Merge pull request #4628 from thaJeztah/24.0_backport_bump_compress

[24.0 backport] vendor: github.com/klauspost/compress v1.17.2

vendor.mod

@@ -57,7 +57,7 @@ require (
 	github.com/golang/protobuf v1.5.2 // indirect
 	github.com/gorilla/mux v1.8.0 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
-	github.com/klauspost/compress v1.16.3 // indirect
+	github.com/klauspost/compress v1.17.2 // indirect
 	github.com/matttproud/golang_protobuf_extensions v1.0.4 // indirect
 	github.com/miekg/pkcs11 v1.1.1 // indirect
 	github.com/moby/sys/symlink v0.2.0 // indirect

vendor.sum

@@ -243,8 +243,8 @@ github.com/julienschmidt/httprouter v1.3.0/go.mod h1:JR6WtHb+2LUe8TCKY3cZOxFyyO8
 github.com/kisielk/errcheck v1.2.0/go.mod h1:/BMXB+zMLi60iA8Vv6Ksmxu/1UDYcXs4uQLJ+jE2L00=
 github.com/kisielk/errcheck v1.5.0/go.mod h1:pFxgyoBC7bSaBwPgfKdkLd5X25qrDl4LWUI2bnpBCr8=
 github.com/kisielk/gotool v1.0.0/go.mod h1:XhKaO+MFFWcvkIS/tQcRk01m1F5IRFswLeQ+oQHNcck=
-github.com/klauspost/compress v1.16.3 h1:XuJt9zzcnaz6a16/OU53ZjWp/v7/42WcR5t2a0PcNQY=
+github.com/klauspost/compress v1.17.2 h1:RlWWUY/Dr4fL8qk9YG7DTZ7PDgME2V4csBXA8L/ixi4=
-github.com/klauspost/compress v1.16.3/go.mod h1:ntbaceVETuRiXiv4DpjP66DpAtAGkEQskQzEyD//IeE=
+github.com/klauspost/compress v1.17.2/go.mod h1:ntbaceVETuRiXiv4DpjP66DpAtAGkEQskQzEyD//IeE=
 github.com/konsorten/go-windows-terminal-sequences v1.0.1/go.mod h1:T0+1ngSBFLxvqU3pZ+m/2kptfBszLMUkC4ZK/EgS/cQ=
 github.com/konsorten/go-windows-terminal-sequences v1.0.3/go.mod h1:T0+1ngSBFLxvqU3pZ+m/2kptfBszLMUkC4ZK/EgS/cQ=
 github.com/kr/logfmt v0.0.0-20140226030751-b84e30acd515/go.mod h1:+0opPa2QZZtGFBFZlji/RkVcI2GknAs/DXo4wKdlNEc=

vendor/github.com/klauspost/compress/.goreleaser.yml

@@ -3,7 +3,7 @@
 before:
   hooks:
     - ./gen.sh
-    - go install mvdan.cc/garble@v0.9.3
+    - go install mvdan.cc/garble@v0.10.1
 
 builds:
   -
@@ -92,16 +92,7 @@ builds:
 archives:
   -
     id: s2-binaries
-    name_template: "s2-{{ .Os }}_{{ .Arch }}_{{ .Version }}"
+    name_template: "s2-{{ .Os }}_{{ .Arch }}{{ if .Arm }}v{{ .Arm }}{{ end }}"
-    replacements:
-      aix: AIX
-      darwin: OSX
-      linux: Linux
-      windows: Windows
-      386: i386
-      amd64: x86_64
-      freebsd: FreeBSD
-      netbsd: NetBSD
     format_overrides:
       - goos: windows
         format: zip
@@ -125,7 +116,7 @@ changelog:
 
 nfpms:
   -
-    file_name_template: "s2_package_{{ .Version }}_{{ .Os }}_{{ .Arch }}"
+    file_name_template: "s2_package__{{ .Os }}_{{ .Arch }}{{ if .Arm }}v{{ .Arm }}{{ end }}"
     vendor: Klaus Post
     homepage: https://github.com/klauspost/compress
     maintainer: Klaus Post <klauspost@gmail.com>
@@ -134,8 +125,3 @@ nfpms:
     formats:
       - deb
       - rpm
-    replacements:
-      darwin: Darwin
-      linux: Linux
-      freebsd: FreeBSD
-      amd64: x86_64

vendor/github.com/klauspost/compress/README.md

@@ -16,6 +16,37 @@ This package provides various compression algorithms.
 
 # changelog
 
+* Sept 19th, 2023 - [v1.17.0](https://github.com/klauspost/compress/releases/tag/v1.17.0)
+	* Add experimental dictionary builder  https://github.com/klauspost/compress/pull/853
+	* Add xerial snappy read/writer https://github.com/klauspost/compress/pull/838
+	* flate: Add limited window compression https://github.com/klauspost/compress/pull/843
+	* s2: Do 2 overlapping match checks https://github.com/klauspost/compress/pull/839
+	* flate: Add amd64 assembly matchlen https://github.com/klauspost/compress/pull/837
+	* gzip: Copy bufio.Reader on Reset by @thatguystone in https://github.com/klauspost/compress/pull/860
+   
+* July 1st, 2023 - [v1.16.7](https://github.com/klauspost/compress/releases/tag/v1.16.7)
+	* zstd: Fix default level first dictionary encode https://github.com/klauspost/compress/pull/829
+	* s2: add GetBufferCapacity() method by @GiedriusS in https://github.com/klauspost/compress/pull/832
+
+* June 13, 2023 - [v1.16.6](https://github.com/klauspost/compress/releases/tag/v1.16.6)
+	* zstd: correctly ignore WithEncoderPadding(1) by @ianlancetaylor in https://github.com/klauspost/compress/pull/806
+	* zstd: Add amd64 match length assembly https://github.com/klauspost/compress/pull/824
+	* gzhttp: Handle informational headers by @rtribotte in https://github.com/klauspost/compress/pull/815
+	* s2: Improve Better compression slightly https://github.com/klauspost/compress/pull/663
+
+* Apr 16, 2023 - [v1.16.5](https://github.com/klauspost/compress/releases/tag/v1.16.5)
+	* zstd: readByte needs to use io.ReadFull by @jnoxon in https://github.com/klauspost/compress/pull/802
+	* gzip: Fix WriterTo after initial read https://github.com/klauspost/compress/pull/804
+
+* Apr 5, 2023 - [v1.16.4](https://github.com/klauspost/compress/releases/tag/v1.16.4)
+	* zstd: Improve zstd best efficiency by @greatroar and @klauspost in https://github.com/klauspost/compress/pull/784
+	* zstd: Respect WithAllLitEntropyCompression https://github.com/klauspost/compress/pull/792
+	* zstd: Fix amd64 not always detecting corrupt data https://github.com/klauspost/compress/pull/785
+	* zstd: Various minor improvements by @greatroar in https://github.com/klauspost/compress/pull/788 https://github.com/klauspost/compress/pull/794 https://github.com/klauspost/compress/pull/795
+	* s2: Fix huge block overflow https://github.com/klauspost/compress/pull/779
+	* s2: Allow CustomEncoder fallback https://github.com/klauspost/compress/pull/780
+	* gzhttp: Suppport ResponseWriter Unwrap() in gzhttp handler by @jgimenez in https://github.com/klauspost/compress/pull/799
+
 * Mar 13, 2023 - [v1.16.1](https://github.com/klauspost/compress/releases/tag/v1.16.1)
 	* zstd: Speed up + improve best encoder by @greatroar in https://github.com/klauspost/compress/pull/776
 	* gzhttp: Add optional [BREACH mitigation](https://github.com/klauspost/compress/tree/master/gzhttp#breach-mitigation). https://github.com/klauspost/compress/pull/762 https://github.com/klauspost/compress/pull/768 https://github.com/klauspost/compress/pull/769 https://github.com/klauspost/compress/pull/770 https://github.com/klauspost/compress/pull/767
@@ -31,6 +62,9 @@ This package provides various compression algorithms.
 	* s2: Support io.ReaderAt in ReadSeeker. https://github.com/klauspost/compress/pull/747
 	* s2c/s2sx: Use concurrent decoding. https://github.com/klauspost/compress/pull/746
 
+<details>
+	<summary>See changes to v1.15.x</summary>
+	
 * Jan 21st, 2023 (v1.15.15)
 	* deflate: Improve level 7-9 by @klauspost in https://github.com/klauspost/compress/pull/739
 	* zstd: Add delta encoding support by @greatroar in https://github.com/klauspost/compress/pull/728
@@ -157,6 +191,8 @@ Stream decompression is now faster on asynchronous, since the goroutine allocati
 
 While the release has been extensively tested, it is recommended to testing when upgrading.
 
+</details>
+
 <details>
 	<summary>See changes to v1.14.x</summary>
 	
@@ -615,6 +651,10 @@ Here are other packages of good quality and pure Go (no cgo wrappers or autoconv
 * [github.com/pierrec/lz4](https://github.com/pierrec/lz4) - strong multithreaded LZ4 compression.
 * [github.com/cosnicolaou/pbzip2](https://github.com/cosnicolaou/pbzip2) - multithreaded bzip2 decompression.
 * [github.com/dsnet/compress](https://github.com/dsnet/compress) - brotli decompression, bzip2 writer.
+* [github.com/ronanh/intcomp](https://github.com/ronanh/intcomp) - Integer compression.
+* [github.com/spenczar/fpc](https://github.com/spenczar/fpc) - Float compression.
+* [github.com/minio/zipindex](https://github.com/minio/zipindex) - External ZIP directory index.
+* [github.com/ybirader/pzip](https://github.com/ybirader/pzip) - Fast concurrent zip archiver and extractor.
 
 # license
 

vendor/github.com/klauspost/compress/SECURITY.md

@@ -0,0 +1,25 @@
+# Security Policy
+
+## Supported Versions
+
+Security updates are applied only to the latest release.
+
+## Vulnerability Definition
+
+A security vulnerability is a bug that with certain input triggers a crash or an infinite loop. Most calls will have varying execution time and only in rare cases will slow operation be considered a security vulnerability.
+
+Corrupted output generally is not considered a security vulnerability, unless independent operations are able to affect each other. Note that not all functionality is re-entrant and safe to use concurrently.
+
+Out-of-memory crashes only applies if the en/decoder uses an abnormal amount of memory, with appropriate options applied, to limit maximum window size, concurrency, etc. However, if you are in doubt you are welcome to file a security issue.
+
+It is assumed that all callers are trusted, meaning internal data exposed through reflection or inspection of returned data structures is not considered a vulnerability.
+
+Vulnerabilities resulting from compiler/assembler errors should be reported upstream. Depending on the severity this package may or may not implement a workaround.
+
+## Reporting a Vulnerability
+
+If you have discovered a security vulnerability in this project, please report it privately. **Do not disclose it as a public issue.** This gives us time to work with you to fix the issue before public exposure, reducing the chance that the exploit will be used before a patch is released.
+
+Please disclose it at [security advisory](https://github.com/klauspost/compress/security/advisories/new). If possible please provide a minimal reproducer. If the issue only applies to a single platform, it would be helpful to provide access to that.
+
+This project is maintained by a team of volunteers on a reasonable-effort basis. As such, vulnerabilities will be disclosed in a best effort base.

vendor/github.com/klauspost/compress/fse/bitwriter.go

@@ -152,12 +152,11 @@ func (b *bitWriter) flushAlign() {
 
 // close will write the alignment bit and write the final byte(s)
 // to the output.
-func (b *bitWriter) close() error {
+func (b *bitWriter) close() {
 	// End mark
 	b.addBits16Clean(1, 1)
 	// flush until next byte.
 	b.flushAlign()
-	return nil
 }
 
 // reset and continue writing by appending to out.

vendor/github.com/klauspost/compress/fse/compress.go

@@ -199,7 +199,8 @@ func (s *Scratch) compress(src []byte) error {
 	c2.flush(s.actualTableLog)
 	c1.flush(s.actualTableLog)
 
-	return s.bw.close()
+	s.bw.close()
+	return nil
 }
 
 // writeCount will write the normalized histogram count to header.

vendor/github.com/klauspost/compress/huff0/bitwriter.go

@@ -13,14 +13,6 @@ type bitWriter struct {
 	out          []byte
 }
 
-// bitMask16 is bitmasks. Has extra to avoid bounds check.
-var bitMask16 = [32]uint16{
-	0, 1, 3, 7, 0xF, 0x1F,
-	0x3F, 0x7F, 0xFF, 0x1FF, 0x3FF, 0x7FF,
-	0xFFF, 0x1FFF, 0x3FFF, 0x7FFF, 0xFFFF, 0xFFFF,
-	0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF,
-	0xFFFF, 0xFFFF} /* up to 16 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) {
@@ -102,10 +94,9 @@ func (b *bitWriter) flushAlign() {
 
