| Message ID | 20220304060724.314582-1-Jinzhou.Su@amd.com |
|---|---|
| Headers | show |
| Series | Add tracer tool for AMD P-State driver | expand |
On Sat, Mar 05, 2022 at 02:49:51AM +0800, Rafael J. Wysocki wrote: > On Fri, Mar 4, 2022 at 7:42 AM Huang Rui <ray.huang@amd.com> wrote: > > > > On Fri, Mar 04, 2022 at 02:07:21PM +0800, Su, Jinzhou (Joe) wrote: > > > Add frequency, mperf, aperf and tsc in the trace. This can be used > > > to debug and tune the performance of AMD P-state driver. > > > > > > Use the time difference between amd_pstate_update to calculate CPU > > > frequency. There could be sleep in arch_freq_get_on_cpu, so do not > > > use it here. > > > > > > Signed-off-by: Jinzhou Su <Jinzhou.Su@amd.com> > > > Co-developed-by: Huang Rui <ray.huang@amd.com> > > > Signed-off-by: Huang Rui <ray.huang@amd.com> > > > > Let's remove "Signed-off-by" of me, just leave "Co-developed-by". > > Actually, they both need to be present (the C-d-b clarifies the S-o-b > meaning), so the above is correct. > OK, I see. Thanks to clarify this. Best Regards, Ray
[AMD Official Use Only] > -----Original Message----- > From: Huang, Ray <Ray.Huang@amd.com> > Sent: Friday, March 4, 2022 2:42 PM > To: Su, Jinzhou (Joe) <Jinzhou.Su@amd.com> > Cc: rjw@rjwysocki.net; linux-pm@vger.kernel.org; > srinivas.pandruvada@linux.intel.com; dsmythies@telus.net; > viresh.kumar@linaro.org; todd.e.brandt@linux.intel.com; linux- > kernel@vger.kernel.org; Sharma, Deepak <Deepak.Sharma@amd.com>; > Deucher, Alexander <Alexander.Deucher@amd.com>; Du, Xiaojian > <Xiaojian.Du@amd.com>; Yuan, Perry <Perry.Yuan@amd.com>; Meng, Li > (Jassmine) <Li.Meng@amd.com> > Subject: Re: [PATCH V2 1/4] cpufreq: amd-pstate: Add more tracepoint for > AMD P-State module > > On Fri, Mar 04, 2022 at 02:07:21PM +0800, Su, Jinzhou (Joe) wrote: > > Add frequency, mperf, aperf and tsc in the trace. This can be used to > > debug and tune the performance of AMD P-state driver. > > > > Use the time difference between amd_pstate_update to calculate CPU > > frequency. There could be sleep in arch_freq_get_on_cpu, so do not use > > it here. > > > > Signed-off-by: Jinzhou Su <Jinzhou.Su@amd.com> > > Co-developed-by: Huang Rui <ray.huang@amd.com> > > Signed-off-by: Huang Rui <ray.huang@amd.com> > > Let's remove "Signed-off-by" of me, just leave "Co-developed-by". > > > --- > > drivers/cpufreq/amd-pstate-trace.h | 22 ++++++++++- > > drivers/cpufreq/amd-pstate.c | 59 +++++++++++++++++++++++++++++- > > 2 files changed, 78 insertions(+), 3 deletions(-) > > > > diff --git a/drivers/cpufreq/amd-pstate-trace.h > > b/drivers/cpufreq/amd-pstate-trace.h > > index 647505957d4f..35f38ae67fb1 100644 > > --- a/drivers/cpufreq/amd-pstate-trace.h > > +++ b/drivers/cpufreq/amd-pstate-trace.h > > @@ -27,6 +27,10 @@ TRACE_EVENT(amd_pstate_perf, > > TP_PROTO(unsigned long min_perf, > > unsigned long target_perf, > > unsigned long capacity, > > + u64 freq, > > + u64 mperf, > > + u64 aperf, > > + u64 tsc, > > unsigned int cpu_id, > > bool changed, > > bool fast_switch > > @@ -35,6 +39,10 @@ TRACE_EVENT(amd_pstate_perf, > > TP_ARGS(min_perf, > > target_perf, > > capacity, > > + freq, > > + mperf, > > + aperf, > > + tsc, > > cpu_id, > > changed, > > fast_switch > > @@ -44,6 +52,10 @@ TRACE_EVENT(amd_pstate_perf, > > __field(unsigned long, min_perf) > > __field(unsigned