xfs: prohibit fstrim in norecovery mode
authorDarrick J. Wong <darrick.wong@oracle.com>
Sat, 23 Mar 2019 01:10:22 +0000 (18:10 -0700)
committerDarrick J. Wong <darrick.wong@oracle.com>
Mon, 25 Mar 2019 15:03:29 +0000 (08:03 -0700)
The xfs fstrim implementation uses the free space btrees to find free
space that can be discarded.  If we haven't recovered the log, the bnobt
will be stale and we absolutely *cannot* use stale metadata to zap the
underlying storage.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
fs/xfs/xfs_discard.c

index 93f07edafd8183a14ca55fae7ffdfad0370ccc89..9ee2a7d02e7059f29c103da1088b7c854ca023bb 100644 (file)
@@ -161,6 +161,14 @@ xfs_ioc_trim(
                return -EPERM;
        if (!blk_queue_discard(q))
                return -EOPNOTSUPP;
+
+       /*
+        * We haven't recovered the log, so we cannot use our bnobt-guided
+        * storage zapping commands.
+        */
+       if (mp->m_flags & XFS_MOUNT_NORECOVERY)
+               return -EROFS;
+
        if (copy_from_user(&range, urange, sizeof(range)))
                return -EFAULT;