On Fri, 2018-07-27 at 11:58 -0600, Keith Busch wrote:
> Instead of PATCH 1/5, how about creating a new timeout return code like
> "BLK_EH_DONT_COMPLETE"?
That sounds like a good idea to me. I think this approach will avoid that we
have to introduce a request queue flag that chooses between the
On Fri, Jul 27, 2018 at 05:14:18PM +, Bart Van Assche wrote:
> On Fri, 2018-07-27 at 11:04 -0600, Keith Busch wrote:
> > On Fri, Jul 27, 2018 at 04:59:34PM +, Bart Van Assche wrote:
> > > On Fri, 2018-07-27 at 10:57 -0600, Keith Busch wrote:
> > > > You skip that code if the driver returns
On Fri, 2018-07-27 at 11:04 -0600, Keith Busch wrote:
> On Fri, Jul 27, 2018 at 04:59:34PM +, Bart Van Assche wrote:
> > On Fri, 2018-07-27 at 10:57 -0600, Keith Busch wrote:
> > > You skip that code if the driver returns BLK_EH_DONT_RESET_TIMER.
> >
> > How about applying the following patch
On Fri, Jul 27, 2018 at 04:59:34PM +, Bart Van Assche wrote:
> On Fri, 2018-07-27 at 10:57 -0600, Keith Busch wrote:
> > You skip that code if the driver returns BLK_EH_DONT_RESET_TIMER.
>
> How about applying the following patch on top of this series?
That works for me if you, but it breaks
On Fri, 2018-07-27 at 10:57 -0600, Keith Busch wrote:
> You skip that code if the driver returns BLK_EH_DONT_RESET_TIMER.
How about applying the following patch on top of this series?
diff --git a/block/blk-mq.c b/block/blk-mq.c
index a97ab5ba9d18..aa66535604fd 100644
--- a/block/blk-mq.c
+++
On Fri, Jul 27, 2018 at 04:51:05PM +, Bart Van Assche wrote:
> On Fri, 2018-07-27 at 10:46 -0600, Keith Busch wrote:
> > On Fri, Jul 27, 2018 at 09:20:42AM -0700, Bart Van Assche wrote:
> > > + ret = req->q->mq_ops->timeout(req, reserved);
> > > + /*
> > > + * BLK_EH_DONT_RESET_TIMER means
On Fri, 2018-07-27 at 10:46 -0600, Keith Busch wrote:
> On Fri, Jul 27, 2018 at 09:20:42AM -0700, Bart Van Assche wrote:
> > + ret = req->q->mq_ops->timeout(req, reserved);
> > + /*
> > +* BLK_EH_DONT_RESET_TIMER means that the block driver either
> > +* completed the request or still
On Fri, Jul 27, 2018 at 09:20:42AM -0700, Bart Van Assche wrote:
> + ret = req->q->mq_ops->timeout(req, reserved);
> + /*
> + * BLK_EH_DONT_RESET_TIMER means that the block driver either
> + * completed the request or still owns the request and will
> + * continue processing
Recently the blk-mq timeout handling code was reworked to avoid that
completions that occur while a timeout handler is in progress get
ignored. However, that rework removed the protection against completions
that occur while a timeout handler is in progress. Fix this by
introducing a new request
