Skip to contentSkip to conversation

block: don't call blk_mq_run_hw_queues() for dead or dying queues

PATCH In review Hannes Reinecke 6 participants 9 messages Mar 26, 2019 – Mar 27, 2019 Block layer

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 block: don't call blk_mq_run_hw_queues() for dead or dying queues

When a queue is dying or dead there is no point in calling blk_mq_run_hw_queues() in blk_mq_unquiesce_queue(); in fact, doing so might crash the machine as the queue structures are in the process of being deleted.

---
 block/blk-mq.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
block/blk-mq.c +2 -1
index a9c181603cbd..b1eeba38bc79 100644
@@ -258,7 +258,8 @@ void blk_mq_unquiesce_queue(struct request_queue *q)
 	blk_queue_flag_clear(QUEUE_FLAG_QUIESCED, q);
 
 	/* dispatch requests which are inserted during quiescing */
-	blk_mq_run_hw_queues(q, true);
+	if (!blk_queue_dying(q) && !blk_queue_dead(q))
+		blk_mq_run_hw_queues(q, true);
 }
 EXPORT_SYMBOL_GPL(blk_mq_unquiesce_queue);
 
Signature · 1 line
2.16.4
#1 lore raw
Quoted 15 lines · Hannes Reinecke

When a queue is dying or dead there is no point in calling blk_mq_run_hw_queues() in blk_mq_unquiesce_queue(); in fact, doing so might crash the machine as the queue structures are in the process of being deleted.

---
  block/blk-mq.c | 3 ++-
  1 file changed, 2 insertions(+), 1 deletion(-)
block/blk-mq.c +0 -0
index a9c181603cbd..b1eeba38bc79 100644

block/blk-mq.c void blk_mq_unquiesce_queue(struct request_queue *q) · line 264

Quoted 8 lines
block/blk-mq.c +2 -1
@@ -258,7 +258,8 @@ void blk_mq_unquiesce_queue(struct request_queue *q)
  	blk_queue_flag_clear(QUEUE_FLAG_QUIESCED, q);
 
  	/* dispatch requests which are inserted during quiescing */
-	blk_mq_run_hw_queues(q, true);
+	if (!blk_queue_dying(q) && !blk_queue_dead(q))
+		blk_mq_run_hw_queues(q, true);
  }
  EXPORT_SYMBOL_GPL(blk_mq_unquiesce_queue);

Hi Hannes,

Please provide more context information. In the "dead" state the queue must be run to make sure that all requests that were queued before the "dead" state get processed. The blk_cleanup_queue() function is responsible for stopping all code that can run the queue after all requests have finished and before destruction of the data structures needed for request processing starts.

Bart.

#2 lore raw
Quoted 26 lines · Bart Van Assche
Quoted 15 lines · Hannes Reinecke

When a queue is dying or dead there is no point in calling blk_mq_run_hw_queues() in blk_mq_unquiesce_queue(); in fact, doing so might crash the machine as the queue structures are in the process of being deleted.

---
  block/blk-mq.c | 3 ++-

  1 file changed, 2 insertions(+), 1 deletion(-)
block/blk-mq.c +0 -0
index a9c181603cbd..b1eeba38bc79 100644

block/blk-mq.c void blk_mq_unquiesce_queue(struct request_queue *q) · line 263

Quoted 7 lines
block/blk-mq.c +2 -1
@@ -258,7 +258,8 @@ void blk_mq_unquiesce_queue(struct request_queue *q)
       blk_queue_flag_clear(QUEUE_FLAG_QUIESCED, q);
       /* dispatch requests which are inserted during quiescing */
-    blk_mq_run_hw_queues(q, true);
+    if (!blk_queue_dying(q) && !blk_queue_dead(q))
+        blk_mq_run_hw_queues(q, true);
   }
   EXPORT_SYMBOL_GPL(blk_mq_unquiesce_queue);

Hi Hannes,

Quoted 2 lines

Please provide more context information. In the "dead" state the queue must be run to make sure that all requests that were queued before the "dead" state get processed. The blk_cleanup_queue() function is responsible for stopping all code that can run the queue after all requests have finished and before destruction of the data structures needed for request processing starts.

