Speed up O_APPEND with remote revalidating

Remote revalidating requires to update file size on every write
on a file opened with O_APPEND. If host FD exists, it can be
used to update the size and skip round trip to the gofer. With
this change, O_APPEND writes with remote revalidating is almost
as fast as exclusive mode:

              BM_Append
VFS1           60.7us
VFS2           56.8us
VFS2 exclusive 14.2us
This change    15.8us

Updates #1792

PiperOrigin-RevId: 369486801

1 files changed, 5 insertions, 4 deletions

diff --git a/pkg/sentry/fsimpl/gofer/regular_file.go b/pkg/sentry/fsimpl/gofer/regular_file.go
index 713f0a480..f0e7bbaf7 100644
--- a/pkg/sentry/fsimpl/gofer/regular_file.go
+++ b/pkg/sentry/fsimpl/gofer/regular_file.go
@@ -204,18 +204,19 @@ func (fd *regularFileFD) pwrite(ctx context.Context, src usermem.IOSequence, off
 	}
 
 	d := fd.dentry()
+
+	d.metadataMu.Lock()
+	defer d.metadataMu.Unlock()
+
 	// If the fd was opened with O_APPEND, make sure the file size is updated.
 	// There is a possible race here if size is modified externally after
 	// metadata cache is updated.
 	if fd.vfsfd.StatusFlags()&linux.O_APPEND != 0 && !d.cachedMetadataAuthoritative() {
-		if err := d.updateFromGetattr(ctx); err != nil {
+		if err := d.refreshSizeLocked(ctx); err != nil {
 			return 0, offset, err
 		}
 	}
 
-	d.metadataMu.Lock()
-	defer d.metadataMu.Unlock()
-
 	// Set offset to file size if the fd was opened with O_APPEND.
 	if fd.vfsfd.StatusFlags()&linux.O_APPEND != 0 {
 		// Holding d.metadataMu is sufficient for reading d.size.