Re: [PATCH 1/7] sched/core: uclamp: add CPU clamp groups accounting

[Peter Zijlstra]

On Fri, Apr 13, 2018 at 11:30:05AM +0200, Peter Zijlstra wrote:
> On Mon, Apr 09, 2018 at 05:56:09PM +0100, Patrick Bellasi wrote:
> > +struct uclamp_group {
> > +	/* Utilization clamp value for tasks on this clamp group */
> > +	int value;
> > +	/* Number of RUNNABLE tasks on this clamp group */
> > +	int tasks;
> > +};
> 
> > +struct uclamp_cpu {
> > +	/* Utilization clamp value for a CPU */
> > +	int value;
> > +	/* Utilization clamp groups affecting this CPU */
> > +	struct uclamp_group group[CONFIG_UCLAMP_GROUPS_COUNT + 1];
> > +};
> 
> > @@ -811,6 +885,11 @@ struct rq {
> >  	unsigned long		cpu_capacity;
> >  	unsigned long		cpu_capacity_orig;
> >  
> > +#ifdef CONFIG_UCLAMP_TASK
> > +	/* util_{min,max} clamp values based on CPU's active tasks */
> > +	struct uclamp_cpu uclamp[UCLAMP_CNT];
> > +#endif
> > +
> >  	struct callback_head	*balance_callback;
> >  
> >  	unsigned char		idle_balance;
> 
> So that is:
> 
> struct rq {
> 
> 	struct uclamp_cpu {
> 		int value;
> 
> 		/* 4 byte hole */
> 
> 		struct uclamp_group {
> 			int value;
> 			int tasks;
> 		} [COUNT + 1];	// size: COUNT+2 * 8 [bytes]
> 
> 	} [2]; // size: 2 * COUNT+2 * 8 [bytes]
> 
> };

Note that you can slightly compress the data structure if you do
something like:

	struct rq {

		int active_clamp[2];

		struct uclamp_group {
			int value;
			int task;
		} [COUNT + 1][2]; // XXX check array order of C
	};

It also avoids the active values being split over 2 lines; and we need
those values every time, even if there are no active clamp tasks on the
system.