 // close will write the alignment bit and write the final byte(s)
 // to the output.
-func (b *bitWriter) close() error {
+func (b *bitWriter) close() {
 	// End mark
 	b.addBits16Clean(1, 1)
 	// flush until next byte.
 	b.flushAlign()
-	return nil
 }

vendor/github.com/klauspost/compress/huff0/compress.go

@@ -227,10 +227,10 @@ func EstimateSizes(in []byte, s *Scratch) (tableSz, dataSz, reuseSz int, err err
 }
 
 func (s *Scratch) compress1X(src []byte) ([]byte, error) {
-	return s.compress1xDo(s.Out, src)
+	return s.compress1xDo(s.Out, src), nil
 }
 
-func (s *Scratch) compress1xDo(dst, src []byte) ([]byte, error) {
+func (s *Scratch) compress1xDo(dst, src []byte) []byte {
 	var bw = bitWriter{out: dst}
 
 	// N is length divisible by 4.
@@ -260,8 +260,8 @@ func (s *Scratch) compress1xDo(dst, src []byte) ([]byte, error) {
 			bw.encTwoSymbols(cTable, tmp[1], tmp[0])
 		}
 	}
-	err := bw.close()
+	bw.close()
-	return bw.out, err
+	return bw.out
 }
 
 var sixZeros [6]byte
@@ -283,12 +283,8 @@ func (s *Scratch) compress4X(src []byte) ([]byte, error) {
 		}
 		src = src[len(toDo):]
 
-		var err error
 		idx := len(s.Out)
-		s.Out, err = s.compress1xDo(s.Out, toDo)
+		s.Out = s.compress1xDo(s.Out, toDo)
-		if err != nil {
-			return nil, err
-		}
 		if len(s.Out)-idx > math.MaxUint16 {
 			// We cannot store the size in the jump table
 			return nil, ErrIncompressible
@@ -315,7 +311,6 @@ func (s *Scratch) compress4Xp(src []byte) ([]byte, error) {
 
 	segmentSize := (len(src) + 3) / 4
 	var wg sync.WaitGroup
-	var errs [4]error
 	wg.Add(4)
 	for i := 0; i < 4; i++ {
 		toDo := src
@@ -326,15 +321,12 @@ func (s *Scratch) compress4Xp(src []byte) ([]byte, error) {
 
 		// Separate goroutine for each block.
 		go func(i int) {
-			s.tmpOut[i], errs[i] = s.compress1xDo(s.tmpOut[i][:0], toDo)
+			s.tmpOut[i] = s.compress1xDo(s.tmpOut[i][:0], toDo)
 			wg.Done()
 		}(i)
 	}
 	wg.Wait()
 	for i := 0; i < 4; i++ {
-		if errs[i] != nil {
-			return nil, errs[i]
-		}
 		o := s.tmpOut[i]
 		if len(o) > math.MaxUint16 {
 			// We cannot store the size in the jump table

vendor/github.com/klauspost/compress/huff0/decompress.go

@@ -253,7 +253,7 @@ func (d *Decoder) decompress1X8Bit(dst, src []byte) ([]byte, error) {
 
 	switch d.actualTableLog {
 	case 8:
-		const shift = 8 - 8
+		const shift = 0
 		for br.off >= 4 {
 			br.fillFast()
 			v := dt[uint8(br.value>>(56+shift))]

vendor/github.com/klauspost/compress/internal/snapref/encode_other.go

@@ -87,18 +87,6 @@ func emitCopy(dst []byte, offset, length int) int {
 	return i + 2
 }
 
-// extendMatch returns the largest k such that k <= len(src) and that
-// src[i:i+k-j] and src[j:k] have the same contents.
-//
-// It assumes that:
-//
-//	0 <= i && i < j && j <= len(src)
-func extendMatch(src []byte, i, j int) int {
-	for ; j < len(src) && src[i] == src[j]; i, j = i+1, j+1 {
-	}
-	return j
-}
-
 func hash(u, shift uint32) uint32 {
 	return (u * 0x1e35a7bd) >> shift
 }

vendor/github.com/klauspost/compress/zstd/README.md

@@ -304,7 +304,7 @@ import "github.com/klauspost/compress/zstd"
 
 // Create a reader that caches decompressors.
 // For this operation type we supply a nil Reader.
-var decoder, _ = zstd.NewReader(nil, WithDecoderConcurrency(0))
+var decoder, _ = zstd.NewReader(nil, zstd.WithDecoderConcurrency(0))
 
 // Decompress a buffer. We don't supply a destination buffer,
 // so it will be allocated by the decoder.

vendor/github.com/klauspost/compress/zstd/bitreader.go

@@ -17,7 +17,6 @@ import (
 // for aligning the input.
 type bitReader struct {
 	in       []byte
-	off      uint   // next byte to read is at in[off - 1]
 	value    uint64 // Maybe use [16]byte, but shifting is awkward.
 	bitsRead uint8
 }
@@ -28,7 +27,6 @@ func (b *bitReader) init(in []byte) error {
 		return errors.New("corrupt stream: too short")
 	}
 	b.in = in
-	b.off = uint(len(in))
 	// The highest bit of the last byte indicates where to start
 	v := in[len(in)-1]
 	if v == 0 {
@@ -69,21 +67,19 @@ func (b *bitReader) fillFast() {
 	if b.bitsRead < 32 {
 		return
 	}
-	// 2 bounds checks.
+	v := b.in[len(b.in)-4:]
-	v := b.in[b.off-4:]
+	b.in = b.in[:len(b.in)-4]
-	v = v[:4]
 	low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
 	b.value = (b.value << 32) | uint64(low)
 	b.bitsRead -= 32
-	b.off -= 4
 }
 
 // fillFastStart() assumes the bitreader is empty and there is at least 8 bytes to read.
 func (b *bitReader) fillFastStart() {
-	// Do single re-slice to avoid bounds checks.
+	v := b.in[len(b.in)-8:]
-	b.value = binary.LittleEndian.Uint64(b.in[b.off-8:])
+	b.in = b.in[:len(b.in)-8]
+	b.value = binary.LittleEndian.Uint64(v)
 	b.bitsRead = 0
-	b.off -= 8
 }
 
 // fill() will make sure at least 32 bits are available.
@@ -91,25 +87,25 @@ func (b *bitReader) fill() {
 	if b.bitsRead < 32 {
 		return
 	}
-	if b.off >= 4 {
+	if len(b.in) >= 4 {
-		v := b.in[b.off-4:]
+		v := b.in[len(b.in)-4:]
-		v = v[:4]
+		b.in = b.in[:len(b.in)-4]
 		low := (uint32(v[0])) | (uint32(v[1]) << 8) | (uint32(v[2]) << 16) | (uint32(v[3]) << 24)
 		b.value = (b.value << 32) | uint64(low)
 		b.bitsRead -= 32
-		b.off -= 4
 		return
 	}
-	for b.off > 0 {
+
-		b.value = (b.value << 8) | uint64(b.in[b.off-1])
+	b.bitsRead -= uint8(8 * len(b.in))
-		b.bitsRead -= 8
+	for len(b.in) > 0 {
-		b.off--
+		b.value = (b.value << 8) | uint64(b.in[len(b.in)-1])
+		b.in = b.in[:len(b.in)-1]
 	}
 }
 
 // finished returns true if all bits have been read from the bit stream.
 func (b *bitReader) finished() bool {
-	return b.off == 0 && b.bitsRead >= 64
+	return len(b.in) == 0 && b.bitsRead >= 64
 }
 
 // overread returns true if more bits have been requested than is on the stream.
@@ -119,7 +115,7 @@ func (b *bitReader) overread() bool {
 
 // remain returns the number of bits remaining.
 func (b *bitReader) remain() uint {
-	return b.off*8 + 64 - uint(b.bitsRead)
+	return 8*uint(len(b.in)) + 64 - uint(b.bitsRead)
 }
 
 // close the bitstream and returns an error if out-of-buffer reads occurred.

vendor/github.com/klauspost/compress/zstd/bitwriter.go

@@ -97,12 +97,11 @@ func (b *bitWriter) flushAlign() {
 
 // close will write the alignment bit and write the final byte(s)
 // to the output.
-func (b *bitWriter) close() error {
+func (b *bitWriter) close() {
 	// End mark
 	b.addBits16Clean(1, 1)
 	// flush until next byte.
 	b.flushAlign()
-	return nil
 }
 
 // reset and continue writing by appending to out.

vendor/github.com/klauspost/compress/zstd/blockdec.go

@@ -592,7 +592,7 @@ func (b *blockDec) prepareSequences(in []byte, hist *history) (err error) {
 				}
 				seq.fse.setRLE(symb)
 				if debugDecoder {
-					printf("RLE set to %+v, code: %v", symb, v)
+					printf("RLE set to 0x%x, code: %v", symb, v)
 				}
 			case compModeFSE:
 				println("Reading table for", tableIndex(i))

vendor/github.com/klauspost/compress/zstd/blockenc.go

@@ -361,14 +361,21 @@ func (b *blockEnc) encodeLits(lits []byte, raw bool) error {
 	if len(lits) >= 1024 {
 		// Use 4 Streams.
 		out, reUsed, err = huff0.Compress4X(lits, b.litEnc)
-	} else if len(lits) > 32 {
+	} else if len(lits) > 16 {
 		// Use 1 stream
 		single = true
 		out, reUsed, err = huff0.Compress1X(lits, b.litEnc)
 	} else {
 		err = huff0.ErrIncompressible
 	}
-
+	if err == nil && len(out)+5 > len(lits) {
+		// If we are close, we may still be worse or equal to raw.
+		var lh literalsHeader
+		lh.setSizes(len(out), len(lits), single)
+		if len(out)+lh.size() >= len(lits) {
+			err = huff0.ErrIncompressible
+		}
+	}
 	switch err {
 	case huff0.ErrIncompressible:
 		if debugEncoder {
@@ -473,7 +480,7 @@ func (b *blockEnc) encode(org []byte, raw, rawAllLits bool) error {
 		return b.encodeLits(b.literals, rawAllLits)
 	}
 	// We want some difference to at least account for the headers.
-	saved := b.size - len(b.literals) - (b.size >> 5)
+	saved := b.size - len(b.literals) - (b.size >> 6)
 	if saved < 16 {
 		if org == nil {
 			return errIncompressible
@@ -503,7 +510,7 @@ func (b *blockEnc) encode(org []byte, raw, rawAllLits bool) error {
 	if len(b.literals) >= 1024 && !raw {
 		// Use 4 Streams.
 		out, reUsed, err = huff0.Compress4X(b.literals, b.litEnc)
-	} else if len(b.literals) > 32 && !raw {
+	} else if len(b.literals) > 16 && !raw {
 		// Use 1 stream
 		single = true
 		out, reUsed, err = huff0.Compress1X(b.literals, b.litEnc)
@@ -511,6 +518,17 @@ func (b *blockEnc) encode(org []byte, raw, rawAllLits bool) error {
 		err = huff0.ErrIncompressible
 	}
 
+	if err == nil && len(out)+5 > len(b.literals) {
+		// If we are close, we may still be worse or equal to raw.
+		var lh literalsHeader
+		lh.setSize(len(b.literals))
+		szRaw := lh.size()
+		lh.setSizes(len(out), len(b.literals), single)
+		szComp := lh.size()
+		if len(out)+szComp >= len(b.literals)+szRaw {
+			err = huff0.ErrIncompressible
+		}
+	}
 	switch err {
 	case huff0.ErrIncompressible:
 		lh.setType(literalsBlockRaw)
@@ -773,16 +791,16 @@ func (b *blockEnc) encode(org []byte, raw, rawAllLits bool) error {
 	ml.flush(mlEnc.actualTableLog)
 	of.flush(ofEnc.actualTableLog)
 	ll.flush(llEnc.actualTableLog)
-	err = wr.close()
+	wr.close()
-	if err != nil {
-		return err
-	}
 	b.output = wr.out
 
+	// Maybe even add a bigger margin.
 	if len(b.output)-3-bhOffset >= b.size {
-		// Maybe even add a bigger margin.
+		// Discard and encode as raw block.
+		b.output = b.encodeRawTo(b.output[:bhOffset], org)
+		b.popOffsets()
 		b.litEnc.Reuse = huff0.ReusePolicyNone
-		return errIncompressible
+		return nil
 	}
 
 	// Size is output minus block header.

