize = 0
+	litRemain := len(s.literals)
+
+	maxBlockSize := maxCompressedBlockSize
+	if s.windowSize < maxBlockSize {
+		maxBlockSize = s.windowSize
+	}
+	for i := range seqs {
+		var ll, mo, ml int
+		if br.off > 4+((maxOffsetBits+16+16)>>3) {
+			// inlined function:
+			// ll, mo, ml = s.nextFast(br, llState, mlState, ofState)
+
+			// Final will not read from stream.
+			var llB, mlB, moB uint8
+			ll, llB = llState.final()
+			ml, mlB = mlState.final()
+			mo, moB = ofState.final()
+
+			// extra bits are stored in reverse order.
+			br.fillFast()
+			mo += br.getBits(moB)
+			if s.maxBits > 32 {
+				br.fillFast()
+			}
+			ml += br.getBits(mlB)
+			ll += br.getBits(llB)
+
+			if moB > 1 {
+				s.prevOffset[2] = s.prevOffset[1]
+				s.prevOffset[1] = s.prevOffset[0]
+				s.prevOffset[0] = mo
+			} else {
+				// mo = s.adjustOffset(mo, ll, moB)
+				// Inlined for rather big speedup
+				if ll == 0 {
+					// There is an exception though, when current sequence's literals_length = 0.
+					// In this case, repeated offsets are shifted by one, so an offset_value of 1 means Repeated_Offset2,
+					// an offset_value of 2 means Repeated_Offset3, and an offset_value of 3 means Repeated_Offset1 - 1_byte.
+					mo++
+				}
+
+				if mo == 0 {
+					mo = s.prevOffset[0]
+				} else {
+					var temp int
+					if mo == 3 {
+						temp = s.prevOffset[0] - 1
+					} else {
+						temp = s.prevOffset[mo]
+					}
+
+					if temp == 0 {
+						// 0 is not valid; input is corrupted; force offset to 1
+						println("WARNING: temp was 0")
+						temp = 1
+					}
+
+					if mo != 1 {
+						s.prevOffset[2] = s.prevOffset[1]
+					}
+					s.prevOffset[1] = s.prevOffset[0]
+					s.prevOffset[0] = temp
+					mo = temp
+				}
+			}
+			br.fillFast()
+		} else {
+			if br.overread() {
+				if debugDecoder {
+					printf("reading sequence %d, exceeded available data\n", i)
+				}
+				return io.ErrUnexpectedEOF
+			}
+			ll, mo, ml = s.next(br, llState, mlState, ofState)
+			br.fill()
+		}
+
+		if debugSequences {
+			println("Seq", i, "Litlen:", ll, "mo:", mo, "(abs) ml:", ml)
+		}
+		// Evaluate.
+		// We might be doing this async, so do it early.
+		if mo == 0 && ml > 0 {
+			return fmt.Errorf("zero matchoff and matchlen (%d) > 0", ml)
+		}
+		if ml > maxMatchLen {
+			return fmt.Errorf("match len (%d) bigger than max allowed length", ml)
+		}
+		s.seqSize += ll + ml
+		if s.seqSize > maxBlockSize {
+			return fmt.Errorf("output (%d) bigger than max block size (%d)", s.seqSize, maxBlockSize)
+		}
+		litRemain -= ll
+		if litRemain < 0 {
+			return fmt.Errorf("unexpected literal count, want %d bytes, but only %d is available", ll, litRemain+ll)
+		}
+		seqs[i] = seqVals{
+			ll: ll,
+			ml: ml,
+			mo: mo,
+		}
+		if i == len(seqs)-1 {
+			// This is the last sequence, so we shouldn't update state.
+			break
+		}
+
+		// Manually inlined, ~ 5-20% faster
+		// Update all 3 states at once. Approx 20% faster.
+		nBits := llState.nbBits() + mlState.nbBits() + ofState.nbBits()
+		if nBits == 0 {
+			llState = llTable[llState.newState()&maxTableMask]
+			mlState = mlTable[mlState.newState()&maxTableMask]
+			ofState = ofTable[ofState.newState()&maxTableMask]
+		} else {
+			bits := br.get32BitsFast(nBits)
+			lowBits := uint16(bits >> ((ofState.nbBits() + mlState.nbBits()) & 31))
+			llState = llTable[(llState.newState()+lowBits)&maxTableMask]
+
+			lowBits = uint16(bits >> (ofState.nbBits() & 31))
+			lowBits &= bitMask[mlState.nbBits()&15]
+			mlState = mlTable[(mlState.newState()+lowBits)&maxTableMask]
+
+			lowBits = uint16(bits) & bitMask[ofState.nbBits()&15]
+			ofState = ofTable[(ofState.newState()+lowBits)&maxTableMask]
+		}
+	}
+	s.seqSize += litRemain
+	if s.seqSize > maxBlockSize {
+		return fmt.Errorf("output (%d) bigger than max block size (%d)", s.seqSize, maxBlockSize)
+	}
+	err := br.close()
+	if err != nil {
+		printf("Closing sequences: %v, %+v\n", err, *br)
+	}
+	return err
+}
+
+// executeSimple handles cases when a dictionary is not used.
+func (s *sequenceDecs) executeSimple(seqs []seqVals, hist []byte) error {
+	// Ensure we have enough output size...
+	if len(s.out)+s.seqSize > cap(s.out) {
+		addBytes := s.seqSize + len(s.out)
+		s.out = append(s.out, make([]byte, addBytes)...)
+		s.out = s.out[:len(s.out)-addBytes]
+	}
+
+	if debugDecoder {
+		printf("Execute %d seqs with literals: %d into %d bytes\n", len(seqs), len(s.literals), s.seqSize)
+	}
+
+	var t = len(s.out)
+	out := s.out[:t+s.seqSize]
+
+	for _, seq := range seqs {
+		// Add literals
+		copy(out[t:], s.literals[:seq.ll])
+		t += seq.ll
+		s.literals = s.literals[seq.ll:]
+
+		// Malformed input
+		if seq.mo > t+len(hist) || seq.mo > s.windowSize {
+			return fmt.Errorf("match offset (%d) bigger than current history (%d)", seq.mo, t+len(hist))
+		}
+
+		// Copy from history.
+		if v := seq.mo - t; v > 0 {
+			// v is the start position in history from end.
+			start := len(hist) - v
+			if seq.ml > v {
+				// Some goes into the current block.
+				// Copy remainder of history
+				copy(out[t:], hist[start:])
+				t += v
+				seq.ml -= v
+			} else {
+				copy(out[t:], hist[start:start+seq.ml])
+				t += seq.ml
+				continue
+			}
+		}
+
+		// We must be in the current buffer now
+		if seq.ml > 0 {
+			start := t - seq.mo
+			if seq.ml <= t-start {
+				// No overlap
+				copy(out[t:], out[start:start+seq.ml])
+				t += seq.ml
+			} else {
+				// Overlapping copy
+				// Extend destination slice and copy one byte at the time.
+				src := out[start : start+seq.ml]
+				dst := out[t:]
+				dst = dst[:len(src)]
+				t += len(src)
+				// Destination is the space we just added.
+				for i := range src {
+					dst[i] = src[i]
+				}
+			}
+		}
+	}
+	// Add final literals
+	copy(out[t:], s.literals)
+	if debugDecoder {
+		t += len(s.literals)
+		if t != len(out) {
+			panic(fmt.Errorf("length mismatch, want %d, got %d, ss: %d", len(out), t, s.seqSize))
+		}
+	}
+	s.out = out
+
+	return nil
+}
diff --git a/vendor/github.com/klauspost/compress/zstd/zip.go b/vendor/github.com/klauspost/compress/zstd/zip.go
index ffffcbc2..29c15c8c 100644
--- a/vendor/github.com/klauspost/compress/zstd/zip.go
+++ b/vendor/github.com/klauspost/compress/zstd/zip.go
@@ -18,26 +18,44 @@ const ZipMethodWinZip = 93
 // See https://pkware.cachefly.net/webdocs/APPNOTE/APPNOTE-6.3.9.TXT
 const ZipMethodPKWare = 20
 
