Skip to contentSkip to conversation

nvmet: loop: kill timeout handler

PATCH New Ming Lei 5 participants 5 messages Apr 15, 2019 – Apr 25, 2019

Subsystems (2)
lore
Share & tools
lore

Ming Lei

author
#0 lore raw

PATCH 1/1 nvmet: loop: kill timeout handler

Firstly it doesn't make sense to handle timeout for loop: 1) for admin queue, the request is always completed in code path of queuing IO. 2) for normal IO request, the timeout on these IOs have been handled by underlying queue already.

Secondly nvme-loop's timeout handler is simply broken, and easy to cause issue: 1) no any sync/protection between timeout and normal completion, and now it is driver's responsibility to deal with that; 2) bad reset implementation, blk_mq_update_nr_hw_queues() is called after all NSs's queue is stopped(quiesced), and easy to trigger deadlock.

So kill the timeout handler.

Cc Keith Busch <keith.busch at intel.com>
Cc James Smart <jsmart2021 at gmail.com>
Cc Hannes Reinecke <hare at suse.de>
Cc Sagi Grimberg <sagi at grimberg.me>
Signed-off-by Ming Lei <ming.lei at redhat.com>
---
 drivers/nvme/target/loop.c | 16 ----------------
 1 file changed, 16 deletions(-)
drivers/nvme/target/loop.c +0 -16
index b9f623ab01f3..c44b425848fb 100644
@@ -129,20 +129,6 @@ static void nvme_loop_execute_work(struct work_struct *work)
 	nvmet_req_execute(&iod->req);
 }
 
-static enum blk_eh_timer_return
-nvme_loop_timeout(struct request *rq, bool reserved)
-{
-	struct nvme_loop_iod *iod = blk_mq_rq_to_pdu(rq);
-
-	/* queue error recovery */
-	nvme_reset_ctrl(&iod->queue->ctrl->ctrl);
-
-	/* fail with DNR on admin cmd timeout */
-	nvme_req(rq)->status = NVME_SC_ABORT_REQ | NVME_SC_DNR;
-
-	return BLK_EH_DONE;
-}
-
 static blk_status_t nvme_loop_queue_rq(struct blk_mq_hw_ctx *hctx,
 		const struct blk_mq_queue_data *bd)
 {
@@ -253,7 +239,6 @@ static const struct blk_mq_ops nvme_loop_mq_ops = {
 	.complete	= nvme_loop_complete_rq,
 	.init_request	= nvme_loop_init_request,
 	.init_hctx	= nvme_loop_init_hctx,
-	.timeout	= nvme_loop_timeout,
 };
 
 static const struct blk_mq_ops nvme_loop_admin_mq_ops = {
@@ -261,7 +246,6 @@ static const struct blk_mq_ops nvme_loop_admin_mq_ops = {
 	.complete	= nvme_loop_complete_rq,
 	.init_request	= nvme_loop_init_request,
 	.init_hctx	= nvme_loop_init_admin_hctx,
-	.timeout	= nvme_loop_timeout,
 };
 
 static void nvme_loop_destroy_admin_queue(struct nvme_loop_ctrl *ctrl)
Signature · 1 line
2.9.5
#1 lore raw

PATCH 1/1 nvmet: loop: kill timeout handler

Quoted 69 lines · Ming Lei

Firstly it doesn't make sense to handle timeout for loop: 1) for admin queue, the request is always completed in code path of queuing IO. 2) for normal IO request, the timeout on these IOs have been handled by underlying queue already.

Secondly nvme-loop's timeout handler is simply broken, and easy to cause issue: 1) no any sync/protection between timeout and normal completion, and now it is driver's responsibility to deal with that; 2) bad reset implementation, blk_mq_update_nr_hw_queues() is called after all NSs's queue is stopped(quiesced), and easy to trigger deadlock.

So kill the timeout handler.

Cc Keith Busch <keith.busch at intel.com>
Cc James Smart <jsmart2021 at gmail.com>
Cc Hannes Reinecke <hare at suse.de>
Cc Sagi Grimberg <sagi at grimberg.me>
Signed-off-by Ming Lei <ming.lei at redhat.com>
---
 drivers/nvme/target/loop.c | 16 ----------------
 1 file changed, 16 deletions(-)
drivers/nvme/target/loop.c +0 -17
index b9f623ab01f3..c44b425848fb 100644
@@ -129,20 +129,6 @@ static void nvme_loop_execute_work(struct work_struct *work)
        nvmet_req_execute(&iod->req);
 }
 
-static enum blk_eh_timer_return
-nvme_loop_timeout(struct request *rq, bool reserved)
-{
-       struct nvme_loop_iod *iod = blk_mq_rq_to_pdu(rq);
-
-       /* queue error recovery */
-       nvme_reset_ctrl(&iod->queue->ctrl->ctrl);
-
-       /* fail with DNR on admin cmd timeout */
-       nvme_req(rq)->status = NVME_SC_ABORT_REQ | NVME_SC_DNR;
-
-       return BLK_EH_DONE;
-}
-
 static blk_status_t nvme_loop_queue_rq(struct blk_mq_hw_ctx *hctx,
                const struct blk_mq_queue_data *bd)
 {
@@ -253,7 +239,6 @@ static const struct blk_mq_ops nvme_loop_mq_ops = {
        .complete       = nvme_loop_complete_rq,
        .init_request   = nvme_loop_init_request,
        .init_hctx      = nvme_loop_init_hctx,
-       .timeout        = nvme_loop_timeout,
 };
 
 static const struct blk_mq_ops nvme_loop_admin_mq_ops = {
@@ -261,7 +246,6 @@ static const struct blk_mq_ops nvme_loop_admin_mq_ops = {
        .complete       = nvme_loop_complete_rq,
        .init_request   = nvme_loop_init_request,
        .init_hctx      = nvme_loop_init_admin_hctx,
-       .timeout        = nvme_loop_timeout,
 };
 
 static void nvme_loop_destroy_admin_queue(struct nvme_loop_ctrl *ctrl)
--

2.9.5

Quoted 4 lines

_______________________________________________
Linux-nvme mailing list
Linux-nvme at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

Ping...

Thanks,
Ming Lei

maintainer
#2 lore raw

PATCH 1/1 nvmet: loop: kill timeout handler

In reply to Ming Lei
Quoted 5 lines · Ming Lei

Firstly it doesn't make sense to handle timeout for loop: 1) for admin queue, the request is always completed in code path of queuing IO. 2) for normal IO request, the timeout on these IOs have been handled by underlying queue already.

Quoted 8 lines

Secondly nvme-loop's timeout handler is simply broken, and easy to cause issue: 1) no any sync/protection between timeout and normal completion, and now it is driver's responsibility to deal with that; 2) bad reset implementation, blk_mq_update_nr_hw_queues() is called after all NSs's queue is stopped(quiesced), and easy to trigger deadlock.

So kill the timeout handler.

Makes sense to me, looks good.

Reviewd-by: Keith Busch <keith.busch at intel.com>