.github/workflows/codeql-analysis.yml

@@ -44,7 +44,7 @@ jobs:
 
     # Initializes the CodeQL tools for scanning.
     - name: Initialize CodeQL
-      uses: github/codeql-action/init@1b1aada464948af03b950897e5eb522f92603cc2 # v3.24.9
+      uses: github/codeql-action/init@f079b8493333aace61c81488f8bd40919487bd9f # v3.25.7
       with:
         languages: ${{ matrix.language }}
         # If you wish to specify custom queries, you can do so here or in a config file.
@@ -58,7 +58,7 @@ jobs:
     # Autobuild attempts to build any compiled languages  (C/C++, C#, or Java).
     # If this step fails, then you should remove it and run the build manually (see below)
     - name: Autobuild
-      uses: github/codeql-action/autobuild@1b1aada464948af03b950897e5eb522f92603cc2 # v3.24.9
+      uses: github/codeql-action/autobuild@f079b8493333aace61c81488f8bd40919487bd9f # v3.25.7
 
     # ℹ️ Command-line programs to run using the OS shell.
     # 📚 See https://docs.github.com/en/actions/using-workflows/workflow-syntax-for-github-actions#jobsjob_idstepsrun
@@ -71,4 +71,4 @@ jobs:
     #   ./location_of_script_within_repo/buildscript.sh
 
     - name: Perform CodeQL Analysis
-      uses: github/codeql-action/analyze@1b1aada464948af03b950897e5eb522f92603cc2 # v3.24.9
+      uses: github/codeql-action/analyze@f079b8493333aace61c81488f8bd40919487bd9f # v3.25.7

.github/workflows/release.yml

@@ -24,7 +24,7 @@ jobs:
           go-version: 1.22.x
       -
         name: Run GoReleaser
-        uses: goreleaser/goreleaser-action@7ec5c2b0c6cdda6e8bbb49444bc797dd33d74dd8 # v5.0.0
+        uses: goreleaser/goreleaser-action@5742e2a039330cbb23ebf35f046f814d4c6ff811 # v5.1.0
         with:
           version: 1.24.x
           args: release --rm-dist

flate/matchlen_amd64.s

@@ -5,7 +5,6 @@
 #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
@@ -17,17 +16,16 @@ TEXT ·matchLen(SB), NOSPLIT, $0-56
 	JB   matchlen_match4_standalone
 
 matchlen_loopback_standalone:
-	MOVQ  (AX)(SI*1), BX
-	XORQ  (CX)(SI*1), BX
-	TESTQ BX, BX
-	JZ    matchlen_loop_standalone
+	MOVQ (AX)(SI*1), BX
+	XORQ (CX)(SI*1), BX
+	JZ   matchlen_loop_standalone
 
 #ifdef GOAMD64_v3
 	TZCNTQ BX, BX
 #else
 	BSFQ BX, BX
 #endif
-	SARQ $0x03, BX
+	SHRL $0x03, BX
 	LEAL (SI)(BX*1), SI
 	JMP  gen_match_len_end
 

internal/godebug/godebug.go

@@ -0,0 +1,44 @@
+// Copyright 2021 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// Package godebug makes the simplified settings in the $GODEBUG environment variable
+// available to packages.
+// Needed since internal/godebug is not available here.
+package godebug
+
+import "os"
+
+func Get(key string) string {
+	s := os.Getenv("GODEBUG")
+	if s == "" {
+		return ""
+	}
+	// Scan the string backward so that later settings are used
+	// and earlier settings are ignored.
+	// Note that a forward scan would cause cached values
+	// to temporarily use the ignored value before being
+	// updated to the "correct" one.
+	end := len(s)
+	eq := -1
+	for i := end - 1; i >= -1; i-- {
+		if i == -1 || s[i] == ',' {
+			if eq >= 0 {
+				name, arg := s[i+1:eq], s[eq+1:end]
+				if name == key {
+					for j := 0; j < len(arg); j++ {
+						if arg[j] == '#' {
+							return arg[:j]
+						}
+					}
+					return arg
+				}
+			}
+			eq = -1
+			end = i
+		} else if s[i] == '=' {
+			eq = i
+		}
+	}
+	return ""
+}