-var zipReaderPool sync.Pool
+// zipReaderPool is the default reader pool.
+var zipReaderPool = sync.Pool{New: func() interface{} {
+	z, err := NewReader(nil, WithDecoderLowmem(true), WithDecoderMaxWindow(128<<20), WithDecoderConcurrency(1))
+	if err != nil {
+		panic(err)
+	}
+	return z
+}}
 
 // newZipReader creates a pooled zip decompressor.
-func newZipReader(r io.Reader) io.ReadCloser {
-	dec, ok := zipReaderPool.Get().(*Decoder)
-	if ok {
-		dec.Reset(r)
-	} else {
-		d, err := NewReader(r, WithDecoderConcurrency(1), WithDecoderLowmem(true))
-		if err != nil {
-			panic(err)
+func newZipReader(opts ...DOption) func(r io.Reader) io.ReadCloser {
+	pool := &zipReaderPool
+	if len(opts) > 0 {
+		opts = append([]DOption{WithDecoderLowmem(true), WithDecoderMaxWindow(128 << 20)}, opts...)
+		// Force concurrency 1
+		opts = append(opts, WithDecoderConcurrency(1))
+		// Create our own pool
+		pool = &sync.Pool{}
+	}
+	return func(r io.Reader) io.ReadCloser {
+		dec, ok := pool.Get().(*Decoder)
+		if ok {
+			dec.Reset(r)
+		} else {
+			d, err := NewReader(r, opts...)
+			if err != nil {
+				panic(err)
+			}
+			dec = d
 		}
-		dec = d
+		return &pooledZipReader{dec: dec, pool: pool}
 	}
-	return &pooledZipReader{dec: dec}
 }
 
 type pooledZipReader struct {
-	mu  sync.Mutex // guards Close and Read
-	dec *Decoder
+	mu   sync.Mutex // guards Close and Read
+	pool *sync.Pool
+	dec  *Decoder
 }
 
 func (r *pooledZipReader) Read(p []byte) (n int, err error) {
@@ -48,8 +66,8 @@ func (r *pooledZipReader) Read(p []byte) (n int, err error) {
 	}
 	dec, err := r.dec.Read(p)
 	if err == io.EOF {
-		err = r.dec.Reset(nil)
-		zipReaderPool.Put(r.dec)
+		r.dec.Reset(nil)
+		r.pool.Put(r.dec)
 		r.dec = nil
 	}
 	return dec, err
@@ -61,7 +79,7 @@ func (r *pooledZipReader) Close() error {
 	var err error
 	if r.dec != nil {
 		err = r.dec.Reset(nil)
-		zipReaderPool.Put(r.dec)
+		r.pool.Put(r.dec)
 		r.dec = nil
 	}
 	return err
@@ -115,6 +133,9 @@ func ZipCompressor(opts ...EOption) func(w io.Writer) (io.WriteCloser, error) {
 
