]> asedeno.scripts.mit.edu Git - linux.git/commitdiff
xfs: check the ir_startino alignment directly
authorDarrick J. Wong <darrick.wong@oracle.com>
Fri, 1 Feb 2019 17:08:50 +0000 (09:08 -0800)
committerDarrick J. Wong <darrick.wong@oracle.com>
Tue, 12 Feb 2019 00:06:38 +0000 (16:06 -0800)
In xchk_iallocbt_rec, check the alignment of ir_startino by converting
the inode cluster block alignment into units of inodes instead of the
other way around (converting ir_startino to blocks).  This prevents us
from tripping over off-by-one errors in ir_startino which are obscured
by the inode -> block conversion.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
fs/xfs/scrub/ialloc.c

index fd431682db0b68e5dc2c0233d09980eca348447b..1c6fef9b3799f877b6223917037b71a0f311f32b 100644 (file)
@@ -265,6 +265,53 @@ xchk_iallocbt_check_freemask(
        return error;
 }
 
+/*
+ * Make sure this inode btree record is aligned properly.  Because a fs block
+ * contains multiple inodes, we check that the inobt record is aligned to the
+ * correct inode, not just the correct block on disk.  This results in a finer
+ * grained corruption check.
+ */
+STATIC void
+xchk_iallocbt_rec_alignment(
+       struct xchk_btree               *bs,
+       struct xfs_inobt_rec_incore     *irec)
+{
+       struct xfs_mount                *mp = bs->sc->mp;
+
+       /*
+        * finobt records have different positioning requirements than inobt
+        * records: each finobt record must have a corresponding inobt record.
+        * That is checked in the xref function, so for now we only catch the
+        * obvious case where the record isn't at all aligned properly.
+        *
+        * Note that if a fs block contains more than a single chunk of inodes,
+        * we will have finobt records only for those chunks containing free
+        * inodes, and therefore expect chunk alignment of finobt records.
+        * Otherwise, we expect that the finobt record is aligned to the
+        * cluster alignment as told by the superblock.
+        */
+       if (bs->cur->bc_btnum == XFS_BTNUM_FINO) {
+               unsigned int    imask;
+
+               imask = min_t(unsigned int, XFS_INODES_PER_CHUNK,
+                               mp->m_cluster_align_inodes) - 1;
+               if (irec->ir_startino & imask)
+                       xchk_btree_set_corrupt(bs->sc, bs->cur, 0);
+               return;
+       }
+
+       /* inobt records must be aligned to cluster and inoalignmnt size. */
+       if (irec->ir_startino & (mp->m_cluster_align_inodes - 1)) {
+               xchk_btree_set_corrupt(bs->sc, bs->cur, 0);
+               return;
+       }
+
+       if (irec->ir_startino & (mp->m_inodes_per_cluster - 1)) {
+               xchk_btree_set_corrupt(bs->sc, bs->cur, 0);
+               return;
+       }
+}
+
 /* Scrub an inobt/finobt record. */
 STATIC int
 xchk_iallocbt_rec(
@@ -277,7 +324,6 @@ xchk_iallocbt_rec(
        uint64_t                        holes;
        xfs_agnumber_t                  agno = bs->cur->bc_private.a.agno;
        xfs_agino_t                     agino;
-       xfs_agblock_t                   agbno;
        xfs_extlen_t                    len;
        int                             holecount;
        int                             i;
@@ -304,11 +350,9 @@ xchk_iallocbt_rec(
                goto out;
        }
 
-       /* Make sure this record is aligned to cluster and inoalignmnt size. */
-       agbno = XFS_AGINO_TO_AGBNO(mp, irec.ir_startino);
-       if ((agbno & (mp->m_cluster_align - 1)) ||
-           (agbno & (mp->m_blocks_per_cluster - 1)))
-               xchk_btree_set_corrupt(bs->sc, bs->cur, 0);
+       xchk_iallocbt_rec_alignment(bs, &irec);
+       if (bs->sc->sm->sm_flags & XFS_SCRUB_OFLAG_CORRUPT)
+               goto out;
 
        iabt->inodes += irec.ir_count;