I have a crash with two processes competing for the same controller:

#0 0xffffffff983d3bcb in sbitmap_any_bit_set (sb=0xffff8a1b874ba0d8)

    at ../lib/sbitmap.c:181
#1  0xffffffff98366c05 in blk_mq_hctx_has_pending (hctx=0xffff8a1b874ba000)
    at ../block/blk-mq.c:66
#2  0xffffffff98366c85 in blk_mq_run_hw_queues (q=0xffff8a1b874ba0d8, async=true) at ../block/blk-mq.c:1292
#3  0xffffffff98366d3a in blk_mq_unquiesce_queue (q=<optimized out>)
    at ../block/blk-mq.c:265
#4  0xffffffffc01f3e0e in nvme_start_queues (ctrl=<optimized out>)
    at ../drivers/nvme/host/core.c:3658
#5  0xffffffffc01e843c in nvme_fc_delete_association (ctrl=0xffff8a1f9be5a000)
    at ../drivers/nvme/host/fc.c:2843
#6  0xffffffffc01e8544 in nvme_fc_delete_association (ctrl=<optimized out>)
    at ../drivers/nvme/host/fc.c:2918
#7  0xffffffffc01e8544 in __nvme_fc_terminate_io (ctrl=0xffff8a1f9be5a000)
    at ../drivers/nvme/host/fc.c:2911
#8  0xffffffffc01e8f09 in nvme_fc_reset_ctrl_work (work=0xffff8a1f9be5a6d0)
    at ../drivers/nvme/host/fc.c:2927
#9  0xffffffff980a224a in process_one_work (worker=0xffff8a1b73934f00, work=0xffff8a1f9be5a6d0) at ../kernel/workqueue.c:2092
#10 0xffffffff980a249b in worker_thread (__worker=0xffff8a1b73934f00)
    at ../kernel/workqueue.c:2226

#7 0xffffffff986d2e9a in wait_for_completion (x=0xffffa48eca88bc40)

    at ../kernel/sched/completion.c:125
#8  0xffffffff980f25ae in __synchronize_srcu (sp=0xffffffff9914fc20 <debugfs_srcu>, do_norm=<optimized out>) at ../kernel/rcu/srcutree.c:851
#9  0xffffffff982d18b1 in debugfs_remove_recursive (dentry=<optimized out>)
    at ../fs/debugfs/inode.c:741

#10 0xffffffff98398ac5 in blk_mq_debugfs_unregister_hctx (hctx=0xffff8a1b7cccc000) at ../block/blk-mq-debugfs.c:897
#11 0xffffffff983661cf in blk_mq_exit_hctx (q=0xffff8a1f825e4040, set=0xffff8a1f9be5a0c0, hctx=0xffff8a1b7cccc000, hctx_idx=2) at ../block/blk-mq.c:1987 #12 0xffffffff9836946a in blk_mq_exit_hw_queues (nr_queue=<optimized out>, set=<optimized out>, q=<optimized out>) at ../block/blk-mq.c:2017 #13 0xffffffff9836946a in blk_mq_free_queue (q=0xffff8a1f825e4040)

    at ../block/blk-mq.c:2506
#14 0xffffffff9835aac5 in blk_cleanup_queue (q=0xffff8a1f825e4040)
    at ../block/blk-core.c:691
#15 0xffffffffc01f5bc8 in nvme_ns_remove (ns=0xffff8a1f819e8f80)
    at ../drivers/nvme/host/core.c:3138
#16 0xffffffffc01f6fea in nvme_validate_ns (ctrl=0xffff8a1f9be5a308, nsid=5)
    at ../drivers/nvme/host/core.c:3164
#17 0xffffffffc01f9053 in nvme_scan_ns_list (nn=<optimized out>, ctrl=<optimized out>) at ../drivers/nvme/host/core.c:3202
#18 0xffffffffc01f9053 in nvme_scan_work (work=<optimized out>)
    at ../drivers/nvme/host/core.c:3280

#19 0xffffffff980a224a in process_one_work (worker=0xffff8a1b7349f6c0, work=0xffff8a1f9be5aba0) at ../kernel/workqueue.c:2092

Point is that the queue is already dead by the time nvme_start_queues() tries to flush existing requests (of which there are none, of course). I had been looking into synchronizing scan_work and reset_work, but then I wasn't sure if that wouldn't deadlock somewhere.

