[v5,14/14] nvme: Use BLK_MQ_S_STOPPED instead of QUEUE_FLAG_STOPPED in blk-mq code

Commit Message

Laurence Oberman Oct. 31, 2016, 1:53 p.m. UTC

----- Original Message -----
> From: "Bart Van Assche" <[e-mail]>
> To: "Jens Axboe" <[e-mail]>
> Cc: "Christoph Hellwig" <[e-mail]>, "James Bottomley" <[e-mail]>, "Martin K. Petersen"
> <[e-mail]>, "Mike Snitzer" <[e-mail]>, "Doug Ledford" <[e-mail]>, "Keith
> Busch" <[e-mail]>, "Ming Lei" <[e-mail]>, "Konrad Rzeszutek Wilk"
> <[e-mail]>, "Roger Pau Monné" <[e-mail]>, "Laurence Oberman" <[e-mail]>,
> [e-mail], [e-mail], [e-mail], [e-mail]
> Sent: Friday, October 28, 2016 8:23:40 PM
> Subject: [PATCH v5 14/14] nvme: Use BLK_MQ_S_STOPPED instead of QUEUE_FLAG_STOPPED in blk-mq code
> 
> Make nvme_requeue_req() check BLK_MQ_S_STOPPED instead of
> QUEUE_FLAG_STOPPED. Remove the QUEUE_FLAG_STOPPED manipulations
> that became superfluous because of this change. Change
> blk_queue_stopped() tests into blk_mq_queue_stopped().
> 
> This patch fixes a race condition: using queue_flag_clear_unlocked()
> is not safe if any other function that manipulates the queue flags
> can be called concurrently, e.g. blk_cleanup_queue().
> 
> Signed-off-by: Bart Van Assche <[e-mail]>
> Cc: Keith Busch <[e-mail]>
> Cc: Christoph Hellwig <[e-mail]>
> Cc: Sagi Grimberg <[e-mail]>
> ---
>  drivers/nvme/host/core.c | 16 ++--------------
>  1 file changed, 2 insertions(+), 14 deletions(-)
> 
> diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
> index fe15d94..45dd237 100644
> --- a/drivers/nvme/host/core.c
> +++ b/drivers/nvme/host/core.c
> @@ -201,13 +201,7 @@ static struct nvme_ns *nvme_get_ns_from_disk(struct
> gendisk *disk)
>  
>  void nvme_requeue_req(struct request *req)
>  {
> -	unsigned long flags;
> -
> -	blk_mq_requeue_request(req, false);
> -	spin_lock_irqsave(req->q->queue_lock, flags);
> -	if (!blk_queue_stopped(req->q))
> -		blk_mq_kick_requeue_list(req->q);
> -	spin_unlock_irqrestore(req->q->queue_lock, flags);
> +	blk_mq_requeue_request(req, !blk_mq_queue_stopped(req->q));
>  }
>  EXPORT_SYMBOL_GPL(nvme_requeue_req);
>  
> @@ -2078,13 +2072,8 @@ void nvme_stop_queues(struct nvme_ctrl *ctrl)
>  	struct nvme_ns *ns;
>  
>  	mutex_lock(&ctrl->namespaces_mutex);
> -	list_for_each_entry(ns, &ctrl->namespaces, list) {
> -		spin_lock_irq(ns->queue->queue_lock);
> -		queue_flag_set(QUEUE_FLAG_STOPPED, ns->queue);
> -		spin_unlock_irq(ns->queue->queue_lock);
> -
> +	list_for_each_entry(ns, &ctrl->namespaces, list)
>  		blk_mq_quiesce_queue(ns->queue);
> -	}
>  	mutex_unlock(&ctrl->namespaces_mutex);
>  }
>  EXPORT_SYMBOL_GPL(nvme_stop_queues);
> @@ -2095,7 +2084,6 @@ void nvme_start_queues(struct nvme_ctrl *ctrl)
>  
>  	mutex_lock(&ctrl->namespaces_mutex);
>  	list_for_each_entry(ns, &ctrl->namespaces, list) {
> -		queue_flag_clear_unlocked(QUEUE_FLAG_STOPPED, ns->queue);
>  		blk_mq_start_stopped_hw_queues(ns->queue, true);
>  		blk_mq_kick_requeue_list(ns->queue);
>  	}
> --
> 2.10.1
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to [e-mail]
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 

Hello Bart

Thanks for all this work.

