Thread (19 messages) 19 messages, 3 authors, 2017-01-27

Re: [PATCH] queue stall with blk-mq-sched

From: Hannes Reinecke <hare@suse.de>
Date: 2017-01-25 11:10:47
Subsystem: block layer, the rest · Maintainers: Jens Axboe, Linus Torvalds

On 01/25/2017 09:07 AM, Hannes Reinecke wrote:
quoted hunk ↗ jump to hunk
On 01/25/2017 08:39 AM, Hannes Reinecke wrote:
quoted
On 01/24/2017 11:06 PM, Jens Axboe wrote:
quoted
On 01/24/2017 12:55 PM, Jens Axboe wrote:
quoted
Try this patch. We only want to bump it for the driver tags, not the
scheduler side.
More complete version, this one actually tested. I think this should fix
your issue, let me know.
Nearly there.
The initial stall is gone, but the test got hung at the 'stonewall'
sequence again:

[global]
bs=4k
ioengine=libaio
iodepth=256
size=4g
direct=1
runtime=60
# directory=/mnt
numjobs=32
group_reporting
cpus_allowed_policy=split
filename=/dev/md127

[seq-read]
rw=read
-> stonewall

[rand-read]
rw=randread
stonewall

Restarting all queues made the fio job continue.
There were 4 queues with state 'restart', and one queue with state 'active'.
So we're missing a queue run somewhere.
I've found the queue stalls are gone with this patch:
diff --git a/block/blk-mq-sched.h b/block/blk-mq-sched.h
index 6b465bc..de5db6c 100644
--- a/block/blk-mq-sched.h
+++ b/block/blk-mq-sched.h
@@ -113,6 +113,15 @@ static inline void blk_mq_sched_put_rq_priv(struct
request_queue *q,
 }

 static inline void
+blk_mq_sched_restart(struct blk_mq_hw_ctx *hctx)
+{
+       if (test_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state)) {
+               clear_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state);
+               blk_mq_run_hw_queue(hctx, true);
+       }
+}
+
+static inline void
 blk_mq_sched_completed_request(struct blk_mq_hw_ctx *hctx, struct
request *rq)
 {
        struct elevator_queue *e = hctx->queue->elevator;
@@ -123,11 +132,6 @@ static inline void blk_mq_sched_put_rq_priv(struct
request_queue *q,
        BUG_ON(rq->internal_tag == -1);

        blk_mq_put_tag(hctx, hctx->sched_tags, rq->mq_ctx,
rq->internal_tag);
-
-       if (test_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state)) {
-               clear_bit(BLK_MQ_S_SCHED_RESTART, &hctx->state);
-               blk_mq_run_hw_queue(hctx, true);
-       }
 }

 static inline void blk_mq_sched_started_request(struct request *rq)
diff --git a/block/blk-mq.c b/block/blk-mq.c
index e872555..63799ad 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -345,6 +345,7 @@ void __blk_mq_finish_request(struct blk_mq_hw_ctx
*hctx, struct blk_mq_ctx *ctx,
                blk_mq_put_tag(hctx, hctx->tags, ctx, rq->tag);
        if (sched_tag != -1)
                blk_mq_sched_completed_request(hctx, rq);
+       blk_mq_sched_restart(hctx);
        blk_queue_exit(q);
 }
Bah.

Not quite. I'm still seeing some queues with state 'restart'.

I've found that I need another patch on top of that:
diff --git a/block/blk-mq.c b/block/blk-mq.c
index e872555..edcbb44 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -753,8 +754,10 @@ static void blk_mq_timeout_work(struct work_struct
*work)

                queue_for_each_hw_ctx(q, hctx, i) {
                        /* the hctx may be unmapped, so check it here */
-                       if (blk_mq_hw_queue_mapped(hctx))
+                       if (blk_mq_hw_queue_mapped(hctx)) {
                                blk_mq_tag_idle(hctx);
+                               blk_mq_sched_restart(hctx);
+                       }
                }
        }
        blk_queue_exit(q);


Reasoning is that in blk_mq_get_tag() we might end up scheduling the
request on another hctx, but the original hctx might still have the
SCHED_RESTART bit set.
Which will never cleared as we complete the request on a different hctx,
so anything we do on the end_request side won't do us any good.

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		   Teamlead Storage & Networking
hare@suse.de			               +49 911 74053 688
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton
HRB 21284 (AG Nürnberg)
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help