Merge tag 'xfs-4.17-merge-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux

Pull xfs updates from Darrick Wong:
 "Here's the first round of fixes for XFS for 4.17.

  The biggest new features this time around are the addition of lazytime
  support, further enhancement of the on-disk inode metadata verifiers,
  and a patch to smooth over some of the AGFL padding problems that have
  intermittently plagued users since 4.5. I forsee sending a second pull
  request next week with further bug fixes and speedups in the online
  scrub code and elsewhere.

  This series has been run through a full xfstests run over the weekend
  and through a quick xfstests run against this morning's master, with
  no major failures reported.

  Summary of changes for this release:

   - Various cleanups and code fixes

   - Implement lazytime as a mount option

   - Convert various on-disk metadata checks from asserts to -EFSCORRUPTED

   - Fix accounting problems with the rmap per-ag reservations

   - Refactorings and cleanups for xfs_log_force

   - Various bugfixes for the reflink code

   - Work around v5 AGFL padding problems to prevent fs shutdowns

   - Establish inode fork verifiers to inspect on-disk metadata
     correctness

   - Various online scrub fixes

   - Fix v5 swapext blowing up on deleted inodes"

* tag 'xfs-4.17-merge-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux: (49 commits)
  xfs: do not log/recover swapext extent owner changes for deleted inodes
  xfs: clean up xfs_mount allocation and dynamic initializers
  xfs: remove dead inode version setting code
  xfs: catch inode allocation state mismatch corruption
  xfs: xfs_scrub_iallocbt_xref_rmap_inodes should use xref_set_corrupt
  xfs: flag inode corruption if parent ptr doesn't get us a real inode
  xfs: don't accept inode buffers with suspicious unlinked chains
  xfs: move inode extent size hint validation to libxfs
  xfs: record inode buf errors as a xref error in inobt scrubber
  xfs: remove xfs_buf parameter from inode scrub methods
  xfs: inode scrubber shouldn't bother with raw checks
  xfs: bmap scrubber should do rmap xref with bmap for sparse files
  xfs: refactor inode buffer verifier error logging
  xfs: refactor inode verifier error logging
  xfs: refactor bmap record validation
  xfs: sanity-check the unused space before trying to use it
  xfs: detect agfl count corruption and reset agfl
  xfs: unwind the try_again loop in xfs_log_force
  xfs: refactor xfs_log_force_lsn
  xfs: minor cleanup for xfs_reflink_end_cow
  ...

1 files changed, 11 insertions, 1 deletions

diff --git a/fs/xfs/scrub/parent.c b/fs/xfs/scrub/parent.c
index 0d3851410c74..1fb88c18d455 100644
--- a/fs/xfs/scrub/parent.c
+++ b/fs/xfs/scrub/parent.c
@@ -167,8 +167,18 @@ xfs_scrub_parent_validate(
 	 * if the parent pointer erroneously points to a file, we
 	 * can't use DONTCACHE here because DONTCACHE inodes can trigger
 	 * immediate inactive cleanup of the inode.
+	 *
+	 * If _iget returns -EINVAL then the parent inode number is garbage
+	 * and the directory is corrupt.  If the _iget returns -EFSCORRUPTED
+	 * or -EFSBADCRC then the parent is corrupt which is a cross
+	 * referencing error.  Any other error is an operational error.
 	 */
-	error = xfs_iget(mp, sc->tp, dnum, 0, 0, &dp);
+	error = xfs_iget(mp, sc->tp, dnum, XFS_IGET_UNTRUSTED, 0, &dp);
+	if (error == -EINVAL) {
+		error = -EFSCORRUPTED;
+		xfs_scrub_fblock_process_error(sc, XFS_DATA_FORK, 0, &error);
+		goto out;
+	}
 	if (!xfs_scrub_fblock_xref_process_error(sc, XFS_DATA_FORK, 0, &error))
 		goto out;
 	if (dp == sc->ip || !S_ISDIR(VFS_I(dp)->i_mode)) {