On Jul 11, 2023 / 11:45, Bart Van Assche wrote:
[...]
>
> Thank you for the detailed report. Does this patch help?
>
> diff --git a/block/mq-deadline.c b/block/mq-deadline.c
> index 6aa5daf7ae32..02a916ba62ee 100644
> --- a/block/mq-deadline.c
> +++ b/block/mq-deadline.c
> @@ -176,7 +176,7 @@ static inline struct request *deadline_from_pos(struct
> dd_per_prio *per_prio,
> * zoned writes, start searching from the start of a zone.
> */
> if (blk_rq_is_seq_zoned_write(rq))
> - pos -= round_down(pos, rq->q->limits.chunk_sectors);
> + pos = round_down(pos, rq->q->limits.chunk_sectors);
>
> while (node) {
> rq = rb_entry_rq(node);
>
Yes, it helps. I confirmed that it avoids the I/O error using a zoned null_blk
device, a QEMU ZNS emulation device and a real ZNS drive. Looks good. If you
post the patch as a formal fix, please feel free to add:
Tested-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@xxxxxxx>