Cheers,

Hannes

Signature · 5 lines
Dr. Hannes Reinecke		   Teamlead Storage & Networking
[email protected]			               +49 911 74053 688
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: Felix Imendörffer, Mary Higgins, Sri Rasiah
HRB 21284 (AG Nürnberg)
#3 lore raw
Quoted 80 lines · Hannes Reinecke
Quoted 20 lines · Bart Van Assche
Quoted 13 lines · Hannes Reinecke

When a queue is dying or dead there is no point in calling blk_mq_run_hw_queues() in blk_mq_unquiesce_queue();in fact, doing so might crash the machine as the queue structures are in the process of being deleted.

 block/blk-mq.c |3 +-+--
 1 file changed, 2 insertions(+-), 1 deletion(-)
block/blk-mq.c +1 -1
index a9c181603cbd..b1eeba38bc79 100644
--- a/block/blk-mq.c
+-+-+- b/block/blk-mq.c
Quoted 4 lines

@@-258,7 +-258,8 @@void blk_mq_unquiesce_queue(struct request_queue *q)
blk_queue_flag_clear(QUEUE_FLAG_QUIESCED, q);>>> /*dispatch requests which are inserted during quiescing */
- blk_mq_run_hw_queues(q, true);>>>+- if (!blk_queue_dying(q) &&!blk_queue_dead(q))
+- blk_mq_run_hw_queues(q, true);>>> }>>> EXPORT_SYMBOL_GPL(blk_mq_unquiesce_queue);>>

Hi Hannes,

Quoted 7 lines

Please provide more context information. In the "dead"state the queue must be run to make sure that all requests that were queued before the "dead"state get processed. The blk_cleanup_queue() function is responsible for stopping all code that can run the queue after all requests have finished and before destruction of the data structures needed for request processing starts.

I have a crash with two processes competing for the same controller:

#0 0xffffffff983d3bcb in sbitmap_any_bit_set (sb=0xffff8a1b874ba0d8)

    at ../lib/sbitmap.c:181
#1  0xffffffff98366c05 in blk_mq_hctx_has_pending (hctx=0xffff8a1b874ba000)
    at ../block/blk-mq.c:66
#2  0xffffffff98366c85 in blk_mq_run_hw_queues (q=0xffff8a1b874ba0d8, 
async=true) at ../block/blk-mq.c:1292
#3  0xffffffff98366d3a in blk_mq_unquiesce_queue (q=<optimized out>
    at ../block/blk-mq.c:265
#4  0xffffffffc01f3e0e in nvme_start_queues (ctrl=<optimized out>
    at ../drivers/nvme/host/core.c:3658
#5  0xffffffffc01e843c in nvme_fc_delete_association 
(ctrl=0xffff8a1f9be5a000)
    at ../drivers/nvme/host/fc.c:2843
#6  0xffffffffc01e8544 in nvme_fc_delete_association (ctrl=<optimized out>
    at ../drivers/nvme/host/fc.c:2918
#7  0xffffffffc01e8544 in __nvme_fc_terminate_io (ctrl=0xffff8a1f9be5a000)
    at ../drivers/nvme/host/fc.c:2911
#8  0xffffffffc01e8f09 in nvme_fc_reset_ctrl_work (work=0xffff8a1f9be5a6d0)
    at ../drivers/nvme/host/fc.c:2927
#9  0xffffffff980a224a in process_one_work (worker=0xffff8a1b73934f00, 
work=0xffff8a1f9be5a6d0) at ../kernel/workqueue.c:2092
#10 0xffffffff980a249b in worker_thread (__worker=0xffff8a1b73934f00)
    at ../kernel/workqueue.c:2226

#7 0xffffffff986d2e9a in wait_for_completion (x=0xffffa48eca88bc40)

    at ../kernel/sched/completion.c:125
#8  0xffffffff980f25ae in __synchronize_srcu (sp=0xffffffff9914fc20 
<debugfs_srcu> do_norm=<optimized out> at ../kernel/rcu/srcutree.c:851
#9  0xffffffff982d18b1 in debugfs_remove_recursive (dentry=<optimized out>
    at ../fs/debugfs/inode.c:741

