xfs: remove OWN_AG rmap when allocating a block from the AGFL
authorDarrick J. Wong <darrick.wong@oracle.com>
Wed, 17 Aug 2016 01:12:57 +0000 (11:12 +1000)
committerDave Chinner <david@fromorbit.com>
Wed, 17 Aug 2016 01:12:57 +0000 (11:12 +1000)
When we're really tight on space, xfs_alloc_ag_vextent_small() can
allocate a block from the AGFL and give it to the caller.  Since the
caller is never the AGFL-fixing method, we must remove the OWN_AG
reverse mapping because it will clash with whatever rmap the caller
wants to set up.  This bug was discovered by running generic/299
repeatedly.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
fs/xfs/libxfs/xfs_alloc.c

index af1a72efa64cf79f6696041a7fa409e01893aeb1..3dd8f1d5449877ed45aa0fb391c1e9048c5c1893 100644 (file)
@@ -1582,6 +1582,7 @@ xfs_alloc_ag_vextent_small(
        xfs_extlen_t    *flenp, /* result length */
        int             *stat)  /* status: 0-freelist, 1-normal/none */
 {
+       struct xfs_owner_info   oinfo;
        int             error;
        xfs_agblock_t   fbno;
        xfs_extlen_t    flen;
@@ -1624,6 +1625,18 @@ xfs_alloc_ag_vextent_small(
                                error0);
                        args->wasfromfl = 1;
                        trace_xfs_alloc_small_freelist(args);
+
+                       /*
+                        * If we're feeding an AGFL block to something that
+                        * doesn't live in the free space, we need to clear
+                        * out the OWN_AG rmap.
+                        */
+                       xfs_rmap_ag_owner(&oinfo, XFS_RMAP_OWN_AG);
+                       error = xfs_rmap_free(args->tp, args->agbp, args->agno,
+                                       fbno, 1, &oinfo);
+                       if (error)
+                               goto error0;
+
                        *stat = 0;
                        return 0;
                }