Commit 4eb4e85c authored by Boris Burkov's avatar Boris Burkov Committed by David Sterba
Browse files

btrfs: retry block group reclaim without infinite loop 



If inc_block_group_ro systematically fails (e.g. due to ETXTBUSY from
swap) or btrfs_relocate_chunk systematically fails (from lack of
space), then this worker becomes an infinite loop.

At the very least, this strands the cleaner thread, but can also result
in hung tasks/RCU stalls on PREEMPT_NONE kernels and if the
reclaim_bgs_lock mutex is not contended.

I believe the best long term fix is to manage reclaim via work queue,
where we queue up a relocation on the triggering condition and re-queue
on failure. In the meantime, this is an easy fix to apply to avoid the
immediate pain.

Fixes: 7e271809 ("btrfs: reinsert BGs failed to reclaim")
CC: stable@vger.kernel.org # 6.6+
Signed-off-by: default avatarBoris Burkov <boris@bur.io>
Reviewed-by: default avatarDavid Sterba <dsterba@suse.com>
Signed-off-by: default avatarDavid Sterba <dsterba@suse.com>
parent f3a5367c

fs/btrfs/block-group.c

+9 −2
Original line number Diff line number Diff line
@@ -1785,6 +1785,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work) 
		container_of(work, struct btrfs_fs_info, reclaim_bgs_work);

	struct btrfs_block_group *bg;

	struct btrfs_space_info *space_info;

	LIST_HEAD(retry_list);



	if (!test_bit(BTRFS_FS_OPEN, &fs_info->flags))

		return;
@@ -1921,8 +1922,11 @@ void btrfs_reclaim_bgs_work(struct work_struct *work) 
		}



next:

		if (ret)

			btrfs_mark_bg_to_reclaim(bg);

		if (ret) {

			/* Refcount held by the reclaim_bgs list after splice. */

			btrfs_get_block_group(bg);

			list_add_tail(&bg->bg_list, &retry_list);

		}

		btrfs_put_block_group(bg);



		mutex_unlock(&fs_info->reclaim_bgs_lock);
@@ -1942,6 +1946,9 @@ void btrfs_reclaim_bgs_work(struct work_struct *work) 
	spin_unlock(&fs_info->unused_bgs_lock);

	mutex_unlock(&fs_info->reclaim_bgs_lock);

end:

	spin_lock(&fs_info->unused_bgs_lock);

	list_splice_tail(&retry_list, &fs_info->reclaim_bgs);

	spin_unlock(&fs_info->unused_bgs_lock);

	btrfs_exclop_finish(fs_info);

	sb_end_write(fs_info->sb);

}