Skip to contentSkip to conversation

btrfs-progs: deal with drop_progress properly in fsck

PATCH In review Josef Bacik 2 participants 2 messages Feb 6, 2019 – Jul 22, 2019 Btrfs file system

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 btrfs-progs: deal with drop_progress properly in fsck

While testing snapshot deletion with dm-log-writes I saw that I was failing the fsck sometimes when the fs was actually in the correct state. This is because we only skip blocks on the same level of root_item->drop_level. If the drop_level < the root level then we could very well walk into nodes that we wouldn't actually walk into on fs mount, because the drop_progress is further ahead in the slot of the root. Instead only process the slots of the nodes that are above the drop_progress key. With this patch in place we no longer improperly fail to check fs'es that have a drop_progress set with a drop_level < root level.

---
 check/main.c | 19 ++++++++++++++++---
 ctree.c      |  6 +++---
 ctree.h      |  2 ++
 3 files changed, 21 insertions(+), 6 deletions(-)
check/main.c +16 -3
index 7547209c5604..89f0ce75e13d 100644
@@ -6325,15 +6325,28 @@ static int run_next_block(struct btrfs_root *root,
 		int level;
 
 		level = btrfs_header_level(buf);
-		for (i = 0; i < nritems; i++) {
+		i = 0;
+
+		/*
+		 * If we have a drop key we need to not walk down any slots we
+		 * would have ignored when mounting the fs.  These blocks are
+		 * technically unreferenced and don't need to be worried about.
+		 */
+		if (ri != NULL && ri->drop_level && level > ri->drop_level) {
+			ret = btrfs_bin_search(buf, &ri->drop_key, level, &i);
+			if (ret && i > 0)
+				i--;
+		}
+
+		for (; i < nritems; i++) {
 			struct extent_record tmpl;
 
 			ptr = btrfs_node_blockptr(buf, i);
 			size = root->fs_info->nodesize;
 			btrfs_node_key_to_cpu(buf, &key, i);
 			if (ri != NULL) {
-				if ((level == ri->drop_level)
-				    && is_dropped_key(&key, &ri->drop_key)) {
+				if ((level == ri->drop_level) &&
+				    is_dropped_key(&key, &ri->drop_key)) {
 					continue;
 				}
 			}
ctree.c +3 -3
index 7cb3f8451542..a086d59d6a1c 100644
@@ -638,8 +638,8 @@ static int generic_bin_search(struct extent_buffer *eb, unsigned long p,
  * simple bin_search frontend that does the right thing for
  * leaves vs nodes
  */
-static int bin_search(struct extent_buffer *eb, const struct btrfs_key *key,
-		      int level, int *slot)
+int btrfs_bin_search(struct extent_buffer *eb, const struct btrfs_key *key,
+		     int level, int *slot)
 {
 	if (level == 0)
 		return generic_bin_search(eb,
@@ -1172,7 +1172,7 @@ again:
 		ret = check_block(root, p, level);
 		if (ret)
 			return -1;
-		ret = bin_search(b, key, level, &slot);
+		ret = btrfs_bin_search(b, key, level, &slot);
 		if (level != 0) {
 			if (ret && slot > 0)
 				slot -= 1;
ctree.h +2 -0
index abc20e283fdc..f31a680eb473 100644
@@ -2616,6 +2616,8 @@ int btrfs_search_slot_for_read(struct btrfs_root *root,
                                const struct btrfs_key *key,
                                struct btrfs_path *p, int find_higher,
                                int return_any);
+int btrfs_bin_search(struct extent_buffer *eb, const struct btrfs_key *key,
+		     int level, int *slot);
 int btrfs_find_item(struct btrfs_root *fs_root, struct btrfs_path *found_path,
 		u64 iobjectid, u64 ioff, u8 key_type,
 		struct btrfs_key *found_key);
Signature · 1 line
2.14.3
#1 lore raw
Quoted 11 lines · Josef Bacik

While testing snapshot deletion with dm-log-writes I saw that I was failing the fsck sometimes when the fs was actually in the correct state. This is because we only skip blocks on the same level of root_item->drop_level. If the drop_level < the root level then we could very well walk into nodes that we wouldn't actually walk into on fs mount, because the drop_progress is further ahead in the slot of the root. Instead only process the slots of the nodes that are above the drop_progress key. With this patch in place we no longer improperly fail to check fs'es that have a drop_progress set with a drop_level < root level.

Quoted 1 line

Now applied to devel, thanks.