Re: [patch 1/3] sched: use jump labels to reduce overhead when bandwidth control is inactive

[Paul Turner <pjt@google.com>]

On 11/08/2011 01:29 AM, Peter Zijlstra wrote:
> On Mon, 2011-11-07 at 20:26 -0800, Paul Turner wrote:
>> @@ -1788,6 +1791,9 @@ static void do_sched_cfs_slack_timer(str
>>    */
>>   static void check_enqueue_throttle(struct cfs_rq *cfs_rq)
>>   {
>> +       if (!cfs_bandwidth_used())
>> +               return;
>> +
>>          /* an active group must be handled by the update_curr()->put() path */
>>          if (!cfs_rq->runtime_enabled || cfs_rq->curr)
>>                  return;
>> @@ -1805,6 +1811,9 @@ static void check_enqueue_throttle(struc
>>   /* conditionally throttle active cfs_rq's from put_prev_entity() */
>>   static void check_cfs_rq_runtime(struct cfs_rq *cfs_rq)
>>   {
>> +       if (!cfs_bandwidth_used())
>> +               return;
>> +
>>          if (likely(!cfs_rq->runtime_enabled || cfs_rq->runtime_remaining>  0))
>>                  return;
>>
>
> does it matter if you pull this out into an inline function like:
>
> static __always_inline void check_enqueue_throttle(struct cfs_rq *cfs_rq)
> {
> 	if (cfs_bandwidth_used())
> 		__check_enqueue_throttle(cfs_rq);
> }
>
> That would avoid the superfluous function call as well.

This this is the only call-site it's being inlined already so there's no 
difference atm.  That said, check_enqueue_throttle is sufficiently small we 
could just add __always_inline to force this to be the default behavior and 
avoid any surprises from this in the future.

Let me know if you'd like me to re-post with this added.

- Paul