#10 0xffffffff98398ac5 in blk_mq_debugfs_unregister_hctx (hctx=0xffff8a1b7cccc000) at ../block/blk-mq-debugfs.c:897 #11 0xffffffff983661cf in blk_mq_exit_hctx (q=0xffff8a1f825e4040, set=0xffff8a1f9be5a0c0, hctx=0xffff8a1b7cccc000, hctx_idx=2) at ../block/blk-mq.c:1987
#12 0xffffffff9836946a in blk_mq_exit_hw_queues (nr_queue=<optimized out> set=<optimized out> q=<optimized out> at ../block/blk-mq.c:2017 #13 0xffffffff9836946a in blk_mq_free_queue (q=0xffff8a1f825e4040)

    at ../block/blk-mq.c:2506
#14 0xffffffff9835aac5 in blk_cleanup_queue (q=0xffff8a1f825e4040)
    at ../block/blk-core.c:691
#15 0xffffffffc01f5bc8 in nvme_ns_remove (ns=0xffff8a1f819e8f80)
    at ../drivers/nvme/host/core.c:3138
#16 0xffffffffc01f6fea in nvme_validate_ns (ctrl=0xffff8a1f9be5a308, nsid=5)
    at ../drivers/nvme/host/core.c:3164
#17 0xffffffffc01f9053 in nvme_scan_ns_list (nn=<optimized out> 
ctrl=<optimized out> at ../drivers/nvme/host/core.c:3202
#18 0xffffffffc01f9053 in nvme_scan_work (work=<optimized out>
    at ../drivers/nvme/host/core.c:3280
Quoted 7 lines

#19 0xffffffff980a224a in process_one_work (worker=0xffff8a1b7349f6c0, work=0xffff8a1f9be5aba0) at ../kernel/workqueue.c:2092

Point is that the queue is already dead by the time nvme_start_queues() tries to flush existing requests (of which there are none, of course). I had been looking into synchronizing scan_work and reset_work, but then I wasn't sure if that wouldn't deadlock somewhere.

James, do you agree that nvme_fc_reset_ctrl_work should be canceled before nvme_ns_remove() is allowed to call blk_cleanup_queue()?

Thanks,

Bart.

#4 lore raw
Quoted 90 lines · Bart Van Assche
Quoted 89 lines · Hannes Reinecke
Quoted 26 lines · Bart Van Assche
Quoted 15 lines · Hannes Reinecke

When a queue is dying or dead there is no point in calling blk_mq_run_hw_queues() in blk_mq_unquiesce_queue(); in fact, doing so might crash the machine as the queue structures are in the process of being deleted.

---
   block/blk-mq.c | 3 ++-
   1 file changed, 2 insertions(+), 1 deletion(-)
block/blk-mq.c +0 -0
index a9c181603cbd..b1eeba38bc79 100644

block/blk-mq.c void blk_mq_unquiesce_queue(struct request_queue *q) · line 263

Quoted 7 lines
block/blk-mq.c +2 -1
@@ -258,7 +258,8 @@ void blk_mq_unquiesce_queue(struct request_queue *q)
       blk_queue_flag_clear(QUEUE_FLAG_QUIESCED, q);
       /* dispatch requests which are inserted during quiescing */
-    blk_mq_run_hw_queues(q, true);
+    if (!blk_queue_dying(q) && !blk_queue_dead(q))
+        blk_mq_run_hw_queues(q, true);
   }
   EXPORT_SYMBOL_GPL(blk_mq_unquiesce_queue);

Hi Hannes,

Quoted 7 lines

Please provide more context information. In the "dead" state the queue must be run to make sure that all requests that were queued before the "dead" state get processed. The blk_cleanup_queue() function is responsible for stopping all code that can run the queue after all requests have finished and before destruction of the data structures needed for request processing starts.

I have a crash with two processes competing for the same controller:

#0 0xffffffff983d3bcb in sbitmap_any_bit_set (sb=0xffff8a1b874ba0d8)

