diff mbox series

[v3,1/5] blk-mq: introduce blk_mq_set_request_complete

Message ID 20210121070330.19701-2-lengchao@huawei.com (mailing list archive)
State New, archived
Headers show
Series avoid double request completion and IO error | expand

Commit Message

Chao Leng Jan. 21, 2021, 7:03 a.m. UTC 
nvme drivers need to set the state of request to MQ_RQ_COMPLETE when
directly complete request in queue_rq.
So add blk_mq_set_request_complete.

Signed-off-by: Chao Leng <lengchao@huawei.com>
---
 include/linux/blk-mq.h | 5 +++++
 1 file changed, 5 insertions(+)

Comments

Christoph Hellwig Jan. 21, 2021, 8:40 a.m. UTC | #1 
On Thu, Jan 21, 2021 at 03:03:26PM +0800, Chao Leng wrote:
> nvme drivers need to set the state of request to MQ_RQ_COMPLETE when
> directly complete request in queue_rq.
> So add blk_mq_set_request_complete.

So I'm not happy with this helper.  It should at least:

 a) be named and documented to only apply for the ->queue_rq faіlure case
 b) check that the request is in MQ_RQ_IDLE state
Chao Leng Jan. 22, 2021, 1:46 a.m. UTC | #2 
On 2021/1/21 16:40, Christoph Hellwig wrote:
> On Thu, Jan 21, 2021 at 03:03:26PM +0800, Chao Leng wrote:
>> nvme drivers need to set the state of request to MQ_RQ_COMPLETE when
>> directly complete request in queue_rq.
>> So add blk_mq_set_request_complete.
> 
> So I'm not happy with this helper.  It should at least:
> 
>   a) be named and documented to only apply for the ->queue_rq faіlure case
Although blk_mq_complete_request_remote can set the markup directly,
blk_mq_complete_request_remote can also use this helper. This helper do
not need special processing for ->queue_rq failure.
>   b) check that the request is in MQ_RQ_IDLE state
No, the request may also be in MQ_RQ_IN_FLIGHT state. Do not need to
care whether the original state is MQ_RQ_IDLE or MQ_RQ_IN_FLIGHT.
> .
>
diff mbox series

Patch

diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h
index 47b021952ac7..fc096b04bb7a 100644
--- a/include/linux/blk-mq.h
+++ b/include/linux/blk-mq.h
@@ -494,6 +494,11 @@  static inline int blk_mq_request_completed(struct request *rq)
 	return blk_mq_rq_state(rq) == MQ_RQ_COMPLETE;
 }
 
+static inline void blk_mq_set_request_complete(struct request *rq)
+{
+	WRITE_ONCE(rq->state, MQ_RQ_COMPLETE);
+}
+
 void blk_mq_start_request(struct request *rq);
 void blk_mq_end_request(struct request *rq, blk_status_t error);
 void __blk_mq_end_request(struct request *rq, blk_status_t error);