Re: [PATCH] btrfs: zoned: fix deadlock between metadata writeback and transaction commit

On Fri Jul 3, 2026 at 2:54 PM JST, Johannes Thumshirn wrote:
> When writing out metadata extent buffers in a zoned filesystem,
> btree_writepages() holds fs_info->zoned_meta_io_lock across the whole
> writeback loop, including the call to btrfs_check_meta_write_pointer() ->
> check_bg_is_active().
>
> For the tree-log block group, check_bg_is_active() may fail to activate
> the zone and fall back to btrfs_zone_finish_one_bg() to free an active
> zone. That path waits for the running transaction to commit while still
> holding zoned_meta_io_lock, but the committer needs that same lock to
> write out the tree extents, so the two tasks deadlock:
>
>   Task A (kworker, metadata writeback)      Task B (fsstress, transaction commit)
>   ------------------------------------      -------------------------------------
>   wb_workfn()                               btrfs_commit_transaction(T)
>    btree_writepages()                        btrfs_write_and_wait_transaction()
>     btrfs_zoned_meta_io_lock()                btrfs_write_marked_extents()
>     btrfs_check_meta_write_pointer()           btree_writepages()
>      check_bg_is_active() [treelog_bg]          btrfs_zoned_meta_io_lock()
>       btrfs_zone_finish_one_bg()               <blocks on zoned_meta_io_lock,
>        btrfs_zone_finish()                      held by Task A>
>         do_zone_finish()
>          btrfs_inc_block_group_ro()
>           btrfs_wait_for_commit()
>            <blocks waiting for commit
>             of transaction T, done by
>             Task B>
>
> The sibling branch in check_bg_is_active() already drops zoned_meta_io_lock
> around do_zone_finish() for this exact reason. Do the same in the tree-log
> branch: release the lock around btrfs_zone_finish_one_bg() and re-acquire
> it afterwards. The lock only protects fs_info->active_{meta,system}_bg,
> which this branch does not touch, and ctx->zoned_bg keeps a reference to
> the block group across the unlock, so nothing is lost while the lock
> is dropped.
>
> This hang occasionally reproduces with fstests generic/475 on a zoned
> btrfs filesystem.

looks good,

Reviewed-by: Naohiro Aota <[email protected]>

Note: I'm thinking to make treelog_bg reserved, and pivot like as
metadata block group. It will be more reliable and naturally aligned
with the metadata block group.

>
> Fixes: 13bb483d32ab ("btrfs: zoned: activate metadata block group on write time")
> Cc: [email protected]
> Signed-off-by: Johannes Thumshirn <[email protected]>
> ---
>  fs/btrfs/zoned.c | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/fs/btrfs/zoned.c b/fs/btrfs/zoned.c
> index 97f06dd01693..44a13ed6b8b2 100644
> --- a/fs/btrfs/zoned.c
> +++ b/fs/btrfs/zoned.c
> @@ -2190,7 +2190,11 @@ static bool check_bg_is_active(struct btrfs_eb_write_context *ctx,
>  
>  	if (fs_info->treelog_bg == block_group->start) {
>  		if (!btrfs_zone_activate(block_group)) {
> -			int ret_fin = btrfs_zone_finish_one_bg(fs_info);
> +			int ret_fin;
> +
> +			btrfs_zoned_meta_io_unlock(fs_info);
> +			ret_fin = btrfs_zone_finish_one_bg(fs_info);
> +			btrfs_zoned_meta_io_lock(fs_info);
>  
>  			if (ret_fin != 1 || !btrfs_zone_activate(block_group))
>  				return false;