      at ../lib/sbitmap.c:181
#1  0xffffffff98366c05 in blk_mq_hctx_has_pending (hctx=0xffff8a1b874ba000)
      at ../block/blk-mq.c:66
#2  0xffffffff98366c85 in blk_mq_run_hw_queues (q=0xffff8a1b874ba0d8,
async=true) at ../block/blk-mq.c:1292
#3  0xffffffff98366d3a in blk_mq_unquiesce_queue (q=<optimized out>)
      at ../block/blk-mq.c:265
#4  0xffffffffc01f3e0e in nvme_start_queues (ctrl=<optimized out>)
      at ../drivers/nvme/host/core.c:3658
#5  0xffffffffc01e843c in nvme_fc_delete_association
(ctrl=0xffff8a1f9be5a000)
      at ../drivers/nvme/host/fc.c:2843
#6  0xffffffffc01e8544 in nvme_fc_delete_association (ctrl=<optimized out>)
      at ../drivers/nvme/host/fc.c:2918
#7  0xffffffffc01e8544 in __nvme_fc_terminate_io (ctrl=0xffff8a1f9be5a000)
      at ../drivers/nvme/host/fc.c:2911
#8  0xffffffffc01e8f09 in nvme_fc_reset_ctrl_work (work=0xffff8a1f9be5a6d0)
      at ../drivers/nvme/host/fc.c:2927
#9  0xffffffff980a224a in process_one_work (worker=0xffff8a1b73934f00,
work=0xffff8a1f9be5a6d0) at ../kernel/workqueue.c:2092
#10 0xffffffff980a249b in worker_thread (__worker=0xffff8a1b73934f00)
      at ../kernel/workqueue.c:2226

#7 0xffffffff986d2e9a in wait_for_completion (x=0xffffa48eca88bc40)

      at ../kernel/sched/completion.c:125
#8  0xffffffff980f25ae in __synchronize_srcu (sp=0xffffffff9914fc20
<debugfs_srcu>, do_norm=<optimized out>) at ../kernel/rcu/srcutree.c:851
#9  0xffffffff982d18b1 in debugfs_remove_recursive (dentry=<optimized out>)
      at ../fs/debugfs/inode.c:741

#10 0xffffffff98398ac5 in blk_mq_debugfs_unregister_hctx (hctx=0xffff8a1b7cccc000) at ../block/blk-mq-debugfs.c:897 #11 0xffffffff983661cf in blk_mq_exit_hctx (q=0xffff8a1f825e4040, set=0xffff8a1f9be5a0c0, hctx=0xffff8a1b7cccc000, hctx_idx=2) at ../block/blk-mq.c:1987
#12 0xffffffff9836946a in blk_mq_exit_hw_queues (nr_queue=<optimized out>, set=<optimized out>, q=<optimized out>) at ../block/blk-mq.c:2017 #13 0xffffffff9836946a in blk_mq_free_queue (q=0xffff8a1f825e4040)

      at ../block/blk-mq.c:2506
#14 0xffffffff9835aac5 in blk_cleanup_queue (q=0xffff8a1f825e4040)
      at ../block/blk-core.c:691
#15 0xffffffffc01f5bc8 in nvme_ns_remove (ns=0xffff8a1f819e8f80)
      at ../drivers/nvme/host/core.c:3138
#16 0xffffffffc01f6fea in nvme_validate_ns (ctrl=0xffff8a1f9be5a308, nsid=5)
      at ../drivers/nvme/host/core.c:3164
#17 0xffffffffc01f9053 in nvme_scan_ns_list (nn=<optimized out>,
ctrl=<optimized out>) at ../drivers/nvme/host/core.c:3202
#18 0xffffffffc01f9053 in nvme_scan_work (work=<optimized out>)
      at ../drivers/nvme/host/core.c:3280

#19 0xffffffff980a224a in process_one_work (worker=0xffff8a1b7349f6c0, work=0xffff8a1f9be5aba0) at ../kernel/workqueue.c:2092

Quoted 8 lines
Quoted 4 lines

Point is that the queue is already dead by the time nvme_start_queues() tries to flush existing requests (of which there are none, of course). I had been looking into synchronizing scan_work and reset_work, but then I wasn't sure if that wouldn't deadlock somewhere.

James, do you agree that nvme_fc_reset_ctrl_work should be canceled before nvme_ns_remove() is allowed to call blk_cleanup_queue()?

Hmm. I would've thought exactly the other way round, namely cancelling/flushing the scan_work element when calling reset; after all, reset definitely takes precedence to scanning the controller (which won't work anyway as the controller is about to be reset...)

Cheers,

Hannes

Signature · 5 lines
Dr. Hannes Reinecke		               zSeries & Storage
[email protected]			               +49 911 74053 688
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: F. Imendörffer, J. Smithard, D. Upmanyu, G. Norton
HRB 21284 (AG Nürnberg)
#5 lore raw
Quoted 97 lines · Hannes Reinecke
Quoted 96 lines · Bart Van Assche
Quoted 94 lines · Hannes Reinecke
Quoted 27 lines · Bart Van Assche
Quoted 15 lines · Hannes Reinecke

When a queue is dying or dead there is no point in calling blk_mq_run_hw_queues() in blk_mq_unquiesce_queue(); in fact, doing so might crash the machine as the queue structures are in the process of being deleted.

