Skip to contentSkip to conversation

btrfs-progs: check/lowmem: Reset path in repair mode to avoid incorrect item from being passed to lowmem check.

PATCH In review Qu Wenruo 3 participants 3 messages May 17, 2019 – Jun 6, 2019 Btrfs file system

lore
Share & tools
lore

Qu Wenruo

author
#0 lore raw

PATCH 1/1 btrfs-progs: check/lowmem: Reset path in repair mode to avoid incorrect item from being passed to lowmem check.

In lowmem mode, we check fs roots and free space cache by iterating each root item and inode item, using btrfs_next_item() and a path pointing to the root tree.

However in repair mode, check_fs_root() can modify the fs root, thus CoWs the tree root, and the old path in check_fs

It could lead to strange behavior, e.g. after repairing a fs tree, the path can point to a fs tree.
Since no ROOT_ITEM exists in fs tree, all remaining trees are skipped in repair mode.

This bug exists from the early time of lowmem mode repair, and is only exposed by recent free space inode check code. (Fs tree inodes are passed to free space inode check, causing false alerts and repair failure).

---
 check/mode-lowmem.c | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
check/mode-lowmem.c +22 -0
index 6d7ae2bc0549..808d6be8db30 100644
@@ -5184,6 +5184,28 @@ int check_fs_roots_lowmem(struct btrfs_fs_info *fs_info)
 			err |= ret;
 		}
 next:
+		/*
+		 * In repair mode, our path is no longer reliable as CoW can
+		 * happen.
+		 * We need to reset our path.
+		 */
+		if (repair) {
+			btrfs_release_path(&path);
+			ret = btrfs_search_slot(NULL, tree_root, &key, &path,
+						0, 0);
+			if (ret < 0) {
+				if (!err)
+					err = ret;
+				goto out;
+			}
+			if (ret > 0) {
+				/* Key not found, but already at next item */
+				if (path.slots[0] <
+				    btrfs_header_nritems(path.nodes[0]))
+					continue;
+				/* falls through to next leaf */
+			}
+		}
 		ret = btrfs_next_item(tree_root, &path);
 		if (ret > 0)
 			goto out;
Signature · 1 line
2.21.0
#1 lore raw
Quoted 12 lines · Qu Wenruo

In lowmem mode, we check fs roots and free space cache by iterating each root item and inode item, using btrfs_next_item() and a path pointing to the root tree.

However in repair mode, check_fs_root() can modify the fs root, thus CoWs the tree root, and the old path in check_fs

It could lead to strange behavior, e.g. after repairing a fs tree, the path can point to a fs tree.
Since no ROOT_ITEM exists in fs tree, all remaining trees are skipped in repair mode.

Quoted 6 lines

This bug exists from the early time of lowmem mode repair, and is only exposed by recent free space inode check code. (Fs tree inodes are passed to free space inode check, causing false alerts and repair failure).

I'll add it to devel, however the lowmem mode of test-check does not work now, so can't really test it.

#2 lore raw
Quoted 18 lines · David Sterba
Quoted 17 lines · Qu Wenruo

In lowmem mode, we check fs roots and free space cache by iterating each root item and inode item, using btrfs_next_item() and a path pointing to the root tree.

However in repair mode, check_fs_root() can modify the fs root, thus CoWs the tree root, and the old path in check_fs

It could lead to strange behavior, e.g. after repairing a fs tree, the path can point to a fs tree.
Since no ROOT_ITEM exists in fs tree, all remaining trees are skipped in repair mode.

This bug exists from the early time of lowmem mode repair, and is only exposed by recent free space inode check code. (Fs tree inodes are passed to free space inode check, causing false alerts and repair failure).

Quoted 5 lines
Quoted 1 line

I'll add it to devel, however the lowmem mode of test-check does not work now, so can't really test it.

With this patch applies, test-fsck runs well for lowmem mode, and that's what this patch should do.

I just tried devel branch it works as expected.

Or did you hit some new bug?

Thanks,
Qu