 // ZipDecompressor returns a decompressor that can be registered with zip libraries.
 // See ZipCompressor for example.
-func ZipDecompressor() func(r io.Reader) io.ReadCloser {
-	return newZipReader
+// Options can be specified. WithDecoderConcurrency(1) is forced,
+// and by default a 128MB maximum decompression window is specified.
+// The window size can be overridden if required.
+func ZipDecompressor(opts ...DOption) func(r io.Reader) io.ReadCloser {
+	return newZipReader(opts...)
 }
diff --git a/vendor/github.com/klauspost/compress/zstd/zstd.go b/vendor/github.com/klauspost/compress/zstd/zstd.go
index c1c90b4a..3eb3f1c8 100644
--- a/vendor/github.com/klauspost/compress/zstd/zstd.go
+++ b/vendor/github.com/klauspost/compress/zstd/zstd.go
@@ -110,17 +110,6 @@ func printf(format string, a ...interface{}) {
 	}
 }
 
-// matchLenFast does matching, but will not match the last up to 7 bytes.
-func matchLenFast(a, b []byte) int {
-	endI := len(a) & (math.MaxInt32 - 7)
-	for i := 0; i < endI; i += 8 {
-		if diff := load64(a, i) ^ load64(b, i); diff != 0 {
-			return i + bits.TrailingZeros64(diff)>>3
-		}
-	}
-	return endI
-}
-
 // matchLen returns the maximum length.
 // a must be the shortest of the two.
 // The function also returns whether all bytes matched.
diff --git a/vendor/github.com/slack-go/slack/README.md b/vendor/github.com/slack-go/slack/README.md
index 39b04ce8..9618aebc 100644
--- a/vendor/github.com/slack-go/slack/README.md
+++ b/vendor/github.com/slack-go/slack/README.md
@@ -39,7 +39,7 @@ func main() {
 	// If you set debugging, it will log all requests to the console
 	// Useful when encountering issues
 	// slack.New("YOUR_TOKEN_HERE", slack.OptionDebug(true))
-	groups, err := api.GetUserGroups(false)
+	groups, err := api.GetUserGroups(slack.GetUserGroupsOptionIncludeUsers(false))
 	if err != nil {
 		fmt.Printf("%s\n", err)
 		return
@@ -86,7 +86,13 @@ See https://github.com/slack-go/slack/blob/master/examples/websocket/websocket.g
 
 See https://github.com/slack-go/slack/blob/master/examples/eventsapi/events.go
 
+## Socketmode Event Handler (Experimental)
 
+When using socket mode, dealing with an event can be pretty lengthy as it requires you to route the event to the right place.
+
+Instead, you can use `SocketmodeHandler` much like you use an HTTP handler to register which event you would like to listen to and what callback function will process that event when it occurs.
+
+See [./examples/socketmode_handler/socketmode_handler.go](./examples/socketmode_handler/socketmode_handler.go)
 ## Contributing
 
 You are more than welcome to contribute to this project.  Fork and
diff --git a/vendor/github.com/slack-go/slack/block_input.go b/vendor/github.com/slack-go/slack/block_input.go
index 087571af..78ffcdb8 100644
--- a/vendor/github.com/slack-go/slack/block_input.go
+++ b/vendor/github.com/slack-go/slack/block_input.go
@@ -19,11 +19,12 @@ func (s InputBlock) BlockType() MessageBlockType {
 }
 