long, target_perf) > > __field(unsigned long, capacity) > > + __field(unsigned long long, freq) > > + __field(unsigned long long, mperf) > > + __field(unsigned long long, aperf) > > + __field(unsigned long long, tsc) > > __field(unsigned int, cpu_id) > > __field(bool, changed) > > __field(bool, fast_switch) > > @@ -53,15 +65,23 @@ TRACE_EVENT(amd_pstate_perf, > > __entry->min_perf = min_perf; > > __entry->target_perf = target_perf; > > __entry->capacity = capacity; > > + __entry->freq = freq; > > + __entry->mperf = mperf; > > + __entry->aperf = aperf; > > + __entry->tsc = tsc; > > __entry->cpu_id = cpu_id; > > __entry->changed = changed; > > __entry->fast_switch = fast_switch; > > ), > > > > - TP_printk("amd_min_perf=%lu amd_des_perf=%lu > amd_max_perf=%lu cpu_id=%u changed=%s fast_switch=%s", > > + TP_printk("amd_min_perf=%lu amd_des_perf=%lu > amd_max_perf=%lu > > +freq=%llu mperf=%llu aperf=%llu tsc=%llu cpu_id=%u changed=%s > > +fast_switch=%s", > > (unsigned long)__entry->min_perf, > > (unsigned long)__entry->target_perf, > > (unsigned long)__entry->capacity, > > + (unsigned long long)__entry->freq, > > + (unsigned long long)__entry->mperf, > > + (unsigned long long)__entry->aperf, > > + (unsigned long long)__entry->tsc, > > (unsigned int)__entry->cpu_id, > > (__entry->changed) ? "true" : "false", > > (__entry->fast_switch) ? "true" : "false" > > diff --git a/drivers/cpufreq/amd-pstate.c > > b/drivers/cpufreq/amd-pstate.c index 9ce75ed11f8e..7be38bc6a673 100644 > > --- a/drivers/cpufreq/amd-pstate.c > > +++ b/drivers/cpufreq/amd-pstate.c > > @@ -65,6 +65,18 @@ MODULE_PARM_DESC(shared_mem, > > > > static struct cpufreq_driver amd_pstate_driver; > > > > +/** > > + * struct amd_aperf_mperf > > + * @aperf: actual performance frequency clock count > > + * @mperf: maximum performance frequency clock count > > + * @tsc: time stamp counter > > + */ > > +struct amd_aperf_mperf { > > + u64 aperf; > > + u64 mperf; > > + u64 tsc; > > +}; > > + > > /** > > * struct amd_cpudata - private CPU data for AMD P-State > > * @cpu: CPU number > > @@ -81,6 +93,9 @@ static struct cpufreq_driver amd_pstate_driver; > > * @min_freq: the frequency that mapped to lowest_perf > > * @nominal_freq: the frequency that mapped to nominal_perf > > * @lowest_nonlinear_freq: the frequency that mapped to > > lowest_nonlinear_perf > > + * @cur: Difference of Aperf/Mperf/tsc count between last and current > > + sample > > + * @prev: Last Aperf/Mperf/tsc count value read from register > > + * @freq: current cpu frequency value > > * @boost_supported: check whether the Processor or SBIOS supports boost > mode > > * > > * The amd_cpudata is key private data for each CPU thread in AMD > > P-State, and @@ -102,6 +117,10 @@ struct amd_cpudata { > > u32 nominal_freq; > > u32 lowest_nonlinear_freq; > > > > + struct amd_aperf_mperf cur; > > + struct amd_aperf_mperf prev; > > + > > + u64 freq; > > bool boost_supported; > > }; > > > > @@ -211,6 +230,39 @@ static inline void amd_pstate_update_perf(struct > amd_cpudata *cpudata, > > max_perf, fast_switch); > > } > > > > +static inline bool amd_pstate_sample(struct amd_cpudata *cpudata) { > > + u64 aperf, mperf, tsc; > > + unsigned long flags; > > + > > + local_irq_save(flags); > > + rdmsrl(MSR_IA32_APERF, aperf); > > + rdmsrl(MSR_IA32_MPERF, mperf); > > + tsc = rdtsc(); > > + > > + if (cpudata->prev.mperf == mperf || cpudata->prev.tsc == tsc) { > > + local_irq_restore(flags); > > + return false; > > + } > > + > > + local_irq_restore(flags); > > + > > + cpudata->cur.aperf = aperf; > > + cpudata->cur.mperf = mperf; > > + cpudata->cur.tsc = tsc; > > + cpudata->cur.aperf -= cpudata->prev.aperf; > > + cpudata->cur.mperf -= cpudata->prev.mperf; > > + cpudata->cur.tsc -= cpudata->prev.tsc; > > + > > + cpudata->prev.aperf = aperf; > > + cpudata->prev.mperf = mperf; > > + cpudata->prev.tsc = tsc; > > + > > + cpudata->freq = div64_u64((cpudata->cur.aperf * cpu_khz), > > +cpudata->cur.mperf); > > + > > + return true; > > +} > > + > > static void amd_pstate_update(struct amd_cpudata *cpudata, u32 > min_perf, > > u32 des_perf, u32 max_perf, bool fast_switch) > { @@ -226,8 > > +278,11 @@ static void amd_pstate_update(struct amd_cpudata *cpudata, > u32 min_perf, > > value &= ~AMD_CPPC_MAX_PERF(~0L); > > value |= AMD_CPPC_MAX_PERF(max_perf); > > > > - trace_amd_pstate_perf(min_perf, des_perf, max_perf, > > - cpudata->cpu, (value != prev), fast_switch); > > + if (trace_amd_pstate_perf_enabled() && > amd_pstate_sample(cpudata)) { > > + trace_amd_pstate_perf(min_perf, des_perf, max_perf, > cpudata->freq, > > + cpudata->cur.mperf, cpudata->cur.aperf, cpudata- > >cur.tsc, > > + cpudata->cpu, (value != prev), fast_switch); > > How about using struct amd_aperf_mperf pointer as one input: > > trace_amd_pstate_perf(min_perf, des_perf, max_perf, &cpudata->cur, ...); > > You can refer the members of struct amd_aperf_mperf in the > amd-pstate-trace.h: > > __entry->mperf = cur->mperf; > __entry->aperf = cur->aperf; > __entry->tsc = cur->tsc; > I prefer the former way. We'd better to split the definition of struct "amd_cpudata" into head file and include it in the trace file with your change. Will do that in the future if needed. > Thanks, > Ray
On Wed, Mar 09, 2022 at 09:23:38AM +0800, Su, Jinzhou (Joe) wrote: > [AMD Official Use Only] > > > -----Original Message----- > > From: Huang, Ray <Ray.Huang@amd.com> > > Sent: Friday, March 4, 2022 2:42 PM > > To: Su, Jinzhou (Joe) <Jinzhou.Su@amd.com> > > Cc: rjw@rjwysocki.net; linux-pm@vger.kernel.org; > > srinivas.pandruvada@linux.intel.com; dsmythies@telus.net; > > viresh.kumar@linaro.org; todd.e.brandt@linux.intel.com; linux- > > kernel@vger.kernel.org; Sharma, Deepak <Deepak.Sharma@amd.com>; > > Deucher, Alexander <Alexander.Deucher@amd.com>; Du, Xiaojian > > <Xiaojian.Du@amd.com>; Yuan, Perry <Perry.Yuan@amd.com>; Meng, Li > > (Jassmine) <Li.Meng@amd.com> > > Subject: Re: [PATCH V2 1/4] cpufreq: amd-pstate: Add more tracepoint for > > AMD P-State module > > > > On Fri, Mar 04, 2022 at 02:07:21PM +0800, Su, Jinzhou (Joe) wrote: > > > Add frequency, mperf, aperf and tsc in the trace. This can be used to > > > debug and tune the performance of AMD P-state driver. > > > > > > Use the time difference between amd_pstate_update to calculate CPU > > > frequency. There could be sleep in arch_freq_get_on_cpu, so do not use > > > it here. > > > > > > Signed-off-by: Jinzhou Su <Jinzhou.Su@amd.com> > > > Co-developed-by: Huang Rui <ray.huang@amd.com> > > > Signed-off-by: Huang Rui <ray.huang@amd.com> > > > > Let's remove "Signed-off-by" of me, just leave "Co-developed-by". > > > > > --- > > > drivers/cpufreq/amd-pstate-trace.h | 22 ++++++++++- > > > drivers/cpufreq/amd-pstate.c | 59 +++++++++++++++++++++++++++++- > > > 2 files changed, 78 insertions(+), 3 deletions(-) > > > > > > diff --git a/drivers/cpufreq/amd-pstate-trace.h > > > b/drivers/cpufreq/amd-pstate-trace.h > > > index 647505957d4f..35f38ae67fb1 100644 > > > --- a/drivers/cpufreq/amd-pstate-trace.h > > > +++ b/drivers/cpufreq/amd-pstate-trace.h > > > @@ -27,6 +27,10 @@ TRACE_EVENT(amd_pstate_perf, > > > TP_PROTO(unsigned long min_perf, > > > unsigned long target_perf, > > > unsigned long capacity, > > > + u64 freq, > > > + u64 mperf, > > > + u64 aperf, > > > + u64 tsc, > > > unsigned int cpu_id, > > > bool changed, > > > bool fast_switch > > > @@ -35,6 +39,10 @@ TRACE_EVENT(amd_pstate_perf, > > > TP_ARGS(min_perf, > > > target_perf, > > > capacity, > > > + freq, > > > + mperf, > > > + aperf, > > > + tsc, > > > cpu_id, > > > changed, > > > fast_switch > > > @@ -44,6 +52,10 @@ TRACE_EVENT(amd_pstate_perf, > > > __field(unsigned long, min_perf) > > > __field(unsigned long, target_perf) > > > __field(unsigned long, capacity) > > > + __field(unsigned long long, freq) > > > + __field(unsigned long long, mperf) > > > + __field(unsigned long long, aperf) > > > + __field(unsigned long long, tsc) > > > __field(unsigned int, cpu_id) > > > __field(bool, changed) > > > __field(bool, fast_switch) > > > @@ -53,15 +65,23 @@ TRACE_EVENT(amd_pstate_perf, > > > __entry->min_perf = min_perf; > > > __entry->target_perf = target_perf; > > > __entry->capacity = capacity; > > > + __entry->freq = freq; > > > + __entry->mperf = mperf; > > > + __entry->aperf = aperf; > > > + __entry->tsc = tsc; > > > __entry->cpu_id = cpu_id; > > > __entry->changed = changed; > > > __entry->fast_switch = fast_switch; > > > ), > > > > > > - TP_printk("amd_min_perf=%lu amd_des_perf=%lu > > amd_max_perf=%lu cpu_id=%u changed=%s fast_switch=%s", > > > + TP_printk("amd_min_perf=%lu amd_des_perf=%lu > > amd_max_perf=%lu > > > +freq=%llu mperf=%llu aperf=%llu tsc=%llu cpu_id=%u changed=%s > > > +fast_switch=%s", > > > (unsigned long)__entry->min_perf, > > > (unsigned long)__entry->target_perf, > > > (unsigned long)__entry->capacity, > > > + (unsigned long long)__entry->freq, > > > + (unsigned long long)__entry->mperf, > > > + (unsigned long long)__entry->aperf, > > > + (unsigned long long)__entry->tsc, > > > (unsigned int)__entry->cpu_id, > > > (__entry->changed) ? "true" : "false", > > > (__entry->fast_switch) ? "true" : "false" > > > diff --git a/drivers/cpufreq/amd-pstate.c > > > b/drivers/cpufreq/amd-pstate.c index 9ce75ed11f8e..7be38bc6a673 100644 > > > --- a/drivers/cpufreq/amd-pstate.c > > > +++ b/drivers/cpufreq/amd-pstate.c > > > @@ -65,6 +65,18 @@ MODULE_PARM_DESC(shared_mem, > > > > > > static struct cpufreq_driver amd_pstate_driver; > > > > > > +/** > > > + * struct amd_aperf_mperf > > > + * @aperf: actual performance frequency clock count > > > + * @mperf: maximum performance frequency clock count > > > + * @tsc: time stamp counter > > > + */ > > > +struct amd_aperf_mperf { > > > + u64 aperf; > > > + u64 mperf; > > > + u64 tsc; > > > +}; > > > + > > > /** > > > * struct amd_cpudata - private CPU data for AMD P-State > > > * @cpu: CPU number > > > @@ -81,6 +93,9 @@ static struct cpufreq_driver amd_pstate_driver; > > > * @min_freq: the frequency that mapped to lowest_perf > > > * @nominal_freq: the frequency that mapped to nominal_perf > > > * @lowest_nonlinear_freq: the frequency that mapped to > > > lowest_nonlinear_perf > > > + * @cur: Difference of Aperf/Mperf/tsc count between last and current > > > + sample > > > + * @prev: Last Aperf/Mperf/tsc count value read from register > > > + * @freq: current cpu frequency value > > > * @boost_supported: check whether the Processor or SBIOS supports boost > > mode > > > * > > > * The amd_cpudata is key private data for each CPU thread in AMD > > > P-State, and @@ -102,6 +117,10 @@ struct amd_cpudata { > > > u32 nominal_freq; > > > u32 lowest_nonlinear_freq; > > > > > > + struct amd_aperf_mperf cur; > > > + struct amd_aperf_mperf prev; > > > + > > > + u64 freq; > > > bool boost_supported; > > > }; > > > > > > @@ -211,6 +230,39 @@ static inline void amd_pstate_update_perf(struct > > amd_cpudata *cpudata, > > > max_perf, fast_switch); > > > } > > > > > > +static inline bool amd_pstate_sample(struct amd_cpudata *cpudata) { > > > + u64 aperf, mperf, tsc; > > > + unsigned long flags; > > > + > > > + local_irq_save(flags); > > > + rdmsrl(MSR_IA32_APERF, aperf); > > > + rdmsrl(MSR_IA32_MPERF, mperf); > > > + tsc = rdtsc(); > > > + > > > + if (cpudata->prev.mperf == mperf || cpudata->prev.tsc == tsc) { > > > + local_irq_restore(flags); > > > + return false; > > > + } > > > + > > > + local_irq_restore(flags); > > > + > > > + cpudata->cur.aperf = aperf; > > > + cpudata->cur.mperf = mperf; > > > + cpudata->cur.tsc = tsc; > > > + cpudata->cur.aperf -= cpudata->prev.aperf; > > > + cpudata->cur.mperf -= cpudata->prev.mperf; > > > + cpudata->cur.tsc -= cpudata->prev.tsc; > > > + > > > + cpudata->prev.aperf = aperf; > > > + cpudata->prev.mperf = mperf; > > > + cpudata->prev.tsc = tsc; > > > + > > > + cpudata->freq = div64_u64((cpudata->cur.aperf * cpu_khz), > > > +cpudata->cur.mperf); > > > + > > > + return true; > > > +} > > > + > > > static void amd_pstate_update(struct amd_cpudata *cpudata, u32 > > min_perf, > > > u32 des_perf, u32 max_perf, bool fast_switch) > > { @@ -226,8 > > > +278,11 @@ static void amd_pstate_update(struct amd_cpudata *cpudata, > > u32 min_perf, > > > value &= ~AMD_CPPC_MAX_PERF(~0L); > > > value |= AMD_CPPC_MAX_PERF(max_perf); > > > > > > - trace_amd_pstate_perf(min_perf, des_perf, max_perf, > > > - cpudata->cpu, (value != prev), fast_switch); > > > + if (trace_amd_pstate_perf_enabled() && > > amd_pstate_sample(cpudata)) { > > > + trace_amd_pstate_perf(min_perf, des_perf, max_perf, > > cpudata->freq, > > > + cpudata->cur.mperf, cpudata->cur.aperf, cpudata- > > >cur.tsc, > > > + cpudata->cpu, (value != prev), fast_switch); > > > > How about using struct amd_aperf_mperf pointer as one input: > > > > trace_amd_pstate_perf(min_perf, des_perf, max_perf, &cpudata->cur, ...); > > > > You can refer the members of struct amd_aperf_mperf in the > > amd-pstate-trace.h: > > > > __entry->mperf = cur->mperf; > > __entry->aperf = cur->aperf; > > __entry->tsc = cur->tsc; > > > > I prefer the former way. We'd better to split the definition of struct "amd_cpudata" into head file and include it in the trace file with your change. Will do that in the future if needed. > Hmm, it should be ok. Because the trace implementation will be moved to include/trace/events/power.h after the shared_mem is enabled. Thanks, Ray