Re: [PATCH v2] block: Limit number of items taken from the I/O scheduler in one go

From: Salman Qazi
Date: Thu Apr 23 2020 - 21:42:09 EST


I remailed it with the changes since v1 added. But just to answer directly:

Changes since v1:

* Removed max_sched_batch.
* Extended the fix to the software queue.
* Use a return value from blk_mq_do_dispatch_sched to indicate if
the dispatch should be rerun.
* Some comments added.

On Thu, Apr 23, 2020 at 2:30 PM Jens Axboe <axboe@xxxxxxxxx> wrote:
>
> On 4/23/20 3:05 PM, Salman Qazi wrote:
> > Flushes bypass the I/O scheduler and get added to hctx->dispatch
> > in blk_mq_sched_bypass_insert. This can happen while a kworker is running
> > hctx->run_work work item and is past the point in
> > blk_mq_sched_dispatch_requests where hctx->dispatch is checked.
> >
> > The blk_mq_do_dispatch_sched call is not guaranteed to end in bounded time,
> > because the I/O scheduler can feed an arbitrary number of commands.
> >
> > Since we have only one hctx->run_work, the commands waiting in
> > hctx->dispatch will wait an arbitrary length of time for run_work to be
> > rerun.
> >
> > A similar phenomenon exists with dispatches from the software queue.
> >
> > The solution is to poll hctx->dispatch in blk_mq_do_dispatch_sched and
> > blk_mq_do_dispatch_ctx and return from the run_work handler and let it
> > rerun.
>
> Any changes since v1? It's customary to put that in here too, below
> the --- lines.
>
> --
> Jens Axboe
>