ext4: fix fdatasync(2) after extent manipulation operations

commit 67a7d5f5 upstream.

Currently, extent manipulation operations such as hole punch, range
zeroing, or extent shifting do not record the fact that file data has
changed and thus fdatasync(2) has a work to do. As a result if we crash
e.g. after a punch hole and fdatasync, user can still possibly see the
punched out data after journal replay. Test generic/392 fails due to
these problems.

Fix the problem by properly marking that file data has changed in these
operations.

Fixes: a4bb6b64Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
[bwh: Backported to 3.2: Only the punch-hole operation is supported, and
 it's in extents.c.]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

fs/ext4/extents.c

@@ -5023,6 +5023,8 @@ int ext4_ext_punch_hole(struct file *file, loff_t offset, loff_t length)
 	ext4_orphan_del(handle, inode);
 	inode->i_mtime = inode->i_ctime = ext4_current_time(inode);
 	ext4_mark_inode_dirty(handle, inode);
+	if (err >= 0)
+		ext4_update_inode_fsync_trans(handle, inode, 1);
 	ext4_journal_stop(handle);
 	return err;
 }