Skip to contentSkip to conversation

btrfs: set update the uuid generation as soon as possible

PATCH In review Josef Bacik 2 participants 2 messages Feb 14, 2020 – Feb 21, 2020 Btrfs file system

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 btrfs: set update the uuid generation as soon as possible

In my EIO stress testing I noticed I was getting forced to rescan the uuid tree pretty often, which was weird. This is because my error injection stuff would sometimes inject an error after log replay but before we loaded the UUID tree. If log replay committed the transaction it wouldn't have updated the uuid tree generation, but the tree was valid and didn't change, so there's no reason to not update the generation here.

Fix this by setting the BTRFS_FS_UPDATE_UUID_TREE_GEN bit immediately after reading all the fs roots if the uuid tree generation matches the fs generation. Then any transaction commits that happen during mount won't screw up our uuid tree state, forcing us to do needless uuid rescans.

Fixes 70f801754728 ("Btrfs: check UUID tree during mount if required")
---
 fs/btrfs/disk-io.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)
fs/btrfs/disk-io.c +13 -2
index e3a2db4d09a6..772cf0fa7c55 100644
@@ -3114,6 +3114,19 @@ int __cold open_ctree(struct super_block *sb, struct btrfs_fs_devices *fs_device
 	if (ret)
 		goto fail_tree_roots;
 
+	/*
+	 * If we have a uuid root and we're not being told to rescan we need to
+	 * check the generation here so we can set the
+	 * BTRFS_FS_UPDATE_UUID_TREE_GEN bit.  Otherwise we could commit the
+	 * transaction during a balance or the log replay without updating the
+	 * uuid generation, and then if we crash we would rescan the uuid tree,
+	 * even though it was perfectly fine.
+	 */
+	if (fs_info->uuid_root && !btrfs_test_opt(fs_info, RESCAN_UUID_TREE) &&
+	    fs_info->generation ==
+	    btrfs_super_uuid_tree_generation(disk_super))
+		set_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags);
+
 	ret = btrfs_verify_dev_extents(fs_info);
 	if (ret) {
 		btrfs_err(fs_info,
@@ -3338,8 +3351,6 @@ int __cold open_ctree(struct super_block *sb, struct btrfs_fs_devices *fs_device
 			close_ctree(fs_info);
 			return ret;
 		}
-	} else {
-		set_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags);
 	}
 	set_bit(BTRFS_FS_OPEN, &fs_info->flags);
 
Signature · 1 line
2.24.1
#1 lore raw
Quoted 8 lines · Josef Bacik

In my EIO stress testing I noticed I was getting forced to rescan the uuid tree pretty often, which was weird. This is because my error injection stuff would sometimes inject an error after log replay but before we loaded the UUID tree. If log replay committed the transaction it wouldn't have updated the uuid tree generation, but the tree was valid and didn't change, so there's no reason to not update the generation here.

Quoted 8 lines

Fix this by setting the BTRFS_FS_UPDATE_UUID_TREE_GEN bit immediately after reading all the fs roots if the uuid tree generation matches the fs generation. Then any transaction commits that happen during mount won't screw up our uuid tree state, forcing us to do needless uuid rescans.

Fixes 70f801754728 ("Btrfs: check UUID tree during mount if required")

Added to misc-next, thanks.