updated syscall usage

vendor/go.etcd.io/bbolt/.go-version

@@ -0,0 +1 @@
+1.22.6

vendor/go.etcd.io/bbolt/Makefile

@@ -41,6 +41,15 @@ coverage:
 	TEST_FREELIST_TYPE=array go test -v -timeout 30m \
 		-coverprofile cover-freelist-array.out -covermode atomic
 
+BOLT_CMD=bbolt
+
+build:
+	go build -o bin/${BOLT_CMD} ./cmd/${BOLT_CMD}
+
+.PHONY: clean
+clean: # Clean binaries
+	rm -f ./bin/${BOLT_CMD}
+
 .PHONY: gofail-enable
 gofail-enable: install-gofail
 	gofail enable .
@@ -61,3 +70,7 @@ test-failpoint:
 	@echo "[failpoint] array freelist test"
 	TEST_FREELIST_TYPE=array go test -v ${TESTFLAGS} -timeout 30m ./tests/failpoint
 
+.PHONY: test-robustness # Running robustness tests requires root permission
+test-robustness:
+	go test -v ${TESTFLAGS} ./tests/dmflakey -test.root
+	go test -v ${TESTFLAGS} ./tests/robustness -test.root

vendor/go.etcd.io/bbolt/README.md

@@ -421,10 +421,19 @@ Prev()   Move to the previous key.
 ```
 
 Each of those functions has a return signature of `(key []byte, value []byte)`.
-When you have iterated to the end of the cursor then `Next()` will return a
-`nil` key.  You must seek to a position using `First()`, `Last()`, or `Seek()`
-before calling `Next()` or `Prev()`. If you do not seek to a position then
-these functions will return a `nil` key.
+You must seek to a position using `First()`, `Last()`, or `Seek()` before calling
+`Next()` or `Prev()`. If you do not seek to a position then these functions will
+return a `nil` key.
+
+When you have iterated to the end of the cursor, then `Next()` will return a
+`nil` key and the cursor still points to the last element if present. When you
+have iterated to the beginning of the cursor, then `Prev()` will return a `nil`
+key and the cursor still points to the first element if present.
+
+If you remove key/value pairs during iteration, the cursor may automatically
+move to the next position if present in current node each time removing a key.
+When you call `c.Next()` after removing a key, it may skip one key/value pair.
+Refer to [pull/611](https://github.com/etcd-io/bbolt/pull/611) to get more detailed info.
 
 During iteration, if the key is non-`nil` but the value is `nil`, that means
 the key refers to a bucket rather than a value.  Use `Bucket.Bucket()` to
@@ -850,6 +859,12 @@ Here are a few things to note when evaluating and using Bolt:
   to grow. However, it's important to note that deleting large chunks of data
   will not allow you to reclaim that space on disk.
 
+* Removing key/values pairs in a bucket during iteration on the bucket using
+  cursor may not work properly. Each time when removing a key/value pair, the
+  cursor may automatically move to the next position if present. When users
+  call `c.Next()` after removing a key, it may skip one key/value pair.
+  Refer to https://github.com/etcd-io/bbolt/pull/611 for more detailed info.
+
   For more information on page allocation, [see this comment][page-allocation].
 
 [page-allocation]: https://github.com/boltdb/bolt/issues/308#issuecomment-74811638

vendor/go.etcd.io/bbolt/bolt_openbsd.go

@@ -1,22 +1,11 @@
 package bbolt
 
 import (
-	"syscall"
-	"unsafe"
-)
-
-const (
-	msAsync      = 1 << iota // perform asynchronous writes
-	msSync                   // perform synchronous writes
-	msInvalidate             // invalidate cached data
+	"golang.org/x/sys/unix"
 )
 
 func msync(db *DB) error {
-	_, _, errno := syscall.Syscall(syscall.SYS_MSYNC, uintptr(unsafe.Pointer(db.data)), uintptr(db.datasz), msInvalidate)
-	if errno != 0 {
-		return errno
-	}
-	return nil
+	return unix.Msync(db.data[:db.datasz], unix.MS_INVALIDATE)
 }
 
 func fdatasync(db *DB) error {

vendor/go.etcd.io/bbolt/bucket.go

@@ -162,12 +162,17 @@ func (b *Bucket) CreateBucket(key []byte) (*Bucket, error) {
 		return nil, ErrBucketNameRequired
 	}
 
+	// Insert into node.
