diff mbox series

[v3] blk-mq: fix io hung due to missing commit_rqs

Message ID 20220726122224.1790882-1-yukuai1@huaweicloud.com (mailing list archive)
State New, archived
Headers show
Series [v3] blk-mq: fix io hung due to missing commit_rqs | expand

Commit Message

Yu Kuai July 26, 2022, 12:22 p.m. UTC
From: Yu Kuai <yukuai3@huawei.com>

Currently, in virtio_scsi, if 'bd->last' is not set to true while
dispatching request, such io will stay in driver's queue, and driver
will wait for block layer to dispatch more rqs. However, if block
layer failed to dispatch more rq, it should trigger commit_rqs to
inform driver.

There is a problem in blk_mq_try_issue_list_directly() that commit_rqs
won't be called:

// assume that queue_depth is set to 1, list contains two rq
blk_mq_try_issue_list_directly
 blk_mq_request_issue_directly
 // dispatch first rq
 // last is false
  __blk_mq_try_issue_directly
   blk_mq_get_dispatch_budget
   // succeed to get first budget
   __blk_mq_issue_directly
    scsi_queue_rq
     cmd->flags |= SCMD_LAST
      virtscsi_queuecommand
       kick = (sc->flags & SCMD_LAST) != 0
       // kick is false, first rq won't issue to disk
 queued++

 blk_mq_request_issue_directly
 // dispatch second rq
  __blk_mq_try_issue_directly
   blk_mq_get_dispatch_budget
   // failed to get second budget
 ret == BLK_STS_RESOURCE
  blk_mq_request_bypass_insert
 // errors is still 0

 if (!list_empty(list) || errors && ...)
  // won't pass, commit_rqs won't be called

In this situation, first rq relied on second rq to dispatch, while
second rq relied on first rq to complete, thus they will both hung.

Fix the problem by also treat 'BLK_STS_*RESOURCE' as 'errors' since
it means that request is not queued successfully.

Same problem exists in blk_mq_dispatch_rq_list(), 'BLK_STS_*RESOURCE'
can't be treated as 'errors' here, fix the problem by calling
commit_rqs if queue_rq return 'BLK_STS_*RESOURCE'.

Fixes: d666ba98f849 ("blk-mq: add mq_ops->commit_rqs()")
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
---
Changes in v3:
 - as poinited out by Ming, v2 will break return value by treating
 BLK_STS_*RESOURCE as errors.
Changes in v2:
 - suggested by Ming, handle blk_mq_dispatch_rq_list() as well.
 - change title and modify commit message.

