| Message ID | 20230629194509.4094455-1-srinivas.pandruvada@linux.intel.com (mailing list archive) |
|---|---|
| State | Mainlined, archived |
| Headers | show |
| Series | [v2] cpufreq: intel_pstate: Fix scaling for hybrid capable system with disabled E-cores | expand |
On Thu, Jun 29, 2023 at 9:45 PM Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> wrote: > > Some system BIOS configuration may provide option to disable E-cores. > As part of this change, CPUID feature for hybrid (Leaf 7 sub leaf 0, > EDX[15] = 0) may not be set. But HWP performance limits will still be > using a scaling factor like any other hybrid enabled system. > > The current check for applying scaling factor will fail when hybrid > CPUID feature is not set. Only way to make sure that scaling should be > applied by checking CPPC nominal frequency and nominal performance. If > CPPC nominal frequency and nominal performance is defined and nominal > frequency is not in multiples of 100MHz of nominal performance, then use > hybrid scaling factor. > > The above check will fail for non hybrid capable systems as they don't > publish nominal frequency field in CPPC, so this function can be used > for all HWP systems without additional cpu model check. > > Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> > --- > v2: > Compile errors reported by kernel test robot and Rafael for the case > when CONFIG_ACPI is not defined > > drivers/cpufreq/intel_pstate.c | 58 ++++++++++++++++++++++++++++------ > 1 file changed, 48 insertions(+), 10 deletions(-) > > diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c > index 2548ec92faa2..7e18999be46a 100644 > --- a/drivers/cpufreq/intel_pstate.c > +++ b/drivers/cpufreq/intel_pstate.c > @@ -302,6 +302,13 @@ static bool hwp_forced __read_mostly; > > static struct cpufreq_driver *intel_pstate_driver __read_mostly; > > +#define HYBRID_SCALING_FACTOR 78741 > + > +static inline int core_get_scaling(void) > +{ > + return 100000; > +} > + > #ifdef CONFIG_ACPI > static bool acpi_ppc; > #endif > @@ -400,6 +407,25 @@ static int intel_pstate_get_cppc_guaranteed(int cpu) > > return cppc_perf.nominal_perf; > } > + > +static int intel_pstate_cppc_get_scaling(int cpu) > +{ > + struct cppc_perf_caps cppc_perf; > + int ret; > + > + ret = cppc_get_perf_caps(cpu, &cppc_perf); > + > + /* > + * Check if nominal frequency is multiples of 100 MHz, if > + * not return hybrid scaling factor. > + */ > + if (!ret && cppc_perf.nominal_perf && cppc_perf.nominal_freq && > + (cppc_perf.nominal_perf * 100 != cppc_perf.nominal_freq)) > + return HYBRID_SCALING_FACTOR; > + > + return core_get_scaling(); > +} > + > #else /* CONFIG_ACPI_CPPC_LIB */ > static inline void intel_pstate_set_itmt_prio(int cpu) > { > @@ -492,6 +518,11 @@ static inline int intel_pstate_get_cppc_guaranteed(int cpu) > { > return -ENOTSUPP; > } > + > +static int intel_pstate_cppc_get_scaling(int cpu) > +{ > + return core_get_scaling(); > +} > #endif /* CONFIG_ACPI_CPPC_LIB */ > > /** > @@ -1895,11 +1926,6 @@ static int core_get_turbo_pstate(int cpu) > return ret; > } > > -static inline int core_get_scaling(void) > -{ > - return 100000; > -} > - > static u64 core_get_val(struct cpudata *cpudata, int pstate) > { > u64 val; > @@ -1936,16 +1962,29 @@ static void hybrid_get_type(void *data) > *cpu_type = get_this_hybrid_cpu_type(); > } > > -static int hybrid_get_cpu_scaling(int cpu) > +static int hwp_get_cpu_scaling(int cpu) > { > u8 cpu_type = 0; > > smp_call_function_single(cpu, hybrid_get_type, &cpu_type, 1); > /* P-cores have a smaller perf level-to-freqency scaling factor. */ > if (cpu_type == 0x40) > - return 78741; > + return HYBRID_SCALING_FACTOR; > > - return core_get_scaling(); > + /* Use default core scaling for E-cores */ > + if (cpu_type == 0x20) > + return core_get_scaling(); > + > + /* > + * If reached here, it means that, this system is either non > + * hybrid system (like Tiger Lake) or hybrid capable system (like > + * Alder Lake or Raptor Lake) with no E cores (CPUID for hybrid > + * support is 0). > + * All non hybrid systems, don't publish nominal_frequency > + * field (means nominal frequency = 0), In that case > + * the legacy core scaling is used. > + */ > + return intel_pstate_cppc_get_scaling(cpu); > } > > static void intel_pstate_set_pstate(struct cpudata *cpu, int pstate) > @@ -3393,8 +3432,7 @@ static int __init intel_pstate_init(void) > if (!default_driver) > default_driver = &intel_pstate; > > - if (boot_cpu_has(X86_FEATURE_HYBRID_CPU)) > - pstate_funcs.get_cpu_scaling = hybrid_get_cpu_scaling; > + pstate_funcs.get_cpu_scaling = hwp_get_cpu_scaling; > > goto hwp_cpu_matched; > } > -- Applied as 6.5-rc material with some adjustments. Please check the bleeding-edge branch. Thanks!