+	// Tip: Use a new variable `newKey` instead of reusing the existing `key` to prevent
+	// it from being marked as leaking, and accordingly cannot be allocated on stack.
+	newKey := cloneBytes(key)
+
 	// Move cursor to correct position.
 	c := b.Cursor()
-	k, _, flags := c.seek(key)
+	k, _, flags := c.seek(newKey)
 
 	// Return an error if there is an existing key.
-	if bytes.Equal(key, k) {
+	if bytes.Equal(newKey, k) {
 		if (flags & bucketLeafFlag) != 0 {
 			return nil, ErrBucketExists
 		}
@@ -182,16 +187,14 @@ func (b *Bucket) CreateBucket(key []byte) (*Bucket, error) {
 	}
 	var value = bucket.write()
 
-	// Insert into node.
-	key = cloneBytes(key)
-	c.node().put(key, key, value, 0, bucketLeafFlag)
+	c.node().put(newKey, newKey, value, 0, bucketLeafFlag)
 
 	// Since subbuckets are not allowed on inline buckets, we need to
 	// dereference the inline page, if it exists. This will cause the bucket
 	// to be treated as a regular, non-inline bucket for the rest of the tx.
 	b.page = nil
 
-	return b.Bucket(key), nil
+	return b.Bucket(newKey), nil
 }
 
 // CreateBucketIfNotExists creates a new bucket if it doesn't already exist and returns a reference to it.
@@ -288,18 +291,23 @@ func (b *Bucket) Put(key []byte, value []byte) error {
 		return ErrValueTooLarge
 	}
 
+	// Insert into node.
+	// Tip: Use a new variable `newKey` instead of reusing the existing `key` to prevent
+	// it from being marked as leaking, and accordingly cannot be allocated on stack.
+	newKey := cloneBytes(key)
+
 	// Move cursor to correct position.
 	c := b.Cursor()
-	k, _, flags := c.seek(key)
+	k, _, flags := c.seek(newKey)
 
 	// Return an error if there is an existing key with a bucket value.
-	if bytes.Equal(key, k) && (flags&bucketLeafFlag) != 0 {
+	if bytes.Equal(newKey, k) && (flags&bucketLeafFlag) != 0 {
 		return ErrIncompatibleValue
 	}
 
-	// Insert into node.
-	key = cloneBytes(key)
-	c.node().put(key, key, value, 0, 0)
+	// gofail: var beforeBucketPut struct{}
+
+	c.node().put(newKey, newKey, value, 0, 0)
 
 	return nil
 }

vendor/go.etcd.io/bbolt/cursor.go

@@ -71,7 +71,7 @@ func (c *Cursor) Last() (key []byte, value []byte) {
 
 	// If this is an empty page (calling Delete may result in empty pages)
 	// we call prev to find the last page that is not empty
-	for len(c.stack) > 0 && c.stack[len(c.stack)-1].count() == 0 {
+	for len(c.stack) > 1 && c.stack[len(c.stack)-1].count() == 0 {
 		c.prev()
 	}
 
@@ -254,6 +254,15 @@ func (c *Cursor) prev() (key []byte, value []byte, flags uint32) {
 			elem.index--
 			break
 		}
+		// If we've hit the beginning, we should stop moving the cursor,
+		// and stay at the first element, so that users can continue to
+		// iterate over the elements in reverse direction by calling `Next`.
+		// We should return nil in such case.