 block/blk-mq.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

Comments

Ming Lei July 27, 2022, 2:11 a.m. UTC | #1
On Tue, Jul 26, 2022 at 08:22:24PM +0800, Yu Kuai wrote:
> From: Yu Kuai <yukuai3@huawei.com>
> 
> Currently, in virtio_scsi, if 'bd->last' is not set to true while
> dispatching request, such io will stay in driver's queue, and driver
> will wait for block layer to dispatch more rqs. However, if block
> layer failed to dispatch more rq, it should trigger commit_rqs to
> inform driver.
> 
> There is a problem in blk_mq_try_issue_list_directly() that commit_rqs
> won't be called:
> 
> // assume that queue_depth is set to 1, list contains two rq
> blk_mq_try_issue_list_directly
>  blk_mq_request_issue_directly
>  // dispatch first rq
>  // last is false
>   __blk_mq_try_issue_directly
>    blk_mq_get_dispatch_budget
>    // succeed to get first budget
>    __blk_mq_issue_directly
>     scsi_queue_rq
>      cmd->flags |= SCMD_LAST
>       virtscsi_queuecommand
>        kick = (sc->flags & SCMD_LAST) != 0
>        // kick is false, first rq won't issue to disk
>  queued++
> 
>  blk_mq_request_issue_directly
>  // dispatch second rq
>   __blk_mq_try_issue_directly
>    blk_mq_get_dispatch_budget
>    // failed to get second budget
>  ret == BLK_STS_RESOURCE
>   blk_mq_request_bypass_insert
>  // errors is still 0
> 
>  if (!list_empty(list) || errors && ...)
>   // won't pass, commit_rqs won't be called
> 
> In this situation, first rq relied on second rq to dispatch, while
> second rq relied on first rq to complete, thus they will both hung.
> 
> Fix the problem by also treat 'BLK_STS_*RESOURCE' as 'errors' since
> it means that request is not queued successfully.
> 
> Same problem exists in blk_mq_dispatch_rq_list(), 'BLK_STS_*RESOURCE'
> can't be treated as 'errors' here, fix the problem by calling
> commit_rqs if queue_rq return 'BLK_STS_*RESOURCE'.
> 
> Fixes: d666ba98f849 ("blk-mq: add mq_ops->commit_rqs()")
> Signed-off-by: Yu Kuai <yukuai3@huawei.com>
> ---
> Changes in v3:
>  - as poinited out by Ming, v2 will break return value by treating
>  BLK_STS_*RESOURCE as errors.
> Changes in v2:
>  - suggested by Ming, handle blk_mq_dispatch_rq_list() as well.
>  - change title and modify commit message.
> 
>  block/blk-mq.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/block/blk-mq.c b/block/blk-mq.c
> index 70177ee74295..7d26b222cbc9 100644
> --- a/block/blk-mq.c
> +++ b/block/blk-mq.c
> @@ -1931,7 +1931,8 @@ bool blk_mq_dispatch_rq_list(struct blk_mq_hw_ctx *hctx, struct list_head *list,
>  	/* If we didn't flush the entire list, we could have told the driver
>  	 * there was more coming, but that turned out to be a lie.
>  	 */
> -	if ((!list_empty(list) || errors) && q->mq_ops->commit_rqs && queued)
> +	if ((!list_empty(list) || errors || needs_resource ||
> +	     ret == BLK_STS_DEV_RESOURCE) && q->mq_ops->commit_rqs && queued)

'ret == BLK_STS_DEV_RESOURCE' may happen before, but 'ret' could be
overridden later.

Thanks,
Ming
Yu Kuai July 27, 2022, 2:26 a.m. UTC | #2
Hi, Ming

在 2022/07/27 10:11, Ming Lei 写道:
> On Tue, Jul 26, 2022 at 08:22:24PM +0800, Yu Kuai wrote:
>> From: Yu Kuai <yukuai3@huawei.com>
>>
>> Currently, in virtio_scsi, if 'bd->last' is not set to true while
>> dispatching request, such io will stay in driver's queue, and driver
>> will wait for block layer to dispatch more rqs. However, if block
>> layer failed to dispatch more rq, it should trigger commit_rqs to
>> inform driver.
>>
>> There is a problem in blk_mq_try_issue_list_directly() that commit_rqs
>> won't be called:
>>
>> // assume that queue_depth is set to 1, list contains two rq
>> blk_mq_try_issue_list_directly
>>   blk_mq_request_issue_directly
>>   // dispatch first rq
>>   // last is false
>>    __blk_mq_try_issue_directly
>>     blk_mq_get_dispatch_budget
>>     // succeed to get first budget
>>     __blk_mq_issue_directly
>>      scsi_queue_rq
>>       cmd->flags |= SCMD_LAST
>>        virtscsi_queuecommand
>>         kick = (sc->flags & SCMD_LAST) != 0
>>         // kick is false, first rq won't issue to disk
>>   queued++
>>
>>   blk_mq_request_issue_directly
>>   // dispatch second rq
>>    __blk_mq_try_issue_directly
>>     blk_mq_get_dispatch_budget
>>     // failed to get second budget
>>   ret == BLK_STS_RESOURCE
>>    blk_mq_request_bypass_insert
>>   // errors is still 0
>>
>>   if (!list_empty(list) || errors && ...)
>>    // won't pass, commit_rqs won't be called
>>
>> In this situation, first rq relied on second rq to dispatch, while
>> second rq relied on first rq to complete, thus they will both hung.
>>
>> Fix the problem by also treat 'BLK_STS_*RESOURCE' as 'errors' since
>> it means that request is not queued successfully.
>>
>> Same problem exists in blk_mq_dispatch_rq_list(), 'BLK_STS_*RESOURCE'
>> can't be treated as 'errors' here, fix the problem by calling
>> commit_rqs if queue_rq return 'BLK_STS_*RESOURCE'.
>>
>> Fixes: d666ba98f849 ("blk-mq: add mq_ops->commit_rqs()")
>> Signed-off-by: Yu Kuai <yukuai3@huawei.com>
>> ---
>> Changes in v3:
>>   - as poinited out by Ming, v2 will break return value by treating
>>   BLK_STS_*RESOURCE as errors.
>> Changes in v2:
>>   - suggested by Ming, handle blk_mq_dispatch_rq_list() as well.
>>   - change title and modify commit message.
>>
>>   block/blk-mq.c | 5 +++--
>>   1 file changed, 3 insertions(+), 2 deletions(-)
>>
>> diff --git a/block/blk-mq.c b/block/blk-mq.c
>> index 70177ee74295..7d26b222cbc9 100644
>> --- a/block/blk-mq.c
>> +++ b/block/blk-mq.c
>> @@ -1931,7 +1931,8 @@ bool blk_mq_dispatch_rq_list(struct blk_mq_hw_ctx *hctx, struct list_head *list,
>>   	/* If we didn't flush the entire list, we could have told the driver
>>   	 * there was more coming, but that turned out to be a lie.
>>   	 */
>> -	if ((!list_empty(list) || errors) && q->mq_ops->commit_rqs && queued)
>> +	if ((!list_empty(list) || errors || needs_resource ||
>> +	     ret == BLK_STS_DEV_RESOURCE) && q->mq_ops->commit_rqs && queued)
> 
> 'ret == BLK_STS_DEV_RESOURCE' may happen before, but 'ret' could be
> overridden later.

