Re: [PATCH 4/4] blk-mq: mark ctx as pending at batch in flush plug path

[Ming Lei]

Hi Jeff,

On Thu, Oct 15, 2015 at 11:26 PM, Jeff Moyer <jmoyer@xxxxxxxxxx> wrote:
> Ming Lei <ming.lei@xxxxxxxxxxxxx> writes:
>
>> Most of times, flush plug should be the hottest I/O path,
>> so mark ctx as pending after all requests in the list are
>> inserted.
>
> Hi, Ming,
>
> Did you see some performance gain from this?

Not done the test yet, but I think it is the correct thing to do because
the flag can be cleared quite frequently from other CPUs, and the numberof
marking pending can be decreased to ~1/16 of previous one.

>
> -Jeff
>
>>
>> Signed-off-by: Ming Lei <ming.lei@xxxxxxxxxxxxx>
>> ---
>>  block/blk-mq.c | 18 +++++++++++++-----
>>  1 file changed, 13 insertions(+), 5 deletions(-)
>>
>> diff --git a/block/blk-mq.c b/block/blk-mq.c
>> index deb5f4c..1c943b9 100644
>> --- a/block/blk-mq.c
>> +++ b/block/blk-mq.c
>> @@ -993,18 +993,25 @@ void blk_mq_delay_queue(struct blk_mq_hw_ctx *hctx, unsigned long msecs)
>>  }
>>  EXPORT_SYMBOL(blk_mq_delay_queue);
>>
>> -static void __blk_mq_insert_request(struct blk_mq_hw_ctx *hctx,
>> -                                 struct request *rq, bool at_head)
>> +static inline void __blk_mq_insert_req_list(struct blk_mq_hw_ctx *hctx,
>> +                                         struct blk_mq_ctx *ctx,
>> +                                         struct request *rq,
>> +                                         bool at_head)
>>  {
>> -     struct blk_mq_ctx *ctx = rq->mq_ctx;
>> -
>>       trace_block_rq_insert(hctx->queue, rq);
>>
>>       if (at_head)
>>               list_add(&rq->queuelist, &ctx->rq_list);
>>       else
>>               list_add_tail(&rq->queuelist, &ctx->rq_list);
>> +}
>> +
>> +static void __blk_mq_insert_request(struct blk_mq_hw_ctx *hctx,
>> +                                 struct request *rq, bool at_head)
>> +{
>> +     struct blk_mq_ctx *ctx = rq->mq_ctx;
>>
>> +     __blk_mq_insert_req_list(hctx, ctx, rq, at_head);
>>       blk_mq_hctx_mark_pending(hctx, ctx);
>>  }
>>
>> @@ -1060,8 +1067,9 @@ static void blk_mq_insert_requests(struct request_queue *q,
>>               rq = list_first_entry(list, struct request, queuelist);
>>               list_del_init(&rq->queuelist);
>>               rq->mq_ctx = ctx;
>> -             __blk_mq_insert_request(hctx, rq, false);
>> +             __blk_mq_insert_req_list(hctx, ctx, rq, false);
>>       }
>> +     blk_mq_hctx_mark_pending(hctx, ctx);
>>       spin_unlock(&ctx->lock);
>>
>>       blk_mq_run_hw_queue(hctx, from_schedule);
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/