vendor/github.com/klauspost/compress/zstd/bytebuf.go

@@ -109,7 +109,7 @@ func (r *readerWrapper) readBig(n int, dst []byte) ([]byte, error) {
 }
 
 func (r *readerWrapper) readByte() (byte, error) {
-	n2, err := r.r.Read(r.tmp[:1])
+	n2, err := io.ReadFull(r.r, r.tmp[:1])
 	if err != nil {
 		if err == io.EOF {
 			err = io.ErrUnexpectedEOF

vendor/github.com/klauspost/compress/zstd/decoder.go

@@ -455,12 +455,7 @@ func (d *Decoder) nextBlock(blocking bool) (ok bool) {
 	}
 
 	if len(next.b) > 0 {
-		n, err := d.current.crc.Write(next.b)
+		d.current.crc.Write(next.b)
-		if err == nil {
-			if n != len(next.b) {
-				d.current.err = io.ErrShortWrite
-			}
-		}
 	}
 	if next.err == nil && next.d != nil && next.d.hasCRC {
 		got := uint32(d.current.crc.Sum64())

vendor/github.com/klauspost/compress/zstd/decoder_options.go

@@ -107,7 +107,7 @@ func WithDecoderDicts(dicts ...[]byte) DOption {
 	}
 }
 
-// WithEncoderDictRaw registers a dictionary that may be used by the decoder.
+// WithDecoderDictRaw registers a dictionary that may be used by the decoder.
 // The slice content can be arbitrary data.
 func WithDecoderDictRaw(id uint32, content []byte) DOption {
 	return func(o *decoderOptions) error {

vendor/github.com/klauspost/compress/zstd/dict.go

@@ -1,10 +1,13 @@
 package zstd
 
 import (
+	"bytes"
 	"encoding/binary"
 	"errors"
 	"fmt"
 	"io"
+	"math"
+	"sort"
 
 	"github.com/klauspost/compress/huff0"
 )
@@ -14,9 +17,8 @@ type dict struct {
 
 	litEnc              *huff0.Scratch
 	llDec, ofDec, mlDec sequenceDec
-	//llEnc, ofEnc, mlEnc []*fseEncoder
+	offsets             [3]int
-	offsets [3]int
+	content             []byte
-	content []byte
 }
 
 const dictMagic = "\x37\xa4\x30\xec"
@@ -159,3 +161,374 @@ func InspectDictionary(b []byte) (interface {
 	d, err := loadDict(b)
 	return d, err
 }
+
+type BuildDictOptions struct {
+	// Dictionary ID.
+	ID uint32
+
+	// Content to use to create dictionary tables.
+	Contents [][]byte
+
+	// History to use for all blocks.
+	History []byte
+
+	// Offsets to use.
+	Offsets [3]int
+
+	// CompatV155 will make the dictionary compatible with Zstd v1.5.5 and earlier.
+	// See https://github.com/facebook/zstd/issues/3724
+	CompatV155 bool
+
+	// Use the specified encoder level.
+	// The dictionary will be built using the specified encoder level,
+	// which will reflect speed and make the dictionary tailored for that level.
+	// If not set SpeedBestCompression will be used.
+	Level EncoderLevel
+
+	// DebugOut will write stats and other details here if set.
+	DebugOut io.Writer
+}
+
+func BuildDict(o BuildDictOptions) ([]byte, error) {
+	initPredefined()
+	hist := o.History
+	contents := o.Contents
+	debug := o.DebugOut != nil
+	println := func(args ...interface{}) {
+		if o.DebugOut != nil {
+			fmt.Fprintln(o.DebugOut, args...)
+		}
+	}
+	printf := func(s string, args ...interface{}) {
+		if o.DebugOut != nil {
+			fmt.Fprintf(o.DebugOut, s, args...)
+		}
+	}
+	print := func(args ...interface{}) {
+		if o.DebugOut != nil {
+			fmt.Fprint(o.DebugOut, args...)
+		}
+	}
+
+	if int64(len(hist)) > dictMaxLength {
+		return nil, fmt.Errorf("dictionary of size %d > %d", len(hist), int64(dictMaxLength))
+	}
+	if len(hist) < 8 {
+		return nil, fmt.Errorf("dictionary of size %d < %d", len(hist), 8)
+	}
+	if len(contents) == 0 {
+		return nil, errors.New("no content provided")
+	}
+	d := dict{
+		id:      o.ID,
+		litEnc:  nil,
+		llDec:   sequenceDec{},
+		ofDec:   sequenceDec{},
+		mlDec:   sequenceDec{},
+		offsets: o.Offsets,
+		content: hist,
+	}
+	block := blockEnc{lowMem: false}
+	block.init()
+	enc := encoder(&bestFastEncoder{fastBase: fastBase{maxMatchOff: int32(maxMatchLen), bufferReset: math.MaxInt32 - int32(maxMatchLen*2), lowMem: false}})
+	if o.Level != 0 {
+		eOpts := encoderOptions{
+			level:      o.Level,
+			blockSize:  maxMatchLen,
+			windowSize: maxMatchLen,
+			dict:       &d,
+			lowMem:     false,
+		}
+		enc = eOpts.encoder()
+	} else {
+		o.Level = SpeedBestCompression
+	}
+	var (
+		remain [256]int
+		ll     [256]int
+		ml     [256]int
+		of     [256]int
+	)
+	addValues := func(dst *[256]int, src []byte) {
+		for _, v := range src {
+			dst[v]++
+		}
+	}
+	addHist := func(dst *[256]int, src *[256]uint32) {
+		for i, v := range src {
+			dst[i] += int(v)
+		}
+	}
+	seqs := 0
+	nUsed := 0
+	litTotal := 0
+	newOffsets := make(map[uint32]int, 1000)
+	for _, b := range contents {
+		block.reset(nil)
+		if len(b) < 8 {
+			continue
+		}
+		nUsed++
+		enc.Reset(&d, true)
+		enc.Encode(&block, b)
+		addValues(&remain, block.literals)
+		litTotal += len(block.literals)
+		seqs += len(block.sequences)
+		block.genCodes()
+		addHist(&ll, block.coders.llEnc.Histogram())
+		addHist(&ml, block.coders.mlEnc.Histogram())
+		addHist(&of, block.coders.ofEnc.Histogram())
+		for i, seq := range block.sequences {
+			if i > 3 {
+				break
+			}
+			offset := seq.offset
+			if offset == 0 {
+				continue
+			}
+			if offset > 3 {
+				newOffsets[offset-3]++
+			} else {
+				newOffsets[uint32(o.Offsets[offset-1])]++
+			}
+		}
+	}
+	// Find most used offsets.
+	var sortedOffsets []uint32
+	for k := range newOffsets {
+		sortedOffsets = append(sortedOffsets, k)
+	}
+	sort.Slice(sortedOffsets, func(i, j int) bool {
+		a, b := sortedOffsets[i], sortedOffsets[j]
+		if a == b {
+			// Prefer the longer offset
+			return sortedOffsets[i] > sortedOffsets[j]
+		}
+		return newOffsets[sortedOffsets[i]] > newOffsets[sortedOffsets[j]]
+	})
+	if len(sortedOffsets) > 3 {
+		if debug {
+			print("Offsets:")
+			for i, v := range sortedOffsets {
+				if i > 20 {
+					break
+				}
+				printf("[%d: %d],", v, newOffsets[v])
+			}
+			println("")
+		}
+
+		sortedOffsets = sortedOffsets[:3]
+	}
+	for i, v := range sortedOffsets {
+		o.Offsets[i] = int(v)
+	}
+	if debug {
+		println("New repeat offsets", o.Offsets)
+	}
+
+	if nUsed == 0 || seqs == 0 {
+		return nil, fmt.Errorf("%d blocks, %d sequences found", nUsed, seqs)
+	}
+	if debug {
+		println("Sequences:", seqs, "Blocks:", nUsed, "Literals:", litTotal)
+	}
+	if seqs/nUsed < 512 {
+		// Use 512 as minimum.
+		nUsed = seqs / 512
+	}
+	copyHist := func(dst *fseEncoder, src *[256]int) ([]byte, error) {
+		hist := dst.Histogram()
+		var maxSym uint8
+		var maxCount int
+		var fakeLength int
+		for i, v := range src {
+			if v > 0 {
+				v = v / nUsed
+				if v == 0 {
+					v = 1
+				}
+			}
+			if v > maxCount {
+				maxCount = v
+			}
+			if v != 0 {
+				maxSym = uint8(i)
+			}
+			fakeLength += v
+			hist[i] = uint32(v)
+		}
+		dst.HistogramFinished(maxSym, maxCount)
+		dst.reUsed = false
+		dst.useRLE = false
+		err := dst.normalizeCount(fakeLength)
+		if err != nil {
+			return nil, err
+		}
+		if debug {
+			println("RAW:", dst.count[:maxSym+1], "NORM:", dst.norm[:maxSym+1], "LEN:", fakeLength)
+		}
+		return dst.writeCount(nil)
+	}
+	if debug {
+		print("Literal lengths: ")
+	}
+	llTable, err := copyHist(block.coders.llEnc, &ll)
+	if err != nil {
+		return nil, err
+	}
+	if debug {
+		print("Match lengths: ")
+	}
+	mlTable, err := copyHist(block.coders.mlEnc, &ml)
+	if err != nil {
+		return nil, err
+	}
+	if debug {
+		print("Offsets: ")
+	}
+	ofTable, err := copyHist(block.coders.ofEnc, &of)
+	if err != nil {
+		return nil, err
+	}
+
+	// Literal table
+	avgSize := litTotal
+	if avgSize > huff0.BlockSizeMax/2 {
+		avgSize = huff0.BlockSizeMax / 2
+	}
+	huffBuff := make([]byte, 0, avgSize)
+	// Target size
+	div := litTotal / avgSize
+	if div < 1 {
+		div = 1
+	}
+	if debug {
+		println("Huffman weights:")
+	}
+	for i, n := range remain[:] {
+		if n > 0 {
+			n = n / div
+			// Allow all entries to be represented.
+			if n == 0 {
+				n = 1
+			}
+			huffBuff = append(huffBuff, bytes.Repeat([]byte{byte(i)}, n)...)
+			if debug {
+				printf("[%d: %d], ", i, n)
+			}
+		}
+	}
+	if o.CompatV155 && remain[255]/div == 0 {
+		huffBuff = append(huffBuff, 255)
+	}
+	scratch := &huff0.Scratch{TableLog: 11}
+	for tries := 0; tries < 255; tries++ {
+		scratch = &huff0.Scratch{TableLog: 11}
+		_, _, err = huff0.Compress1X(huffBuff, scratch)
+		if err == nil {
+			break
+		}
+		if debug {
+			printf("Try %d: Huffman error: %v\n", tries+1, err)
+		}
+		huffBuff = huffBuff[:0]
+		if tries == 250 {
+			if debug {
+				println("Huffman: Bailing out with predefined table")
+			}
+
+			// Bail out.... Just generate something
+			huffBuff = append(huffBuff, bytes.Repeat([]byte{255}, 10000)...)
+			for i := 0; i < 128; i++ {
+				huffBuff = append(huffBuff, byte(i))
+			}
+			continue
+		}
+		if errors.Is(err, huff0.ErrIncompressible) {
+			// Try truncating least common.
+			for i, n := range remain[:] {
+				if n > 0 {
+					n = n / (div * (i + 1))
+					if n > 0 {
+						huffBuff = append(huffBuff, bytes.Repeat([]byte{byte(i)}, n)...)
+					}
+				}
+			}
+			if o.CompatV155 && len(huffBuff) > 0 && huffBuff[len(huffBuff)-1] != 255 {
+				huffBuff = append(huffBuff, 255)
+			}
+			if len(huffBuff) == 0 {
+				huffBuff = append(huffBuff, 0, 255)
+			}
+		}
+		if errors.Is(err, huff0.ErrUseRLE) {
+			for i, n := range remain[:] {
+				n = n / (div * (i + 1))
+				// Allow all entries to be represented.
+				if n == 0 {
+					n = 1
+				}
+				huffBuff = append(huffBuff, bytes.Repeat([]byte{byte(i)}, n)...)
+			}
+		}
+	}
+
+	var out bytes.Buffer
+	out.Write([]byte(dictMagic))
+	out.Write(binary.LittleEndian.AppendUint32(nil, o.ID))
+	out.Write(scratch.OutTable)
+	if debug {
+		println("huff table:", len(scratch.OutTable), "bytes")
+		println("of table:", len(ofTable), "bytes")
+		println("ml table:", len(mlTable), "bytes")
+		println("ll table:", len(llTable), "bytes")
+	}
+	out.Write(ofTable)
+	out.Write(mlTable)
+	out.Write(llTable)
+	out.Write(binary.LittleEndian.AppendUint32(nil, uint32(o.Offsets[0])))
+	out.Write(binary.LittleEndian.AppendUint32(nil, uint32(o.Offsets[1])))
+	out.Write(binary.LittleEndian.AppendUint32(nil, uint32(o.Offsets[2])))
+	out.Write(hist)
+	if debug {
+		_, err := loadDict(out.Bytes())
+		if err != nil {
+			panic(err)
+		}
+		i, err := InspectDictionary(out.Bytes())
+		if err != nil {
+			panic(err)
+		}
+		println("ID:", i.ID())
+		println("Content size:", i.ContentSize())
+		println("Encoder:", i.LitEncoder() != nil)
+		println("Offsets:", i.Offsets())
+		var totalSize int
+		for _, b := range contents {
+			totalSize += len(b)
+		}
+
+		encWith := func(opts ...EOption) int {
+			enc, err := NewWriter(nil, opts...)
+			if err != nil {
+				panic(err)
+			}
+			defer enc.Close()
+			var dst []byte
+			var totalSize int
+			for _, b := range contents {
+				dst = enc.EncodeAll(b, dst[:0])
+				totalSize += len(dst)
+			}
+			return totalSize
+		}
+		plain := encWith(WithEncoderLevel(o.Level))
+		withDict := encWith(WithEncoderLevel(o.Level), WithEncoderDict(out.Bytes()))
+		println("Input size:", totalSize)
+		println("Plain Compressed:", plain)
+		println("Dict Compressed:", withDict)
+		println("Saved:", plain-withDict, (plain-withDict)/len(contents), "bytes per input (rounded down)")
+	}
+	return out.Bytes(), nil
+}

vendor/github.com/klauspost/compress/zstd/enc_base.go

@@ -144,6 +144,7 @@ func (e *fastBase) resetBase(d *dict, singleBlock bool) {
 	} else {
 		e.crc.Reset()
 	}
+	e.blk.dictLitEnc = nil
 	if d != nil {
 		low := e.lowMem
 		if singleBlock {

vendor/github.com/klauspost/compress/zstd/enc_best.go

@@ -34,7 +34,7 @@ type match struct {
 	est    int32
 }
 
-const highScore = 25000
+const highScore = maxMatchLen * 8
 
 // estBits will estimate output bits from predefined tables.
 func (m *match) estBits(bitsPerByte int32) {
@@ -159,7 +159,6 @@ func (e *bestFastEncoder) Encode(blk *blockEnc, src []byte) {
 