 From what I see, if queue_rq return 'BLK_STS_RESOURCE' or
'BLK_STS_DEV_RESOURCE', the loop will break, so that 'ret' could not
be overridden. 'BLK_STS_ZONE_RESOURCE' can be overridden, however,
needs_resource can cover that case.

Thanks,
Kuai
> 
> Thanks,
> Ming
> 
> .
>
Ming Lei July 27, 2022, 3:08 a.m. UTC | #3
On Tue, Jul 26, 2022 at 08:22:24PM +0800, Yu Kuai wrote:
> From: Yu Kuai <yukuai3@huawei.com>
> 
> Currently, in virtio_scsi, if 'bd->last' is not set to true while
> dispatching request, such io will stay in driver's queue, and driver
> will wait for block layer to dispatch more rqs. However, if block
> layer failed to dispatch more rq, it should trigger commit_rqs to
> inform driver.
> 
> There is a problem in blk_mq_try_issue_list_directly() that commit_rqs
> won't be called:
> 
> // assume that queue_depth is set to 1, list contains two rq
> blk_mq_try_issue_list_directly
>  blk_mq_request_issue_directly
>  // dispatch first rq
>  // last is false
>   __blk_mq_try_issue_directly
>    blk_mq_get_dispatch_budget
>    // succeed to get first budget
>    __blk_mq_issue_directly
>     scsi_queue_rq
>      cmd->flags |= SCMD_LAST
>       virtscsi_queuecommand
>        kick = (sc->flags & SCMD_LAST) != 0
>        // kick is false, first rq won't issue to disk
>  queued++
> 
>  blk_mq_request_issue_directly
>  // dispatch second rq
>   __blk_mq_try_issue_directly
>    blk_mq_get_dispatch_budget
>    // failed to get second budget
>  ret == BLK_STS_RESOURCE
>   blk_mq_request_bypass_insert
>  // errors is still 0
> 
>  if (!list_empty(list) || errors && ...)
>   // won't pass, commit_rqs won't be called
> 
> In this situation, first rq relied on second rq to dispatch, while
> second rq relied on first rq to complete, thus they will both hung.
> 
> Fix the problem by also treat 'BLK_STS_*RESOURCE' as 'errors' since
> it means that request is not queued successfully.
> 
> Same problem exists in blk_mq_dispatch_rq_list(), 'BLK_STS_*RESOURCE'
> can't be treated as 'errors' here, fix the problem by calling
> commit_rqs if queue_rq return 'BLK_STS_*RESOURCE'.
> 
> Fixes: d666ba98f849 ("blk-mq: add mq_ops->commit_rqs()")
> Signed-off-by: Yu Kuai <yukuai3@huawei.com>

Reviewed-by: Ming Lei <ming.lei@redhat.com>

Thanks,
Ming
Yu Kuai July 29, 2022, 6:35 a.m. UTC | #4
Hi, Jens

Can you please consider this patch for v5.20?