 // NewInputBlock returns a new instance of an input block
-func NewInputBlock(blockID string, label *TextBlockObject, element BlockElement) *InputBlock {
+func NewInputBlock(blockID string, label, hint *TextBlockObject, element BlockElement) *InputBlock {
 	return &InputBlock{
 		Type:    MBTInput,
 		BlockID: blockID,
 		Label:   label,
 		Element: element,
+		Hint:    hint,
 	}
 }
diff --git a/vendor/github.com/slack-go/slack/bookmarks.go b/vendor/github.com/slack-go/slack/bookmarks.go
new file mode 100644
index 00000000..78753507
--- /dev/null
+++ b/vendor/github.com/slack-go/slack/bookmarks.go
@@ -0,0 +1,159 @@
+package slack
+
+import (
+	"context"
+	"net/url"
+)
+
+type Bookmark struct {
+	ID        string   `json:"id"`
+	ChannelID string   `json:"channel_id"`
+	Title     string   `json:"title"`
+	Link      string   `json:"link"`
+	Emoji     string   `json:"emoji"`
+	IconURL   string   `json:"icon_url"`
+	Type      string   `json:"type"`
+	Created   JSONTime `json:"date_created"`
+	Updated   JSONTime `json:"date_updated"`
+	Rank      string   `json:"rank"`
+
+	LastUpdatedByUserID string `json:"last_updated_by_user_id"`
+	LastUpdatedByTeamID string `json:"last_updated_by_team_id"`
+
+	ShortcutID string `json:"shortcut_id"`
+	EntityID   string `json:"entity_id"`
+	AppID      string `json:"app_id"`
+}
+
+type AddBookmarkParameters struct {
+	Title    string // A required title for the bookmark
+	Type     string // A required type for the bookmark
+	Link     string // URL required for type:link
+	Emoji    string // An optional emoji
+	EntityID string
+	ParentID string
+}
+
+type EditBookmarkParameters struct {
+	Title *string // Change the title. Set to "" to clear
+	Emoji *string // Change the emoji. Set to "" to clear
+	Link  string  // Change the link
+}
+
+type addBookmarkResponse struct {
+	Bookmark Bookmark `json:"bookmark"`
+	SlackResponse
+}
+
+type editBookmarkResponse struct {
+	Bookmark Bookmark `json:"bookmark"`
+	SlackResponse
+}
+
+type listBookmarksResponse struct {
+	Bookmarks []Bookmark `json:"bookmarks"`
+	SlackResponse
+}
+
+// AddBookmark adds a bookmark in a channel
+func (api *Client) AddBookmark(channelID string, params AddBookmarkParameters) (Bookmark, error) {
+	return api.AddBookmarkContext(context.Background(), channelID, params)
+}
+
+// AddBookmarkContext adds a bookmark in a channel with a custom context
+func (api *Client) AddBookmarkContext(ctx context.Context, channelID string, params AddBookmarkParameters) (Bookmark, error) {
+	values := url.Values{
+		"channel_id": {channelID},
+		"token":      {api.token},
+		"title":      {params.Title},
+		"type":       {params.Type},
+	}
+	if params.Link != "" {
+		values.Set("link", params.Link)
+	}
+	if params.Emoji != "" {
+		values.Set("emoji", params.Emoji)
+	}
+	if params.EntityID != "" {
+		values.Set("entity_id", params.EntityID)
+	}
+	if params.ParentID != "" {
+		values.Set("parent_id", params.ParentID)
+	}
+
+	response := &addBookmarkResponse{}
+	if err := api.postMethod(ctx, "bookmarks.add", values, response); err != nil {
+		return Bookmark{}, err
+	}
+
+	return response.Bookmark, response.Err()
+}
+
+// RemoveBookmark removes a bookmark from a channel
+func (api *Client) RemoveBookmark(channelID, bookmarkID string) error {
+	return api.RemoveBookmarkContext(context.Background(), channelID, bookmarkID)
+}
+
+// RemoveBookmarkContext removes a bookmark from a channel with a custom context
+func (api *Client) RemoveBookmarkContext(ctx context.Context, channelID, bookmarkID string) error {
+	values := url.Values{
+		"channel_id":  {channelID},
+		"token":       {api.token},
+		"bookmark_id": {bookmarkID},
+	}
+
+	response := &SlackResponse{}
+	if err := api.postMethod(ctx, "bookmarks.remove", values, response); err != nil {
+		return err
+	}
+
+	return response.Err()
+}
+
+// ListBookmarks returns all bookmarks for a channel.
+func (api *Client) ListBookmarks(channelID string) ([]Bookmark, error) {
+	return api.ListBookmarksContext(context.Background(), channelID)
+}
+
+// ListBookmarksContext returns all bookmarks for a channel with a custom context.
+func (api *Client) ListBookmarksContext(ctx context.Context, channelID string) ([]Bookmark, error) {
+	values := url.Values{
+		"channel_id": {channelID},
+		"token":      {api.token},
+	}
+
+	response := &listBookmarksResponse{}
+	err := api.postMethod(ctx, "bookmarks.list", values, response)
+	if err != nil {
+		return nil, err
+	}
+	return response.Bookmarks, response.Err()
+}
+
+func (api *Client) EditBookmark(channelID, bookmarkID string, params EditBookmarkParameters) (Bookmark, error) {
+	return api.EditBookmarkContext(context.Background(), channelID, bookmarkID, params)
+}
+
+func (api *Client) EditBookmarkContext(ctx context.Context, channelID, bookmarkID string, params EditBookmarkParameters) (Bookmark, error) {
+	values := url.Values{
+		"channel_id":  {channelID},
+		"token":       {api.token},
+		"bookmark_id": {bookmarkID},
+	}
+	if params.Link != "" {
+		values.Set("link", params.Link)
+	}
+	if params.Emoji != nil {
+		values.Set("emoji", *params.Emoji)
+	}
+	if params.Title != nil {
+		values.Set("title", *params.Title)
+	}
+
+	response := &editBookmarkResponse{}
+	if err := api.postMethod(ctx, "bookmarks.edit", values, response); err != nil {
+		return Bookmark{}, err
+	}
+
+	return response.Bookmark, response.Err()
+}
diff --git a/vendor/github.com/slack-go/slack/misc.go b/vendor/github.com/slack-go/slack/misc.go
index 804724d7..bb99f2c4 100644
--- a/vendor/github.com/slack-go/slack/misc.go
+++ b/vendor/github.com/slack-go/slack/misc.go
@@ -18,8 +18,6 @@ import (
 	"strconv"
 	"strings"
 	"time"
-
-	"github.com/slack-go/slack/internal/misc"
 )
 
 // SlackResponse handles parsing out errors from the web api.
@@ -299,7 +297,7 @@ func checkStatusCode(resp *http.Response, d Debug) error {
 	// Slack seems to send an HTML body along with 5xx error codes. Don't parse it.
 	if resp.StatusCode != http.StatusOK {
 		logResponse(resp, d)
-		return misc.StatusCodeError{Code: resp.StatusCode, Status: resp.Status}
+		return StatusCodeError{Code: resp.StatusCode, Status: resp.Status}
 	}
 