 	// nextEmit is where in src the next emitLiteral should start from.
 	nextEmit := s
-	cv := load6432(src, s)
 
 	// Relative offsets
 	offset1 := int32(blk.recentOffsets[0])
@@ -173,7 +172,6 @@ func (e *bestFastEncoder) Encode(blk *blockEnc, src []byte) {
 		blk.literals = append(blk.literals, src[nextEmit:until]...)
 		s.litLen = uint32(until - nextEmit)
 	}
-	_ = addLiterals
 
 	if debugEncoder {
 		println("recent offsets:", blk.recentOffsets)
@@ -188,7 +186,9 @@ encodeLoop:
 			panic("offset0 was 0")
 		}
 
-		const goodEnough = 100
+		const goodEnough = 250
+
+		cv := load6432(src, s)
 
 		nextHashL := hashLen(cv, bestLongTableBits, bestLongLen)
 		nextHashS := hashLen(cv, bestShortTableBits, bestShortLen)
@@ -197,15 +197,50 @@ encodeLoop:
 
 		// Set m to a match at offset if it looks like that will improve compression.
 		improve := func(m *match, offset int32, s int32, first uint32, rep int32) {
-			if s-offset >= e.maxMatchOff || load3232(src, offset) != first {
+			delta := s - offset
+			if delta >= e.maxMatchOff || delta <= 0 || load3232(src, offset) != first {
 				return
 			}
 			if debugAsserts {
+				if offset >= s {
+					panic(fmt.Sprintf("offset: %d - s:%d - rep: %d - cur :%d - max: %d", offset, s, rep, e.cur, e.maxMatchOff))
+				}
 				if !bytes.Equal(src[s:s+4], src[offset:offset+4]) {
 					panic(fmt.Sprintf("first match mismatch: %v != %v, first: %08x", src[s:s+4], src[offset:offset+4], first))
 				}
 			}
-			cand := match{offset: offset, s: s, length: 4 + e.matchlen(s+4, offset+4, src), rep: rep}
+			// Try to quick reject if we already have a long match.
+			if m.length > 16 {
+				left := len(src) - int(m.s+m.length)
+				// If we are too close to the end, keep as is.
+				if left <= 0 {
+					return
+				}
+				checkLen := m.length - (s - m.s) - 8
+				if left > 2 && checkLen > 4 {
+					// Check 4 bytes, 4 bytes from the end of the current match.
+					a := load3232(src, offset+checkLen)
+					b := load3232(src, s+checkLen)
+					if a != b {
+						return
+					}
+				}
+			}
+			l := 4 + e.matchlen(s+4, offset+4, src)
+			if rep < 0 {
+				// Extend candidate match backwards as far as possible.
+				tMin := s - e.maxMatchOff
+				if tMin < 0 {
+					tMin = 0
+				}
+				for offset > tMin && s > nextEmit && src[offset-1] == src[s-1] && l < maxMatchLength {
+					s--
+					offset--
+					l++
+				}
+			}
+
+			cand := match{offset: offset, s: s, length: l, rep: rep}
 			cand.estBits(bitsPerByte)
 			if m.est >= highScore || cand.est-m.est+(cand.s-m.s)*bitsPerByte>>10 < 0 {
 				*m = cand
@@ -219,17 +254,29 @@ encodeLoop:
 		improve(&best, candidateS.prev-e.cur, s, uint32(cv), -1)
 
 		if canRepeat && best.length < goodEnough {
-			cv32 := uint32(cv >> 8)
+			if s == nextEmit {
-			spp := s + 1
+				// Check repeats straight after a match.
-			improve(&best, spp-offset1, spp, cv32, 1)
+				improve(&best, s-offset2, s, uint32(cv), 1|4)
-			improve(&best, spp-offset2, spp, cv32, 2)
+				improve(&best, s-offset3, s, uint32(cv), 2|4)
-			improve(&best, spp-offset3, spp, cv32, 3)
+				if offset1 > 1 {
-			if best.length > 0 {
+					improve(&best, s-(offset1-1), s, uint32(cv), 3|4)
-				cv32 = uint32(cv >> 24)
+				}
-				spp += 2
+			}
+
+			// If either no match or a non-repeat match, check at + 1
+			if best.rep <= 0 {
+				cv32 := uint32(cv >> 8)
+				spp := s + 1
 				improve(&best, spp-offset1, spp, cv32, 1)
 				improve(&best, spp-offset2, spp, cv32, 2)
 				improve(&best, spp-offset3, spp, cv32, 3)
+				if best.rep < 0 {
+					cv32 = uint32(cv >> 24)
+					spp += 2
+					improve(&best, spp-offset1, spp, cv32, 1)
+					improve(&best, spp-offset2, spp, cv32, 2)
+					improve(&best, spp-offset3, spp, cv32, 3)
+				}
 			}
 		}
 		// Load next and check...
@@ -244,41 +291,44 @@ encodeLoop:
 				if s >= sLimit {
 					break encodeLoop
 				}
-				cv = load6432(src, s)
 				continue
 			}
 
-			s++
 			candidateS = e.table[hashLen(cv>>8, bestShortTableBits, bestShortLen)]
-			cv = load6432(src, s)
+			cv = load6432(src, s+1)
-			cv2 := load6432(src, s+1)
+			cv2 := load6432(src, s+2)
 			candidateL = e.longTable[hashLen(cv, bestLongTableBits, bestLongLen)]
 			candidateL2 := e.longTable[hashLen(cv2, bestLongTableBits, bestLongLen)]
 
 			// Short at s+1
-			improve(&best, candidateS.offset-e.cur, s, uint32(cv), -1)
+			improve(&best, candidateS.offset-e.cur, s+1, uint32(cv), -1)
 			// Long at s+1, s+2
-			improve(&best, candidateL.offset-e.cur, s, uint32(cv), -1)
+			improve(&best, candidateL.offset-e.cur, s+1, uint32(cv), -1)
-			improve(&best, candidateL.prev-e.cur, s, uint32(cv), -1)
+			improve(&best, candidateL.prev-e.cur, s+1, uint32(cv), -1)
-			improve(&best, candidateL2.offset-e.cur, s+1, uint32(cv2), -1)
+			improve(&best, candidateL2.offset-e.cur, s+2, uint32(cv2), -1)
-			improve(&best, candidateL2.prev-e.cur, s+1, uint32(cv2), -1)
+			improve(&best, candidateL2.prev-e.cur, s+2, uint32(cv2), -1)
 			if false {
 				// Short at s+3.
 				// Too often worse...
-				improve(&best, e.table[hashLen(cv2>>8, bestShortTableBits, bestShortLen)].offset-e.cur, s+2, uint32(cv2>>8), -1)
+				improve(&best, e.table[hashLen(cv2>>8, bestShortTableBits, bestShortLen)].offset-e.cur, s+3, uint32(cv2>>8), -1)
 			}
-			// See if we can find a better match by checking where the current best ends.
+
-			// Use that offset to see if we can find a better full match.
+			// Start check at a fixed offset to allow for a few mismatches.
-			if sAt := best.s + best.length; sAt < sLimit {
+			// For this compression level 2 yields the best results.
-				nextHashL := hashLen(load6432(src, sAt), bestLongTableBits, bestLongLen)
+			// We cannot do this if we have already indexed this position.
-				candidateEnd := e.longTable[nextHashL]
+			const skipBeginning = 2
-				// Start check at a fixed offset to allow for a few mismatches.
+			if best.s > s-skipBeginning {
-				// For this compression level 2 yields the best results.
+				// See if we can find a better match by checking where the current best ends.
-				const skipBeginning = 2
+				// Use that offset to see if we can find a better full match.
-				if pos := candidateEnd.offset - e.cur - best.length + skipBeginning; pos >= 0 {
+				if sAt := best.s + best.length; sAt < sLimit {
-					improve(&best, pos, best.s+skipBeginning, load3232(src, best.s+skipBeginning), -1)
+					nextHashL := hashLen(load6432(src, sAt), bestLongTableBits, bestLongLen)
-					if pos := candidateEnd.prev - e.cur - best.length + skipBeginning; pos >= 0 {
+					candidateEnd := e.longTable[nextHashL]
-						improve(&best, pos, best.s+skipBeginning, load3232(src, best.s+skipBeginning), -1)
+
+					if off := candidateEnd.offset - e.cur - best.length + skipBeginning; off >= 0 {
+						improve(&best, off, best.s+skipBeginning, load3232(src, best.s+skipBeginning), -1)
+						if off := candidateEnd.prev - e.cur - best.length + skipBeginning; off >= 0 {
+							improve(&best, off, best.s+skipBeginning, load3232(src, best.s+skipBeginning), -1)
+						}
 					}
 				}
 			}
@@ -292,51 +342,34 @@ encodeLoop:
 
 		// We have a match, we can store the forward value
 		if best.rep > 0 {
-			s = best.s
 			var seq seq
 			seq.matchLen = uint32(best.length - zstdMinMatch)
-
+			if debugAsserts && s < nextEmit {
-			// We might be able to match backwards.
+				panic("s < nextEmit")
-			// Extend as long as we can.
-			start := best.s
-			// We end the search early, so we don't risk 0 literals
-			// and have to do special offset treatment.
-			startLimit := nextEmit + 1
-
-			tMin := s - e.maxMatchOff
-			if tMin < 0 {
-				tMin = 0
 			}
-			repIndex := best.offset
+			addLiterals(&seq, best.s)
-			for repIndex > tMin && start > startLimit && src[repIndex-1] == src[start-1] && seq.matchLen < maxMatchLength-zstdMinMatch-1 {
-				repIndex--
-				start--
-				seq.matchLen++
-			}
-			addLiterals(&seq, start)
 
-			// rep 0
+			// Repeat. If bit 4 is set, this is a non-lit repeat.
-			seq.offset = uint32(best.rep)
+			seq.offset = uint32(best.rep & 3)
 			if debugSequences {
 				println("repeat sequence", seq, "next s:", s)
 			}
 			blk.sequences = append(blk.sequences, seq)
 
-			// Index match start+1 (long) -> s - 1
+			// Index old s + 1 -> s - 1
-			index0 := s
+			index0 := s + 1
 			s = best.s + best.length
 
 			nextEmit = s
 			if s >= sLimit {
 				if debugEncoder {
 					println("repeat ended", s, best.length)
-
 				}
 				break encodeLoop
 			}
 			// Index skipped...
 			off := index0 + e.cur
-			for index0 < s-1 {
+			for index0 < s {
 				cv0 := load6432(src, index0)
 				h0 := hashLen(cv0, bestLongTableBits, bestLongLen)
 				h1 := hashLen(cv0, bestShortTableBits, bestShortLen)
@@ -346,17 +379,19 @@ encodeLoop:
 				index0++
 			}
 			switch best.rep {
-			case 2:
+			case 2, 4 | 1:
 				offset1, offset2 = offset2, offset1
-			case 3:
+			case 3, 4 | 2:
 				offset1, offset2, offset3 = offset3, offset1, offset2
+			case 4 | 3:
+				offset1, offset2, offset3 = offset1-1, offset1, offset2
 			}
-			cv = load6432(src, s)
 			continue
 		}
 
 		// A 4-byte match has been found. Update recent offsets.
 		// We'll later see if more than 4 bytes.
+		index0 := s + 1
 		s = best.s
 		t := best.offset
 		offset1, offset2, offset3 = s-t, offset1, offset2
@@ -369,22 +404,9 @@ encodeLoop:
 			panic("invalid offset")
 		}
 