Thanks,
Kuai

在 2022/07/27 11:08, Ming Lei 写道:
> On Tue, Jul 26, 2022 at 08:22:24PM +0800, Yu Kuai wrote:
>> From: Yu Kuai <yukuai3@huawei.com>
>>
>> Currently, in virtio_scsi, if 'bd->last' is not set to true while
>> dispatching request, such io will stay in driver's queue, and driver
>> will wait for block layer to dispatch more rqs. However, if block
>> layer failed to dispatch more rq, it should trigger commit_rqs to
>> inform driver.
>>
>> There is a problem in blk_mq_try_issue_list_directly() that commit_rqs
>> won't be called:
>>
>> // assume that queue_depth is set to 1, list contains two rq
>> blk_mq_try_issue_list_directly
>>   blk_mq_request_issue_directly
>>   // dispatch first rq
>>   // last is false
>>    __blk_mq_try_issue_directly
>>     blk_mq_get_dispatch_budget
>>     // succeed to get first budget
>>     __blk_mq_issue_directly
>>      scsi_queue_rq
>>       cmd->flags |= SCMD_LAST
>>        virtscsi_queuecommand
>>         kick = (sc->flags & SCMD_LAST) != 0
>>         // kick is false, first rq won't issue to disk
>>   queued++
>>
>>   blk_mq_request_issue_directly
>>   // dispatch second rq
>>    __blk_mq_try_issue_directly
>>     blk_mq_get_dispatch_budget
>>     // failed to get second budget
>>   ret == BLK_STS_RESOURCE
>>    blk_mq_request_bypass_insert
>>   // errors is still 0
>>
>>   if (!list_empty(list) || errors && ...)
>>    // won't pass, commit_rqs won't be called
>>
>> In this situation, first rq relied on second rq to dispatch, while
>> second rq relied on first rq to complete, thus they will both hung.
>>
>> Fix the problem by also treat 'BLK_STS_*RESOURCE' as 'errors' since
>> it means that request is not queued successfully.
>>
>> Same problem exists in blk_mq_dispatch_rq_list(), 'BLK_STS_*RESOURCE'
>> can't be treated as 'errors' here, fix the problem by calling
>> commit_rqs if queue_rq return 'BLK_STS_*RESOURCE'.
>>
>> Fixes: d666ba98f849 ("blk-mq: add mq_ops->commit_rqs()")
>> Signed-off-by: Yu Kuai <yukuai3@huawei.com>
> 
> Reviewed-by: Ming Lei <ming.lei@redhat.com>
> 
> Thanks,
> Ming
> 
> .
>
Yu Kuai Aug. 13, 2022, 6 a.m. UTC | #5
在 2022/07/29 14:35, Yu Kuai 写道:
> Hi, Jens
> 
> Can you please consider this patch for v5.20?
> 
friendly ping ...
> Thanks,
> Kuai
> 
> 在 2022/07/27 11:08, Ming Lei 写道:
>> On Tue, Jul 26, 2022 at 08:22:24PM +0800, Yu Kuai wrote:
>>> From: Yu Kuai <yukuai3@huawei.com>
>>>
>>> Currently, in virtio_scsi, if 'bd->last' is not set to true while
>>> dispatching request, such io will stay in driver's queue, and driver
>>> will wait for block layer to dispatch more rqs. However, if block
>>> layer failed to dispatch more rq, it should trigger commit_rqs to
>>> inform driver.
>>>
>>> There is a problem in blk_mq_try_issue_list_directly() that commit_rqs
>>> won't be called:
>>>
>>> // assume that queue_depth is set to 1, list contains two rq
>>> blk_mq_try_issue_list_directly
>>>   blk_mq_request_issue_directly
>>>   // dispatch first rq
>>>   // last is false
>>>    __blk_mq_try_issue_directly
>>>     blk_mq_get_dispatch_budget
>>>     // succeed to get first budget
>>>     __blk_mq_issue_directly
>>>      scsi_queue_rq
>>>       cmd->flags |= SCMD_LAST
>>>        virtscsi_queuecommand
>>>         kick = (sc->flags & SCMD_LAST) != 0
>>>         // kick is false, first rq won't issue to disk
>>>   queued++
>>>
>>>   blk_mq_request_issue_directly
>>>   // dispatch second rq
>>>    __blk_mq_try_issue_directly
>>>     blk_mq_get_dispatch_budget
>>>     // failed to get second budget
>>>   ret == BLK_STS_RESOURCE
>>>    blk_mq_request_bypass_insert
>>>   // errors is still 0
>>>
>>>   if (!list_empty(list) || errors && ...)
>>>    // won't pass, commit_rqs won't be called
>>>
>>> In this situation, first rq relied on second rq to dispatch, while
>>> second rq relied on first rq to complete, thus they will both hung.
>>>
>>> Fix the problem by also treat 'BLK_STS_*RESOURCE' as 'errors' since
>>> it means that request is not queued successfully.
>>>
>>> Same problem exists in blk_mq_dispatch_rq_list(), 'BLK_STS_*RESOURCE'
>>> can't be treated as 'errors' here, fix the problem by calling
>>> commit_rqs if queue_rq return 'BLK_STS_*RESOURCE'.
>>>
>>> Fixes: d666ba98f849 ("blk-mq: add mq_ops->commit_rqs()")
>>> Signed-off-by: Yu Kuai <yukuai3@huawei.com>
>>
>> Reviewed-by: Ming Lei <ming.lei@redhat.com>
>>
>> Thanks,
>> Ming
>>
>> .
>>
> 
> .
>
Jens Axboe Aug. 19, 2022, 8:56 p.m. UTC | #6
On Tue, 26 Jul 2022 20:22:24 +0800, Yu Kuai wrote:
> From: Yu Kuai <yukuai3@huawei.com>
> 
> Currently, in virtio_scsi, if 'bd->last' is not set to true while
> dispatching request, such io will stay in driver's queue, and driver
> will wait for block layer to dispatch more rqs. However, if block
> layer failed to dispatch more rq, it should trigger commit_rqs to
> inform driver.
> 
> [...]

