| Message ID | 20231013013016.197102-1-jeremy.linton@arm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | arm64: cpufeature: Display the set of cores with a feature | expand |
On Thu, Oct 12, 2023 at 08:30:16PM -0500, Jeremy Linton wrote: > The AMU feature can be enabled on a subset of the cores in a system. > Because of that, it prints a message for each core as it is detected. > This becomes tedious when there are hundreds of cores. Instead, for > CPU features which can be enabled on a subset of the present cores, > lets wait until update_cpu_capabilities() and print the subset of cores > the feature was enabled on. > > Signed-off-by: Jeremy Linton <jeremy.linton@arm.com> > --- > arch/arm64/include/asm/cpufeature.h | 2 ++ > arch/arm64/kernel/cpufeature.c | 16 +++++++++++++--- > 2 files changed, 15 insertions(+), 3 deletions(-) That's pretty neat. Are there any other caps we could switch over to this new mechanism as well? We have a weird bodge, for example, in has_hw_dbm() to print a "detected: ..." message and I wonder whether following the example of the AMU would be better? Will
On Fri, Oct 13, 2023 at 09:29:48AM +0100, Will Deacon wrote: > That's pretty neat. Are there any other caps we could switch over to this > new mechanism as well? We have a weird bodge, for example, in has_hw_dbm() > to print a "detected: ..." message and I wonder whether following the > example of the AMU would be better? There's 32 bit EL0, though that's a bit different...
Hi, On Thursday 12 Oct 2023 at 20:30:16 (-0500), Jeremy Linton wrote: > The AMU feature can be enabled on a subset of the cores in a system. > Because of that, it prints a message for each core as it is detected. > This becomes tedious when there are hundreds of cores. Instead, for > CPU features which can be enabled on a subset of the present cores, > lets wait until update_cpu_capabilities() and print the subset of cores > the feature was enabled on. > > Signed-off-by: Jeremy Linton <jeremy.linton@arm.com> > --- > arch/arm64/include/asm/cpufeature.h | 2 ++ > arch/arm64/kernel/cpufeature.c | 16 +++++++++++++--- > 2 files changed, 15 insertions(+), 3 deletions(-) > > diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h > index 5bba39376055..19b4d001d845 100644 > --- a/arch/arm64/include/asm/cpufeature.h > +++ b/arch/arm64/include/asm/cpufeature.h > @@ -23,6 +23,7 @@ > #include <linux/bug.h> > #include <linux/jump_label.h> > #include <linux/kernel.h> > +#include <linux/cpumask.h> > > /* > * CPU feature register tracking > @@ -380,6 +381,7 @@ struct arm64_cpu_capabilities { > * method is robust against being called multiple times. > */ > const struct arm64_cpu_capabilities *match_list; > + const struct cpumask *cpus; > }; > > static inline int cpucap_default_scope(const struct arm64_cpu_capabilities *cap) > diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c > index 444a73c2e638..18711e35924c 100644 > --- a/arch/arm64/kernel/cpufeature.c > +++ b/arch/arm64/kernel/cpufeature.c > @@ -1944,8 +1944,6 @@ int get_cpu_with_amu_feat(void) > static void cpu_amu_enable(struct arm64_cpu_capabilities const *cap) > { > if (has_cpuid_feature(cap, SCOPE_LOCAL_CPU)) { > - pr_info("detected CPU%d: Activity Monitors Unit (AMU)\n", > - smp_processor_id()); > cpumask_set_cpu(smp_processor_id(), &amu_cpus); > > /* 0 reference values signal broken/disabled counters */ > @@ -2411,10 +2409,12 @@ static const struct arm64_cpu_capabilities arm64_features[] = { > * message to be shown until at least one CPU is detected to > * support the feature. > */ > + .desc = "Activity Monitors Unit (AMU)", > .capability = ARM64_HAS_AMU_EXTN, > .type = ARM64_CPUCAP_WEAK_LOCAL_CPU_FEATURE, > .matches = has_amu, > .cpu_enable = cpu_amu_enable, > + .cpus = &amu_cpus, > ARM64_CPUID_FIELDS(ID_AA64PFR0_EL1, AMU, IMP) > }, > #endif /* CONFIG_ARM64_AMU_EXTN */ > @@ -2981,7 +2981,7 @@ static void update_cpu_capabilities(u16 scope_mask) > !caps->matches(caps, cpucap_default_scope(caps))) > continue; > > - if (caps->desc) > + if (caps->desc && !caps->cpus) > pr_info("detected: %s\n", caps->desc); > > __set_bit(caps->capability, system_cpucaps); > @@ -3330,6 +3330,7 @@ unsigned long cpu_get_elf_hwcap2(void) > > static void __init setup_system_capabilities(void) > { > + int i; > /* > * We have finalised the system-wide safe feature > * registers, finalise the capabilities that depend > @@ -3338,6 +3339,15 @@ static void __init setup_system_capabilities(void) > */ > update_cpu_capabilities(SCOPE_SYSTEM); > enable_cpu_capabilities(SCOPE_ALL & ~SCOPE_BOOT_CPU); > + > + for (i = 0; i < ARM64_NCAPS; i++) { > + const struct arm64_cpu_capabilities *caps = cpucap_ptrs[i]; > + > + if (caps && caps->cpus && caps->desc && > + cpumask_any(caps->cpus) < nr_cpu_ids) > + pr_info("detected: %s on CPU%*pbl\n", > + caps->desc, cpumask_pr_args(caps->cpus)); > + } > } > > void __init setup_cpu_features(void) > -- > 2.41.0 > It looks very nice to me! I gave it a run on an RD-N2 FVP so FWIW: Reviewed-by: Ionela Voinescu <ionela.voinescu@arm.com> Tested-by: Ionela Voinescu <ionela.voinescu@arm.com> The same could be used for DBM, but I'm not sure if creating a mask of supporting CPUs is worth it, for this purpose only. Hope it helps, Ionela.