-		// Extend the n-byte match as long as possible.
-		l := best.length
-
-		// Extend backwards
-		tMin := s - e.maxMatchOff
-		if tMin < 0 {
-			tMin = 0
-		}
-		for t > tMin && s > nextEmit && src[t-1] == src[s-1] && l < maxMatchLength {
-			s--
-			t--
-			l++
-		}
-
 		// Write our sequence
 		var seq seq
+		l := best.length
 		seq.litLen = uint32(s - nextEmit)
 		seq.matchLen = uint32(l - zstdMinMatch)
 		if seq.litLen > 0 {
@@ -401,10 +423,8 @@ encodeLoop:
 			break encodeLoop
 		}
 
-		// Index match start+1 (long) -> s - 1
+		// Index old s + 1 -> s - 1
-		index0 := s - l + 1
+		for index0 < s {
-		// every entry
-		for index0 < s-1 {
 			cv0 := load6432(src, index0)
 			h0 := hashLen(cv0, bestLongTableBits, bestLongLen)
 			h1 := hashLen(cv0, bestShortTableBits, bestShortLen)
@@ -413,50 +433,6 @@ encodeLoop:
 			e.table[h1] = prevEntry{offset: off, prev: e.table[h1].offset}
 			index0++
 		}
-
-		cv = load6432(src, s)
-		if !canRepeat {
-			continue
-		}
-
-		// Check offset 2
-		for {
-			o2 := s - offset2
-			if load3232(src, o2) != uint32(cv) {
-				// Do regular search
-				break
-			}
-
-			// Store this, since we have it.
-			nextHashS := hashLen(cv, bestShortTableBits, bestShortLen)
-			nextHashL := hashLen(cv, bestLongTableBits, bestLongLen)
-
-			// We have at least 4 byte match.
-			// No need to check backwards. We come straight from a match
-			l := 4 + e.matchlen(s+4, o2+4, src)
-
-			e.longTable[nextHashL] = prevEntry{offset: s + e.cur, prev: e.longTable[nextHashL].offset}
-			e.table[nextHashS] = prevEntry{offset: s + e.cur, prev: e.table[nextHashS].offset}
-			seq.matchLen = uint32(l) - zstdMinMatch
-			seq.litLen = 0
-
-			// Since litlen is always 0, this is offset 1.
-			seq.offset = 1
-			s += l
-			nextEmit = s
-			if debugSequences {
-				println("sequence", seq, "next s:", s)
-			}
-			blk.sequences = append(blk.sequences, seq)
-
-			// Swap offset 1 and 2.
-			offset1, offset2 = offset2, offset1
-			if s >= sLimit {
-				// Finished
-				break encodeLoop
-			}
-			cv = load6432(src, s)
-		}
 	}
 
 	if int(nextEmit) < len(src) {

vendor/github.com/klauspost/compress/zstd/enc_dfast.go

@@ -1084,7 +1084,7 @@ func (e *doubleFastEncoderDict) Reset(d *dict, singleBlock bool) {
 			}
 		}
 		e.lastDictID = d.id
-		e.allDirty = true
+		allDirty = true
 	}
 	// Reset table to initial state
 	e.cur = e.maxMatchOff

vendor/github.com/klauspost/compress/zstd/enc_fast.go

@@ -133,8 +133,7 @@ encodeLoop:
 			if canRepeat && repIndex >= 0 && load3232(src, repIndex) == uint32(cv>>16) {
 				// Consider history as well.
 				var seq seq
-				var length int32
+				length := 4 + e.matchlen(s+6, repIndex+4, src)
-				length = 4 + e.matchlen(s+6, repIndex+4, src)
 				seq.matchLen = uint32(length - zstdMinMatch)
 
 				// We might be able to match backwards.
@@ -645,8 +644,7 @@ encodeLoop:
 			if canRepeat && repIndex >= 0 && load3232(src, repIndex) == uint32(cv>>16) {
 				// Consider history as well.
 				var seq seq
-				var length int32
+				length := 4 + e.matchlen(s+6, repIndex+4, src)
-				length = 4 + e.matchlen(s+6, repIndex+4, src)
 
 				seq.matchLen = uint32(length - zstdMinMatch)
 
@@ -831,13 +829,12 @@ func (e *fastEncoderDict) Reset(d *dict, singleBlock bool) {
 		}
 		if true {
 			end := e.maxMatchOff + int32(len(d.content)) - 8
-			for i := e.maxMatchOff; i < end; i += 3 {
+			for i := e.maxMatchOff; i < end; i += 2 {
 				const hashLog = tableBits
 
 				cv := load6432(d.content, i-e.maxMatchOff)
-				nextHash := hashLen(cv, hashLog, tableFastHashLen)      // 0 -> 5
+				nextHash := hashLen(cv, hashLog, tableFastHashLen)     // 0 -> 6
-				nextHash1 := hashLen(cv>>8, hashLog, tableFastHashLen)  // 1 -> 6
+				nextHash1 := hashLen(cv>>8, hashLog, tableFastHashLen) // 1 -> 7
-				nextHash2 := hashLen(cv>>16, hashLog, tableFastHashLen) // 2 -> 7
 				e.dictTable[nextHash] = tableEntry{
 					val:    uint32(cv),
 					offset: i,
@@ -846,10 +843,6 @@ func (e *fastEncoderDict) Reset(d *dict, singleBlock bool) {
 					val:    uint32(cv >> 8),
 					offset: i + 1,
 				}
-				e.dictTable[nextHash2] = tableEntry{
-					val:    uint32(cv >> 16),
-					offset: i + 2,
-				}
 			}
 		}
 		e.lastDictID = d.id

vendor/github.com/klauspost/compress/zstd/encoder.go

@@ -227,10 +227,7 @@ func (e *Encoder) nextBlock(final bool) error {
 			DictID:        e.o.dict.ID(),
 		}
 
-		dst, err := fh.appendTo(tmp[:0])
+		dst := fh.appendTo(tmp[:0])
-		if err != nil {
-			return err
-		}
 		s.headerWritten = true
 		s.wWg.Wait()
 		var n2 int
@@ -277,23 +274,9 @@ func (e *Encoder) nextBlock(final bool) error {
 			s.eofWritten = true
 		}
 
-		err := errIncompressible
+		s.err = blk.encode(src, e.o.noEntropy, !e.o.allLitEntropy)
-		// If we got the exact same number of literals as input,
+		if s.err != nil {
-		// assume the literals cannot be compressed.
+			return s.err
-		if len(src) != len(blk.literals) || len(src) != e.o.blockSize {
-			err = blk.encode(src, e.o.noEntropy, !e.o.allLitEntropy)
-		}
-		switch err {
-		case errIncompressible:
-			if debugEncoder {
-				println("Storing incompressible block as raw")
-			}
-			blk.encodeRaw(src)
-			// In fast mode, we do not transfer offsets, so we don't have to deal with changing the.
-		case nil:
-		default:
-			s.err = err
-			return err
 		}
 		_, s.err = s.w.Write(blk.output)
 		s.nWritten += int64(len(blk.output))
@@ -343,22 +326,8 @@ func (e *Encoder) nextBlock(final bool) error {
 				}
 				s.wWg.Done()
 			}()
-			err := errIncompressible
+			s.writeErr = blk.encode(src, e.o.noEntropy, !e.o.allLitEntropy)
-			// If we got the exact same number of literals as input,
+			if s.writeErr != nil {
-			// assume the literals cannot be compressed.
-			if len(src) != len(blk.literals) || len(src) != e.o.blockSize {
-				err = blk.encode(src, e.o.noEntropy, !e.o.allLitEntropy)
-			}
-			switch err {
-			case errIncompressible:
-				if debugEncoder {
-					println("Storing incompressible block as raw")
-				}
-				blk.encodeRaw(src)
-				// In fast mode, we do not transfer offsets, so we don't have to deal with changing the.
-			case nil:
-			default:
-				s.writeErr = err
 				return
 			}
 			_, s.writeErr = s.w.Write(blk.output)
@@ -511,7 +480,7 @@ func (e *Encoder) EncodeAll(src, dst []byte) []byte {
 				Checksum: false,
 				DictID:   0,
 			}
-			dst, _ = fh.appendTo(dst)
+			dst = fh.appendTo(dst)
 
 			// Write raw block as last one only.
 			var blk blockHeader
@@ -546,10 +515,7 @@ func (e *Encoder) EncodeAll(src, dst []byte) []byte {
 	if len(dst) == 0 && cap(dst) == 0 && len(src) < 1<<20 && !e.o.lowMem {
 		dst = make([]byte, 0, len(src))
 	}
-	dst, err := fh.appendTo(dst)
+	dst = fh.appendTo(dst)
-	if err != nil {
-		panic(err)
-	}
 
 	// If we can do everything in one block, prefer that.
 	if len(src) <= e.o.blockSize {
@@ -568,25 +534,15 @@ func (e *Encoder) EncodeAll(src, dst []byte) []byte {
 
 		// If we got the exact same number of literals as input,
 		// assume the literals cannot be compressed.
-		err := errIncompressible
 		oldout := blk.output
-		if len(blk.literals) != len(src) || len(src) != e.o.blockSize {
+		// Output directly to dst
-			// Output directly to dst
+		blk.output = dst
-			blk.output = dst
-			err = blk.encode(src, e.o.noEntropy, !e.o.allLitEntropy)
-		}
 
-		switch err {
+		err := blk.encode(src, e.o.noEntropy, !e.o.allLitEntropy)
-		case errIncompressible:
+		if err != nil {
-			if debugEncoder {
-				println("Storing incompressible block as raw")
-			}
-			dst = blk.encodeRawTo(dst, src)
-		case nil:
-			dst = blk.output
-		default:
 			panic(err)
 		}
+		dst = blk.output
 		blk.output = oldout
 	} else {
 		enc.Reset(e.o.dict, false)
@@ -605,25 +561,11 @@ func (e *Encoder) EncodeAll(src, dst []byte) []byte {
 			if len(src) == 0 {
 				blk.last = true
 			}
-			err := errIncompressible
+			err := blk.encode(todo, e.o.noEntropy, !e.o.allLitEntropy)
-			// If we got the exact same number of literals as input,
+			if err != nil {
-			// assume the literals cannot be compressed.
-			if len(blk.literals) != len(todo) || len(todo) != e.o.blockSize {
-				err = blk.encode(todo, e.o.noEntropy, !e.o.allLitEntropy)
-			}
-
-			switch err {
-			case errIncompressible:
-				if debugEncoder {
-					println("Storing incompressible block as raw")
-				}
-				dst = blk.encodeRawTo(dst, todo)
-				blk.popOffsets()
-			case nil:
-				dst = append(dst, blk.output...)
-			default:
 				panic(err)
 			}
+			dst = append(dst, blk.output...)
 			blk.reset(nil)
 		}
 	}
@@ -633,6 +575,7 @@ func (e *Encoder) EncodeAll(src, dst []byte) []byte {
 	// Add padding with content from crypto/rand.Reader
 	if e.o.pad > 0 {
 		add := calcSkippableFrame(int64(len(dst)), int64(e.o.pad))
+		var err error
 		dst, err = skippableFrame(dst, add, rand.Reader)
 		if err != nil {
 			panic(err)

vendor/github.com/klauspost/compress/zstd/encoder_options.go

@@ -39,7 +39,7 @@ func (o *encoderOptions) setDefault() {
 		blockSize:     maxCompressedBlockSize,
 		windowSize:    8 << 20,
 		level:         SpeedDefault,
-		allLitEntropy: true,
+		allLitEntropy: false,
 		lowMem:        false,
 	}
 }
@@ -129,7 +129,7 @@ func WithEncoderPadding(n int) EOption {
 		}
 		// No need to waste our time.
 		if n == 1 {
-			o.pad = 0
+			n = 0
 		}
 		if n > 1<<30 {
 			return fmt.Errorf("padding must less than 1GB (1<<30 bytes) ")
@@ -238,7 +238,7 @@ func WithEncoderLevel(l EncoderLevel) EOption {
 			}
 		}
 		if !o.customALEntropy {
-			o.allLitEntropy = l > SpeedFastest
+			o.allLitEntropy = l > SpeedDefault
 		}
 
 		return nil

vendor/github.com/klauspost/compress/zstd/framedec.go

@@ -73,20 +73,20 @@ func (d *frameDec) reset(br byteBuffer) error {
 		switch err {
 		case io.EOF, io.ErrUnexpectedEOF:
 			return io.EOF
-		default:
-			return err
 		case nil:
 			signature[0] = b[0]
+		default:
+			return err
 		}
 		// Read the rest, don't allow io.ErrUnexpectedEOF
 		b, err = br.readSmall(3)
 		switch err {
 		case io.EOF:
 			return io.EOF
-		default:
-			return err
 		case nil:
 			copy(signature[1:], b)
+		default:
+			return err
 		}
 
 		if string(signature[1:4]) != skippableFrameMagic || signature[0]&0xf0 != 0x50 {
@@ -293,13 +293,9 @@ func (d *frameDec) next(block *blockDec) error {
 	return nil
 }
 
