Re: [PATCH 4/4] dm: support I/O polling

[Mike Snitzer <snitzer@redhat.com>]

On Thu, Mar 04 2021 at 10:01am -0500,
Jeff Moyer <jmoyer@redhat.com> wrote:

> Hi, Mikulas,
> 
> Mikulas Patocka <mpatocka@redhat.com> writes:
> 
> > On Wed, 3 Mar 2021, JeffleXu wrote:
> >
> >> 
> >> 
> >> On 3/3/21 3:05 AM, Mikulas Patocka wrote:
> >> 
> >> > Support I/O polling if submit_bio_noacct_mq_direct returned non-empty
> >> > cookie.
> >> > 
> >> > Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
> >> > 
> >> > ---
> >> >  drivers/md/dm.c |    5 +++++
> >> >  1 file changed, 5 insertions(+)
> >> > 
> >> > Index: linux-2.6/drivers/md/dm.c
> >> > ===================================================================
> >> > --- linux-2.6.orig/drivers/md/dm.c	2021-03-02 19:26:34.000000000 +0100
> >> > +++ linux-2.6/drivers/md/dm.c	2021-03-02 19:26:34.000000000 +0100
> >> > @@ -1682,6 +1682,11 @@ static void __split_and_process_bio(stru
> >> >  		}
> >> >  	}
> >> >  
> >> > +	if (ci.poll_cookie != BLK_QC_T_NONE) {
> >> > +		while (atomic_read(&ci.io->io_count) > 1 &&
> >> > +		       blk_poll(ci.poll_queue, ci.poll_cookie, true)) ;
> >> > +	}
> >> > +
> >> >  	/* drop the extra reference count */
> >> >  	dec_pending(ci.io, errno_to_blk_status(error));
> >> >  }
> >> 
> >> It seems that the general idea of your design is to
> >> 1) submit *one* split bio
> >> 2) blk_poll(), waiting the previously submitted split bio complets
> >
> > No, I submit all the bios and poll for the last one.
> 
> What happens if the last bio completes first?  It looks like you will
> call blk_poll with a cookie that already completed, and I'm pretty sure
> that's invalid.

In addition, I'm concerned this approach to have DM internalize IO
polling is a non-starter.

I just don't think this approach adheres to the io_uring + IO polling
interface.. it never returns a cookie to upper layers... so there is
really no opportunity for standard io_uring + IO polling interface to
work is there?

But Heinz and Mikulas are about to kick off some fio io_uring + hipri=1
(io polling) testing of Jeffle's latest v5 patchset:
https://patchwork.kernel.org/project/dm-devel/list/?series=442075

compared to Mikulas' patchset:
https://patchwork.kernel.org/project/dm-devel/list/?series=440719

We should have definitive answers soon enough, just using Jeffle's fio
config (with hipri=1 for IO polling) that was documented here:
https://listman.redhat.com/archives/dm-devel/2020-October/msg00129.html

Mike