Jeremy Linton <jeremy.linton@arm.com> writes: > The AMU feature can be enabled on a subset of the cores in a system. > Because of that, it prints a message for each core as it is detected. > This becomes tedious when there are hundreds of cores. Instead, for > CPU features which can be enabled on a subset of the present cores, > lets wait until update_cpu_capabilities() and print the subset of cores > the feature was enabled on. > > Signed-off-by: Jeremy Linton <jeremy.linton@arm.com> > --- > arch/arm64/include/asm/cpufeature.h | 2 ++ > arch/arm64/kernel/cpufeature.c | 16 +++++++++++++--- > 2 files changed, 15 insertions(+), 3 deletions(-) > > diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h > index 5bba39376055..19b4d001d845 100644 > --- a/arch/arm64/include/asm/cpufeature.h > +++ b/arch/arm64/include/asm/cpufeature.h > @@ -23,6 +23,7 @@ > #include <linux/bug.h> > #include <linux/jump_label.h> > #include <linux/kernel.h> > +#include <linux/cpumask.h> > > /* > * CPU feature register tracking > @@ -380,6 +381,7 @@ struct arm64_cpu_capabilities { > * method is robust against being called multiple times. > */ > const struct arm64_cpu_capabilities *match_list; > + const struct cpumask *cpus; > }; > > static inline int cpucap_default_scope(const struct arm64_cpu_capabilities *cap) > diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c > index 444a73c2e638..18711e35924c 100644 > --- a/arch/arm64/kernel/cpufeature.c > +++ b/arch/arm64/kernel/cpufeature.c > @@ -1944,8 +1944,6 @@ int get_cpu_with_amu_feat(void) > static void cpu_amu_enable(struct arm64_cpu_capabilities const *cap) > { > if (has_cpuid_feature(cap, SCOPE_LOCAL_CPU)) { > - pr_info("detected CPU%d: Activity Monitors Unit (AMU)\n", > - smp_processor_id()); > cpumask_set_cpu(smp_processor_id(), &amu_cpus); > > /* 0 reference values signal broken/disabled counters */ > @@ -2411,10 +2409,12 @@ static const struct arm64_cpu_capabilities arm64_features[] = { > * message to be shown until at least one CPU is detected to > * support the feature. > */ > + .desc = "Activity Monitors Unit (AMU)", > .capability = ARM64_HAS_AMU_EXTN, > .type = ARM64_CPUCAP_WEAK_LOCAL_CPU_FEATURE, > .matches = has_amu, > .cpu_enable = cpu_amu_enable, > + .cpus = &amu_cpus, > ARM64_CPUID_FIELDS(ID_AA64PFR0_EL1, AMU, IMP) > }, > #endif /* CONFIG_ARM64_AMU_EXTN */ > @@ -2981,7 +2981,7 @@ static void update_cpu_capabilities(u16 scope_mask) > !caps->matches(caps, cpucap_default_scope(caps))) > continue; > > - if (caps->desc) > + if (caps->desc && !caps->cpus) > pr_info("detected: %s\n", caps->desc); > > __set_bit(caps->capability, system_cpucaps); > @@ -3330,6 +3330,7 @@ unsigned long cpu_get_elf_hwcap2(void) > > static void __init setup_system_capabilities(void) > { > + int i; > /* > * We have finalised the system-wide safe feature > * registers, finalise the capabilities that depend > @@ -3338,6 +3339,15 @@ static void __init setup_system_capabilities(void) > */ > update_cpu_capabilities(SCOPE_SYSTEM); > enable_cpu_capabilities(SCOPE_ALL & ~SCOPE_BOOT_CPU); > + > + for (i = 0; i < ARM64_NCAPS; i++) { > + const struct arm64_cpu_capabilities *caps = cpucap_ptrs[i]; > + > + if (caps && caps->cpus && caps->desc && > + cpumask_any(caps->cpus) < nr_cpu_ids) > + pr_info("detected: %s on CPU%*pbl\n", > + caps->desc, cpumask_pr_args(caps->cpus)); > + } > } > > void __init setup_cpu_features(void) Thanks, Jeremy for putting this patch together. The duplicated prints bothered me too every time I stared at the boot logs. Fwiw,, Reviewed-by: Punit Agrawal <punit.agrawal@bytedance.com> Tested-by: Punit Agrawal <punit.agrawal@bytedance.com> The patch was verified on an N2 based platform.
diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h index 5bba39376055..19b4d001d845 100644 --- a/arch/arm64/include/asm/cpufeature.h +++ b/arch/arm64/include/asm/cpufeature.h @@ -23,6 +23,7 @@ #include <linux/bug.h> #include <linux/jump_label.h> #include <linux/kernel.h> +#include <linux/cpumask.h> /* * CPU feature register tracking @@ -380,6 +381,7 @@ struct arm64_cpu_capabilities { * method is robust against being called multiple times. */ const struct arm64_cpu_capabilities *match_list; + const struct cpumask *cpus; }; static inline int cpucap_default_scope(const struct arm64_cpu_capabilities *cap) diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c index 444a73c2e638..18711e35924c 100644 --- a/arch/arm64/kernel/cpufeature.c +++ b/arch/arm64/kernel/cpufeature.c @@ -1944,8 +1944,6 @@ int get_cpu_with_amu_feat(void) static void cpu_amu_enable(struct arm64_cpu_capabilities const *cap) { if (has_cpuid_feature(cap, SCOPE_LOCAL_CPU)) { - pr_info("detected CPU%d: Activity Monitors Unit (AMU)\n", - smp_processor_id()); cpumask_set_cpu(smp_processor_id(), &amu_cpus); /* 0 reference values signal broken/disabled counters */ @@ -2411,10 +2409,12 @@ static const struct arm64_cpu_capabilities arm64_features[] = { * message to be shown until at least one CPU is detected to * support the feature. */ + .desc = "Activity Monitors Unit (AMU)", .capability = ARM64_HAS_AMU_EXTN, .type = ARM64_CPUCAP_WEAK_LOCAL_CPU_FEATURE, .matches = has_amu, .cpu_enable = cpu_amu_enable, + .cpus = &amu_cpus, ARM64_CPUID_FIELDS(ID_AA64PFR0_EL1, AMU, IMP) }, #endif /* CONFIG_ARM64_AMU_EXTN */ @@ -2981,7 +2981,7 @@ static void update_cpu_capabilities(u16 scope_mask) !caps->matches(caps, cpucap_default_scope(caps))) continue; - if (caps->desc) + if (caps->desc && !caps->cpus) pr_info("detected: %s\n", caps->desc); __set_bit(caps->capability, system_cpucaps); @@ -3330,6 +3330,7 @@ unsigned long cpu_get_elf_hwcap2(void) static void __init setup_system_capabilities(void) { + int i; /* * We have finalised the system-wide safe feature * registers, finalise the capabilities that depend @@ -3338,6 +3339,15 @@ static void __init setup_system_capabilities(void) */ update_cpu_capabilities(SCOPE_SYSTEM); enable_cpu_capabilities(SCOPE_ALL & ~SCOPE_BOOT_CPU); + + for (i = 0; i < ARM64_NCAPS; i++) { + const struct arm64_cpu_capabilities *caps = cpucap_ptrs[i]; + + if (caps && caps->cpus && caps->desc && + cpumask_any(caps->cpus) < nr_cpu_ids) + pr_info("detected: %s on CPU%*pbl\n", + caps->desc, cpumask_pr_args(caps->cpus)); + } } void __init setup_cpu_features(void)
The AMU feature can be enabled on a subset of the cores in a system. Because of that, it prints a message for each core as it is detected. This becomes tedious when there are hundreds of cores. Instead, for CPU features which can be enabled on a subset of the present cores, lets wait until update_cpu_capabilities() and print the subset of cores the feature was enabled on. Signed-off-by: Jeremy Linton <jeremy.linton@arm.com> --- arch/arm64/include/asm/cpufeature.h | 2 ++ arch/arm64/kernel/cpufeature.c | 16 +++++++++++++--- 2 files changed, 15 insertions(+), 3 deletions(-)