-// checkCRC will check the checksum if the frame has one.
+// checkCRC will check the checksum, assuming the frame has one.
 // Will return ErrCRCMismatch if crc check failed, otherwise nil.
 func (d *frameDec) checkCRC() error {
-	if !d.HasCheckSum {
-		return nil
-	}
-
 	// We can overwrite upper tmp now
 	buf, err := d.rawInput.readSmall(4)
 	if err != nil {
@@ -307,10 +303,6 @@ func (d *frameDec) checkCRC() error {
 		return err
 	}
 
-	if d.o.ignoreChecksum {
-		return nil
-	}
-
 	want := binary.LittleEndian.Uint32(buf[:4])
 	got := uint32(d.crc.Sum64())
 
@@ -326,17 +318,13 @@ func (d *frameDec) checkCRC() error {
 	return nil
 }
 
-// consumeCRC reads the checksum data if the frame has one.
+// consumeCRC skips over the checksum, assuming the frame has one.
 func (d *frameDec) consumeCRC() error {
-	if d.HasCheckSum {
+	_, err := d.rawInput.readSmall(4)
-		_, err := d.rawInput.readSmall(4)
+	if err != nil {
-		if err != nil {
+		println("CRC missing?", err)
-			println("CRC missing?", err)
-			return err
-		}
 	}
-
+	return err
-	return nil
 }
 
 // runDecoder will run the decoder for the remainder of the frame.
@@ -415,15 +403,8 @@ func (d *frameDec) runDecoder(dst []byte, dec *blockDec) ([]byte, error) {
 			if d.o.ignoreChecksum {
 				err = d.consumeCRC()
 			} else {
-				var n int
+				d.crc.Write(dst[crcStart:])
-				n, err = d.crc.Write(dst[crcStart:])
+				err = d.checkCRC()
-				if err == nil {
-					if n != len(dst)-crcStart {
-						err = io.ErrShortWrite
-					} else {
-						err = d.checkCRC()
-					}
-				}
 			}
 		}
 	}

vendor/github.com/klauspost/compress/zstd/frameenc.go

@@ -22,7 +22,7 @@ type frameHeader struct {
 
 const maxHeaderSize = 14
 
-func (f frameHeader) appendTo(dst []byte) ([]byte, error) {
+func (f frameHeader) appendTo(dst []byte) []byte {
 	dst = append(dst, frameMagic...)
 	var fhd uint8
 	if f.Checksum {
@@ -88,7 +88,7 @@ func (f frameHeader) appendTo(dst []byte) ([]byte, error) {
 	default:
 		panic("invalid fcs")
 	}
-	return dst, nil
+	return dst
 }
 
 const skippableFrameHeader = 4 + 4

vendor/github.com/klauspost/compress/zstd/matchlen_amd64.go

@@ -0,0 +1,16 @@
+//go:build amd64 && !appengine && !noasm && gc
+// +build amd64,!appengine,!noasm,gc
+
+// Copyright 2019+ Klaus Post. All rights reserved.
+// License information can be found in the LICENSE file.
+
+package zstd
+
+// matchLen returns how many bytes match in a and b
+//
+// It assumes that:
+//
+//	len(a) <= len(b) and len(a) > 0
+//
+//go:noescape
+func matchLen(a []byte, b []byte) int

vendor/github.com/klauspost/compress/zstd/matchlen_amd64.s

@@ -0,0 +1,68 @@
+// Copied from S2 implementation.
+
+//go:build !appengine && !noasm && gc && !noasm
+
+#include "textflag.h"
+
+// func matchLen(a []byte, b []byte) int
+// Requires: BMI
+TEXT ·matchLen(SB), NOSPLIT, $0-56
+	MOVQ a_base+0(FP), AX
+	MOVQ b_base+24(FP), CX
+	MOVQ a_len+8(FP), DX
+
+	// matchLen
+	XORL SI, SI
+	CMPL DX, $0x08
+	JB   matchlen_match4_standalone
+
+matchlen_loopback_standalone:
+	MOVQ  (AX)(SI*1), BX
+	XORQ  (CX)(SI*1), BX
+	TESTQ BX, BX
+	JZ    matchlen_loop_standalone
+
+#ifdef GOAMD64_v3
+	TZCNTQ BX, BX
+#else
+	BSFQ BX, BX
+#endif
+	SARQ $0x03, BX
+	LEAL (SI)(BX*1), SI
+	JMP  gen_match_len_end
+
+matchlen_loop_standalone:
+	LEAL -8(DX), DX
+	LEAL 8(SI), SI
+	CMPL DX, $0x08
+	JAE  matchlen_loopback_standalone
+
+matchlen_match4_standalone:
+	CMPL DX, $0x04
+	JB   matchlen_match2_standalone
+	MOVL (AX)(SI*1), BX
+	CMPL (CX)(SI*1), BX
+	JNE  matchlen_match2_standalone
+	LEAL -4(DX), DX
+	LEAL 4(SI), SI
+
+matchlen_match2_standalone:
+	CMPL DX, $0x02
+	JB   matchlen_match1_standalone
+	MOVW (AX)(SI*1), BX
+	CMPW (CX)(SI*1), BX
+	JNE  matchlen_match1_standalone
+	LEAL -2(DX), DX
+	LEAL 2(SI), SI
+
+matchlen_match1_standalone:
+	CMPL DX, $0x01
+	JB   gen_match_len_end
+	MOVB (AX)(SI*1), BL
+	CMPB (CX)(SI*1), BL
+	JNE  gen_match_len_end
+	INCL SI
+
+gen_match_len_end:
+	MOVQ SI, ret+48(FP)
+	RET

vendor/github.com/klauspost/compress/zstd/matchlen_generic.go

@@ -0,0 +1,33 @@
+//go:build !amd64 || appengine || !gc || noasm
+// +build !amd64 appengine !gc noasm
+
+// Copyright 2019+ Klaus Post. All rights reserved.
+// License information can be found in the LICENSE file.
+
+package zstd
+
+import (
+	"encoding/binary"
+	"math/bits"
+)
+
+// matchLen returns the maximum common prefix length of a and b.
+// a must be the shortest of the two.
+func matchLen(a, b []byte) (n int) {
+	for ; len(a) >= 8 && len(b) >= 8; a, b = a[8:], b[8:] {
+		diff := binary.LittleEndian.Uint64(a) ^ binary.LittleEndian.Uint64(b)
+		if diff != 0 {
+			return n + bits.TrailingZeros64(diff)>>3
+		}
+		n += 8
+	}
+
+	for i := range a {
+		if a[i] != b[i] {
+			break
+		}
+		n++
+	}
+	return n
+
+}

vendor/github.com/klauspost/compress/zstd/seqdec.go

@@ -236,13 +236,16 @@ func (s *sequenceDecs) decodeSync(hist []byte) error {
 		maxBlockSize = s.windowSize
 	}
 
+	if debugDecoder {
+		println("decodeSync: decoding", seqs, "sequences", br.remain(), "bits remain on stream")
+	}
 	for i := seqs - 1; i >= 0; i-- {
 		if br.overread() {
-			printf("reading sequence %d, exceeded available data\n", seqs-i)
+			printf("reading sequence %d, exceeded available data. Overread by %d\n", seqs-i, -br.remain())
 			return io.ErrUnexpectedEOF
 		}
 		var ll, mo, ml int
-		if br.off > 4+((maxOffsetBits+16+16)>>3) {
+		if len(br.in) > 4+((maxOffsetBits+16+16)>>3) {
 			// inlined function:
 			// ll, mo, ml = s.nextFast(br, llState, mlState, ofState)
 
@@ -449,18 +452,13 @@ func (s *sequenceDecs) next(br *bitReader, llState, mlState, ofState decSymbol)
 
 	// extra bits are stored in reverse order.
 	br.fill()
-	if s.maxBits <= 32 {
+	mo += br.getBits(moB)
-		mo += br.getBits(moB)
+	if s.maxBits > 32 {
-		ml += br.getBits(mlB)
-		ll += br.getBits(llB)
-	} else {
-		mo += br.getBits(moB)
 		br.fill()
-		// matchlength+literal length, max 32 bits
-		ml += br.getBits(mlB)
-		ll += br.getBits(llB)
-
 	}
+	// matchlength+literal length, max 32 bits
+	ml += br.getBits(mlB)
+	ll += br.getBits(llB)
 	mo = s.adjustOffset(mo, ll, moB)
 	return
 }

vendor/github.com/klauspost/compress/zstd/seqdec_amd64.go

@@ -5,6 +5,7 @@ package zstd
 
 import (
 	"fmt"
+	"io"
 
 	"github.com/klauspost/compress/internal/cpuinfo"
 )
@@ -134,6 +135,9 @@ func (s *sequenceDecs) decodeSyncSimple(hist []byte) (bool, error) {
 		return true, fmt.Errorf("unexpected literal count, want %d bytes, but only %d is available",
 			ctx.ll, ctx.litRemain+ctx.ll)
 
+	case errorOverread:
+		return true, io.ErrUnexpectedEOF
+
 	case errorNotEnoughSpace:
 		size := ctx.outPosition + ctx.ll + ctx.ml
 		if debugDecoder {
@@ -202,6 +206,9 @@ const errorNotEnoughLiterals = 4
 // error reported when capacity of `out` is too small
 const errorNotEnoughSpace = 5
 
+// error reported when bits are overread.
+const errorOverread = 6
+
 // sequenceDecs_decode implements the main loop of sequenceDecs in x86 asm.
 //
 // Please refer to seqdec_generic.go for the reference implementation.
@@ -247,6 +254,10 @@ func (s *sequenceDecs) decode(seqs []seqVals) error {
 		litRemain: len(s.literals),
 	}
 
+	if debugDecoder {
+		println("decode: decoding", len(seqs), "sequences", br.remain(), "bits remain on stream")
+	}
+
 	s.seqSize = 0
 	lte56bits := s.maxBits+s.offsets.fse.actualTableLog+s.matchLengths.fse.actualTableLog+s.litLengths.fse.actualTableLog <= 56
 	var errCode int
@@ -277,6 +288,8 @@ func (s *sequenceDecs) decode(seqs []seqVals) error {
 		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)
+		case errorOverread:
+			return io.ErrUnexpectedEOF
 		}
 
 		return fmt.Errorf("sequenceDecs_decode_amd64 returned erronous code %d", errCode)
@@ -291,6 +304,9 @@ func (s *sequenceDecs) decode(seqs []seqVals) error {
 	if s.seqSize > maxBlockSize {
 		return fmt.Errorf("output bigger than max block size (%d)", maxBlockSize)
 	}
+	if debugDecoder {
+		println("decode: ", br.remain(), "bits remain on stream. code:", errCode)
+	}
 	err := br.close()
 	if err != nil {
 		printf("Closing sequences: %v, %+v\n", err, *br)

vendor/github.com/klauspost/compress/zstd/seqdec_amd64.s

@@ -5,11 +5,11 @@
 // 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    br+8(FP), CX
-	MOVQ    32(AX), DX
+	MOVQ    24(CX), DX
-	MOVBQZX 40(AX), BX
+	MOVBQZX 32(CX), BX
-	MOVQ    24(AX), SI
+	MOVQ    (CX), AX
-	MOVQ    (AX), AX
+	MOVQ    8(CX), SI
 	ADDQ    SI, AX
 	MOVQ    AX, (SP)
 	MOVQ    ctx+16(FP), AX
@@ -38,7 +38,7 @@ sequenceDecs_decode_amd64_main_loop:
 
 sequenceDecs_decode_amd64_fill_byte_by_byte:
 	CMPQ    SI, $0x00
-	JLE     sequenceDecs_decode_amd64_fill_end
+	JLE     sequenceDecs_decode_amd64_fill_check_overread
 	CMPQ    BX, $0x07
 	JLE     sequenceDecs_decode_amd64_fill_end
 	SHLQ    $0x08, DX
@@ -49,6 +49,10 @@ sequenceDecs_decode_amd64_fill_byte_by_byte:
 	ORQ     AX, DX
 	JMP     sequenceDecs_decode_amd64_fill_byte_by_byte
 
+sequenceDecs_decode_amd64_fill_check_overread:
+	CMPQ BX, $0x40
+	JA   error_overread
+
 sequenceDecs_decode_amd64_fill_end:
 	// Update offset
 	MOVQ  R9, AX
@@ -105,7 +109,7 @@ sequenceDecs_decode_amd64_ml_update_zero:
 
 sequenceDecs_decode_amd64_fill_2_byte_by_byte:
 	CMPQ    SI, $0x00
-	JLE     sequenceDecs_decode_amd64_fill_2_end
+	JLE     sequenceDecs_decode_amd64_fill_2_check_overread
 	CMPQ    BX, $0x07
 	JLE     sequenceDecs_decode_amd64_fill_2_end
 	SHLQ    $0x08, DX
@@ -116,6 +120,10 @@ sequenceDecs_decode_amd64_fill_2_byte_by_byte:
 	ORQ     AX, DX
 	JMP     sequenceDecs_decode_amd64_fill_2_byte_by_byte
 
