Skip to contentSkip to conversation

xfs_repair: don't fail on INCOMPLETE attrs in leaf blocks

PATCH Superseded Darrick J. Wong 2 participants 5 messages Mar 16 – Mar 18

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 xfs_repair: don't fail on INCOMPLETE attrs in leaf blocks

From: Darrick J. Wong <[email protected]>

It's far too drastic to delete the entire attr fork because doing that destroys things like security labels. The kernel won't show incomplete attrs so it's not a big deal.

---
 repair/attr_repair.c |   22 +++++++++++++++++-----
 1 file changed, 17 insertions(+), 5 deletions(-)
repair/attr_repair.c +17 -5
index 50159b9a533875..fe4089026cae74 100644
@@ -571,7 +571,13 @@ process_leaf_attr_remote(
 	    !libxfs_attr_namecheck(entry->flags, remotep->name,
 				   remotep->namelen) ||
 	    be32_to_cpu(entry->hashval) != computed ||
-	    be32_to_cpu(entry->hashval) < last_hashval ||
+	    be32_to_cpu(entry->hashval) < last_hashval) {
+		do_warn(
+	_("inconsistent remote attribute entry %d in attr block %u, ino %" PRIu64 "\n"), i, da_bno, ino);
+		return -1;
+	}
+
+	if (!(entry->flags & XFS_ATTR_INCOMPLETE) &&
 	    be32_to_cpu(remotep->valueblk) == 0) {
 		do_warn(
 	_("inconsistent remote attribute entry %d in attr block %u, ino %" PRIu64 "\n"), i, da_bno, ino);
@@ -592,6 +598,9 @@ process_leaf_attr_remote(
 		return -1;
 	}
 
+	if (entry->flags & XFS_ATTR_INCOMPLETE)
+		goto out;
+
 	value = malloc(be32_to_cpu(remotep->valuelen));
 	if (value == NULL) {
 		do_warn(
@@ -708,12 +717,15 @@ process_leaf_attr_block(
 		}
 
 		if (entry->flags & XFS_ATTR_INCOMPLETE) {
-			/* we are inconsistent state. get rid of us */
-			do_warn(
+			/*
+			 * Warn about incomplete xattrs but don't zap the
+			 * entire attr fork because that causes loss of
+			 * security labels.  The kernel can handle stray
+			 * incomplete attr entries.
+			 */
+			do_log(
 	_("attribute entry #%d in attr block %u, inode %" PRIu64 " is INCOMPLETE\n"),
 				i, da_bno, ino);
-			clearit = 1;
-			break;
 		}
 
 		/* mark the entry used */
#1 lore raw
Quoted 5 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

It's far too drastic to delete the entire attr fork because doing that destroys things like security labels. The kernel won't show incomplete attrs so it's not a big deal.

This sounds a bit lax. Incomplete attributes are very much expected given how the non-logged attr code works. Maybe reword this a bit to not make it sound like we just ignore a problem?

#2 lore raw
Quoted 10 lines · Christoph Hellwig
Quoted 5 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

It's far too drastic to delete the entire attr fork because doing that destroys things like security labels. The kernel won't show incomplete attrs so it's not a big deal.

This sounds a bit lax. Incomplete attributes are very much expected given how the non-logged attr code works. Maybe reword this a bit to not make it sound like we just ignore a problem?

Ok will do.

--D

#3 lore raw

PATCH v2 1/1 xfs_repair: don't fail on INCOMPLETE attrs in leaf blocks

In reply to Darrick J. Wong

From: Darrick J. Wong <[email protected]>

While trying to fix problems in generic/753, I noticed test failures on account of xfs_repair:

attribute entry #4 in attr block 0, inode 131 is INCOMPLETE problem with attribute contents in inode 131 would clear attr fork
bad nblocks 4 for inode 131, would reset to 0
bad anextents 1 for inode 131, would reset to 0

Looking at the dumped filesystem, inode 131 is a linked file, and the "incomplete" xattr was clearly part of an xfs_attr_set operation that failed midway through because the induced log shutdown prevented xfs from finishing the creation of a remote xattr. This kind of thing is expected, but instead xfs_repair deletes the entire attr fork!

It's far too drastic to delete every xattr because doing that destroys things like security labels. The kernel won't show incomplete attrs so it's not a big deal to leave them attached to the file. Note that xfs_scrub can fix such things.

---
v2: improve commit message to explain how we got to this acceptable place

---
 repair/attr_repair.c |   22 +++++++++++++++++-----
 1 file changed, 17 insertions(+), 5 deletions(-)
repair/attr_repair.c +17 -5
index 50159b9a533875..fe4089026cae74 100644
@@ -571,7 +571,13 @@ process_leaf_attr_remote(
 	    !libxfs_attr_namecheck(entry->flags, remotep->name,
 				   remotep->namelen) ||
 	    be32_to_cpu(entry->hashval) != computed ||
-	    be32_to_cpu(entry->hashval) < last_hashval ||
+	    be32_to_cpu(entry->hashval) < last_hashval) {
+		do_warn(
+	_("inconsistent remote attribute entry %d in attr block %u, ino %" PRIu64 "\n"), i, da_bno, ino);
+		return -1;
+	}
+
+	if (!(entry->flags & XFS_ATTR_INCOMPLETE) &&
 	    be32_to_cpu(remotep->valueblk) == 0) {
 		do_warn(
 	_("inconsistent remote attribute entry %d in attr block %u, ino %" PRIu64 "\n"), i, da_bno, ino);
@@ -592,6 +598,9 @@ process_leaf_attr_remote(
 		return -1;
 	}
 
+	if (entry->flags & XFS_ATTR_INCOMPLETE)
+		goto out;
+
 	value = malloc(be32_to_cpu(remotep->valuelen));
 	if (value == NULL) {
 		do_warn(
@@ -708,12 +717,15 @@ process_leaf_attr_block(
 		}
 
 		if (entry->flags & XFS_ATTR_INCOMPLETE) {
-			/* we are inconsistent state. get rid of us */
-			do_warn(
+			/*
+			 * Warn about incomplete xattrs but don't zap the
+			 * entire attr fork because that causes loss of
+			 * security labels.  The kernel can handle stray
+			 * incomplete attr entries.
+			 */
+			do_log(
 	_("attribute entry #%d in attr block %u, inode %" PRIu64 " is INCOMPLETE\n"),
 				i, da_bno, ino);
-			clearit = 1;
-			break;
 		}
 
 		/* mark the entry used */