From: "Nicholas A. Bellinger" <nab@linux-iscsi.org>
To: Alexander Gordeev <agordeev@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>,
Mike Christie <mchristie@fusionio.com>,
James Bottomley <James.Bottomley@HansenPartnership.com>,
Tejun Heo <tj@kernel.org>,
linux-kernel@vger.kernel.org, linux-ide@vger.kernel.org,
Jeff Garzik <jgarzik@pobox.com>,
linux-scsi <linux-scsi@vger.kernel.org>
Subject: Re: [PATCH RESEND 0/1] AHCI: Optimize interrupt processing
Date: Fri, 20 Sep 2013 13:41:29 -0700 [thread overview]
Message-ID: <1379709689.31806.6.camel@haakon3.risingtidesystems.com> (raw)
In-Reply-To: <20130920151915.GC10507@dhcp-26-207.brq.redhat.com>
Hi Alexander!
Apologies for the long delay on this follow-up.. Comments below.
On Fri, 2013-09-20 at 17:19 +0200, Alexander Gordeev wrote:
> On Thu, Aug 15, 2013 at 07:19:29PM -0700, Nicholas A. Bellinger wrote:
> > Ok, here's a bit better idea of what is going on now..
> >
> > The problem is that blkdev_issue_flush() -> blk_mq_make_request() ->
> > __blk_mq_alloc_request() allocates the first tag, which calls
> > blk_insert_flush() -> blk_flush_complete_seq() -> blk_flush_kick() ->
> > mq_flush_work() -> blk_mq_alloc_request() to allocate a second tag for
> > the struct request that actually gets dispatched into scsi-mq as a
> > SYCHRONIZE_CACHE command..
> >
> > I'm not exactly sure why this double tag usage of struct request is
> > occurring, but AFAICT it does happen for every flush, and is not
> > specific to the blkdev_issue_flush() codepath.. I'm sure that Jens can
> > fill us in on that bit. ;)
>
> I also played with the double tag using a reserved tag (below).
>
> While it fixes 'fdisk /dev/sda' issue when trying to 'mount /dev/sda1 /mnt'
> what appears to be a call to bio->bi_end_io() from the free'd bio hits in.
>
> Not sure if I should pursue the root cause until the whole double-tag
> thingy is confirmed.
>
> Jens?
>
>
> diff --git a/block/blk-mq.c b/block/blk-mq.c
> index 6fc1df3..81794dc 100644
> --- a/block/blk-mq.c
> +++ b/block/blk-mq.c
> @@ -874,14 +874,14 @@ static void blk_mq_make_request(struct request_queue *q, struct bio *bio)
> hctx = q->mq_ops->map_queue(q, ctx->cpu);
>
> trace_block_getrq(q, bio, rw);
> - rq = __blk_mq_alloc_request(hctx, GFP_ATOMIC, false);
> + rq = __blk_mq_alloc_request(hctx, GFP_ATOMIC, is_flush_fua);
> if (likely(rq))
> blk_mq_rq_ctx_init(ctx, rq, rw);
> else {
> blk_mq_put_ctx(ctx);
> trace_block_sleeprq(q, bio, rw);
> rq = blk_mq_alloc_request_pinned(q, rw, __GFP_WAIT|GFP_ATOMIC,
> - false);
> + is_flush_fua);
> ctx = rq->mq_ctx;
> hctx = q->mq_ops->map_queue(q, ctx->cpu);
> }
So this is what I ended up doing as well, and does address the specific
bug with queue_depth=1.
> @@ -1317,6 +1317,9 @@ struct request_queue *blk_mq_init_queue(struct blk_mq_reg *reg,
> reg->queue_depth = BLK_MQ_MAX_DEPTH;
> }
>
> + reg->queue_depth++;
> + reg->reserved_tags++;
> +
> ctx = alloc_percpu(struct blk_mq_ctx);
> if (!ctx)
> return ERR_PTR(-ENOMEM);
>
I was actually setting this within scsi_mq_alloc_queue(), but given that
the queue_depth=1 issue is independent of scsi-mq, this does make more
sense.
Also, these extra increments should probably happen only when the passed
queue_depth == 1 && reserved_tags == 0.
Other than that minor nit.
