[AMD Official Use Only - General]
Hi Wyes:
> -----Original Message-----
> From: Karny, Wyes <Wyes.Karny@xxxxxxx>
> Sent: Monday, October 9, 2023 2:19 PM
> To: Meng, Li (Jassmine) <Li.Meng@xxxxxxx>
> Cc: Karny, Wyes <Wyes.Karny@xxxxxxx>; Rafael J . Wysocki
> <rafael.j.wysocki@xxxxxxxxx>; Huang, Ray <Ray.Huang@xxxxxxx>; linux-
> pm@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; x86@xxxxxxxxxx; linux-
> acpi@xxxxxxxxxxxxxxx; Shuah Khan <skhan@xxxxxxxxxxxxxxxxxxx>; linux-
> kselftest@xxxxxxxxxxxxxxx; Fontenot, Nathan
> <Nathan.Fontenot@xxxxxxx>; Sharma, Deepak
> <Deepak.Sharma@xxxxxxx>; Deucher, Alexander
> <Alexander.Deucher@xxxxxxx>; Limonciello, Mario
> <Mario.Limonciello@xxxxxxx>; Huang, Shimmer
> <Shimmer.Huang@xxxxxxx>; Yuan, Perry <Perry.Yuan@xxxxxxx>; Du,
> Xiaojian <Xiaojian.Du@xxxxxxx>; Viresh Kumar <viresh.kumar@xxxxxxxxxx>;
> Borislav Petkov <bp@xxxxxxxxx>
> Subject: Re: [PATCH V8 5/7] cpufreq: amd-pstate: Update amd-pstate
> preferred core ranking dynamically
>
> Hi Meng Li,
>
> On 09 Oct 10:49, Meng Li wrote:
> > Preferred core rankings can be changed dynamically by the platform
> > based on the workload and platform conditions and accounting for
> > thermals and aging.
> > When this occurs, cpu priority need to be set.
> >
> > Reviewed-by: Mario Limonciello <mario.limonciello@xxxxxxx>
> > Reviewed-by: Wyes Karny <wyes.karny@xxxxxxx>
> > Reviewed-by: Huang Rui <ray.huang@xxxxxxx>
> > Signed-off-by: Meng Li <li.meng@xxxxxxx>
> > ---
> > drivers/cpufreq/amd-pstate.c | 34
> ++++++++++++++++++++++++++++++++--
> > include/linux/amd-pstate.h | 6 ++++++
> > 2 files changed, 38 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/cpufreq/amd-pstate.c
> > b/drivers/cpufreq/amd-pstate.c index 6ac8939fce5a..d3369247c6c9 100644
> > --- a/drivers/cpufreq/amd-pstate.c
> > +++ b/drivers/cpufreq/amd-pstate.c
> > @@ -313,6 +313,7 @@ static int pstate_init_perf(struct amd_cpudata
> *cpudata)
> > WRITE_ONCE(cpudata->nominal_perf,
> AMD_CPPC_NOMINAL_PERF(cap1));
> > WRITE_ONCE(cpudata->lowest_nonlinear_perf,
> AMD_CPPC_LOWNONLIN_PERF(cap1));
> > WRITE_ONCE(cpudata->lowest_perf,
> AMD_CPPC_LOWEST_PERF(cap1));
> > + WRITE_ONCE(cpudata->prefcore_ranking,
> AMD_CPPC_HIGHEST_PERF(cap1));
> >
> > return 0;
> > }
> > @@ -334,6 +335,7 @@ static int cppc_init_perf(struct amd_cpudata
> *cpudata)
> > WRITE_ONCE(cpudata->lowest_nonlinear_perf,
> > cppc_perf.lowest_nonlinear_perf);
> > WRITE_ONCE(cpudata->lowest_perf, cppc_perf.lowest_perf);
> > + WRITE_ONCE(cpudata->prefcore_ranking, cppc_perf.highest_perf);
> >
> > if (cppc_state == AMD_PSTATE_ACTIVE)
> > return 0;
> > @@ -540,7 +542,7 @@ static void amd_pstate_adjust_perf(unsigned int
> cpu,
> > if (target_perf < capacity)
> > des_perf = DIV_ROUND_UP(cap_perf * target_perf,
> capacity);
> >
> > - min_perf = READ_ONCE(cpudata->highest_perf);
> > + min_perf = READ_ONCE(cpudata->lowest_perf);
>
> This seems to be a fix. So, this could be a separate patch.
[Meng, Li (Jassmine)] Thanks, I will modify it.