+		// Refer to https://github.com/etcd-io/bbolt/issues/733
+		if len(c.stack) == 1 {
+			c.first()
+			return nil, nil, 0
+		}
 		c.stack = c.stack[:i]
 	}
 

vendor/go.etcd.io/bbolt/db.go

@@ -57,6 +57,12 @@ const (
 // All data access is performed through transactions which can be obtained through the DB.
 // All the functions on DB will return a ErrDatabaseNotOpen if accessed before Open() is called.
 type DB struct {
+	// Put `stats` at the first field to ensure it's 64-bit aligned. Note that
+	// the first word in an allocated struct can be relied upon to be 64-bit
+	// aligned. Refer to https://pkg.go.dev/sync/atomic#pkg-note-BUG. Also
+	// refer to discussion in https://github.com/etcd-io/bbolt/issues/577.
+	stats Stats
+
 	// When enabled, the database will perform a Check() after every commit.
 	// A panic is issued if the database is in an inconsistent state. This
 	// flag has a large performance impact so it should only be used for
@@ -147,7 +153,6 @@ type DB struct {
 	opened   bool
 	rwtx     *Tx
 	txs      []*Tx
-	stats    Stats
 
 	freelist     *freelist
 	freelistLoad sync.Once
@@ -424,7 +429,7 @@ func (db *DB) hasSyncedFreelist() bool {
 
 // mmap opens the underlying memory-mapped file and initializes the meta references.
 // minsz is the minimum size that the new mmap can be.
-func (db *DB) mmap(minsz int) error {
+func (db *DB) mmap(minsz int) (err error) {
 	db.mmaplock.Lock()
 	defer db.mmaplock.Unlock()
 
@@ -459,17 +464,27 @@ func (db *DB) mmap(minsz int) error {
 	}
 
 	// Unmap existing data before continuing.
-	if err := db.munmap(); err != nil {
+	if err = db.munmap(); err != nil {
 		return err
 	}
 
 	// Memory-map the data file as a byte slice.
 	// gofail: var mapError string
 	// return errors.New(mapError)
-	if err := mmap(db, size); err != nil {
+	if err = mmap(db, size); err != nil {
 		return err
 	}
 
+	// Perform unmmap on any error to reset all data fields:
+	// dataref, data, datasz, meta0 and meta1.
+	defer func() {
+		if err != nil {
+			if unmapErr := db.munmap(); unmapErr != nil {
+				err = fmt.Errorf("%w; rollback unmap also failed: %v", err, unmapErr)
+			}
+		}
+	}()
+
 	if db.Mlock {
 		// Don't allow swapping of data file
 		if err := db.mlock(fileSize); err != nil {
@@ -509,7 +524,7 @@ func (db *DB) munmap() error {
 	// gofail: var unmapError string
 	// return errors.New(unmapError)
 	if err := munmap(db); err != nil {
-		return fmt.Errorf("unmap error: " + err.Error())
+		return fmt.Errorf("unmap error: %v", err.Error())
 	}
 
 	return nil
@@ -553,15 +568,19 @@ func (db *DB) mmapSize(size int) (int, error) {
 }
 
 func (db *DB) munlock(fileSize int) error {
+	// gofail: var munlockError string
+	// return errors.New(munlockError)
 	if err := munlock(db, fileSize); err != nil {
-		return fmt.Errorf("munlock error: " + err.Error())
+		return fmt.Errorf("munlock error: %v", err.Error())
 	}
 	return nil
 }
 
 func (db *DB) mlock(fileSize int) error {
+	// gofail: var mlockError string
+	// return errors.New(mlockError)
 	if err := mlock(db, fileSize); err != nil {
-		return fmt.Errorf("mlock error: " + err.Error())
+		return fmt.Errorf("mlock error: %v", err.Error())
 	}
 	return nil
 }
@@ -649,9 +668,10 @@ func (db *DB) close() error {
 	// Clear ops.
 	db.ops.writeAt = nil
 
+	var errs []error
 	// Close the mmap.
 	if err := db.munmap(); err != nil {
-		return err
+		errs = append(errs, err)
 	}
 
 	// Close file handles.