---
   block/blk-mq.c | 3 ++-

   1 file changed, 2 insertions(+), 1 deletion(-)
block/blk-mq.c +0 -0
index a9c181603cbd..b1eeba38bc79 100644

block/blk-mq.c void blk_mq_unquiesce_queue(struct · line 264

Quoted 8 lines
block/blk-mq.c +2 -1
@@ -258,7 +258,8 @@ void blk_mq_unquiesce_queue(struct 
 request_queue *q)
        blk_queue_flag_clear(QUEUE_FLAG_QUIESCED, q);
        /* dispatch requests which are inserted during quiescing */
-    blk_mq_run_hw_queues(q, true);
+    if (!blk_queue_dying(q) && !blk_queue_dead(q))
+        blk_mq_run_hw_queues(q, true);
    }
    EXPORT_SYMBOL_GPL(blk_mq_unquiesce_queue);

Hi Hannes,

Quoted 7 lines

Please provide more context information. In the "dead" state the queue must be run to make sure that all requests that were queued before the "dead" state get processed. The blk_cleanup_queue() function is responsible for stopping all code that can run the queue after all requests have finished and before destruction of the data structures needed for request processing starts.

I have a crash with two processes competing for the same controller:

#0  0xffffffff983d3bcb in sbitmap_any_bit_set (sb=0xffff8a1b874ba0d8) at ../lib/sbitmap.c:181
#1  0xffffffff98366c05 in blk_mq_hctx_has_pending (hctx=0xffff8a1b874ba000) at ../block/blk-mq.c:66
#2  0xffffffff98366c85 in blk_mq_run_hw_queues (q=0xffff8a1b874ba0d8, async=true) at ../block/blk-mq.c:1292
#3  0xffffffff98366d3a in blk_mq_unquiesce_queue (q=<optimized out>) at ../block/blk-mq.c:265
#4  0xffffffffc01f3e0e in nvme_start_queues (ctrl=<optimized out>) at ../drivers/nvme/host/core.c:3658
#5  0xffffffffc01e843c in nvme_fc_delete_association (ctrl=0xffff8a1f9be5a000)
      at ../drivers/nvme/host/fc.c:2843
#6  0xffffffffc01e8544 in nvme_fc_delete_association (ctrl=<optimized out>) at ../drivers/nvme/host/fc.c:2918
#7  0xffffffffc01e8544 in __nvme_fc_terminate_io (ctrl=0xffff8a1f9be5a000) at ../drivers/nvme/host/fc.c:2911
#8  0xffffffffc01e8f09 in nvme_fc_reset_ctrl_work (work=0xffff8a1f9be5a6d0) at ../drivers/nvme/host/fc.c:2927
#9  0xffffffff980a224a in process_one_work (worker=0xffff8a1b73934f00, work=0xffff8a1f9be5a6d0) at ../kernel/workqueue.c:2092
#10 0xffffffff980a249b in worker_thread (__worker=0xffff8a1b73934f00) at ../kernel/workqueue.c:2226

