Re: [PATCH v7 04/15] PM / EM: add support for other devices than CPUs in Energy Model

[Quentin Perret <qperret@google.com>]

Hey Lukasz,

On Monday 11 May 2020 at 12:19:01 (+0100), Lukasz Luba wrote:
<snip>
> @@ -27,12 +29,15 @@ struct em_perf_state {
>   * em_perf_domain - Performance domain
>   * @table:		List of performance states, in ascending order
>   * @nr_perf_states:	Number of performance states
> - * @cpus:		Cpumask covering the CPUs of the domain
> + * @cpus:		Cpumask covering the CPUs of the domain. It's here
> + *			for performance reasons to avoid potential cache
> + *			misses during energy calculations in the scheduler

And because that saves a pointer, and simplifies allocating/freeing that
memory region :)

<snip>
> diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
> index 5b8a1566526a..9cc7f2973600 100644
> --- a/kernel/power/energy_model.c
> +++ b/kernel/power/energy_model.c
> @@ -2,8 +2,9 @@
>  /*
>   * Energy Model of CPUs

Should this comment change too?

<snip>
> -static void em_debug_create_pd(struct em_perf_domain *pd, int cpu)
> +static void em_debug_create_pd(struct device *dev)
>  {
>  	struct dentry *d;
> -	char name[8];
>  	int i;
>  
> -	snprintf(name, sizeof(name), "pd%d", cpu);
> -
>  	/* Create the directory of the performance domain */
> -	d = debugfs_create_dir(name, rootdir);
> +	d = debugfs_create_dir(dev_name(dev), rootdir);

So what will be the name for the perf domain of CPUs now? cpuX?

<snip>
> @@ -142,8 +149,8 @@ em_create_pd(struct device *dev, int nr_states, struct em_data_callback *cb,
>  		 */
>  		opp_eff = freq / power;
>  		if (opp_eff >= prev_opp_eff)
> -			pr_warn("pd%d: hertz/watts ratio non-monotonically decreasing: em_perf_state %d >= em_perf_state%d\n",
> -					cpu, i, i - 1);
> +			dev_dbg(dev, "EM: hertz/watts ratio non-monotonically decreasing: em_perf_state %d >= em_perf_state%d\n",
> +					i, i - 1);

It feels like changing from warn to debug doesn't really belong to this
patch no?

<snip>
> @@ -216,47 +274,50 @@ int em_dev_register_perf_domain(struct device *dev, unsigned int nr_states,
>  	 */
>  	mutex_lock(&em_pd_mutex);
>  
> -	for_each_cpu(cpu, span) {
> -		/* Make sure we don't register again an existing domain. */
> -		if (READ_ONCE(per_cpu(em_data, cpu))) {
> -			ret = -EEXIST;
> -			goto unlock;
> -		}
> +	if (dev->em_pd) {
> +		ret = -EEXIST;
> +		goto unlock;
> +	}
>  
> -		/*
> -		 * All CPUs of a domain must have the same micro-architecture
> -		 * since they all share the same table.
> -		 */
> -		cap = arch_scale_cpu_capacity(cpu);
> -		if (prev_cap && prev_cap != cap) {
> -			pr_err("CPUs of %*pbl must have the same capacity\n",
> -							cpumask_pr_args(span));
> +	if (_is_cpu_device(dev)) {

Something like

	if (!_is_cpu_device(dev))
		goto device;

would limit the diff a bit, but that may just be personal taste.

But appart from these nits, the patch LGTM.

Thanks,
Quentin