 	return nil
diff --git a/vendor/github.com/slack-go/slack/remotefiles.go b/vendor/github.com/slack-go/slack/remotefiles.go
new file mode 100644
index 00000000..8a908a8f
--- /dev/null
+++ b/vendor/github.com/slack-go/slack/remotefiles.go
@@ -0,0 +1,316 @@
+package slack
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"net/url"
+	"strconv"
+	"strings"
+)
+
+const (
+	DEFAULT_REMOTE_FILES_CHANNEL = ""
+	DEFAULT_REMOTE_FILES_TS_FROM = 0
+	DEFAULT_REMOTE_FILES_TS_TO   = -1
+	DEFAULT_REMOTE_FILES_COUNT   = 100
+)
+
+// RemoteFile contains all the information for a remote file
+// For more details:
+// https://api.slack.com/messaging/files/remote
+type RemoteFile struct {
+	ID              string   `json:"id"`
+	Created         JSONTime `json:"created"`
+	Timestamp       JSONTime `json:"timestamp"`
+	Name            string   `json:"name"`
+	Title           string   `json:"title"`
+	Mimetype        string   `json:"mimetype"`
+	Filetype        string   `json:"filetype"`
+	PrettyType      string   `json:"pretty_type"`
+	User            string   `json:"user"`
+	Editable        bool     `json:"editable"`
+	Size            int      `json:"size"`
+	Mode            string   `json:"mode"`
+	IsExternal      bool     `json:"is_external"`
+	ExternalType    string   `json:"external_type"`
+	IsPublic        bool     `json:"is_public"`
+	PublicURLShared bool     `json:"public_url_shared"`
+	DisplayAsBot    bool     `json:"display_as_bot"`
+	Username        string   `json:"username"`
+	URLPrivate      string   `json:"url_private"`
+	Permalink       string   `json:"permalink"`
+	CommentsCount   int      `json:"comments_count"`
+	IsStarred       bool     `json:"is_starred"`
+	Shares          Share    `json:"shares"`
+	Channels        []string `json:"channels"`
+	Groups          []string `json:"groups"`
+	IMs             []string `json:"ims"`
+	ExternalID      string   `json:"external_id"`
+	ExternalURL     string   `json:"external_url"`
+	HasRichPreview  bool     `json:"has_rich_preview"`
+}
+
+// RemoteFileParameters contains required and optional parameters for a remote file.
+//
+// ExternalID is a user defined GUID, ExternalURL is where the remote file can be accessed,
+// and Title is the name of the file.
+//
+// For more details:
+// https://api.slack.com/methods/files.remote.add
+type RemoteFileParameters struct {
+	ExternalID            string // required
+	ExternalURL           string // required
+	Title                 string // required
+	Filetype              string
+	IndexableFileContents string
+	PreviewImage          string
+	PreviewImageReader    io.Reader
+}
+
+// ListRemoteFilesParameters contains arguments for the ListRemoteFiles method.
+// For more details:
+// https://api.slack.com/methods/files.remote.list
+type ListRemoteFilesParameters struct {
+	Channel       string
+	Cursor        string
+	Limit         int
+	TimestampFrom JSONTime
+	TimestampTo   JSONTime
+}
+
+type remoteFileResponseFull struct {
+	RemoteFile `json:"file"`
+	Paging     `json:"paging"`
+	Files      []RemoteFile `json:"files"`
+	SlackResponse
+}
+
+func (api *Client) remoteFileRequest(ctx context.Context, path string, values url.Values) (*remoteFileResponseFull, error) {
+	response := &remoteFileResponseFull{}
+	err := api.postMethod(ctx, path, values, response)
+	if err != nil {
+		return nil, err
+	}
+
+	return response, response.Err()
+}
+
+// AddRemoteFile adds a remote file. Unlike regular files, remote files must be explicitly shared.
+// For more details:
+// https://api.slack.com/methods/files.remote.add
+func (api *Client) AddRemoteFile(params RemoteFileParameters) (*RemoteFile, error) {
+	return api.AddRemoteFileContext(context.Background(), params)
+}
+
+// AddRemoteFileContext adds a remote file and setting a custom context
+// For more details see the AddRemoteFile documentation.
+func (api *Client) AddRemoteFileContext(ctx context.Context, params RemoteFileParameters) (remotefile *RemoteFile, err error) {
+	if params.ExternalID == "" || params.ExternalURL == "" || params.Title == "" {
+		return nil, ErrParametersMissing
+	}
+	response := &remoteFileResponseFull{}
+	values := url.Values{
+		"token":        {api.token},
+		"external_id":  {params.ExternalID},
+		"external_url": {params.ExternalURL},
+		"title":        {params.Title},
+	}
+	if params.Filetype != "" {
+		values.Add("filetype", params.Filetype)
+	}
+	if params.IndexableFileContents != "" {
+		values.Add("indexable_file_contents", params.IndexableFileContents)
+	}
+	if params.PreviewImage != "" {
+		err = postLocalWithMultipartResponse(ctx, api.httpclient, api.endpoint+"files.remote.add", params.PreviewImage, "preview_image", api.token, values, response, api)
+	} else if params.PreviewImageReader != nil {
+		err = postWithMultipartResponse(ctx, api.httpclient, api.endpoint+"files.remote.add", "preview.png", "preview_image", api.token, values, params.PreviewImageReader, response, api)
+	} else {
+		response, err = api.remoteFileRequest(ctx, "files.remote.add", values)
+	}
+
+	if err != nil {
+		return nil, err
+	}
+
+	return &response.RemoteFile, response.Err()
+}
+
+// ListRemoteFiles retrieves all remote files according to the parameters given. Uses cursor based pagination.
+// For more details:
+// https://api.slack.com/methods/files.remote.list
+func (api *Client) ListRemoteFiles(params ListRemoteFilesParameters) ([]RemoteFile, error) {
+	return api.ListRemoteFilesContext(context.Background(), params)
+}
+
+// ListRemoteFilesContext retrieves all remote files according to the parameters given with a custom context. Uses cursor based pagination.
+// For more details see the ListRemoteFiles documentation.
+func (api *Client) ListRemoteFilesContext(ctx context.Context, params ListRemoteFilesParameters) ([]RemoteFile, error) {
+	values := url.Values{
+		"token": {api.token},
+	}
+	if params.Channel != DEFAULT_REMOTE_FILES_CHANNEL {
+		values.Add("channel", params.Channel)
+	}