Applied all 14 patches, also corrected the part of the xen-blkfront.c blkif_recover patch in patchv5-5/14.


Ran multiple read/write buffered and directio tests via RDMA/SRP and mlx5 (100Gbit) with max_sectors_kb set to 1024, 2048, 4096 and 8196
Ran multiple read/write buffered and directio tests via RDMA/SRP and mlx4 (56Gbit)  with max_sectors_kb set to 1024, 2048, 4096 and 8196
Reset the SRP hosts multiple times with multipath set to no_path_retry queue
Ran basic NVME read/write testing with no hot plug disconnects on multiple block sizes

All tests passed.

For the series:
Tested-by: Laurence Oberman <[e-mail]>
--
To unsubscribe from this list: send the line "unsubscribe linux-block" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Comments

Bart Van Assche Oct. 31, 2016, 1:59 p.m. UTC | #1

On 10/31/2016 06:53 AM, Laurence Oberman wrote:
> Ran multiple read/write buffered and directio tests via RDMA/SRP and mlx5 (100Gbit) with max_sectors_kb set to 1024, 2048, 4096 and 8196
> Ran multiple read/write buffered and directio tests via RDMA/SRP and mlx4 (56Gbit)  with max_sectors_kb set to 1024, 2048, 4096 and 8196
> Reset the SRP hosts multiple times with multipath set to no_path_retry queue
> Ran basic NVME read/write testing with no hot plug disconnects on multiple block sizes
>
> All tests passed.
>
> For the series:
> Tested-by: Laurence Oberman <[e-mail]>

Hello Laurence,

Thanks for having tested this version of this patch series again so quickly!

Bart.
--
To unsubscribe from this list: send the line "unsubscribe linux-block" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Bart Van Assche Oct. 31, 2016, 3:10 p.m. UTC | #2

On Mon, 2016-10-31 at 09:53 -0400, Laurence Oberman wrote:
> Applied all 14 patches, also corrected the part of the xen-blkfront.c 

> blkif_recover patch in patchv5-5/14.

> 

> diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-

> blkfront.c

> index 9908597..60fff99 100644

> --- a/drivers/block/xen-blkfront.c

> +++ b/drivers/block/xen-blkfront.c

> @@ -2045,6 +2045,7 @@ static int blkif_recover(struct blkfront_info

> *info)

>                  BUG_ON(req->nr_phys_segments > segs);

>                  blk_mq_requeue_request(req);

>          }

> +        blk_mq_start_stopped_hw_queues(infrq,

> true);                    *** Corrected

>          blk_mq_kick_requeue_list(infrq);

>  

>          while ((bio = bio_list_pop(&infbio_list)) != NULL) {


Hello Laurence,

Sorry for the build failure. The way you changed xen-blkfront is indeed
what I intended. Apparently I forgot to enable Xen in my kernel config
...

Bart.

Patch

diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
index 9908597..60fff99 100644
--- a/drivers/block/xen-blkfront.c
+++ b/drivers/block/xen-blkfront.c
@@ -2045,6 +2045,7 @@  static int blkif_recover(struct blkfront_info *info)
                 BUG_ON(req->nr_phys_segments > segs);
                 blk_mq_requeue_request(req);
         }
+        blk_mq_start_stopped_hw_queues(infrq, true);                    *** Corrected
         blk_mq_kick_requeue_list(infrq);
 
         while ((bio = bio_list_pop(&infbio_list)) != NULL) {