>
> > if (_min_perf < capacity)
> > min_perf = DIV_ROUND_UP(cap_perf * _min_perf, capacity);
> >
> > @@ -760,6 +762,32 @@ static void amd_pstate_init_prefcore(struct
> amd_cpudata *cpudata)
> > }
> > }
> >
> > +static void amd_pstate_update_highest_perf(unsigned int cpu) {
> > + struct cpufreq_policy *policy;
> > + struct amd_cpudata *cpudata;
> > + u32 prev_high = 0, cur_high = 0;
> > + int ret;
> > +
> > + if ((!amd_pstate_prefcore) || (!cpudata->hw_prefcore))
> > + return;
> > +
> > + ret = amd_pstate_get_highest_perf(cpu, &cur_high);
> > + if (ret)
> > + return;
> > +
> > + policy = cpufreq_cpu_get(cpu);
> > + cpudata = policy->driver_data;
> > + prev_high = READ_ONCE(cpudata->prefcore_ranking);
> > +
> > + if (prev_high != cur_high) {
> > + WRITE_ONCE(cpudata->prefcore_ranking, cur_high);
> > + sched_set_itmt_core_prio(cur_high, cpu);
> > + }
> > +
> > + cpufreq_cpu_put(policy);
> > +}
> > +
> > static int amd_pstate_cpu_init(struct cpufreq_policy *policy) {
> > int min_freq, max_freq, nominal_freq, lowest_nonlinear_freq, ret;
> @@
> > -926,7 +954,7 @@ static ssize_t show_amd_pstate_highest_perf(struct
> cpufreq_policy *policy,
> > u32 perf;
> > struct amd_cpudata *cpudata = policy->driver_data;
> >
> > - perf = READ_ONCE(cpudata->highest_perf);
> > + perf = READ_ONCE(cpudata->prefcore_ranking);
>
> I think this should show cpudata->highest_perf.
[Meng, Li (Jassmine)] Thanks, I will modify it.
Add a new function for prefcore_ranking.
>
> Thanks,
> Wyes
> >
> > return sysfs_emit(buf, "%u\n", perf); } @@ -1502,6 +1530,7 @@
> > static struct cpufreq_driver amd_pstate_driver = {
> > .suspend = amd_pstate_cpu_suspend,
> > .resume = amd_pstate_cpu_resume,
> > .set_boost = amd_pstate_set_boost,
> > + .update_highest_perf = amd_pstate_update_highest_perf,
> > .name = "amd-pstate",
> > .attr = amd_pstate_attr,
> > };
> > @@ -1516,6 +1545,7 @@ static struct cpufreq_driver
> amd_pstate_epp_driver = {
> > .online = amd_pstate_epp_cpu_online,
> > .suspend = amd_pstate_epp_suspend,
> > .resume = amd_pstate_epp_resume,
> > + .update_highest_perf = amd_pstate_update_highest_perf,
> > .name = "amd-pstate-epp",
> > .attr = amd_pstate_epp_attr,
> > };
> > diff --git a/include/linux/amd-pstate.h b/include/linux/amd-pstate.h
> > index 87e140e9e6db..426822612373 100644
> > --- a/include/linux/amd-pstate.h
> > +++ b/include/linux/amd-pstate.h
> > @@ -39,11 +39,16 @@ struct amd_aperf_mperf {
> > * @cppc_req_cached: cached performance request hints
> > * @highest_perf: the maximum performance an individual processor may
> reach,
> > * assuming ideal conditions
> > + * For platforms that do not support the preferred core
> feature, the
> > + * highest_pef may be configured with 166 or 255, to avoid
> max frequency
> > + * calculated wrongly. we take the fixed value as the
> highest_perf.
> > * @nominal_perf: the maximum sustained performance level of the
> processor,
> > * assuming ideal operating conditions
> > * @lowest_nonlinear_perf: the lowest performance level at which
> nonlinear power
> > * savings are achieved
> > * @lowest_perf: the absolute lowest performance level of the
> > processor
> > + * @prefcore_ranking: the preferred core ranking, the higher value
> indicates a higher
> > + * priority.
> > * @max_freq: the frequency that mapped to highest_perf
> > * @min_freq: the frequency that mapped to lowest_perf
> > * @nominal_freq: the frequency that mapped to nominal_perf @@ -73,6
> > +78,7 @@ struct amd_cpudata {
> > u32 nominal_perf;
> > u32 lowest_nonlinear_perf;
> > u32 lowest_perf;
> > + u32 prefcore_ranking;
> >
> > u32 max_freq;
> > u32 min_freq;
> > --
> > 2.34.1
> >
