| Message ID | 20230526215410.2435674-15-irogers@google.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | PMU refactoring and improvements | expand |
Em Fri, May 26, 2023 at 02:53:49PM -0700, Ian Rogers escreveu: > __evlist__add_default adds a cycles event to a typically empty evlist > and was extended for hybrid with evlist__add_default_hybrid, as more > than 1 PMU was necessary. Rather than have dedicated logic for the > cycles event, this change switches to parsing 'cycles:P' which will > handle wildcarding the PMUs appropriately for hybrid. I think I reported this earlier, but at this point 'perf test python' breaks, I fixed it in the tmp.perf-tools-next: 19: 'import perf' in python : FAILED! ⬢[acme@toolbox perf-tools-next]$ git log --oneline -1 fe4f622c4fc7a02a (HEAD) perf evlist: Remove __evlist__add_default ⬢[acme@toolbox perf-tools-next]$ ⬢[acme@toolbox perf-tools-next]$ perf test -v python Couldn't bump rlimit(MEMLOCK), failures may take place when creating BPF maps, etc 19: 'import perf' in python : --- start --- test child forked, pid 2976621 python usage test: "echo "import sys ; sys.path.append('/tmp/build/perf-tools-next/python'); import perf" | '/usr/bin/python3' " Traceback (most recent call last): File "<stdin>", line 1, in <module> ImportError: /tmp/build/perf-tools-next/python/perf.cpython-310-x86_64-linux-gnu.so: undefined symbol: parse_event test child finished with -1 ---- end ---- 'import perf' in python: FAILED! ⬢[acme@toolbox perf-tools-next]$ Probably there will be a few more cases in the next patches, please check. - Arnaldo > Signed-off-by: Ian Rogers <irogers@google.com> > Reviewed-by: Kan Liang <kan.liang@linux.intel.com> > --- > tools/perf/arch/x86/util/evsel.c | 20 -------------- > tools/perf/builtin-record.c | 13 +++------ > tools/perf/builtin-top.c | 10 ++++--- > tools/perf/util/evlist-hybrid.c | 25 ----------------- > tools/perf/util/evlist-hybrid.h | 1 - > tools/perf/util/evlist.c | 22 ++++++--------- > tools/perf/util/evlist.h | 7 ----- > tools/perf/util/evsel.c | 46 -------------------------------- > tools/perf/util/evsel.h | 3 --- > 9 files changed, 17 insertions(+), 130 deletions(-) > > diff --git a/tools/perf/arch/x86/util/evsel.c b/tools/perf/arch/x86/util/evsel.c > index ea3972d785d1..153cdca94cd4 100644 > --- a/tools/perf/arch/x86/util/evsel.c > +++ b/tools/perf/arch/x86/util/evsel.c > @@ -16,26 +16,6 @@ void arch_evsel__set_sample_weight(struct evsel *evsel) > evsel__set_sample_bit(evsel, WEIGHT_STRUCT); > } > > -void arch_evsel__fixup_new_cycles(struct perf_event_attr *attr) > -{ > - struct perf_env env = { .total_mem = 0, } ; > - > - if (!perf_env__cpuid(&env)) > - return; > - > - /* > - * On AMD, precise cycles event sampling internally uses IBS pmu. > - * But IBS does not have filtering capabilities and perf by default > - * sets exclude_guest = 1. This makes IBS pmu event init fail and > - * thus perf ends up doing non-precise sampling. Avoid it by clearing > - * exclude_guest. > - */ > - if (env.cpuid && strstarts(env.cpuid, "AuthenticAMD")) > - attr->exclude_guest = 0; > - > - free(env.cpuid); > -} > - > /* Check whether the evsel's PMU supports the perf metrics */ > bool evsel__sys_has_perf_metrics(const struct evsel *evsel) > { > diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c > index 88f7b4241153..d80b54a6f450 100644 > --- a/tools/perf/builtin-record.c > +++ b/tools/perf/builtin-record.c > @@ -4161,18 +4161,11 @@ int cmd_record(int argc, const char **argv) > record.opts.tail_synthesize = true; > > if (rec->evlist->core.nr_entries == 0) { > - if (perf_pmu__has_hybrid()) { > - err = evlist__add_default_hybrid(rec->evlist, > - !record.opts.no_samples); > - } else { > - err = __evlist__add_default(rec->evlist, > - !record.opts.no_samples); > - } > + bool can_profile_kernel = perf_event_paranoid_check(1); > > - if (err < 0) { > - pr_err("Not enough memory for event selector list\n"); > + err = parse_event(rec->evlist, can_profile_kernel ? "cycles:P" : "cycles:Pu"); > + if (err) > goto out; > - } > } > > if (rec->opts.target.tid && !rec->opts.no_inherit_set) > diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c > index 48ee49e95c5e..27a7f068207d 100644 > --- a/tools/perf/builtin-top.c > +++ b/tools/perf/builtin-top.c > @@ -1653,10 +1653,12 @@ int cmd_top(int argc, const char **argv) > if (annotate_check_args(&top.annotation_opts) < 0) > goto out_delete_evlist; > > - if (!top.evlist->core.nr_entries && > - evlist__add_default(top.evlist) < 0) { > - pr_err("Not enough memory for event selector list\n"); > - goto out_delete_evlist; > + if (!top.evlist->core.nr_entries) { > + bool can_profile_kernel = perf_event_paranoid_check(1); > + int err = parse_event(top.evlist, can_profile_kernel ? "cycles:P" : "cycles:Pu"); > + > + if (err) > + goto out_delete_evlist; > } > > status = evswitch__init(&top.evswitch, top.evlist, stderr); > diff --git a/tools/perf/util/evlist-hybrid.c b/tools/perf/util/evlist-hybrid.c > index 0f59c80f27b2..64f78d06fe19 100644 > --- a/tools/perf/util/evlist-hybrid.c > +++ b/tools/perf/util/evlist-hybrid.c > @@ -16,31 +16,6 @@ > #include <perf/evsel.h> > #include <perf/cpumap.h> > > -int evlist__add_default_hybrid(struct evlist *evlist, bool precise) > -{ > - struct evsel *evsel; > - struct perf_pmu *pmu; > - __u64 config; > - struct perf_cpu_map *cpus; > - > - perf_pmu__for_each_hybrid_pmu(pmu) { > - config = PERF_COUNT_HW_CPU_CYCLES | > - ((__u64)pmu->type << PERF_PMU_TYPE_SHIFT); > - evsel = evsel__new_cycles(precise, PERF_TYPE_HARDWARE, > - config); > - if (!evsel) > - return -ENOMEM; > - > - cpus = perf_cpu_map__get(pmu->cpus); > - evsel->core.cpus = cpus; > - evsel->core.own_cpus = perf_cpu_map__get(cpus); > - evsel->pmu_name = strdup(pmu->name); > - evlist__add(evlist, evsel); > - } > - > - return 0; > -} > - > bool evlist__has_hybrid(struct evlist *evlist) > { > struct evsel *evsel; > diff --git a/tools/perf/util/evlist-hybrid.h b/tools/perf/util/evlist-hybrid.h > index 4b000eda6626..0cded76eb344 100644 > --- a/tools/perf/util/evlist-hybrid.h > +++ b/tools/perf/util/evlist-hybrid.h > @@ -7,7 +7,6 @@ > #include "evlist.h" > #include <unistd.h> > > -int evlist__add_default_hybrid(struct evlist *evlist, bool precise); > bool evlist__has_hybrid(struct evlist *evlist); > > #endif /* __PERF_EVLIST_HYBRID_H */ > diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c > index 9dfa977193b3..63f8821a5395 100644 > --- a/tools/perf/util/evlist.c > +++ b/tools/perf/util/evlist.c > @@ -93,8 +93,15 @@ struct evlist *evlist__new(void) > struct evlist *evlist__new_default(void) > { > struct evlist *evlist = evlist__new(); > + bool can_profile_kernel; > + int err; > + > + if (!evlist) > + return NULL; > > - if (evlist && evlist__add_default(evlist)) { > + can_profile_kernel = perf_event_paranoid_check(1); > + err = parse_event(evlist, can_profile_kernel ? "cycles:P" : "cycles:Pu"); > + if (err) { > evlist__delete(evlist); > evlist = NULL; > } > @@ -237,19 +244,6 @@ static void evlist__set_leader(struct evlist *evlist) > perf_evlist__set_leader(&evlist->core); > } > > -int __evlist__add_default(struct evlist *evlist, bool precise) > -{ > - struct evsel *evsel; > - > - evsel = evsel__new_cycles(precise, PERF_TYPE_HARDWARE, > - PERF_COUNT_HW_CPU_CYCLES); > - if (evsel == NULL) > - return -ENOMEM; > - > - evlist__add(evlist, evsel); > - return 0; > -} > - > static struct evsel *evlist__dummy_event(struct evlist *evlist) > { > struct perf_event_attr attr = { > diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h > index 5e7ff44f3043..664c6bf7b3e0 100644 > --- a/tools/perf/util/evlist.h > +++ b/tools/perf/util/evlist.h > @@ -100,13 +100,6 @@ void evlist__delete(struct evlist *evlist); > void evlist__add(struct evlist *evlist, struct evsel *entry); > void evlist__remove(struct evlist *evlist, struct evsel *evsel); > > -int __evlist__add_default(struct evlist *evlist, bool precise); > - > -static inline int evlist__add_default(struct evlist *evlist) > -{ > - return __evlist__add_default(evlist, true); > -} > - > int evlist__add_attrs(struct evlist *evlist, struct perf_event_attr *attrs, size_t nr_attrs); > > int __evlist__add_default_attrs(struct evlist *evlist, > diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c > index 8c8f371ea2b5..1df8f967d2eb 100644 > --- a/tools/perf/util/evsel.c > +++ b/tools/perf/util/evsel.c > @@ -316,48 +316,6 @@ struct evsel *evsel__new_idx(struct perf_event_attr *attr, int idx) > return evsel; > } > > -static bool perf_event_can_profile_kernel(void) > -{ > - return perf_event_paranoid_check(1); > -} > - > -struct evsel *evsel__new_cycles(bool precise __maybe_unused, __u32 type, __u64 config) > -{ > - struct perf_event_attr attr = { > - .type = type, > - .config = config, > - .exclude_kernel = !perf_event_can_profile_kernel(), > - }; > - struct evsel *evsel; > - > - event_attr_init(&attr); > - > - /* > - * Now let the usual logic to set up the perf_event_attr defaults > - * to kick in when we return and before perf_evsel__open() is called. > - */ > - evsel = evsel__new(&attr); > - if (evsel == NULL) > - goto out; > - > - arch_evsel__fixup_new_cycles(&evsel->core.attr); > - > - evsel->precise_max = true; > - > - /* use asprintf() because free(evsel) assumes name is allocated */ > - if (asprintf(&evsel->name, "cycles%s%s%.*s", > - (attr.precise_ip || attr.exclude_kernel) ? ":" : "", > - attr.exclude_kernel ? "u" : "", > - attr.precise_ip ? attr.precise_ip + 1 : 0, "ppp") < 0) > - goto error_free; > -out: > - return evsel; > -error_free: > - evsel__delete(evsel); > - evsel = NULL; > - goto out; > -} > - > int copy_config_terms(struct list_head *dst, struct list_head *src) > { > struct evsel_config_term *pos, *tmp; > @@ -1131,10 +1089,6 @@ void __weak arch_evsel__set_sample_weight(struct evsel *evsel) > evsel__set_sample_bit(evsel, WEIGHT); > } > > -void __weak arch_evsel__fixup_new_cycles(struct perf_event_attr *attr __maybe_unused) > -{ > -} > - > void __weak arch__post_evsel_config(struct evsel *evsel __maybe_unused, > struct perf_event_attr *attr __maybe_unused) > { > diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h > index df8928745fc6..429b172cc94d 100644 > --- a/tools/perf/util/evsel.h > +++ b/tools/perf/util/evsel.h > @@ -243,8 +243,6 @@ static inline