colblk: use crlib/crbytes.CommonPrefix

Update crlib and use its faster implementation for finding the longest
shared prefix of two byte slices.

go.mod

@@ -4,7 +4,7 @@ require (
  github.com/DataDog/zstd v1.5.6-0.20230824185856-869dae002e5e
  github.com/HdrHistogram/hdrhistogram-go v1.1.2
  github.com/cespare/xxhash/v2 v2.2.0
- github.com/cockroachdb/crlib v0.0.0-20240729155931-991150b7e290
+ github.com/cockroachdb/crlib v0.0.0-20240816115810-1c502cdb7c1d
  github.com/cockroachdb/datadriven v1.0.3-0.20240530155848-7682d40af056
  github.com/cockroachdb/errors v1.11.3
  github.com/cockroachdb/fifo v0.0.0-20240606204812-0bbfbd93a7ce

go.sum

@@ -71,8 +71,8 @@ github.com/chzyer/readline v0.0.0-20180603132655-2972be24d48e/go.mod h1:nSuG5e5P
 github.com/chzyer/test v0.0.0-20180213035817-a1ea475d72b1/go.mod h1:Q3SI9o4m/ZMnBNeIyt5eFwwo7qiLfzFZmjNmxjkiQlU=
 github.com/client9/misspell v0.3.4/go.mod h1:qj6jICC3Q7zFZvVWo7KLAzC3yx5G7kyvSDkc90ppPyw=
 github.com/cncf/udpa/go v0.0.0-20191209042840-269d4d468f6f/go.mod h1:M8M6+tZqaGXZJjfX53e64911xZQV5JYwmTeXPW+k8Sc=
-github.com/cockroachdb/crlib v0.0.0-20240729155931-991150b7e290 h1:oJGWhlrgtcPSzJEgFHo6VIQW2TCFR/OiByQr6yzjDkU=
-github.com/cockroachdb/crlib v0.0.0-20240729155931-991150b7e290/go.mod h1:Gq51ZeKaFCXk6QwuGM0w1dnaOqc/F5zKT2zA9D6Xeac=
+github.com/cockroachdb/crlib v0.0.0-20240816115810-1c502cdb7c1d h1:IgQRpDBWe+tThngFMwhWpRJ/oEe5ZUJHq5nBNqeiIeI=
+github.com/cockroachdb/crlib v0.0.0-20240816115810-1c502cdb7c1d/go.mod h1:Gq51ZeKaFCXk6QwuGM0w1dnaOqc/F5zKT2zA9D6Xeac=
 github.com/cockroachdb/datadriven v1.0.3-0.20240530155848-7682d40af056 h1:slXychO2uDM6hYRu4c0pD0udNI8uObfeKN6UInWViS8=
 github.com/cockroachdb/datadriven v1.0.3-0.20240530155848-7682d40af056/go.mod h1:a9RdTaap04u637JoCzcUoIcDmvwSUtcUFtT/C3kJlTU=
 github.com/cockroachdb/errors v1.11.3 h1:5bA+k2Y6r+oz/6Z/RFlNeVCesGARKuC6YymtcDrbC/I=

sstable/colblk/block_test.go

@@ -15,6 +15,7 @@ import (
  "testing"
  "time"
 
+ "github.com/cockroachdb/crlib/crbytes"
  "github.com/cockroachdb/datadriven"
  "golang.org/x/exp/rand"
 )
@@ -244,7 +245,7 @@ func buildBlock(schema []testColumnSpec, rows int, data []interface{}) []byte {
  for r, v := range colData {
  sharedPrefix := 0
  if r > 0 {
- sharedPrefix = bytesSharedPrefix(colData[r-1], v)
+ sharedPrefix = crbytes.CommonPrefix(colData[r-1], v)
  }
  pbb.Put(v, sharedPrefix)
  }

sstable/colblk/cockroach_test.go

@@ -15,6 +15,7 @@ import (
  "time"
  "unsafe"
 
+ "github.com/cockroachdb/crlib/crbytes"
  "github.com/cockroachdb/pebble/internal/base"
  "github.com/cockroachdb/pebble/internal/crdbtest"
  "github.com/cockroachdb/pebble/sstable/block"
@@ -62,7 +63,7 @@ func (kw *cockroachKeyWriter) ComparePrev(key []byte) KeyComparison {
  lp := kw.prefixes.LastKey()
  var cmpv KeyComparison
  cmpv.PrefixLen = int32(crdbtest.Split(key)) // TODO(jackson): Inline
- cmpv.CommonPrefixLen = int32(bytesSharedPrefix(lp, key[:cmpv.PrefixLen]))
+ cmpv.CommonPrefixLen = int32(crbytes.CommonPrefix(lp, key[:cmpv.PrefixLen]))
  if cmpv.CommonPrefixLen == cmpv.PrefixLen {
  cmpv.UserKeyComparison = int32(crdbtest.CompareSuffixes(key[cmpv.PrefixLen:], kw.prevSuffix))
  return cmpv

sstable/colblk/data_block.go

@@ -15,6 +15,7 @@ import (
  "sync"
  "unsafe"
 
