| Message ID | 20200528153441.3501777-1-kbusch@kernel.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [PATCHv3,1/2] blk-mq: export __blk_mq_complete_request | expand |
On 5/28/20 9:34 AM, Keith Busch wrote: > For when drivers have a need to bypass error injection. Acked-by: Jens Axboe <axboe@kernel.dk Assuming this goes in through the NVMe tree.
On 5/28/20 8:34 AM, Keith Busch wrote: > For when drivers have a need to bypass error injection. > > Signed-off-by: Keith Busch <kbusch@kernel.org> > --- > v2->v3: Use _GPL export > > block/blk-mq.c | 3 ++- > include/linux/blk-mq.h | 1 + > 2 files changed, 3 insertions(+), 1 deletion(-) > > diff --git a/block/blk-mq.c b/block/blk-mq.c > index cac11945f602..e62559ac7c45 100644 > --- a/block/blk-mq.c > +++ b/block/blk-mq.c > @@ -556,7 +556,7 @@ static void __blk_mq_complete_request_remote(void *data) > q->mq_ops->complete(rq); > } > > -static void __blk_mq_complete_request(struct request *rq) > +void __blk_mq_complete_request(struct request *rq) > { > struct blk_mq_ctx *ctx = rq->mq_ctx; > struct request_queue *q = rq->q; > @@ -602,6 +602,7 @@ static void __blk_mq_complete_request(struct request *rq) > } > put_cpu(); > } > +EXPORT_SYMBOL_GPL(__blk_mq_complete_request); > > static void hctx_unlock(struct blk_mq_hw_ctx *hctx, int srcu_idx) > __releases(hctx->srcu) > diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h > index d7307795439a..cfe7eac3764e 100644 > --- a/include/linux/blk-mq.h > +++ b/include/linux/blk-mq.h > @@ -494,6 +494,7 @@ void blk_mq_requeue_request(struct request *rq, bool kick_requeue_list); > void blk_mq_kick_requeue_list(struct request_queue *q); > void blk_mq_delay_kick_requeue_list(struct request_queue *q, unsigned long msecs); > bool blk_mq_complete_request(struct request *rq); > +void __blk_mq_complete_request(struct request *rq); > bool blk_mq_bio_list_merge(struct request_queue *q, struct list_head *list, > struct bio *bio, unsigned int nr_segs); > bool blk_mq_queue_stopped(struct request_queue *q); The null block driver requires this as well. Otherwise, the below would hang forever because timeout handler keeps failing. # echo 100 > /sys/kernel/debug/fail_io_timeout/probability # echo 1000 > /sys/kernel/debug/fail_io_timeout/times # echo 1 > /sys/block/nullb0/io-timeout-fail # dd if=/dev/zero of=/dev/nullb0 bs=512 count=1 oflag=direct With below patch, the timeout handler is able to complete the IO. diff --git a/drivers/block/null_blk_main.c b/drivers/block/null_blk_main.c index ce9e33603a4d..ba107b264fab 100644 --- a/drivers/block/null_blk_main.c +++ b/drivers/block/null_blk_main.c @@ -1397,7 +1397,7 @@ static bool should_requeue_request(struct request *rq) static enum blk_eh_timer_return null_timeout_rq(struct request *rq, bool res) { pr_info("rq %p timed out\n", rq); - blk_mq_complete_request(rq); + __blk_mq_complete_request(rq); return BLK_EH_DONE; } I would wait for the new exported API name until I send out the above. Perhaps there are more drivers that require such kind of change, but in more specific corner case. Dongli Zhang
On Thu, May 28, 2020 at 08:34:40AM -0700, Keith Busch wrote: > For when drivers have a need to bypass error injection. > > Signed-off-by: Keith Busch <kbusch@kernel.org> Reviewed-by: Daniel Wagner <dwagner@suse.de>
On Thu, May 28, 2020 at 09:36:23AM -0600, Jens Axboe wrote: > On 5/28/20 9:34 AM, Keith Busch wrote: > > For when drivers have a need to bypass error injection. > > Acked-by: Jens Axboe <axboe@kernel.dk > > Assuming this goes in through the NVMe tree. Given that other drivers will need this as well, and the nvme queue for 5.8 is empty at the moment I think you should pick up the next version once it has better naming and documentation.
On 5/29/20 6:27 AM, Christoph Hellwig wrote: > On Thu, May 28, 2020 at 09:36:23AM -0600, Jens Axboe wrote: >> On 5/28/20 9:34 AM, Keith Busch wrote: >>> For when drivers have a need to bypass error injection. >> >> Acked-by: Jens Axboe <axboe@kernel.dk >> >> Assuming this goes in through the NVMe tree. > > Given that other drivers will need this as well, and the nvme queue for > 5.8 is empty at the moment I think you should pick up the next version > once it has better naming and documentation. Sure
diff --git a/block/blk-mq.c b/block/blk-mq.c index cac11945f602..e62559ac7c45 100644 --- a/block/blk-mq.c +++ b/block/blk-mq.c @@ -556,7 +556,7 @@ static void __blk_mq_complete_request_remote(void *data) q->mq_ops->complete(rq); } -static void __blk_mq_complete_request(struct request *rq) +void __blk_mq_complete_request(struct request *rq) { struct blk_mq_ctx *ctx = rq->mq_ctx; struct request_queue *q = rq->q; @@ -602,6 +602,7 @@ static void __blk_mq_complete_request(struct request *rq) } put_cpu(); } +EXPORT_SYMBOL_GPL(__blk_mq_complete_request); static void hctx_unlock(struct blk_mq_hw_ctx *hctx, int srcu_idx) __releases(hctx->srcu) diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h index d7307795439a..cfe7eac3764e 100644 --- a/include/linux/blk-mq.h +++ b/include/linux/blk-mq.h @@ -494,6 +494,7 @@ void blk_mq_requeue_request(struct request *rq, bool kick_requeue_list); void blk_mq_kick_requeue_list(struct request_queue *q); void blk_mq_delay_kick_requeue_list(struct request_queue *q, unsigned long msecs); bool blk_mq_complete_request(struct request *rq); +void __blk_mq_complete_request(struct request *rq); bool blk_mq_bio_list_merge(struct request_queue *q, struct list_head *list, struct bio *bio, unsigned int nr_segs); bool blk_mq_queue_stopped(struct request_queue *q);
For when drivers have a need to bypass error injection. Signed-off-by: Keith Busch <kbusch@kernel.org> --- v2->v3: Use _GPL export block/blk-mq.c | 3 ++- include/linux/blk-mq.h | 1 + 2 files changed, 3 insertions(+), 1 deletion(-)