[PATCH] blk-mq: don't call ktime_get_ns() if we don't need it

[PATCH] blk-mq: don't call ktime_get_ns() if we don't need it

[Jens Axboe]

We only need the request fields and the end_io time if we have stats
enabled, or if we have a scheduler attached as those may use it for
completion time stats.

Signed-off-by: Jens Axboe <axboe@kernel.dk>
---
 block/blk-mq.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/block/blk-mq.c b/block/blk-mq.c
index 7dcef565dc0f..2b458e9da946 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -316,7 +316,8 @@ static struct request *blk_mq_rq_ctx_init(struct blk_mq_alloc_data *data,
 	RB_CLEAR_NODE(&rq->rb_node);
 	rq->rq_disk = NULL;
 	rq->part = NULL;
-	rq->start_time_ns = ktime_get_ns();
+	if (data->q->elevator)
+		rq->start_time_ns = ktime_get_ns();
 	rq->io_start_time_ns = 0;
 	rq->nr_phys_segments = 0;
 #if defined(CONFIG_BLK_DEV_INTEGRITY)
@@ -522,12 +523,18 @@ EXPORT_SYMBOL_GPL(blk_mq_free_request);
 
 inline void __blk_mq_end_request(struct request *rq, blk_status_t error)
 {
-	u64 now = ktime_get_ns();
+	u64 now = 0;
 
+	/*
+	 * Only read ktime if we need it, which is if stats are enabled
+	 * or we have a scheduler attached that may use it.
+	 */
 	if (rq->rq_flags & RQF_STATS) {
 		blk_mq_poll_stats_start(rq->q);
+		now = ktime_get_ns();
 		blk_stat_add(rq, now);
-	}
+	} else if (rq->q->elevator)
+		now = ktime_get_ns();
 
 	if (rq->internal_tag != -1)
 		blk_mq_sched_completed_request(rq, now);
-- 
2.17.1

-- 
Jens Axboe

Re: [PATCH] blk-mq: don't call ktime_get_ns() if we don't need it

[Christoph Hellwig]

On Fri, Nov 30, 2018 at 08:56:25AM -0700, Jens Axboe wrote:
> We only need the request fields and the end_io time if we have stats
> enabled, or if we have a scheduler attached as those may use it for
> completion time stats.
> 
> Signed-off-by: Jens Axboe <axboe@kernel.dk>
> ---
>  block/blk-mq.c | 13 ++++++++++---
>  1 file changed, 10 insertions(+), 3 deletions(-)
> 
> diff --git a/block/blk-mq.c b/block/blk-mq.c
> index 7dcef565dc0f..2b458e9da946 100644
> --- a/block/blk-mq.c
> +++ b/block/blk-mq.c
> @@ -316,7 +316,8 @@ static struct request *blk_mq_rq_ctx_init(struct blk_mq_alloc_data *data,
>  	RB_CLEAR_NODE(&rq->rb_node);
>  	rq->rq_disk = NULL;
>  	rq->part = NULL;
> -	rq->start_time_ns = ktime_get_ns();
> +	if (data->q->elevator)
> +		rq->start_time_ns = ktime_get_ns();

We also seem to use start_time_ns it in blk_account_io_done.

Also update it in attempt_merge, which might have some odd results
for completly uninitialized fields.

Re: [PATCH] blk-mq: don't call ktime_get_ns() if we don't need it

[Jens Axboe]

On 11/30/18 10:27 AM, Christoph Hellwig wrote:
> On Fri, Nov 30, 2018 at 08:56:25AM -0700, Jens Axboe wrote:
>> We only need the request fields and the end_io time if we have stats
>> enabled, or if we have a scheduler attached as those may use it for
>> completion time stats.
>>
>> Signed-off-by: Jens Axboe <axboe@kernel.dk>
>> ---
>>  block/blk-mq.c | 13 ++++++++++---
>>  1 file changed, 10 insertions(+), 3 deletions(-)
>>
>> diff --git a/block/blk-mq.c b/block/blk-mq.c
>> index 7dcef565dc0f..2b458e9da946 100644
>> --- a/block/blk-mq.c
>> +++ b/block/blk-mq.c
>> @@ -316,7 +316,8 @@ static struct request *blk_mq_rq_ctx_init(struct blk_mq_alloc_data *data,
>>  	RB_CLEAR_NODE(&rq->rb_node);
>>  	rq->rq_disk = NULL;
>>  	rq->part = NULL;
>> -	rq->start_time_ns = ktime_get_ns();
>> +	if (data->q->elevator)
>> +		rq->start_time_ns = ktime_get_ns();
> 
> We also seem to use start_time_ns it in blk_account_io_done.

But only if RQF_STATS is set, in which case we are passing it in correctly.

> Also update it in attempt_merge, which might have some odd results
> for completly uninitialized fields.

I did notice these, but it's pointless to change those, as you'd then have
to check validity. Nobody cares at that point...

-- 
Jens Axboe

Re: [PATCH] blk-mq: don't call ktime_get_ns() if we don't need it

[Jens Axboe]

On 11/30/18 10:29 AM, Jens Axboe wrote:
> On 11/30/18 10:27 AM, Christoph Hellwig wrote:
>> On Fri, Nov 30, 2018 at 08:56:25AM -0700, Jens Axboe wrote:
>>> We only need the request fields and the end_io time if we have stats
>>> enabled, or if we have a scheduler attached as those may use it for
>>> completion time stats.
>>>
>>> Signed-off-by: Jens Axboe <axboe@kernel.dk>
>>> ---
>>>  block/blk-mq.c | 13 ++++++++++---
>>>  1 file changed, 10 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/block/blk-mq.c b/block/blk-mq.c
>>> index 7dcef565dc0f..2b458e9da946 100644
>>> --- a/block/blk-mq.c
>>> +++ b/block/blk-mq.c
>>> @@ -316,7 +316,8 @@ static struct request *blk_mq_rq_ctx_init(struct blk_mq_alloc_data *data,
>>>  	RB_CLEAR_NODE(&rq->rb_node);
>>>  	rq->rq_disk = NULL;
>>>  	rq->part = NULL;
>>> -	rq->start_time_ns = ktime_get_ns();
>>> +	if (data->q->elevator)
>>> +		rq->start_time_ns = ktime_get_ns();
>>
>> We also seem to use start_time_ns it in blk_account_io_done.
> 
> But only if RQF_STATS is set, in which case we are passing it in correctly.

Oh, needs a stats check for this one, for some reason I was thinking
of the end_io path. Will update.

-- 
Jens Axboe