Applied, thanks!

[1/1] blk-mq: fix io hung due to missing commit_rqs
      (no commit info)

Best regards,
diff mbox series

Patch

diff --git a/block/blk-mq.c b/block/blk-mq.c
index 70177ee74295..7d26b222cbc9 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1931,7 +1931,8 @@  bool blk_mq_dispatch_rq_list(struct blk_mq_hw_ctx *hctx, struct list_head *list,
 	/* If we didn't flush the entire list, we could have told the driver
 	 * there was more coming, but that turned out to be a lie.
 	 */
-	if ((!list_empty(list) || errors) && q->mq_ops->commit_rqs && queued)
+	if ((!list_empty(list) || errors || needs_resource ||
+	     ret == BLK_STS_DEV_RESOURCE) && q->mq_ops->commit_rqs && queued)
 		q->mq_ops->commit_rqs(hctx);
 	/*
 	 * Any items that need requeuing? Stuff them into hctx->dispatch,
@@ -2680,6 +2681,7 @@  void blk_mq_try_issue_list_directly(struct blk_mq_hw_ctx *hctx,
 		list_del_init(&rq->queuelist);
 		ret = blk_mq_request_issue_directly(rq, list_empty(list));
 		if (ret != BLK_STS_OK) {
+			errors++;
 			if (ret == BLK_STS_RESOURCE ||
 					ret == BLK_STS_DEV_RESOURCE) {
 				blk_mq_request_bypass_insert(rq, false,
@@ -2687,7 +2689,6 @@  void blk_mq_try_issue_list_directly(struct blk_mq_hw_ctx *hctx,
 				break;
 			}
 			blk_mq_end_request(rq, ret);
-			errors++;
 		} else
 			queued++;
 	}