Merge branch 'xiang90-tx_write'

README.md

@@ -426,8 +426,11 @@ func (*Bucket) DeleteBucket(key []byte) error
 Bolt is a single file so it's easy to backup. You can use the `Tx.WriteTo()`
 function to write a consistent view of the database to a writer. If you call
 this from a read-only transaction, it will perform a hot backup and not block
-your other database reads and writes. It will also use `O_DIRECT` when available
-to prevent page cache trashing.
+your other database reads and writes.
+
+By default, it will use a regular file handle which will utilize the operating
+system's page cache. See the [`Tx`](https://godoc.org/github.com/boltdb/bolt#Tx)
+documentation for information about optimizing for larger-than-RAM datasets. 
 
 One common use case is to backup over HTTP so you can use tools like `cURL` to
 do database backups:

bolt_linux.go

@@ -4,8 +4,6 @@ import (
 	"syscall"
 )
 
-var odirect = syscall.O_DIRECT
-
 // fdatasync flushes written data to a file descriptor.
 func fdatasync(db *DB) error {
 	return syscall.Fdatasync(int(db.file.Fd()))

bolt_openbsd.go

@@ -11,8 +11,6 @@ const (
 	msInvalidate             // invalidate cached data
 )
 
-var odirect int
-
 func msync(db *DB) error {
 	_, _, errno := syscall.Syscall(syscall.SYS_MSYNC, uintptr(unsafe.Pointer(db.data)), uintptr(db.datasz), msInvalidate)
 	if errno != 0 {

bolt_windows.go

@@ -40,8 +40,6 @@ func unlockFileEx(h syscall.Handle, reserved, locklow, lockhigh uint32, ol *sysc
 	return nil
 }
 
-var odirect int
-
 // fdatasync flushes written data to a file descriptor.
 func fdatasync(db *DB) error {
 	return db.file.Sync()

boltsync_unix.go

@@ -2,8 +2,6 @@
 
 package bolt
 
-var odirect int
-
 // fdatasync flushes written data to a file descriptor.
 func fdatasync(db *DB) error {
 	return db.file.Sync()

tx.go

@@ -29,6 +29,14 @@ type Tx struct {
 	pages          map[pgid]*page
 	stats          TxStats
 	commitHandlers []func()
+
+	// WriteFlag specifies the flag for write-related methods like WriteTo().
+	// Tx opens the database file with the specified flag to copy the data.
+	//
+	// By default, the flag is unset, which works well for mostly in-memory
+	// workloads. For databases that are much larger than available RAM,
+	// set the flag to syscall.O_DIRECT to avoid trashing the page cache.
+	WriteFlag int
 }
 
 // init initializes the transaction.
@@ -272,21 +280,18 @@ func (tx *Tx) Copy(w io.Writer) error {
 // WriteTo writes the entire database to a writer.
 // If err == nil then exactly tx.Size() bytes will be written into the writer.
 func (tx *Tx) WriteTo(w io.Writer) (n int64, err error) {
-	// Attempt to open reader directly.
-	var f *os.File
-	if f, err = os.OpenFile(tx.db.path, os.O_RDONLY|odirect, 0); err != nil {
-		// Fallback to a regular open if that doesn't work.
-		if f, err = os.OpenFile(tx.db.path, os.O_RDONLY, 0); err != nil {
+	// Attempt to open reader with WriteFlag
+	f, err := os.OpenFile(tx.db.path, os.O_RDONLY|tx.WriteFlag, 0)
+	if err != nil {
 		return 0, err
 	}
-	}
+	defer f.Close()
 
 	// Copy the meta pages.
 	tx.db.metalock.Lock()
 	n, err = io.CopyN(w, f, int64(tx.db.pageSize*2))
 	tx.db.metalock.Unlock()
 	if err != nil {
-		_ = f.Close()
 		return n, fmt.Errorf("meta copy: %s", err)
 	}
 
@@ -294,7 +299,6 @@ func (tx *Tx) WriteTo(w io.Writer) (n int64, err error) {
 	wn, err := io.CopyN(w, f, tx.Size()-int64(tx.db.pageSize*2))
 	n += wn
 	if err != nil {
-		_ = f.Close()
 		return n, err
 	}