+sequenceDecs_decode_amd64_fill_2_check_overread:
+	CMPQ BX, $0x40
+	JA   error_overread
+
 sequenceDecs_decode_amd64_fill_2_end:
 	// Update literal length
 	MOVQ  DI, AX
@@ -293,9 +301,9 @@ sequenceDecs_decode_amd64_match_len_ofs_ok:
 	MOVQ R12, 152(AX)
 	MOVQ R13, 160(AX)
 	MOVQ br+8(FP), AX
-	MOVQ DX, 32(AX)
+	MOVQ DX, 24(AX)
-	MOVB BL, 40(AX)
+	MOVB BL, 32(AX)
-	MOVQ SI, 24(AX)
+	MOVQ SI, 8(AX)
 
 	// Return success
 	MOVQ $0x00000000, ret+24(FP)
@@ -320,14 +328,19 @@ error_not_enough_literals:
 	MOVQ $0x00000004, ret+24(FP)
 	RET
 
+	// Return with overread error
+error_overread:
+	MOVQ $0x00000006, 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    br+8(FP), CX
-	MOVQ    32(AX), DX
+	MOVQ    24(CX), DX
-	MOVBQZX 40(AX), BX
+	MOVBQZX 32(CX), BX
-	MOVQ    24(AX), SI
+	MOVQ    (CX), AX
-	MOVQ    (AX), AX
+	MOVQ    8(CX), SI
 	ADDQ    SI, AX
 	MOVQ    AX, (SP)
 	MOVQ    ctx+16(FP), AX
@@ -356,7 +369,7 @@ sequenceDecs_decode_56_amd64_main_loop:
 
 sequenceDecs_decode_56_amd64_fill_byte_by_byte:
 	CMPQ    SI, $0x00
-	JLE     sequenceDecs_decode_56_amd64_fill_end
+	JLE     sequenceDecs_decode_56_amd64_fill_check_overread
 	CMPQ    BX, $0x07
 	JLE     sequenceDecs_decode_56_amd64_fill_end
 	SHLQ    $0x08, DX
@@ -367,6 +380,10 @@ sequenceDecs_decode_56_amd64_fill_byte_by_byte:
 	ORQ     AX, DX
 	JMP     sequenceDecs_decode_56_amd64_fill_byte_by_byte
 
+sequenceDecs_decode_56_amd64_fill_check_overread:
+	CMPQ BX, $0x40
+	JA   error_overread
+
 sequenceDecs_decode_56_amd64_fill_end:
 	// Update offset
 	MOVQ  R9, AX
@@ -586,9 +603,9 @@ sequenceDecs_decode_56_amd64_match_len_ofs_ok:
 	MOVQ R12, 152(AX)
 	MOVQ R13, 160(AX)
 	MOVQ br+8(FP), AX
-	MOVQ DX, 32(AX)
+	MOVQ DX, 24(AX)
-	MOVB BL, 40(AX)
+	MOVB BL, 32(AX)
-	MOVQ SI, 24(AX)
+	MOVQ SI, 8(AX)
 
 	// Return success
 	MOVQ $0x00000000, ret+24(FP)
@@ -613,14 +630,19 @@ error_not_enough_literals:
 	MOVQ $0x00000004, ret+24(FP)
 	RET
 
+	// Return with overread error
+error_overread:
+	MOVQ $0x00000006, 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    br+8(FP), BX
-	MOVQ    32(CX), AX
+	MOVQ    24(BX), AX
-	MOVBQZX 40(CX), DX
+	MOVBQZX 32(BX), DX
-	MOVQ    24(CX), BX
+	MOVQ    (BX), CX
-	MOVQ    (CX), CX
+	MOVQ    8(BX), BX
 	ADDQ    BX, CX
 	MOVQ    CX, (SP)
 	MOVQ    ctx+16(FP), CX
@@ -649,7 +671,7 @@ sequenceDecs_decode_bmi2_main_loop:
 
 sequenceDecs_decode_bmi2_fill_byte_by_byte:
 	CMPQ    BX, $0x00
-	JLE     sequenceDecs_decode_bmi2_fill_end
+	JLE     sequenceDecs_decode_bmi2_fill_check_overread
 	CMPQ    DX, $0x07
 	JLE     sequenceDecs_decode_bmi2_fill_end
 	SHLQ    $0x08, AX
@@ -660,6 +682,10 @@ sequenceDecs_decode_bmi2_fill_byte_by_byte:
 	ORQ     CX, AX
 	JMP     sequenceDecs_decode_bmi2_fill_byte_by_byte
 
+sequenceDecs_decode_bmi2_fill_check_overread:
+	CMPQ DX, $0x40
+	JA   error_overread
+
 sequenceDecs_decode_bmi2_fill_end:
 	// Update offset
 	MOVQ   $0x00000808, CX
@@ -700,7 +726,7 @@ sequenceDecs_decode_bmi2_fill_end:
 
 sequenceDecs_decode_bmi2_fill_2_byte_by_byte:
 	CMPQ    BX, $0x00
-	JLE     sequenceDecs_decode_bmi2_fill_2_end
+	JLE     sequenceDecs_decode_bmi2_fill_2_check_overread
 	CMPQ    DX, $0x07
 	JLE     sequenceDecs_decode_bmi2_fill_2_end
 	SHLQ    $0x08, AX
@@ -711,6 +737,10 @@ sequenceDecs_decode_bmi2_fill_2_byte_by_byte:
 	ORQ     CX, AX
 	JMP     sequenceDecs_decode_bmi2_fill_2_byte_by_byte
 
+sequenceDecs_decode_bmi2_fill_2_check_overread:
+	CMPQ DX, $0x40
+	JA   error_overread
+
 sequenceDecs_decode_bmi2_fill_2_end:
 	// Update literal length
 	MOVQ   $0x00000808, CX
@@ -862,9 +892,9 @@ sequenceDecs_decode_bmi2_match_len_ofs_ok:
 	MOVQ R11, 152(CX)
 	MOVQ R12, 160(CX)
 	MOVQ br+8(FP), CX
-	MOVQ AX, 32(CX)
+	MOVQ AX, 24(CX)
-	MOVB DL, 40(CX)
+	MOVB DL, 32(CX)
-	MOVQ BX, 24(CX)
+	MOVQ BX, 8(CX)
 
 	// Return success
 	MOVQ $0x00000000, ret+24(FP)
@@ -889,14 +919,19 @@ error_not_enough_literals:
 	MOVQ $0x00000004, ret+24(FP)
 	RET
 
+	// Return with overread error
+error_overread:
+	MOVQ $0x00000006, 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    br+8(FP), BX
-	MOVQ    32(CX), AX
+	MOVQ    24(BX), AX
-	MOVBQZX 40(CX), DX
+	MOVBQZX 32(BX), DX
-	MOVQ    24(CX), BX
+	MOVQ    (BX), CX
-	MOVQ    (CX), CX
+	MOVQ    8(BX), BX
 	ADDQ    BX, CX
 	MOVQ    CX, (SP)
 	MOVQ    ctx+16(FP), CX
@@ -925,7 +960,7 @@ sequenceDecs_decode_56_bmi2_main_loop:
 
 sequenceDecs_decode_56_bmi2_fill_byte_by_byte:
 	CMPQ    BX, $0x00
-	JLE     sequenceDecs_decode_56_bmi2_fill_end
+	JLE     sequenceDecs_decode_56_bmi2_fill_check_overread
 	CMPQ    DX, $0x07
 	JLE     sequenceDecs_decode_56_bmi2_fill_end
 	SHLQ    $0x08, AX
@@ -936,6 +971,10 @@ sequenceDecs_decode_56_bmi2_fill_byte_by_byte:
 	ORQ     CX, AX
 	JMP     sequenceDecs_decode_56_bmi2_fill_byte_by_byte
 
+sequenceDecs_decode_56_bmi2_fill_check_overread:
+	CMPQ DX, $0x40
+	JA   error_overread
+
 sequenceDecs_decode_56_bmi2_fill_end:
 	// Update offset
 	MOVQ   $0x00000808, CX
@@ -1113,9 +1152,9 @@ sequenceDecs_decode_56_bmi2_match_len_ofs_ok:
 	MOVQ R11, 152(CX)
 	MOVQ R12, 160(CX)
 	MOVQ br+8(FP), CX
-	MOVQ AX, 32(CX)
+	MOVQ AX, 24(CX)
-	MOVB DL, 40(CX)
+	MOVB DL, 32(CX)
-	MOVQ BX, 24(CX)
+	MOVQ BX, 8(CX)
 
 	// Return success
 	MOVQ $0x00000000, ret+24(FP)
@@ -1140,6 +1179,11 @@ error_not_enough_literals:
 	MOVQ $0x00000004, ret+24(FP)
 	RET
 
+	// Return with overread error
+error_overread:
+	MOVQ $0x00000006, ret+24(FP)
+	RET
+
 // func sequenceDecs_executeSimple_amd64(ctx *executeAsmContext) bool
 // Requires: SSE
 TEXT ·sequenceDecs_executeSimple_amd64(SB), $8-9
@@ -1753,11 +1797,11 @@ empty_seqs:
 // 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    br+8(FP), CX
-	MOVQ    32(AX), DX
+	MOVQ    24(CX), DX
-	MOVBQZX 40(AX), BX
+	MOVBQZX 32(CX), BX
-	MOVQ    24(AX), SI
+	MOVQ    (CX), AX
-	MOVQ    (AX), AX
+	MOVQ    8(CX), SI
 	ADDQ    SI, AX
 	MOVQ    AX, (SP)
 	MOVQ    ctx+16(FP), AX
@@ -1804,7 +1848,7 @@ sequenceDecs_decodeSync_amd64_main_loop:
 
 sequenceDecs_decodeSync_amd64_fill_byte_by_byte:
 	CMPQ    SI, $0x00
-	JLE     sequenceDecs_decodeSync_amd64_fill_end
+	JLE     sequenceDecs_decodeSync_amd64_fill_check_overread
 	CMPQ    BX, $0x07
 	JLE     sequenceDecs_decodeSync_amd64_fill_end
 	SHLQ    $0x08, DX
@@ -1815,6 +1859,10 @@ sequenceDecs_decodeSync_amd64_fill_byte_by_byte:
 	ORQ     AX, DX
 	JMP     sequenceDecs_decodeSync_amd64_fill_byte_by_byte
 
+sequenceDecs_decodeSync_amd64_fill_check_overread:
+	CMPQ BX, $0x40
+	JA   error_overread
+
 sequenceDecs_decodeSync_amd64_fill_end:
 	// Update offset
 	MOVQ  R9, AX
@@ -1871,7 +1919,7 @@ sequenceDecs_decodeSync_amd64_ml_update_zero:
 
 sequenceDecs_decodeSync_amd64_fill_2_byte_by_byte:
 	CMPQ    SI, $0x00
-	JLE     sequenceDecs_decodeSync_amd64_fill_2_end
+	JLE     sequenceDecs_decodeSync_amd64_fill_2_check_overread
 	CMPQ    BX, $0x07
 	JLE     sequenceDecs_decodeSync_amd64_fill_2_end
 	SHLQ    $0x08, DX
@@ -1882,6 +1930,10 @@ sequenceDecs_decodeSync_amd64_fill_2_byte_by_byte:
 	ORQ     AX, DX
 	JMP     sequenceDecs_decodeSync_amd64_fill_2_byte_by_byte
 
+sequenceDecs_decodeSync_amd64_fill_2_check_overread:
+	CMPQ BX, $0x40
+	JA   error_overread
+
 sequenceDecs_decodeSync_amd64_fill_2_end:
 	// Update literal length
 	MOVQ  DI, AX
@@ -2243,9 +2295,9 @@ handle_loop:
 
 loop_finished:
 	MOVQ br+8(FP), AX
-	MOVQ DX, 32(AX)
+	MOVQ DX, 24(AX)
-	MOVB BL, 40(AX)
+	MOVB BL, 32(AX)
-	MOVQ SI, 24(AX)
+	MOVQ SI, 8(AX)
 
 	// Update the context
 	MOVQ ctx+16(FP), AX
@@ -2291,6 +2343,11 @@ error_not_enough_literals:
 	MOVQ $0x00000004, ret+24(FP)
 	RET
 
+	// Return with overread error
+error_overread:
+	MOVQ $0x00000006, ret+24(FP)
+	RET
+
 	// Return with not enough output space error
 error_not_enough_space:
 	MOVQ ctx+16(FP), AX
@@ -2305,11 +2362,11 @@ error_not_enough_space:
 // 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    br+8(FP), BX
-	MOVQ    32(CX), AX
+	MOVQ    24(BX), AX
-	MOVBQZX 40(CX), DX
+	MOVBQZX 32(BX), DX
-	MOVQ    24(CX), BX
+	MOVQ    (BX), CX
-	MOVQ    (CX), CX
+	MOVQ    8(BX), BX
 	ADDQ    BX, CX
 	MOVQ    CX, (SP)
 	MOVQ    ctx+16(FP), CX
@@ -2356,7 +2413,7 @@ sequenceDecs_decodeSync_bmi2_main_loop:
 
 sequenceDecs_decodeSync_bmi2_fill_byte_by_byte:
 	CMPQ    BX, $0x00
-	JLE     sequenceDecs_decodeSync_bmi2_fill_end
+	JLE     sequenceDecs_decodeSync_bmi2_fill_check_overread
 	CMPQ    DX, $0x07
 	JLE     sequenceDecs_decodeSync_bmi2_fill_end
 	SHLQ    $0x08, AX
@@ -2367,6 +2424,10 @@ sequenceDecs_decodeSync_bmi2_fill_byte_by_byte:
 	ORQ     CX, AX
 	JMP     sequenceDecs_decodeSync_bmi2_fill_byte_by_byte
 
