Re: [RFC PATCH 4/7] sched: cpufreq: Remove magic 1.25 headroom from apply_dvfs_headroom()

[Peter Zijlstra]

On Mon, Aug 28, 2023 at 12:32:00AM +0100, Qais Yousef wrote:
> Instead of the magical 1.25 headroom, use the new approximate_util_avg()
> to provide headroom based on the dvfs_update_delay; which is the period
> at which the cpufreq governor will send DVFS updates to the hardware.
> 
> Add a new percpu dvfs_update_delay that can be cheaply accessed whenever
> apply_dvfs_headroom() is called. We expect cpufreq governors that rely
> on util to drive its DVFS logic/algorithm to populate these percpu
> variables. schedutil is the only such governor at the moment.
> 
> Signed-off-by: Qais Yousef (Google) <qyousef@xxxxxxxxxxx>
> ---
>  kernel/sched/core.c              |  3 ++-
>  kernel/sched/cpufreq_schedutil.c | 10 +++++++++-
>  kernel/sched/sched.h             | 25 ++++++++++++++-----------
>  3 files changed, 25 insertions(+), 13 deletions(-)
> 
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 602e369753a3..f56eb44745a8 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -116,6 +116,7 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(sched_util_est_se_tp);
>  EXPORT_TRACEPOINT_SYMBOL_GPL(sched_update_nr_running_tp);
>  
>  DEFINE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues);
> +DEFINE_PER_CPU_SHARED_ALIGNED(u64, dvfs_update_delay);

This makes no sense, why are you using SHARED_ALIGNED and thus wasting
an entire cacheline for the one variable?