+	if params.TimestampFrom != DEFAULT_REMOTE_FILES_TS_FROM {
+		values.Add("ts_from", strconv.FormatInt(int64(params.TimestampFrom), 10))
+	}
+	if params.TimestampTo != DEFAULT_REMOTE_FILES_TS_TO {
+		values.Add("ts_to", strconv.FormatInt(int64(params.TimestampTo), 10))
+	}
+	if params.Limit != DEFAULT_REMOTE_FILES_COUNT {
+		values.Add("limit", strconv.Itoa(params.Limit))
+	}
+	if params.Cursor != "" {
+		values.Add("cursor", params.Cursor)
+	}
+
+	response, err := api.remoteFileRequest(ctx, "files.remote.list", values)
+	if err != nil {
+		return nil, err
+	}
+
+	params.Cursor = response.SlackResponse.ResponseMetadata.Cursor
+
+	return response.Files, nil
+}
+
+// GetRemoteFileInfo retrieves the complete remote file information.
+// For more details:
+// https://api.slack.com/methods/files.remote.info
+func (api *Client) GetRemoteFileInfo(externalID, fileID string) (remotefile *RemoteFile, err error) {
+	return api.GetRemoteFileInfoContext(context.Background(), externalID, fileID)
+}
+
+// GetRemoteFileInfoContext retrieves the complete remote file information given with a custom context.
+// For more details see the GetRemoteFileInfo documentation.
+func (api *Client) GetRemoteFileInfoContext(ctx context.Context, externalID, fileID string) (remotefile *RemoteFile, err error) {
+	if fileID == "" && externalID == "" {
+		return nil, fmt.Errorf("either externalID or fileID is required")
+	}
+	if fileID != "" && externalID != "" {
+		return nil, fmt.Errorf("don't provide both externalID and fileID")
+	}
+	values := url.Values{
+		"token": {api.token},
+	}
+	if fileID != "" {
+		values.Add("file", fileID)
+	}
+	if externalID != "" {
+		values.Add("external_id", externalID)
+	}
+	response, err := api.remoteFileRequest(ctx, "files.remote.info", values)
+	if err != nil {
+		return nil, err
+	}
+	return &response.RemoteFile, err
+}
+
+// ShareRemoteFile shares a remote file to channels
+// For more details:
+// https://api.slack.com/methods/files.remote.share
+func (api *Client) ShareRemoteFile(channels []string, externalID, fileID string) (file *RemoteFile, err error) {
+	return api.ShareRemoteFileContext(context.Background(), channels, externalID, fileID)
+}
+
+// ShareRemoteFileContext shares a remote file to channels with a custom context.
+// For more details see the ShareRemoteFile documentation.
+func (api *Client) ShareRemoteFileContext(ctx context.Context, channels []string, externalID, fileID string) (file *RemoteFile, err error) {
+	if channels == nil || len(channels) == 0 {
+		return nil, ErrParametersMissing
+	}
+	if fileID == "" && externalID == "" {
+		return nil, fmt.Errorf("either externalID or fileID is required")
+	}
+	values := url.Values{
+		"token":    {api.token},
+		"channels": {strings.Join(channels, ",")},
+	}
+	if fileID != "" {
+		values.Add("file", fileID)
+	}
+	if externalID != "" {
+		values.Add("external_id", externalID)
+	}
+	response, err := api.remoteFileRequest(ctx, "files.remote.share", values)
+	if err != nil {
+		return nil, err
+	}
+	return &response.RemoteFile, err
+}
+
+// UpdateRemoteFile updates a remote file
+// For more details:
+// https://api.slack.com/methods/files.remote.update
+func (api *Client) UpdateRemoteFile(fileID string, params RemoteFileParameters) (remotefile *RemoteFile, err error) {
+	return api.UpdateRemoteFileContext(context.Background(), fileID, params)
+}
+
+// UpdateRemoteFileContext updates a remote file with a custom context
+// For more details see the UpdateRemoteFile documentation.
+func (api *Client) UpdateRemoteFileContext(ctx context.Context, fileID string, params RemoteFileParameters) (remotefile *RemoteFile, err error) {
+	response := &remoteFileResponseFull{}
+	values := url.Values{
+		"token": {api.token},
+	}
+	if fileID != "" {
+		values.Add("file", fileID)
+	}
+	if params.ExternalID != "" {
+		values.Add("external_id", params.ExternalID)
+	}
+	if params.ExternalURL != "" {
+		values.Add("external_url", params.ExternalURL)
+	}
+	if params.Title != "" {
+		values.Add("title", params.Title)
+	}
+	if params.Filetype != "" {
+		values.Add("filetype", params.Filetype)
+	}
+	if params.IndexableFileContents != "" {
+		values.Add("indexable_file_contents", params.IndexableFileContents)
+	}
+	if params.PreviewImageReader != nil {
+		err = postWithMultipartResponse(ctx, api.httpclient, api.endpoint+"files.remote.update", "preview.png", "preview_image", api.token, values, params.PreviewImageReader, response, api)
+	} else {
+		response, err = api.remoteFileRequest(ctx, "files.remote.update", values)
+	}
+
+	if err != nil {
+		return nil, err
+	}
+
+	return &response.RemoteFile, response.Err()
+}
+
+// RemoveRemoteFile removes a remote file.
+// For more details:
+// https://api.slack.com/methods/files.remote.remove
+func (api *Client) RemoveRemoteFile(externalID, fileID string) (err error) {
+	return api.RemoveRemoteFileContext(context.Background(), externalID, fileID)
+}
+
+// RemoveRemoteFileContext removes a remote file with a custom context
+// For more information see the RemoveRemoteFiles documentation.
+func (api *Client) RemoveRemoteFileContext(ctx context.Context, externalID, fileID string) (err error) {
+	if fileID == "" && externalID == "" {
+		return fmt.Errorf("either externalID or fileID is required")
+	}
+	if fileID != "" && externalID != "" {
+		return fmt.Errorf("don't provide both externalID and fileID")
+	}
+	values := url.Values{
+		"token": {api.token},
+	}
+	if fileID != "" {
+		values.Add("file", fileID)
+	}
+	if externalID != "" {
+		values.Add("external_id", externalID)
+	}
+	_, err = api.remoteFileRequest(ctx, "files.remote.remove", values)
+	return err
+}
diff --git a/vendor/github.com/slack-go/slack/internal/misc/misc.go b/vendor/github.com/slack-go/slack/status_code_error.go
index eab8cdd8..7347137a 100644
--- a/vendor/github.com/slack-go/slack/internal/misc/misc.go
+++ b/vendor/github.com/slack-go/slack/status_code_error.go
@@ -1,4 +1,4 @@
-package misc
+package slack
 
