In ocfs2, actually we don't allow any direct write pass i_size,
see the function ocfs2_prepare_inode_for_write. So we don't
need the bogus simple_setsize.
Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
written = generic_file_direct_write(iocb, iov, &nr_segs, *ppos,
ppos, count, ocount);
if (written < 0) {
- /*
- * direct write may have instantiated a few
- * blocks outside i_size. Trim these off again.
- * Don't need i_size_read because we hold i_mutex.
- *
- * XXX(truncate): this looks buggy because ocfs2 did not
- * actually implement ->truncate. Take a look at
- * the new truncate sequence and update this accordingly
- */
- if (*ppos + count > inode->i_size)
- truncate_setsize(inode, inode->i_size);
ret = written;
goto out_dio;
}