struct evsel *evsel__newtp(const char *sys, const char *name) > } > #endif > > -struct evsel *evsel__new_cycles(bool precise, __u32 type, __u64 config); > - > #ifdef HAVE_LIBTRACEEVENT > struct tep_event *event_format__new(const char *sys, const char *name); > #endif > @@ -312,7 +310,6 @@ void __evsel__reset_sample_bit(struct evsel *evsel, enum perf_event_sample_forma > void evsel__set_sample_id(struct evsel *evsel, bool use_sample_identifier); > > void arch_evsel__set_sample_weight(struct evsel *evsel); > -void arch_evsel__fixup_new_cycles(struct perf_event_attr *attr); > void arch__post_evsel_config(struct evsel *evsel, struct perf_event_attr *attr); > > int evsel__set_filter(struct evsel *evsel, const char *filter); > -- > 2.41.0.rc0.172.g3f132b7071-goog >
On Fri, May 26, 2023 at 6:45 PM Arnaldo Carvalho de Melo <acme@kernel.org> wrote: > > Em Fri, May 26, 2023 at 02:53:49PM -0700, Ian Rogers escreveu: > > __evlist__add_default adds a cycles event to a typically empty evlist > > and was extended for hybrid with evlist__add_default_hybrid, as more > > than 1 PMU was necessary. Rather than have dedicated logic for the > > cycles event, this change switches to parsing 'cycles:P' which will > > handle wildcarding the PMUs appropriately for hybrid. > > I think I reported this earlier, but at this point 'perf test python' > breaks, I fixed it in the tmp.perf-tools-next: > > 19: 'import perf' in python : FAILED! > ⬢[acme@toolbox perf-tools-next]$ git log --oneline -1 > fe4f622c4fc7a02a (HEAD) perf evlist: Remove __evlist__add_default > ⬢[acme@toolbox perf-tools-next]$ > ⬢[acme@toolbox perf-tools-next]$ perf test -v python > Couldn't bump rlimit(MEMLOCK), failures may take place when creating BPF maps, etc > 19: 'import perf' in python : > --- start --- > test child forked, pid 2976621 > python usage test: "echo "import sys ; sys.path.append('/tmp/build/perf-tools-next/python'); import perf" | '/usr/bin/python3' " > Traceback (most recent call last): > File "<stdin>", line 1, in <module> > ImportError: /tmp/build/perf-tools-next/python/perf.cpython-310-x86_64-linux-gnu.so: undefined symbol: parse_event > test child finished with -1 > ---- end ---- > 'import perf' in python: FAILED! > ⬢[acme@toolbox perf-tools-next]$ > > Probably there will be a few more cases in the next patches, please > check. I'll rebase and resend. I needed to add: https://lore.kernel.org/lkml/20230527055517.2711487-1-irogers@google.com/ to repro the failure. The test was passing without it. Thanks, Ian > - Arnaldo > > > Signed-off-by: Ian Rogers <irogers@google.com> > > Reviewed-by: Kan Liang <kan.liang@linux.intel.com> > > --- > > tools/perf/arch/x86/util/evsel.c | 20 -------------- > > tools/perf/builtin-record.c | 13 +++------ > > tools/perf/builtin-top.c | 10 ++++--- > > tools/perf/util/evlist-hybrid.c | 25 ----------------- > > tools/perf/util/evlist-hybrid.h | 1 - > > tools/perf/util/evlist.c | 22 ++++++--------- > > tools/perf/util/evlist.h | 7 ----- > > tools/perf/util/evsel.c | 46 -------------------------------- > > tools/perf/util/evsel.h | 3 --- > > 9 files changed, 17 insertions(+), 130 deletions(-) > > > > diff --git a/tools/perf/arch/x86/util/evsel.c b/tools/perf/arch/x86/util/evsel.c > > index ea3972d785d1..153cdca94cd4 100644 > > --- a/tools/perf/arch/x86/util/evsel.c > > +++ b/tools/perf/arch/x86/util/evsel.c > > @@ -16,26 +16,6 @@ void arch_evsel__set_sample_weight(struct