Re: [PATCH 2/6] block: allow blk_execute_rq_nowait() to be calledfrom IRQ context

[Tejun Heo]

Hey, Jens.

On Wed, Oct 26, 2011 at 10:11:00AM +0200, Jens Axboe wrote:
> On 2011-10-26 03:02, Tejun Heo wrote:
> > Currently blk_execute_rq_nowait() directly calls __blk_run_queue() and
> > thus must be called from sleepable context.  This patch updates the
> > function such that it can be called from non-sleepable context and
> > schedules async execution in such cases.  This will be used to
> > unexport elv_add_request().
> > 
> > While at it, add FIXME comment for REQ_TYPE_PM_RESUME special case.
> > 
> > -v2: hch pointed out that blk_execute_rq_nowait() can be hot path for
> >      some drivers.  Retained direct execution from sleepable context.
> 
> Ugh, this looks nasty:
> 
> > +	bool may_sleep = !preempt_count() && !irqs_disabled();
> 
> please don't ever do that. Pass the context in instead.

Hmmm... I don't know.  This is strictly for optimization in block
layer proper.  It's kinda nasty to expose that, especially when the
interface is named _nowait and we already have separate request
allocation API.  Let's leave this alone for now.  The only offending
driver is ide anyway.

> > +	/*
> > +	 * Some drivers beat this path pretty hard.  As an optimization, if
> > +	 * we're being called from sleepable context, run @q directly.
> > +	 */
> > +	if (may_sleep) {
> > +		__blk_run_queue(q);
> > +		/*
> > +		 * The queue is stopped so it won't be run.
> > +		 * FIXME: Please kill me along with REQ_TYPE_PM_RESUME.
> > +		 */
> > +		if (rq->cmd_type == REQ_TYPE_PM_RESUME)
> > +			q->request_fn(q);
> 
> This is very nasty as well.

That's what the current code has.  We really need to take out those
odd request types.

Thanks.

-- 
tejun
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/