#7  0xffffffff986d2e9a in wait_for_completion (x=0xffffa48eca88bc40) at ../kernel/sched/completion.c:125
#8  0xffffffff980f25ae in __synchronize_srcu (sp=0xffffffff9914fc20 <debugfs_srcu>, do_norm=<optimized out>) at ../kernel/rcu/srcutree.c:851 #9  0xffffffff982d18b1 in debugfs_remove_recursive (dentry=<optimized out>) at ../fs/debugfs/inode.c:741
#10 0xffffffff98398ac5 in blk_mq_debugfs_unregister_hctx (hctx=0xffff8a1b7cccc000) at ../block/blk-mq-debugfs.c:897
#11 0xffffffff983661cf in blk_mq_exit_hctx (q=0xffff8a1f825e4040, set=0xffff8a1f9be5a0c0, hctx=0xffff8a1b7cccc000, hctx_idx=2) at ../block/blk-mq.c:1987
#12 0xffffffff9836946a in blk_mq_exit_hw_queues (nr_queue=<optimized out>, set=<optimized out>, q=<optimized out>) at ../block/blk-mq.c:2017 #13 0xffffffff9836946a in blk_mq_free_queue (q=0xffff8a1f825e4040) at ../block/blk-mq.c:2506
#14 0xffffffff9835aac5 in blk_cleanup_queue (q=0xffff8a1f825e4040) at ../block/blk-core.c:691
#15 0xffffffffc01f5bc8 in nvme_ns_remove (ns=0xffff8a1f819e8f80) at ../drivers/nvme/host/core.c:3138
#16 0xffffffffc01f6fea in nvme_validate_ns (ctrl=0xffff8a1f9be5a308, nsid=5) at ../drivers/nvme/host/core.c:3164
#17 0xffffffffc01f9053 in nvme_scan_ns_list (nn=<optimized out>, ctrl=<optimized out>) at ../drivers/nvme/host/core.c:3202
#18 0xffffffffc01f9053 in nvme_scan_work (work=<optimized out>) at ../drivers/nvme/host/core.c:3280
#19 0xffffffff980a224a in process_one_work (worker=0xffff8a1b7349f6c0, work=0xffff8a1f9be5aba0) at ../kernel/workqueue.c:2092

Point is that the queue is already dead by the time nvme_start_queues() tries to flush existing requests (of which there are none, of course). I had been looking into synchronizing scan_work and reset_work, but then I wasn't sure if that wouldn't deadlock somewhere.

Quoted 8 lines
Quoted 3 lines

James, do you agree that nvme_fc_reset_ctrl_work should be canceled before nvme_ns_remove() is allowed to call blk_cleanup_queue()?

Hmm. I would've thought exactly the other way round, namely cancelling/flushing the scan_work element when calling reset; after all, reset definitely takes precedence to scanning the controller (which won't work anyway as the controller is about to be reset...)

Cheers,

Hannes

Cancelling/flushing scan_work before starting reset won't work. Hannes is, correct, reset must take precedent.

The scenario is a controller that was recently connected, thus scan work started, and while the scan thread is running, there's an error or connectivity is lost.  The errors force the reset - which must run to terminate any outstanding requests, including those issued by the scan thread.  This has to happen or we're going to hang the scan thread as it's synchronous io.  The nvme_fc_delete_association->nvme_start_queues() sequence says the reset path froze the queues, cancelled all outstanding io and is unfreezing the queues to allow io to pass again, so that any new io on the request queue can be rejected while the controller is not connected (ioctls or subsequent io from the scan thread while the controller is unconnected as well as anything that was pending on the io queues from the fs that should be bounced back for multipath).  Meanwhile, the scan thread got the error on the io request, and since it's the identify command that failed, it's deleting the namespace at the same time the transport is unfreezing the queues.

Given the transport nvme_start_queues() routine has no idea what ns-queues exist, or what state the ns queues are in, it seems like a synchronization issue on the ns queue itself, that can't be solved by scheduling transport work elements.

I wonder if it should be something like this:

b    2019-03-26 10:47:55.081252967 -0700 renamed from a    2019-03-26 10:46:08.576056741 -0700 +0 -0
@@ -3870,8 +3870,10 @@ void nvme_start_queues(struct nvme_ctrl

     struct nvme_ns *ns;

     down_read(&ctrl->namespaces_rwsem);
-    list_for_each_entry(ns, &ctrl->namespaces, list) -        blk_mq_unquiesce_queue(ns->queue);
+    list_for_each_entry(ns, &ctrl->namespaces, list) { +        if (!test_bit(NVME_NS_REMOVING, &ns->flags)) +            blk_mq_unquiesce_queue(ns->queue); +    }
     up_read(&ctrl->namespaces_rwsem);
 }
 EXPORT_SYMBOL_GPL(nvme_start_queues);

which if valid, would also mean the same check should be in nvme_stop_queues() and perhaps elsewhere.

-- james

#6 lore raw
Quoted 24 lines · James Smart

Cancelling/flushing scan_work before starting reset won't work. Hannes is, correct, reset must take precedent.