 import (
 	"fmt"
diff --git a/vendor/github.com/slack-go/slack/views.go b/vendor/github.com/slack-go/slack/views.go
index a3a1bd05..e6a96178 100644
--- a/vendor/github.com/slack-go/slack/views.go
+++ b/vendor/github.com/slack-go/slack/views.go
@@ -18,24 +18,25 @@ type ViewState struct {
 
 type View struct {
 	SlackResponse
-	ID              string           `json:"id"`
-	TeamID          string           `json:"team_id"`
-	Type            ViewType         `json:"type"`
-	Title           *TextBlockObject `json:"title"`
-	Close           *TextBlockObject `json:"close"`
-	Submit          *TextBlockObject `json:"submit"`
-	Blocks          Blocks           `json:"blocks"`
-	PrivateMetadata string           `json:"private_metadata"`
-	CallbackID      string           `json:"callback_id"`
-	State           *ViewState       `json:"state"`
-	Hash            string           `json:"hash"`
-	ClearOnClose    bool             `json:"clear_on_close"`
-	NotifyOnClose   bool             `json:"notify_on_close"`
-	RootViewID      string           `json:"root_view_id"`
-	PreviousViewID  string           `json:"previous_view_id"`
-	AppID           string           `json:"app_id"`
-	ExternalID      string           `json:"external_id"`
-	BotID           string           `json:"bot_id"`
+	ID                 string           `json:"id"`
+	TeamID             string           `json:"team_id"`
+	Type               ViewType         `json:"type"`
+	Title              *TextBlockObject `json:"title"`
+	Close              *TextBlockObject `json:"close"`
+	Submit             *TextBlockObject `json:"submit"`
+	Blocks             Blocks           `json:"blocks"`
+	PrivateMetadata    string           `json:"private_metadata"`
+	CallbackID         string           `json:"callback_id"`
+	State              *ViewState       `json:"state"`
+	Hash               string           `json:"hash"`
+	ClearOnClose       bool             `json:"clear_on_close"`
+	NotifyOnClose      bool             `json:"notify_on_close"`
+	RootViewID         string           `json:"root_view_id"`
+	PreviousViewID     string           `json:"previous_view_id"`
+	AppID              string           `json:"app_id"`
+	ExternalID         string           `json:"external_id"`
+	BotID              string           `json:"bot_id"`
+	AppInstalledTeamID string           `json:"app_installed_team_id"`
 }
 
 type ViewSubmissionCallbackResponseURL struct {
diff --git a/vendor/github.com/slack-go/slack/webhooks.go b/vendor/github.com/slack-go/slack/webhooks.go
index 15097f03..2f8fb47f 100644
--- a/vendor/github.com/slack-go/slack/webhooks.go
+++ b/vendor/github.com/slack-go/slack/webhooks.go
@@ -5,6 +5,8 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
+	"io"
+	"io/ioutil"
 	"net/http"
 )
 
@@ -21,6 +23,7 @@ type WebhookMessage struct {
 	ResponseType    string       `json:"response_type,omitempty"`
 	ReplaceOriginal bool         `json:"replace_original,omitempty"`
 	DeleteOriginal  bool         `json:"delete_original,omitempty"`
+	ReplyBroadcast  bool         `json:"reply_broadcast,omitempty"`
 }
 
 func PostWebhook(url string, msg *WebhookMessage) error {
@@ -51,7 +54,10 @@ func PostWebhookCustomHTTPContext(ctx context.Context, url string, httpClient *h
 	if err != nil {
 		return fmt.Errorf("failed to post webhook: %w", err)
 	}
-	defer resp.Body.Close()
+	defer func() {
+		io.Copy(ioutil.Discard, resp.Body)
+		resp.Body.Close()
+	}()
 
 	return checkStatusCode(resp, discard{})
 }
diff --git a/vendor/github.com/slack-go/slack/websocket_managed_conn.go b/vendor/github.com/slack-go/slack/websocket_managed_conn.go
index 92536171..14dea219 100644
--- a/vendor/github.com/slack-go/slack/websocket_managed_conn.go
+++ b/vendor/github.com/slack-go/slack/websocket_managed_conn.go
@@ -13,7 +13,6 @@ import (
 
 	"github.com/slack-go/slack/internal/backoff"
 	"github.com/slack-go/slack/internal/errorsx"
-	"github.com/slack-go/slack/internal/misc"
 	"github.com/slack-go/slack/internal/timex"
 )
 
