On Fri, May 28, 2021 at 02:26:31PM +0200, Jan Kara wrote:
> On Fri 28-05-21 11:20:55, Ming Lei wrote:
> > Commit 6e6fcbc27e77 ("blk-mq: support batching dispatch in case of io")
> > starts to support io batching submission by using hctx->dispatch_busy.
> >
> > However, blk_mq_update_dispatch_busy() isn't changed to update hctx->dispatch_busy
> > in that commit, so fix the issue by updating hctx->dispatch_busy in case
> > of real scheduler.
> >
> > Reported-by: Jan Kara <jack@xxxxxxx>
> > Fixes: 6e6fcbc27e77 ("blk-mq: support batching dispatch in case of io")
> > Signed-off-by: Ming Lei <ming.lei@xxxxxxxxxx>
> > ---
> > block/blk-mq.c | 3 ---
> > 1 file changed, 3 deletions(-)
>
> Looks good to me. You can add:
>
> Reviewed-by: Jan Kara <jack@xxxxxxx>
>
> BTW: Do you plan to submit also your improvement to
> __blk_mq_do_dispatch_sched() to update dispatch_busy during the fetching
> requests from the scheduler to avoid draining all requests from the IO
> scheduler?
I understand that kind of change isn't needed. When more requests are
dequeued, hctx->dispatch_busy will be updated, then __blk_mq_do_dispatch_sched()
won't dequeue at batch any more if either .queue_rq() returns
STS_RESOURCE or running out of driver tag/budget.
Or do you still see related issues after this patch is applied?
Thanks,
Ming
