ocfs2: Use buffer IO if we are appending a file.
authorTao Ma <tao.ma@oracle.com>
Thu, 10 Sep 2009 07:28:47 +0000 (15:28 +0800)
committerJoel Becker <joel.becker@oracle.com>
Wed, 23 Sep 2009 08:54:49 +0000 (01:54 -0700)
In ocfs2_file_aio_write, we will prevent direct io if
we find that we are appending(changing i_size) and call
generic_file_aio_write_nolock. But actually O_DIRECT flag
is there and this function will call generic_file_direct_write
eventually which will update i_size and leave di->i_size
alone. The bug is
http://oss.oracle.com/bugzilla/show_bug.cgi?id=1173.

So this patch let ocfs2_direct_IO returns 0 directly if we
are appending so that buffered write will be called and
di->i_size get updated successfully. And this is also
what we want in ocfs2_file_aio_write.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
fs/ocfs2/aops.c

index 33e03c5511274756fecc13cf4a61a3b476790cc3..72e76062a900d2555fac2dae44ef325a5d8aa897 100644 (file)
@@ -690,6 +690,10 @@ static ssize_t ocfs2_direct_IO(int rw,
        if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL)
                return 0;
 
+       /* Fallback to buffered I/O if we are appending. */
+       if (i_size_read(inode) <= offset)
+               return 0;
+
        ret = blockdev_direct_IO_no_locking(rw, iocb, inode,
                                            inode->i_sb->s_bdev, iov, offset,
                                            nr_segs,