+sequenceDecs_decodeSync_bmi2_fill_check_overread:
+	CMPQ DX, $0x40
+	JA   error_overread
+
 sequenceDecs_decodeSync_bmi2_fill_end:
 	// Update offset
 	MOVQ   $0x00000808, CX
@@ -2407,7 +2468,7 @@ sequenceDecs_decodeSync_bmi2_fill_end:
 
 sequenceDecs_decodeSync_bmi2_fill_2_byte_by_byte:
 	CMPQ    BX, $0x00
-	JLE     sequenceDecs_decodeSync_bmi2_fill_2_end
+	JLE     sequenceDecs_decodeSync_bmi2_fill_2_check_overread
 	CMPQ    DX, $0x07
 	JLE     sequenceDecs_decodeSync_bmi2_fill_2_end
 	SHLQ    $0x08, AX
@@ -2418,6 +2479,10 @@ sequenceDecs_decodeSync_bmi2_fill_2_byte_by_byte:
 	ORQ     CX, AX
 	JMP     sequenceDecs_decodeSync_bmi2_fill_2_byte_by_byte
 
+sequenceDecs_decodeSync_bmi2_fill_2_check_overread:
+	CMPQ DX, $0x40
+	JA   error_overread
+
 sequenceDecs_decodeSync_bmi2_fill_2_end:
 	// Update literal length
 	MOVQ   $0x00000808, CX
@@ -2753,9 +2818,9 @@ handle_loop:
 
 loop_finished:
 	MOVQ br+8(FP), CX
-	MOVQ AX, 32(CX)
+	MOVQ AX, 24(CX)
-	MOVB DL, 40(CX)
+	MOVB DL, 32(CX)
-	MOVQ BX, 24(CX)
+	MOVQ BX, 8(CX)
 
 	// Update the context
 	MOVQ ctx+16(FP), AX
@@ -2801,6 +2866,11 @@ error_not_enough_literals:
 	MOVQ $0x00000004, ret+24(FP)
 	RET
 
+	// Return with overread error
+error_overread:
+	MOVQ $0x00000006, ret+24(FP)
+	RET
+
 	// Return with not enough output space error
 error_not_enough_space:
 	MOVQ ctx+16(FP), AX
@@ -2815,11 +2885,11 @@ error_not_enough_space:
 // 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    br+8(FP), CX
-	MOVQ    32(AX), DX
+	MOVQ    24(CX), DX
-	MOVBQZX 40(AX), BX
+	MOVBQZX 32(CX), BX
-	MOVQ    24(AX), SI
+	MOVQ    (CX), AX
-	MOVQ    (AX), AX
+	MOVQ    8(CX), SI
 	ADDQ    SI, AX
 	MOVQ    AX, (SP)
 	MOVQ    ctx+16(FP), AX
@@ -2866,7 +2936,7 @@ sequenceDecs_decodeSync_safe_amd64_main_loop:
 
 sequenceDecs_decodeSync_safe_amd64_fill_byte_by_byte:
 	CMPQ    SI, $0x00
-	JLE     sequenceDecs_decodeSync_safe_amd64_fill_end
+	JLE     sequenceDecs_decodeSync_safe_amd64_fill_check_overread
 	CMPQ    BX, $0x07
 	JLE     sequenceDecs_decodeSync_safe_amd64_fill_end
 	SHLQ    $0x08, DX
@@ -2877,6 +2947,10 @@ sequenceDecs_decodeSync_safe_amd64_fill_byte_by_byte:
 	ORQ     AX, DX
 	JMP     sequenceDecs_decodeSync_safe_amd64_fill_byte_by_byte
 
+sequenceDecs_decodeSync_safe_amd64_fill_check_overread:
+	CMPQ BX, $0x40
+	JA   error_overread
+
 sequenceDecs_decodeSync_safe_amd64_fill_end:
 	// Update offset
 	MOVQ  R9, AX
@@ -2933,7 +3007,7 @@ sequenceDecs_decodeSync_safe_amd64_ml_update_zero:
 
 sequenceDecs_decodeSync_safe_amd64_fill_2_byte_by_byte:
 	CMPQ    SI, $0x00
-	JLE     sequenceDecs_decodeSync_safe_amd64_fill_2_end
+	JLE     sequenceDecs_decodeSync_safe_amd64_fill_2_check_overread
 	CMPQ    BX, $0x07
 	JLE     sequenceDecs_decodeSync_safe_amd64_fill_2_end
 	SHLQ    $0x08, DX
@@ -2944,6 +3018,10 @@ sequenceDecs_decodeSync_safe_amd64_fill_2_byte_by_byte:
 	ORQ     AX, DX
 	JMP     sequenceDecs_decodeSync_safe_amd64_fill_2_byte_by_byte
 
+sequenceDecs_decodeSync_safe_amd64_fill_2_check_overread:
+	CMPQ BX, $0x40
+	JA   error_overread
+
 sequenceDecs_decodeSync_safe_amd64_fill_2_end:
 	// Update literal length
 	MOVQ  DI, AX
@@ -3407,9 +3485,9 @@ handle_loop:
 
 loop_finished:
 	MOVQ br+8(FP), AX
-	MOVQ DX, 32(AX)
+	MOVQ DX, 24(AX)
-	MOVB BL, 40(AX)
+	MOVB BL, 32(AX)
-	MOVQ SI, 24(AX)
+	MOVQ SI, 8(AX)
 
 	// Update the context
 	MOVQ ctx+16(FP), AX
@@ -3455,6 +3533,11 @@ error_not_enough_literals:
 	MOVQ $0x00000004, ret+24(FP)
 	RET
 
+	// Return with overread error
+error_overread:
+	MOVQ $0x00000006, ret+24(FP)
+	RET
+
 	// Return with not enough output space error
 error_not_enough_space:
 	MOVQ ctx+16(FP), AX
@@ -3469,11 +3552,11 @@ error_not_enough_space:
 // 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    br+8(FP), BX
-	MOVQ    32(CX), AX
+	MOVQ    24(BX), AX
-	MOVBQZX 40(CX), DX
+	MOVBQZX 32(BX), DX
-	MOVQ    24(CX), BX
+	MOVQ    (BX), CX
-	MOVQ    (CX), CX
+	MOVQ    8(BX), BX
 	ADDQ    BX, CX
 	MOVQ    CX, (SP)
 	MOVQ    ctx+16(FP), CX
@@ -3520,7 +3603,7 @@ sequenceDecs_decodeSync_safe_bmi2_main_loop:
 
 sequenceDecs_decodeSync_safe_bmi2_fill_byte_by_byte:
 	CMPQ    BX, $0x00
-	JLE     sequenceDecs_decodeSync_safe_bmi2_fill_end
+	JLE     sequenceDecs_decodeSync_safe_bmi2_fill_check_overread
 	CMPQ    DX, $0x07
 	JLE     sequenceDecs_decodeSync_safe_bmi2_fill_end
 	SHLQ    $0x08, AX
@@ -3531,6 +3614,10 @@ sequenceDecs_decodeSync_safe_bmi2_fill_byte_by_byte:
 	ORQ     CX, AX
 	JMP     sequenceDecs_decodeSync_safe_bmi2_fill_byte_by_byte
 
+sequenceDecs_decodeSync_safe_bmi2_fill_check_overread:
+	CMPQ DX, $0x40
+	JA   error_overread
+
 sequenceDecs_decodeSync_safe_bmi2_fill_end:
 	// Update offset
 	MOVQ   $0x00000808, CX
@@ -3571,7 +3658,7 @@ sequenceDecs_decodeSync_safe_bmi2_fill_end:
 
 sequenceDecs_decodeSync_safe_bmi2_fill_2_byte_by_byte:
 	CMPQ    BX, $0x00
-	JLE     sequenceDecs_decodeSync_safe_bmi2_fill_2_end
+	JLE     sequenceDecs_decodeSync_safe_bmi2_fill_2_check_overread
 	CMPQ    DX, $0x07
 	JLE     sequenceDecs_decodeSync_safe_bmi2_fill_2_end
 	SHLQ    $0x08, AX
@@ -3582,6 +3669,10 @@ sequenceDecs_decodeSync_safe_bmi2_fill_2_byte_by_byte:
 	ORQ     CX, AX
 	JMP     sequenceDecs_decodeSync_safe_bmi2_fill_2_byte_by_byte
 
+sequenceDecs_decodeSync_safe_bmi2_fill_2_check_overread:
+	CMPQ DX, $0x40
+	JA   error_overread
+
 sequenceDecs_decodeSync_safe_bmi2_fill_2_end:
 	// Update literal length
 	MOVQ   $0x00000808, CX
@@ -4019,9 +4110,9 @@ handle_loop:
 
 loop_finished:
 	MOVQ br+8(FP), CX
-	MOVQ AX, 32(CX)
+	MOVQ AX, 24(CX)
-	MOVB DL, 40(CX)
+	MOVB DL, 32(CX)
-	MOVQ BX, 24(CX)
+	MOVQ BX, 8(CX)
 
 	// Update the context
 	MOVQ ctx+16(FP), AX
@@ -4067,6 +4158,11 @@ error_not_enough_literals:
 	MOVQ $0x00000004, ret+24(FP)
 	RET
 
+	// Return with overread error
+error_overread:
+	MOVQ $0x00000006, ret+24(FP)
+	RET
+
 	// Return with not enough output space error
 error_not_enough_space:
 	MOVQ ctx+16(FP), AX

vendor/github.com/klauspost/compress/zstd/seqdec_generic.go

@@ -29,7 +29,7 @@ func (s *sequenceDecs) decode(seqs []seqVals) error {
 	}
 	for i := range seqs {
 		var ll, mo, ml int
-		if br.off > 4+((maxOffsetBits+16+16)>>3) {
+		if len(br.in) > 4+((maxOffsetBits+16+16)>>3) {
 			// inlined function:
 			// ll, mo, ml = s.nextFast(br, llState, mlState, ofState)
 

vendor/github.com/klauspost/compress/zstd/snappy.go

@@ -95,10 +95,9 @@ func (r *SnappyConverter) Convert(in io.Reader, w io.Writer) (int64, error) {
 	var written int64
 	var readHeader bool
 	{
-		var header []byte
+		header := frameHeader{WindowSize: snappyMaxBlockSize}.appendTo(r.buf[:0])
-		var n int
-		header, r.err = frameHeader{WindowSize: snappyMaxBlockSize}.appendTo(r.buf[:0])
 
+		var n int
 		n, r.err = w.Write(header)
 		if r.err != nil {
 			return written, r.err

vendor/github.com/klauspost/compress/zstd/zstd.go

@@ -9,7 +9,6 @@ import (
 	"errors"
 	"log"
 	"math"
-	"math/bits"
 )
 
 // enable debug printing
@@ -106,33 +105,12 @@ func printf(format string, a ...interface{}) {
 	}
 }
 
-// matchLen returns the maximum common prefix length of a and b.
-// a must be the shortest of the two.
-func matchLen(a, b []byte) (n int) {
-	for ; len(a) >= 8 && len(b) >= 8; a, b = a[8:], b[8:] {
-		diff := binary.LittleEndian.Uint64(a) ^ binary.LittleEndian.Uint64(b)
-		if diff != 0 {
-			return n + bits.TrailingZeros64(diff)>>3
-		}
-		n += 8
-	}
-
-	for i := range a {
-		if a[i] != b[i] {
-			break
-		}
-		n++
-	}
-	return n
-
-}
-
 func load3232(b []byte, i int32) uint32 {
-	return binary.LittleEndian.Uint32(b[i:])
+	return binary.LittleEndian.Uint32(b[:len(b):len(b)][i:])
 }
 
 func load6432(b []byte, i int32) uint64 {
-	return binary.LittleEndian.Uint64(b[i:])
+	return binary.LittleEndian.Uint64(b[:len(b):len(b)][i:])
 }
 
 type byter interface {

vendor/modules.txt

@@ -138,7 +138,7 @@ github.com/imdario/mergo
 # github.com/inconshreveable/mousetrap v1.1.0
 ## explicit; go 1.18
 github.com/inconshreveable/mousetrap
-# github.com/klauspost/compress v1.16.3
+# github.com/klauspost/compress v1.17.2
 ## explicit; go 1.18
 github.com/klauspost/compress
 github.com/klauspost/compress/fse