The scenario is a controller that was recently connected, thus scan work started, and while the scan thread is running, there's an error or connectivity is lost.  The errors force the reset - which must run to terminate any outstanding requests, including those issued by the scan thread.  This has to happen or we're going to hang the scan thread as it's synchronous io.  The nvme_fc_delete_association->nvme_start_queues() sequence says the reset path froze the queues, cancelled all outstanding io and is unfreezing the queues to allow io to pass again, so that any new io on the request queue can be rejected while the controller is not connected (ioctls or subsequent io from the scan thread while the controller is unconnected as well as anything that was pending on the io queues from the fs that should be bounced back for multipath).  Meanwhile, the scan thread got the error on the io request, and since it's the identify command that failed, it's deleting the namespace at the same time the transport is unfreezing the queues.

Given the transport nvme_start_queues() routine has no idea what ns-queues exist, or what state the ns queues are in, it seems like a synchronization issue on the ns queue itself, that can't be solved by scheduling transport work elements.

I wonder if it should be something like this:

b    2019-03-26 10:47:55.081252967 -0700 renamed from a    2019-03-26 10:46:08.576056741 -0700 +4 -2
@@ -3870,8 +3870,10 @@ void nvme_start_queues(struct nvme_ctrl
      struct nvme_ns *ns;
 
      down_read(&ctrl->namespaces_rwsem);
-    list_for_each_entry(ns, &ctrl->namespaces, list)
-        blk_mq_unquiesce_queue(ns->queue);
+    list_for_each_entry(ns, &ctrl->namespaces, list) {
+        if (!test_bit(NVME_NS_REMOVING, &ns->flags))
+            blk_mq_unquiesce_queue(ns->queue);
+    }
      up_read(&ctrl->namespaces_rwsem);
  }
  EXPORT_SYMBOL_GPL(nvme_start_queues);
Quoted 2 lines

which if valid, would also mean the same check should be in nvme_stop_queues() and perhaps elsewhere.

Well, if the namespace wouldn't be part of the list in the first place once it's been set to 'REMOVING'... lemme check.

Cheers,

Hannes

#7 lore raw
In reply to Hannes Reinecke
Quoted 15 lines · Hannes Reinecke

When a queue is dying or dead there is no point in calling blk_mq_run_hw_queues() in blk_mq_unquiesce_queue(); in fact, doing so might crash the machine as the queue structures are in the process of being deleted.

---
 block/blk-mq.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
block/blk-mq.c +0 -0
index a9c181603cbd..b1eeba38bc79 100644

block/blk-mq.c void blk_mq_unquiesce_queue(struct request_queue *q) · line 264

Quoted 8 lines
block/blk-mq.c +2 -1
@@ -258,7 +258,8 @@ void blk_mq_unquiesce_queue(struct request_queue *q)
 	blk_queue_flag_clear(QUEUE_FLAG_QUIESCED, q);
 
 	/* dispatch requests which are inserted during quiescing */
-	blk_mq_run_hw_queues(q, true);
+	if (!blk_queue_dying(q) && !blk_queue_dead(q))
+		blk_mq_run_hw_queues(q, true);
 }
 EXPORT_SYMBOL_GPL(blk_mq_unquiesce_queue);

We actually still have to unquieces the queue to flush entered requests to a failed completion after killing that queue, like in nvme_set_queue_dying().

#8 lore raw
In reply to Hannes Reinecke
Quoted 15 lines · Hannes Reinecke

When a queue is dying or dead there is no point in calling blk_mq_run_hw_queues() in blk_mq_unquiesce_queue(); in fact, doing so might crash the machine as the queue structures are in the process of being deleted.

---
 block/blk-mq.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
block/blk-mq.c +0 -0
index a9c181603cbd..b1eeba38bc79 100644

block/blk-mq.c void blk_mq_unquiesce_queue(struct request_queue *q) · line 264

Quoted 8 lines
block/blk-mq.c +2 -1
@@ -258,7 +258,8 @@ void blk_mq_unquiesce_queue(struct request_queue *q)
 	blk_queue_flag_clear(QUEUE_FLAG_QUIESCED, q);
 
 	/* dispatch requests which are inserted during quiescing */
-	blk_mq_run_hw_queues(q, true);
+	if (!blk_queue_dying(q) && !blk_queue_dead(q))
+		blk_mq_run_hw_queues(q, true);
 }
 EXPORT_SYMBOL_GPL(blk_mq_unquiesce_queue);

Per previous protocol in legacy io path, the LLD should handle requests when queue is dying.

Also now, the request queue has become frozen before setting DEAD, so not necessary to add the DEAD check before run queue.

Thanks,
Ming