+ "github.com/cockroachdb/crlib/crbytes"
  "github.com/cockroachdb/errors"
  "github.com/cockroachdb/pebble/internal/base"
  "github.com/cockroachdb/pebble/internal/binfmt"
@@ -163,7 +164,7 @@ func (w *defaultKeyWriter) ComparePrev(key []byte) KeyComparison {
 
  var cmpv KeyComparison
  cmpv.PrefixLen = int32(w.comparer.Split(key))
- cmpv.CommonPrefixLen = int32(bytesSharedPrefix(lp, key[:cmpv.PrefixLen]))
+ cmpv.CommonPrefixLen = int32(crbytes.CommonPrefix(lp, key[:cmpv.PrefixLen]))
  if len(lp) == 0 {
  // The first key has no previous key to compare to.
  return cmpv

sstable/colblk/prefix_bytes.go

@@ -6,14 +6,14 @@ package colblk
 
 import (
  "bytes"
- "encoding/binary"
  "fmt"
  "io"
  "math/bits"
  "slices"
  "strings"
  "unsafe"
 
+ "github.com/cockroachdb/crlib/crbytes"
  "github.com/cockroachdb/errors"
  "github.com/cockroachdb/pebble/internal/binfmt"
  "github.com/cockroachdb/pebble/internal/invariants"
@@ -757,9 +757,9 @@ func (b *PrefixBytesBuilder) Put(key []byte, bytesSharedWithPrev int) {
  if bytes.Compare(key, b.data[len(b.data)-prev.lastKeyLen:]) < 0 {
  panic(errors.AssertionFailedf("keys must be added in order: %q < %q", key, b.data[len(b.data)-prev.lastKeyLen:]))
  }
- if bytesSharedWithPrev != bytesSharedPrefix(key, b.data[len(b.data)-prev.lastKeyLen:]) {
+ if bytesSharedWithPrev != crbytes.CommonPrefix(key, b.data[len(b.data)-prev.lastKeyLen:]) {
  panic(errors.AssertionFailedf("bytesSharedWithPrev %d != %d", bytesSharedWithPrev,
- bytesSharedPrefix(key, b.data[len(b.data)-prev.lastKeyLen:])))
+ crbytes.CommonPrefix(key, b.data[len(b.data)-prev.lastKeyLen:])))
  }
  }
  }
@@ -1091,19 +1091,3 @@ func (b bundleCalc) offsetIndexByBundleIndex(bi int) int {
 func (b bundleCalc) bundleCount(rows int) int {
  return 1 + (rows-1)>>b.bundleShift
 }
-
-// bytesSharedPrefix returns the length of the shared prefix between a and b.
-func bytesSharedPrefix(a, b []byte) int {
- asUint64 := func(data []byte, i int) uint64 {
- return binary.LittleEndian.Uint64(data[i:])
- }
- var shared int
- n := min(len(a), len(b))
- for shared < n-7 && asUint64(a, shared) == asUint64(b, shared) {
- shared += 8
- }
- for shared < n && a[shared] == b[shared] {
- shared++
- }
- return shared
-}

sstable/colblk/prefix_bytes_test.go

@@ -13,6 +13,7 @@ import (
  "testing"
  "time"
 
+ "github.com/cockroachdb/crlib/crbytes"
  "github.com/cockroachdb/datadriven"
  "github.com/cockroachdb/pebble/internal/binfmt"
  "github.com/cockroachdb/pebble/internal/invariants"
@@ -46,7 +47,7 @@ func TestPrefixBytes(t *testing.T) {
  for _, k := range inputKeys {
  keyPrefixLenSharedWithPrev := len(k)
  if builder.nKeys > 0 {
- keyPrefixLenSharedWithPrev = bytesSharedPrefix(builder.LastKey(), k)
+ keyPrefixLenSharedWithPrev = crbytes.CommonPrefix(builder.LastKey(), k)
  }
  p := []byte(k)
  builder.Put(p, keyPrefixLenSharedWithPrev)
@@ -149,7 +150,7 @@ func TestPrefixBytesRandomized(t *testing.T) {
  for i := 0; i < len(userKeys); i++ {
  keyPrefixLenSharedWithPrev := 0
  if i > 0 {
- keyPrefixLenSharedWithPrev = bytesSharedPrefix(userKeys[i-1], userKeys[i])
+ keyPrefixLenSharedWithPrev = crbytes.CommonPrefix(userKeys[i-1], userKeys[i])
  }
  pbb.Put(userKeys[i], keyPrefixLenSharedWithPrev)
  }
@@ -247,7 +248,7 @@ func BenchmarkPrefixBytes(b *testing.B) {
  for i := 0; i < n; i++ {
  keyPrefixLenSharedWithPrev := 0
  if i > 0 {
- keyPrefixLenSharedWithPrev = bytesSharedPrefix(userKeys[i-1], userKeys[i])
+ keyPrefixLenSharedWithPrev = crbytes.CommonPrefix(userKeys[i-1], userKeys[i])
  }
  pbb.Put(userKeys[i], keyPrefixLenSharedWithPrev)
  }