s2/decode_arm64.s

@@ -60,7 +60,7 @@
 //
 // The d variable is implicitly R_DST - R_DBASE,  and len(dst)-d is R_DEND - R_DST.
 // The s variable is implicitly R_SRC - R_SBASE, and len(src)-s is R_SEND - R_SRC.
-TEXT ·s2Decode(SB), NOSPLIT, $56-64
+TEXT ·s2Decode(SB), NOSPLIT, $56-56
 	// Initialize R_SRC, R_DST and R_DBASE-R_SEND.
 	MOVD dst_base+0(FP), R_DBASE
 	MOVD dst_len+8(FP), R_DLEN

s2/index.go

@@ -17,6 +17,8 @@ const (
 	S2IndexHeader   = "s2idx\x00"
 	S2IndexTrailer  = "\x00xdi2s"
 	maxIndexEntries = 1 << 16
+	// If distance is less than this, we do not add the entry.
+	minIndexDist = 1 << 20
 )
 
 // Index represents an S2/Snappy index.
@@ -72,6 +74,10 @@ func (i *Index) add(compressedOffset, uncompressedOffset int64) error {
 		if latest.compressedOffset > compressedOffset {
 			return fmt.Errorf("internal error: Earlier compressed received (%d > %d)", latest.uncompressedOffset, uncompressedOffset)
 		}
+		if latest.uncompressedOffset+minIndexDist > uncompressedOffset {
+			// Only add entry if distance is large enough.
+			return nil
+		}
 	}
 	i.info = append(i.info, struct {
 		compressedOffset   int64
@@ -122,7 +128,7 @@ func (i *Index) Find(offset int64) (compressedOff, uncompressedOff int64, err er
 
 // reduce to stay below maxIndexEntries
 func (i *Index) reduce() {
-	if len(i.info) < maxIndexEntries && i.estBlockUncomp >= 1<<20 {
+	if len(i.info) < maxIndexEntries && i.estBlockUncomp >= minIndexDist {
 		return
 	}
 
@@ -132,7 +138,7 @@ func (i *Index) reduce() {
 	j := 0
 
 	// Each block should be at least 1MB, but don't reduce below 1000 entries.
-	for i.estBlockUncomp*(int64(removeN)+1) < 1<<20 && len(i.info)/(removeN+1) > 1000 {
+	for i.estBlockUncomp*(int64(removeN)+1) < minIndexDist && len(i.info)/(removeN+1) > 1000 {
 		removeN++
 	}
 	for idx := 0; idx < len(src); idx++ {

s2/s2.go

@@ -109,7 +109,11 @@ const (
 	chunkTypeStreamIdentifier = 0xff
 )
 
-var crcTable = crc32.MakeTable(crc32.Castagnoli)
+var (
+	crcTable              = crc32.MakeTable(crc32.Castagnoli)
+	magicChunkSnappyBytes = []byte(magicChunkSnappy) // Can be passed to functions where it escapes.
+	magicChunkBytes       = []byte(magicChunk)       // Can be passed to functions where it escapes.
+)
 
 // crc implements the checksum specified in section 3 of
 // https://github.com/google/snappy/blob/master/framing_format.txt

s2/writer.go

@@ -239,6 +239,9 @@ func (w *Writer) ReadFrom(r io.Reader) (n int64, err error) {
 			}
 		}
 		if n2 == 0 {
+			if cap(inbuf) >= w.obufLen {
+				w.buffers.Put(inbuf)
+			}
 			break
 		}
 		n += int64(n2)
@@ -314,9 +317,9 @@ func (w *Writer) AddSkippableBlock(id uint8, data []byte) (err error) {
 		hWriter := make(chan result)
 		w.output <- hWriter
 		if w.snappy {
-			hWriter <- result{startOffset: w.uncompWritten, b: []byte(magicChunkSnappy)}
+			hWriter <- result{startOffset: w.uncompWritten, b: magicChunkSnappyBytes}
 		} else {
-			hWriter <- result{startOffset: w.uncompWritten, b: []byte(magicChunk)}
+			hWriter <- result{startOffset: w.uncompWritten, b: magicChunkBytes}
 		}
 	}
 
