2018-08-08 08:22:56

by jianchao.wang

Subject: [PATCH] blk-mq: count the hctx as active before allocate tag

Currently, we count the hctx as active after allocate tag
successfully. The other shared-tags users could exhaust the tags
and starve a non-active and waiting-tag hctx. Count the hctx as
active before try to allocate tag to fix this.

Signed-off-by: Jianchao Wang <[email protected]>
---
block/blk-mq.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/block/blk-mq.c b/block/blk-mq.c
index ae44e85..acd40eb 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -369,6 +369,8 @@ static struct request *blk_mq_get_request(struct request_queue *q,
e->type->ops.mq.limit_depth(op, data);
}

+ blk_mq_tag_busy(data.hctx);
+
tag = blk_mq_get_tag(data);
if (tag == BLK_MQ_TAG_FAIL) {
if (put_ctx_on_error) {
@@ -972,6 +974,7 @@ bool blk_mq_get_driver_tag(struct request *rq, struct blk_mq_hw_ctx **hctx,
.hctx = blk_mq_map_queue(rq->q, rq->mq_ctx->cpu),
.flags = wait ? 0 : BLK_MQ_REQ_NOWAIT,
};
+ bool shared;

might_sleep_if(wait);

@@ -981,9 +984,10 @@ bool blk_mq_get_driver_tag(struct request *rq, struct blk_mq_hw_ctx **hctx,
if (blk_mq_tag_is_reserved(data.hctx->sched_tags, rq->internal_tag))
data.flags |= BLK_MQ_REQ_RESERVED;

+ shared = blk_mq_tag_busy(data.hctx);
rq->tag = blk_mq_get_tag(&data);
if (rq->tag >= 0) {
- if (blk_mq_tag_busy(data.hctx)) {
+ if (shared) {
rq->rq_flags |= RQF_MQ_INFLIGHT;
atomic_inc(&data.hctx->nr_active);
}
--
2.7.4

2018-08-08 09:25:26

by Ming Lei

Subject: Re: [PATCH] blk-mq: count the hctx as active before allocate tag

On Wed, Aug 8, 2018 at 4:22 PM, Jianchao Wang
<[email protected]> wrote:
> Currently, we count the hctx as active after allocate tag
> successfully. The other shared-tags users could exhaust the tags
> and starve a non-active and waiting-tag hctx. Count the hctx as
> active before try to allocate tag to fix this.

This patch looks fine.

>
> Signed-off-by: Jianchao Wang <[email protected]>
> ---
> block/blk-mq.c | 6 +++++-
> 1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/block/blk-mq.c b/block/blk-mq.c
> index ae44e85..acd40eb 100644
> --- a/block/blk-mq.c
> +++ b/block/blk-mq.c
> @@ -369,6 +369,8 @@ static struct request *blk_mq_get_request(struct request_queue *q,
> e->type->ops.mq.limit_depth(op, data);
> }
>
> + blk_mq_tag_busy(data.hctx);
> +

blk_mq_tag_busy is only required for driver tag.

Thanks,
Ming Lei

2018-08-08 09:34:26

by jianchao.wang

Subject: Re: [PATCH] blk-mq: count the hctx as active before allocate tag

On 08/08/2018 05:24 PM, Ming Lei wrote:
> On Wed, Aug 8, 2018 at 4:22 PM, Jianchao Wang
> <[email protected]> wrote:
>> Currently, we count the hctx as active after allocate tag
>> successfully. The other shared-tags users could exhaust the tags
>> and starve a non-active and waiting-tag hctx. Count the hctx as
>> active before try to allocate tag to fix this.
>
> This patch looks fine.
>
>>
>> Signed-off-by: Jianchao Wang <[email protected]>
>> ---
>> block/blk-mq.c | 6 +++++-
>> 1 file changed, 5 insertions(+), 1 deletion(-)
>>
>> diff --git a/block/blk-mq.c b/block/blk-mq.c
>> index ae44e85..acd40eb 100644
>> --- a/block/blk-mq.c
>> +++ b/block/blk-mq.c
>> @@ -369,6 +369,8 @@ static struct request *blk_mq_get_request(struct request_queue *q,
>> e->type->ops.mq.limit_depth(op, data);
>> }
>>
>> + blk_mq_tag_busy(data.hctx);
>> +
>
> blk_mq_tag_busy is only required for driver tag.
>

Yes, I will change it and post the V2.

Thanks
Jianchao