Reviewed-by: Nicholas Bellinger <nab@linux-iscsi.org>
next prev parent reply other threads:[~2013-09-20 20:33 UTC|newest]
Thread overview: 70+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-05-21 19:00 Alexander Gordeev
2013-05-21 19:00 ` [PATCH RESEND 1/1] " Alexander Gordeev
2013-05-21 23:50 ` [PATCH RESEND 0/1] " Tejun Heo
2013-05-22 14:39 ` Alexander Gordeev
2013-05-22 17:03 ` Jens Axboe
2013-07-11 10:26 ` Alexander Gordeev
2013-07-11 23:00 ` Nicholas A. Bellinger
2013-07-12 7:46 ` Alexander Gordeev
2013-07-13 5:20 ` Nicholas A. Bellinger
2013-07-16 18:32 ` Alexander Gordeev
2013-07-16 21:38 ` Nicholas A. Bellinger
2013-07-17 16:19 ` Alexander Gordeev
2013-07-18 18:51 ` Nicholas A. Bellinger
2013-07-18 19:12 ` Mike Christie
2013-07-19 0:23 ` Nicholas A. Bellinger
2013-07-19 0:30 ` Jens Axboe
2013-07-19 1:03 ` Nicholas A. Bellinger
2013-07-19 6:34 ` Nicholas A. Bellinger
2013-07-19 15:33 ` James Bottomley
2013-07-19 21:01 ` Nicholas A. Bellinger
2013-07-20 4:56 ` Nicholas A. Bellinger
2013-07-20 14:48 ` Mike Christie
2013-07-20 22:14 ` Nicholas A. Bellinger
2013-07-20 23:57 ` Jens Axboe
2013-08-09 19:15 ` Alexander Gordeev
2013-08-09 20:17 ` Nicholas A. Bellinger
2013-08-15 16:23 ` Alexander Gordeev
2013-08-16 2:19 ` Nicholas A. Bellinger
2013-08-16 16:41 ` Alexander Gordeev
2013-08-16 17:46 ` Nicholas A. Bellinger
2013-08-28 15:56 ` Alexander Gordeev
2013-09-20 15:19 ` Alexander Gordeev
2013-09-20 20:41 ` Nicholas A. Bellinger [this message]
2013-10-03 11:06 ` Christoph Hellwig
2013-10-07 14:44 ` Alexander Gordeev
2013-07-22 15:03 ` Alexander Gordeev
2013-07-22 21:10 ` Nicholas A. Bellinger
2013-07-25 10:16 ` Alexander Gordeev
2013-07-25 22:08 ` Nicholas A. Bellinger
2013-07-26 2:09 ` Jens Axboe
2013-07-26 21:14 ` Nicholas A. Bellinger
2013-07-27 0:43 ` Nicholas A. Bellinger
2013-07-29 11:18 ` Alexander Gordeev
2013-07-29 14:08 ` Jens Axboe
2013-07-29 19:19 ` Nicholas A. Bellinger
2013-07-31 4:16 ` Marc C
2013-07-31 10:23 ` Tejun Heo
2013-07-29 11:50 ` Tejun Heo
2013-07-29 19:11 ` Nicholas A. Bellinger
2013-07-29 11:46 ` Tejun Heo
2013-07-29 14:03 ` Jens Axboe
2013-08-09 8:23 ` Alexander Gordeev
2013-08-09 14:15 ` Tejun Heo
2013-08-09 14:24 ` Jens Axboe
2013-08-09 15:07 ` Alexander Gordeev
2013-08-09 15:52 ` Jens Axboe
2013-08-09 16:46 ` Alexander Gordeev
2013-08-09 17:07 ` Jens Axboe
2013-08-12 15:21 ` Alexander Gordeev
2013-07-29 7:28 ` Hannes Reinecke
2013-07-31 17:11 ` Alexander Gordeev
2013-07-19 15:58 ` Mike Christie
2013-07-19 21:05 ` Nicholas A. Bellinger
2013-07-18 19:14 ` Nicholas A. Bellinger
2013-07-18 21:21 ` Jens Axboe
2014-09-11 12:42 ` Alexander Gordeev
2014-09-11 12:44 ` [PATCH v2] " Alexander Gordeev
2014-09-13 4:43 ` Tejun Heo
2014-09-11 13:36 ` [PATCH RESEND 0/1] " Bartlomiej Zolnierkiewicz
2014-09-13 4:46 ` Tejun Heo
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1379709689.31806.6.camel@haakon3.risingtidesystems.com \
--to=nab@linux-iscsi.org \
--cc=James.Bottomley@HansenPartnership.com \
--cc=agordeev@redhat.com \
--cc=axboe@kernel.dk \
--cc=jgarzik@pobox.com \
--cc=linux-ide@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-scsi@vger.kernel.org \
--cc=mchristie@fusionio.com \
--cc=tj@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®