@@ -660,18 +680,22 @@ func (db *DB) close() error {
 		if !db.readOnly {
 			// Unlock the file.
 			if err := funlock(db); err != nil {
-				return fmt.Errorf("bolt.Close(): funlock error: %w", err)
+				errs = append(errs, fmt.Errorf("bolt.Close(): funlock error: %w", err))
 			}
 		}
 
 		// Close the file descriptor.
 		if err := db.file.Close(); err != nil {
-			return fmt.Errorf("db file close: %s", err)
+			errs = append(errs, fmt.Errorf("db file close: %w", err))
 		}
 		db.file = nil
 	}
 
 	db.path = ""
+
+	if len(errs) > 0 {
+		return errs[0]
+	}
 	return nil
 }
 
@@ -1135,6 +1159,8 @@ func (db *DB) grow(sz int) error {
 	// https://github.com/boltdb/bolt/issues/284
 	if !db.NoGrowSync && !db.readOnly {
 		if runtime.GOOS != "windows" {
+			// gofail: var resizeFileError string
+			// return errors.New(resizeFileError)
 			if err := db.file.Truncate(int64(sz)); err != nil {
 				return fmt.Errorf("file resize error: %s", err)
 			}
@@ -1263,6 +1289,12 @@ var DefaultOptions = &Options{
 
 // Stats represents statistics about the database.
 type Stats struct {
+	// Put `TxStats` at the first field to ensure it's 64-bit aligned. Note
+	// that the first word in an allocated struct can be relied upon to be
+	// 64-bit aligned. Refer to https://pkg.go.dev/sync/atomic#pkg-note-BUG.
+	// Also refer to discussion in https://github.com/etcd-io/bbolt/issues/577.
+	TxStats TxStats // global, ongoing stats.
+
 	// Freelist stats
 	FreePageN     int // total number of free pages on the freelist
 	PendingPageN  int // total number of pending pages on the freelist
@@ -1272,8 +1304,6 @@ type Stats struct {
 	// Transaction stats
 	TxN     int // total number of started read transactions
 	OpenTxN int // number of currently open read transactions
-
-	TxStats TxStats // global, ongoing stats.
 }
 
 // Sub calculates and returns the difference between two sets of database stats.

vendor/go.etcd.io/bbolt/freelist.go

@@ -252,6 +252,14 @@ func (f *freelist) rollback(txid txid) {
 	}
 	// Remove pages from pending list and mark as free if allocated by txid.
 	delete(f.pending, txid)
+
+	// Remove pgids which are allocated by this txid
+	for pgid, tid := range f.allocs {
+		if tid == txid {
+			delete(f.allocs, pgid)
+		}
+	}
+
 	f.mergeSpans(m)
 }
 
@@ -282,9 +290,8 @@ func (f *freelist) read(p *page) {
 	if count == 0 {
 		f.ids = nil
 	} else {
-		var ids []pgid
-		data := unsafeIndex(unsafe.Pointer(p), unsafe.Sizeof(*p), unsafe.Sizeof(ids[0]), idx)
-		unsafeSlice(unsafe.Pointer(&ids), data, count)
+		data := unsafeIndex(unsafe.Pointer(p), unsafe.Sizeof(*p), unsafe.Sizeof(pgid(0)), idx)
+		ids := unsafe.Slice((*pgid)(data), count)
 
 		// copy the ids, so we don't modify on the freelist page directly
 		idsCopy := make([]pgid, count)
@@ -322,15 +329,13 @@ func (f *freelist) write(p *page) error {
 		p.count = uint16(l)
 	} else if l < 0xFFFF {
