By prioritizing sync I/Os, btrfs calls btrfs_map_block() for blocks allocated later before calling the function allocated earlier. By the disorder of calling btrfs_map_block(), syncing on I/Os on larger LBAs sometime wait for I/Os on smaller LBAs. Since active checksum worker is limited to some specified number, it is possible to wait for non-starting checksum on smaller LBAs. In such situation, transactions are stucked waiting for I/Os on smaller LBAs to finish, which is never finished. This situation can be reproduced by e.g. fstests btrfs/073. To avoid such disordering, disable sync IO prioritization for now. In the future, it will be reworked to finish checksumming of I/Os on smaller I/Os on committing a transaction. Signed-off-by: Naohiro Aota <naota@xxxxxxxxx> --- fs/btrfs/disk-io.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 00fa6aca9bb5..f79abd5e6b3a 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -807,7 +807,7 @@ blk_status_t btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct bio *bio, async->status = 0; - if (op_is_sync(bio->bi_opf)) + if (op_is_sync(bio->bi_opf) && !btrfs_fs_incompat(fs_info, HMZONED)) btrfs_set_work_high_priority(&async->work); btrfs_queue_work(fs_info->workers, &async->work); -- 2.18.0 -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