@@ -127,7 +126,7 @@ func (rtm *RTM) connect(connectionCount int, useRTMStart bool) (*Info, *websocke
 		}
 
 		switch actual := err.(type) {
-		case misc.StatusCodeError:
+		case StatusCodeError:
 			if actual.Code == http.StatusNotFound {
 				rtm.Debugf("invalid auth when connecting with RTM: %s", err)
 				rtm.IncomingEvents <- RTMEvent{"invalid_auth", &InvalidAuthEvent{}}
@@ -475,7 +474,7 @@ func (rtm *RTM) handleEvent(typeStr string, event json.RawMessage) {
 	v, exists := EventMapping[typeStr]
 	if !exists {
 		rtm.Debugf("RTM Error - received unmapped event %q: %s\n", typeStr, string(event))
-		err := fmt.Errorf("RTM Error: Received unmapped event %q: %s", typeStr, string(event))
+		err := fmt.Errorf("RTM Error: Received unmapped event %q", typeStr)
 		rtm.IncomingEvents <- RTMEvent{"unmarshalling_error", &UnmarshallingErrorEvent{err}}
 		return
 	}
@@ -484,7 +483,7 @@ func (rtm *RTM) handleEvent(typeStr string, event json.RawMessage) {
 	err := json.Unmarshal(event, recvEvent)
 	if err != nil {
 		rtm.Debugf("RTM Error, could not unmarshall event %q: %s\n", typeStr, string(event))
-		err := fmt.Errorf("RTM Error: Could not unmarshall event %q: %s", typeStr, string(event))
+		err := fmt.Errorf("RTM Error: Could not unmarshall event %q", typeStr)
 		rtm.IncomingEvents <- RTMEvent{"unmarshalling_error", &UnmarshallingErrorEvent{err}}
 		return
 	}
diff --git a/vendor/github.com/slack-go/slack/websocket_subteam.go b/vendor/github.com/slack-go/slack/websocket_subteam.go
index a23b274c..2f618733 100644
--- a/vendor/github.com/slack-go/slack/websocket_subteam.go
+++ b/vendor/github.com/slack-go/slack/websocket_subteam.go
@@ -14,9 +14,9 @@ type SubteamMembersChangedEvent struct {
 	DatePreviousUpdate JSONTime `json:"date_previous_update"`
 	DateUpdate         JSONTime `json:"date_update"`
 	AddedUsers         []string `json:"added_users"`
-	AddedUsersCount    string   `json:"added_users_count"`
+	AddedUsersCount    int      `json:"added_users_count"`
 	RemovedUsers       []string `json:"removed_users"`
-	RemovedUsersCount  string   `json:"removed_users_count"`
+	RemovedUsersCount  int      `json:"removed_users_count"`
 }
 
 // SubteamSelfAddedEvent represents an event of you have been added to a User Group
diff --git a/vendor/github.com/stretchr/testify/assert/assertion_compare_can_convert.go b/vendor/github.com/stretchr/testify/assert/assertion_compare_can_convert.go
index df22c47f..da867903 100644
--- a/vendor/github.com/stretchr/testify/assert/assertion_compare_can_convert.go
+++ b/vendor/github.com/stretchr/testify/assert/assertion_compare_can_convert.go
@@ -9,7 +9,7 @@ package assert
 
 import "reflect"
 
-// Wrapper around reflect.Value.CanConvert, for compatability
+// Wrapper around reflect.Value.CanConvert, for compatibility
 // reasons.
 func canConvert(value reflect.Value, to reflect.Type) bool {
 	return value.CanConvert(to)
diff --git a/vendor/github.com/stretchr/testify/suite/suite.go b/vendor/github.com/stretchr/testify/suite/suite.go
index b9b5d1c5..1c402e8d 100644
--- a/vendor/github.com/stretchr/testify/suite/suite.go
+++ b/vendor/github.com/stretchr/testify/suite/suite.go
@@ -7,6 +7,7 @@ import (
 	"reflect"
 	"regexp"
 	"runtime/debug"
+	"sync"
 	"testing"
 	"time"
 
@@ -21,17 +22,22 @@ var matchMethod = flag.String("testify.m", "", "regular expression to select tes
 // retrieving the current *testing.T context.
 type Suite struct {
 	*assert.Assertions
+	mu      sync.RWMutex
 	require *require.Assertions
 	t       *testing.T
 }
 
 // T retrieves the current *testing.T context.
 func (suite *Suite) T() *testing.T {
+	suite.mu.RLock()
+	defer suite.mu.RUnlock()
 	return suite.t
 }
 
 // SetT sets the current *testing.T context.
 func (suite *Suite) SetT(t *testing.T) {
+	suite.mu.Lock()
+	defer suite.mu.Unlock()
 	suite.t = t
 	suite.Assertions = assert.New(t)
 	suite.require = require.New(t)
@@ -39,6 +45,8 @@ func (suite *Suite) SetT(t *testing.T) {
 
 // Require returns a require context for suite.
 func (suite *Suite) Require() *require.Assertions {
+	suite.mu.Lock()
+	defer suite.mu.Unlock()
 	if suite.require == nil {
 		suite.require = require.New(suite.T())
 	}
@@ -51,6 +59,8 @@ func (suite *Suite) Require() *require.Assertions {
 // assert.Assertions with require.Assertions), this method is provided so you
 // can call `suite.Assert().NoError()`.
 func (suite *Suite) Assert() *assert.Assertions {
+	suite.mu.Lock()
+	defer suite.mu.Unlock()
 	if suite.Assertions == nil {
 		suite.Assertions = assert.New(suite.T())
 	}