xfs: switch to NOFS allocation under i_lock in xfs_attr_rmtval_get
authorChristoph Hellwig <hch@infradead.org>
Sat, 18 Jul 2009 22:15:01 +0000 (18:15 -0400)
committerFelix Blyakher <felixb@sgi.com>
Wed, 12 Aug 2009 06:08:01 +0000 (01:08 -0500)
xfs_attr_rmtval_get is always called with i_lock held, but i_lock is taken
in reclaim context so all allocations under it must avoid recursions into
the filesystem.

Reported by the new reclaim context tracing in lockdep.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Felix Blyakher <felixb@sgi.com>
Signed-off-by: Felix Blyakher <felixb@sgi.com>
fs/xfs/xfs_attr.c

index bfb583791fe2585e92541f4c60e85245d59d9862..4ece1906bd41c0d801f8b66a41efb48bd3bcad6d 100644 (file)
@@ -2010,7 +2010,9 @@ xfs_attr_rmtval_get(xfs_da_args_t *args)
                        dblkno = XFS_FSB_TO_DADDR(mp, map[i].br_startblock);
                        blkcnt = XFS_FSB_TO_BB(mp, map[i].br_blockcount);
                        error = xfs_read_buf(mp, mp->m_ddev_targp, dblkno,
-                                            blkcnt, XFS_BUF_LOCK, &bp);
+                                            blkcnt,
+                                            XFS_BUF_LOCK | XBF_DONT_BLOCK,
+                                            &bp);
                        if (error)
                                return(error);