diff mbox series

[3/3] xfs: scrub should mark a directory corrupt if any entries cannot be iget'd

Message ID 160704438289.736504.15952269053640029711.stgit@magnolia (mailing list archive)
State Accepted
Headers show
Series xfs: random fixes for 5.11 | expand

Commit Message

Darrick J. Wong Dec. 4, 2020, 1:13 a.m. UTC 
From: Darrick J. Wong <darrick.wong@oracle.com>

It's possible that xfs_iget can return EINVAL for inodes that the inobt
thinks are free, or ENOENT for inodes that look free.  If this is the
case, mark the directory corrupt immediately when we check ftype.  Note
that we already check the ftype of the '.' and '..' entries, so we
can skip the iget part since we already know the inode type for '.' and
we have a separate parent pointer scrubber for '..'.

Fixes: a5c46e5e8912 ("xfs: scrub directory metadata")
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/scrub/dir.c |   21 ++++++++++++++++++---
 1 file changed, 18 insertions(+), 3 deletions(-)

Comments

Christoph Hellwig Dec. 4, 2020, 10:11 a.m. UTC | #1 
On Thu, Dec 03, 2020 at 05:13:02PM -0800, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> It's possible that xfs_iget can return EINVAL for inodes that the inobt
> thinks are free, or ENOENT for inodes that look free.  If this is the
> case, mark the directory corrupt immediately when we check ftype.  Note
> that we already check the ftype of the '.' and '..' entries, so we
> can skip the iget part since we already know the inode type for '.' and
> we have a separate parent pointer scrubber for '..'.

Looks good,

Reviewed-by: Christoph Hellwig <hch@lst.de>

But I wonder if we need the EINVAL vs ENOENT distinction to start
with or if we could return a single coherent error code from iget.
diff mbox series

Patch

diff --git a/fs/xfs/scrub/dir.c b/fs/xfs/scrub/dir.c
index b045e95c2ea7..178b3455a170 100644
--- a/fs/xfs/scrub/dir.c
+++ b/fs/xfs/scrub/dir.c
@@ -66,8 +66,18 @@  xchk_dir_check_ftype(
 	 * eofblocks cleanup (which allocates what would be a nested
 	 * transaction), we can't use DONTCACHE here because DONTCACHE
 	 * inodes can trigger immediate inactive cleanup of the inode.
+	 *
+	 * If _iget returns -EINVAL or -ENOENT then the child inode number is
+	 * garbage and the directory is corrupt.  If the _iget returns
+	 * -EFSCORRUPTED or -EFSBADCRC then the child is corrupt which is a
+	 *  cross referencing error.  Any other error is an operational error.
 	 */
 	error = xfs_iget(mp, sdc->sc->tp, inum, 0, 0, &ip);
+	if (error == -EINVAL || error == -ENOENT) {
+		error = -EFSCORRUPTED;
+		xchk_fblock_process_error(sdc->sc, XFS_DATA_FORK, 0, &error);
+		goto out;
+	}
 	if (!xchk_fblock_xref_process_error(sdc->sc, XFS_DATA_FORK, offset,
 			&error))
 		goto out;
@@ -105,6 +115,7 @@  xchk_dir_actor(
 	struct xfs_name		xname;
 	xfs_ino_t		lookup_ino;
 	xfs_dablk_t		offset;
+	bool			checked_ftype = false;
 	int			error = 0;
 
 	sdc = container_of(dir_iter, struct xchk_dir_ctx, dir_iter);
@@ -133,6 +144,7 @@  xchk_dir_actor(
 		if (xfs_sb_version_hasftype(&mp->m_sb) && type != DT_DIR)
 			xchk_fblock_set_corrupt(sdc->sc, XFS_DATA_FORK,
 					offset);
+		checked_ftype = true;
 		if (ino != ip->i_ino)
 			xchk_fblock_set_corrupt(sdc->sc, XFS_DATA_FORK,
 					offset);
@@ -144,6 +156,7 @@  xchk_dir_actor(
 		if (xfs_sb_version_hasftype(&mp->m_sb) && type != DT_DIR)
 			xchk_fblock_set_corrupt(sdc->sc, XFS_DATA_FORK,
 					offset);
+		checked_ftype = true;
 		if (ip->i_ino == mp->m_sb.sb_rootino && ino != ip->i_ino)
 			xchk_fblock_set_corrupt(sdc->sc, XFS_DATA_FORK,
 					offset);
@@ -167,9 +180,11 @@  xchk_dir_actor(
 	}
 
 	/* Verify the file type.  This function absorbs error codes. */
-	error = xchk_dir_check_ftype(sdc, offset, lookup_ino, type);
-	if (error)
-		goto out;
+	if (!checked_ftype) {
+		error = xchk_dir_check_ftype(sdc, offset, lookup_ino, type);
+		if (error)
+			goto out;
+	}
 out:
 	/*
 	 * A negative error code returned here is supposed to cause the