 		p.count = uint16(l)
-		var ids []pgid
 		data := unsafeAdd(unsafe.Pointer(p), unsafe.Sizeof(*p))
-		unsafeSlice(unsafe.Pointer(&ids), data, l)
+		ids := unsafe.Slice((*pgid)(data), l)
 		f.copyall(ids)
 	} else {
 		p.count = 0xFFFF
-		var ids []pgid
 		data := unsafeAdd(unsafe.Pointer(p), unsafe.Sizeof(*p))
-		unsafeSlice(unsafe.Pointer(&ids), data, l+1)
+		ids := unsafe.Slice((*pgid)(data), l+1)
 		ids[0] = pgid(l)
 		f.copyall(ids[1:])
 	}

vendor/go.etcd.io/bbolt/page.go

@@ -74,9 +74,8 @@ func (p *page) leafPageElements() []leafPageElement {
 	if p.count == 0 {
 		return nil
 	}
-	var elems []leafPageElement
 	data := unsafeAdd(unsafe.Pointer(p), unsafe.Sizeof(*p))
-	unsafeSlice(unsafe.Pointer(&elems), data, int(p.count))
+	elems := unsafe.Slice((*leafPageElement)(data), int(p.count))
 	return elems
 }
 
@@ -91,9 +90,8 @@ func (p *page) branchPageElements() []branchPageElement {
 	if p.count == 0 {
 		return nil
 	}
-	var elems []branchPageElement
 	data := unsafeAdd(unsafe.Pointer(p), unsafe.Sizeof(*p))
-	unsafeSlice(unsafe.Pointer(&elems), data, int(p.count))
+	elems := unsafe.Slice((*branchPageElement)(data), int(p.count))
 	return elems
 }
 

vendor/go.etcd.io/bbolt/tx.go

@@ -1,6 +1,7 @@
 package bbolt
 
 import (
+	"errors"
 	"fmt"
 	"io"
 	"os"
@@ -185,6 +186,10 @@ func (tx *Tx) Commit() error {
 
 	// If the high water mark has moved up then attempt to grow the database.
 	if tx.meta.pgid > opgid {
+		_ = errors.New("")
+		// gofail: var lackOfDiskSpace string
+		// tx.rollback()
+		// return errors.New(lackOfDiskSpace)
 		if err := tx.db.grow(int(tx.meta.pgid+1) * tx.db.pageSize); err != nil {
 			tx.rollback()
 			return err
@@ -470,6 +475,7 @@ func (tx *Tx) write() error {
 
 	// Ignore file sync if flag is set on DB.
 	if !tx.db.NoSync || IgnoreNoSync {
+		// gofail: var beforeSyncDataPages struct{}
 		if err := fdatasync(tx.db); err != nil {
 			return err
 		}
@@ -507,6 +513,7 @@ func (tx *Tx) writeMeta() error {
 		return err
 	}
 	if !tx.db.NoSync || IgnoreNoSync {
+		// gofail: var beforeSyncMetaPage struct{}
 		if err := fdatasync(tx.db); err != nil {
 			return err
 		}

vendor/go.etcd.io/bbolt/unsafe.go

@@ -1,7 +1,6 @@
 package bbolt
 
 import (
-	"reflect"
 	"unsafe"
 )
 
@@ -26,14 +25,3 @@ func unsafeByteSlice(base unsafe.Pointer, offset uintptr, i, j int) []byte {
 	// all), so this is believed to be correct.
 	return (*[maxAllocSize]byte)(unsafeAdd(base, offset))[i:j:j]
 }
-
-// unsafeSlice modifies the data, len, and cap of a slice variable pointed to by
-// the slice parameter.  This helper should be used over other direct
-// manipulation of reflect.SliceHeader to prevent misuse, namely, converting
-// from reflect.SliceHeader to a Go slice type.
-func unsafeSlice(slice, data unsafe.Pointer, len int) {
-	s := (*reflect.SliceHeader)(slice)
-	s.Data = uintptr(data)
-	s.Cap = len
-	s.Len = len
-}