2017-11-01 16:52:31

by Bart Van Assche

[permalink] [raw]
Subject: Re: [PATCH 1/1] [RFC] blk-mq: fix queue stalling on shared hctx restart

On Mon, 2017-10-23 at 17:12 +0200, Roman Penyaev wrote:
> On Fri, Oct 20, 2017 at 10:05 PM, Bart Van Assche wrote:
> > Commit 6d8c6c0f97ad is something I came up with to fix queue lockups in the
> > SCSI and dm-mq drivers.
>
> You mean fairness? (some hctx get less amount of chances to be restarted).
> That's why you need to restart them in RR fashion, right?

Commit 6d8c6c0f97ad not only improved fairness but also fixed a lockup scenario.
After a driver tag has been freed not only the hw queues of the request queue for
which a request completed must be reexamined but also the hw queues associated
with the other request queues that share the same tag set.

Bart.?&ן6?}?ӟ7??4????S?㜷n5??9?^??M4?M{_???1? ????z?n?۾x߭??e?f????m?l??ۣ?jנ??????