On Tue, Oct 30, 2018 at 10:20:38PM +1100, Dave Chinner wrote:
> From: Dave Chinner <dchinner@xxxxxxxxxx>
>
> I noticed phase 4 writing back lots of inode buffers during recent
> testing. The recent rework of clear_inode() in commit 0724d0f4cb53
> ("xfs_repair: clear_dinode should simply clear, not check contents")
> accidentally caught a call to clear_inode_unlinked() as well,
> resulting in all inodes being marked dirty whether then needed
> updating or not.
>
> Fix it by reverting the erroneous hunk and adding warnings so taht
> this corruption is no longer silently fixed.
>
> Signed-off-by: Dave Chinner <dchinner@xxxxxxxxxx>
Oops... that explains a lot. :/
Reviewed-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
--D
> ---
> repair/dinode.c | 12 +++++++++---
> 1 file changed, 9 insertions(+), 3 deletions(-)
>
> diff --git a/repair/dinode.c b/repair/dinode.c
> index 379f85cf1268..90400128d4bb 100644
> --- a/repair/dinode.c
> +++ b/repair/dinode.c
> @@ -2675,9 +2675,15 @@ _("bad (negative) size %" PRId64 " on inode %" PRIu64 "\n"),
> * we're going to find. check_dups is set to 1 only during
> * phase 4. Ugly.
> */
> - if (check_dups && !no_modify) {
> - clear_dinode_unlinked(mp, dino);
> - *dirty += 1;
> + if (check_dups && clear_dinode_unlinked(mp, dino)) {
> + if (no_modify) {
> + do_warn(
> + _("Would clear unlinked_next in inode %" PRIu64 "\n"), lino);
> + } else {
> + do_warn(
> + _("Cleared unlinked_next in inode %" PRIu64 "\n"), lino);
> + *dirty += 1;
> + }
> }
>
> /* set type and map type info */
> --
> 2.19.1
>