@@ -370,9 +373,9 @@ func (w *Writer) EncodeBuffer(buf []byte) (err error) {
 		hWriter := make(chan result)
 		w.output <- hWriter
 		if w.snappy {
-			hWriter <- result{startOffset: w.uncompWritten, b: []byte(magicChunkSnappy)}
+			hWriter <- result{startOffset: w.uncompWritten, b: magicChunkSnappyBytes}
 		} else {
-			hWriter <- result{startOffset: w.uncompWritten, b: []byte(magicChunk)}
+			hWriter <- result{startOffset: w.uncompWritten, b: magicChunkBytes}
 		}
 	}
 
@@ -478,9 +481,9 @@ func (w *Writer) write(p []byte) (nRet int, errRet error) {
 			hWriter := make(chan result)
 			w.output <- hWriter
 			if w.snappy {
-				hWriter <- result{startOffset: w.uncompWritten, b: []byte(magicChunkSnappy)}
+				hWriter <- result{startOffset: w.uncompWritten, b: magicChunkSnappyBytes}
 			} else {
-				hWriter <- result{startOffset: w.uncompWritten, b: []byte(magicChunk)}
+				hWriter <- result{startOffset: w.uncompWritten, b: magicChunkBytes}
 			}
 		}
 
@@ -560,6 +563,9 @@ func (w *Writer) writeFull(inbuf []byte) (errRet error) {
 
 	if w.concurrency == 1 {
 		_, err := w.writeSync(inbuf[obufHeaderLen:])
+		if cap(inbuf) >= w.obufLen {
+			w.buffers.Put(inbuf)
+		}
 		return err
 	}
 
@@ -569,9 +575,9 @@ func (w *Writer) writeFull(inbuf []byte) (errRet error) {
 		hWriter := make(chan result)
 		w.output <- hWriter
 		if w.snappy {
-			hWriter <- result{startOffset: w.uncompWritten, b: []byte(magicChunkSnappy)}
+			hWriter <- result{startOffset: w.uncompWritten, b: magicChunkSnappyBytes}
 		} else {
-			hWriter <- result{startOffset: w.uncompWritten, b: []byte(magicChunk)}
+			hWriter <- result{startOffset: w.uncompWritten, b: magicChunkBytes}
 		}
 	}
 
@@ -637,9 +643,9 @@ func (w *Writer) writeSync(p []byte) (nRet int, errRet error) {
 		var n int
 		var err error
 		if w.snappy {
-			n, err = w.writer.Write([]byte(magicChunkSnappy))
+			n, err = w.writer.Write(magicChunkSnappyBytes)
 		} else {
-			n, err = w.writer.Write([]byte(magicChunk))
+			n, err = w.writer.Write(magicChunkBytes)
 		}
 		if err != nil {
 			return 0, w.err(err)

s2/writer_test.go

@@ -603,6 +603,36 @@ func BenchmarkWriterRandom(b *testing.B) {
 	}
 }
 
+func BenchmarkReadFromRandom(b *testing.B) {
+	rng := rand.New(rand.NewSource(1))
+	// Make max window so we never get matches.
+	data := make([]byte, 8<<20)
+	for i := range data {
+		data[i] = uint8(rng.Intn(256))
+	}
+
+	for name, opts := range testOptions(b) {
+		w := NewWriter(io.Discard, opts...)
+		in := bytes.NewReader(data)
+		w.ReadFrom(in)
+		b.Run(name, func(b *testing.B) {
+			b.ResetTimer()
+			b.ReportAllocs()
+			b.SetBytes(int64(len(data)))
+			for i := 0; i < b.N; i++ {
+				in.Reset(data)
+				_, err := w.ReadFrom(in)
+				if err != nil {
+					b.Fatal(err)
+				}
+			}
+			// Flush output
+			w.Flush()
+		})
+		w.Close()
+	}
+}
+
 func BenchmarkIndexFind(b *testing.B) {
 	fatalErr := func(t testing.TB, err error) {
 		if err != nil {