On Fri, 2023-06-30 at 10:59 +0200, Rafael J. Wysocki wrote: > On Thu, Jun 29, 2023 at 9:45 PM Srinivas Pandruvada > <srinivas.pandruvada@linux.intel.com> wrote: > > > > [...] > > goto hwp_cpu_matched; > > } > > -- > > Applied as 6.5-rc material with some adjustments. Please check the > bleeding-edge branch. Looks good. Thanks, Srinivas > > Thanks!
diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index 2548ec92faa2..7e18999be46a 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -302,6 +302,13 @@ static bool hwp_forced __read_mostly; static struct cpufreq_driver *intel_pstate_driver __read_mostly; +#define HYBRID_SCALING_FACTOR 78741 + +static inline int core_get_scaling(void) +{ + return 100000; +} + #ifdef CONFIG_ACPI static bool acpi_ppc; #endif @@ -400,6 +407,25 @@ static int intel_pstate_get_cppc_guaranteed(int cpu) return cppc_perf.nominal_perf; } + +static int intel_pstate_cppc_get_scaling(int cpu) +{ + struct cppc_perf_caps cppc_perf; + int ret; + + ret = cppc_get_perf_caps(cpu, &cppc_perf); + + /* + * Check if nominal frequency is multiples of 100 MHz, if + * not return hybrid scaling factor. + */ + if (!ret && cppc_perf.nominal_perf && cppc_perf.nominal_freq && + (cppc_perf.nominal_perf * 100 != cppc_perf.nominal_freq)) + return HYBRID_SCALING_FACTOR; + + return core_get_scaling(); +} + #else /* CONFIG_ACPI_CPPC_LIB */ static inline void intel_pstate_set_itmt_prio(int cpu) { @@ -492,6 +518,11 @@ static inline int intel_pstate_get_cppc_guaranteed(int cpu) { return -ENOTSUPP; } + +static int intel_pstate_cppc_get_scaling(int cpu) +{ + return core_get_scaling(); +} #endif /* CONFIG_ACPI_CPPC_LIB */ /** @@ -1895,11 +1926,6 @@ static int core_get_turbo_pstate(int cpu) return ret; } -static inline int core_get_scaling(void) -{ - return 100000; -} - static u64 core_get_val(struct cpudata *cpudata, int pstate) { u64 val; @@ -1936,16 +1962,29 @@ static void hybrid_get_type(void *data) *cpu_type = get_this_hybrid_cpu_type(); } -static int hybrid_get_cpu_scaling(int cpu) +static int hwp_get_cpu_scaling(int cpu) { u8 cpu_type = 0; smp_call_function_single(cpu, hybrid_get_type, &cpu_type, 1); /* P-cores have a smaller perf level-to-freqency scaling factor. */ if (cpu_type == 0x40) - return 78741; + return HYBRID_SCALING_FACTOR; - return core_get_scaling(); + /* Use default core scaling for E-cores */ + if (cpu_type == 0x20) + return core_get_scaling(); + + /* + * If reached here, it means that, this system is either non + * hybrid system (like Tiger Lake) or hybrid capable system (like + * Alder Lake or Raptor Lake) with no E cores (CPUID for hybrid + * support is 0). + * All non hybrid systems, don't publish nominal_frequency + * field (means nominal frequency = 0), In that case + * the legacy core scaling is used. + */ + return intel_pstate_cppc_get_scaling(cpu); } static void intel_pstate_set_pstate(struct cpudata *cpu, int pstate) @@ -3393,8 +3432,7 @@ static int __init intel_pstate_init(void) if (!default_driver) default_driver = &intel_pstate; - if (boot_cpu_has(X86_FEATURE_HYBRID_CPU)) - pstate_funcs.get_cpu_scaling = hybrid_get_cpu_scaling; + pstate_funcs.get_cpu_scaling = hwp_get_cpu_scaling; goto hwp_cpu_matched; }
Some system BIOS configuration may provide option to disable E-cores. As part of this change, CPUID feature for hybrid (Leaf 7 sub leaf 0, EDX[15] = 0) may not be set. But HWP performance limits will still be using a scaling factor like any other hybrid enabled system. The current check for applying scaling factor will fail when hybrid CPUID feature is not set. Only way to make sure that scaling should be applied by checking CPPC nominal frequency and nominal performance. If CPPC nominal frequency and nominal performance is defined and nominal frequency is not in multiples of 100MHz of nominal performance, then use hybrid scaling factor. The above check will fail for non hybrid capable systems as they don't publish nominal frequency field in CPPC, so this function can be used for all HWP systems without additional cpu model check. Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> --- v2: Compile errors reported by kernel test robot and Rafael for the case when CONFIG_ACPI is not defined drivers/cpufreq/intel_pstate.c | 58 ++++++++++++++++++++++++++++------ 1 file changed, 48 insertions(+), 10 deletions(-)