evsel *evsel) > > evsel__set_sample_bit(evsel, WEIGHT_STRUCT); > > } > > > > -void arch_evsel__fixup_new_cycles(struct perf_event_attr *attr) > > -{ > > - struct perf_env env = { .total_mem = 0, } ; > > - > > - if (!perf_env__cpuid(&env)) > > - return; > > - > > - /* > > - * On AMD, precise cycles event sampling internally uses IBS pmu. > > - * But IBS does not have filtering capabilities and perf by default > > - * sets exclude_guest = 1. This makes IBS pmu event init fail and > > - * thus perf ends up doing non-precise sampling. Avoid it by clearing > > - * exclude_guest. > > - */ > > - if (env.cpuid && strstarts(env.cpuid, "AuthenticAMD")) > > - attr->exclude_guest = 0; > > - > > - free(env.cpuid); > > -} > > - > > /* Check whether the evsel's PMU supports the perf metrics */ > > bool evsel__sys_has_perf_metrics(const struct evsel *evsel) > > { > > diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c > > index 88f7b4241153..d80b54a6f450 100644 > > --- a/tools/perf/builtin-record.c > > +++ b/tools/perf/builtin-record.c > > @@ -4161,18 +4161,11 @@ int cmd_record(int argc, const char **argv) > > record.opts.tail_synthesize = true; > > > > if (rec->evlist->core.nr_entries == 0) { > > - if (perf_pmu__has_hybrid()) { > > - err = evlist__add_default_hybrid(rec->evlist, > > - !record.opts.no_samples); > > - } else { > > - err = __evlist__add_default(rec->evlist, > > - !record.opts.no_samples); > > - } > > + bool can_profile_kernel = perf_event_paranoid_check(1); > > > > - if (err < 0) { > > - pr_err("Not enough memory for event selector list\n"); > > + err = parse_event(rec->evlist, can_profile_kernel ? "cycles:P" : "cycles:Pu"); > > + if (err) > > goto out; > > - } > > } > > > > if (rec->opts.target.tid && !rec->opts.no_inherit_set) > > diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c > > index 48ee49e95c5e..27a7f068207d 100644 > > --- a/tools/perf/builtin-top.c > > +++ b/tools/perf/builtin-top.c > > @@ -1653,10 +1653,12 @@ int cmd_top(int argc, const char **argv) > > if (annotate_check_args(&top.annotation_opts) < 0) > > goto out_delete_evlist; > > > > - if (!top.evlist->core.nr_entries && > > - evlist__add_default(top.evlist) < 0) { > > - pr_err("Not enough memory for event selector list\n"); > > - goto out_delete_evlist; > > + if (!top.evlist->core.nr_entries) { > > + bool can_profile_kernel = perf_event_paranoid_check(1); > > + int err = parse_event(top.evlist, can_profile_kernel ? "cycles:P" : "cycles:Pu"); > > + > > + if (err) > > + goto out_delete_evlist; > > } > > > > status = evswitch__init(&top.evswitch, top.evlist, stderr); > > diff --git a/tools/perf/util/evlist-hybrid.c b/tools/perf/util/evlist-hybrid.c > > index 0f59c80f27b2..64f78d06fe19 100644 > > --- a/tools/perf/util/evlist-hybrid.c > > +++ b/tools/perf/util/evlist-hybrid.c > > @@ -16,31 +16,6 @@ > > #include <perf/evsel.h> > > #include <perf/cpumap.h> > > > > -int evlist__add_default_hybrid(struct evlist *evlist, bool precise) > > -{ > > - struct evsel *evsel; > > - struct perf_pmu *pmu; > > - __u64 config; > > - struct perf_cpu_map *cpus; > > - > > - perf_pmu__for_each_hybrid_pmu(pmu) { > > - config = PERF_COUNT_HW_CPU_CYCLES | > > - ((__u64)pmu->type << PERF_PMU_TYPE_SHIFT); > > - evsel = evsel__new_cycles(precise, PERF_TYPE_HARDWARE, > > - config); > > - if (!evsel) > > - return -ENOMEM; > > - > > - cpus = perf_cpu_map__get(pmu->cpus); > > - evsel->core.cpus = cpus; > > - evsel->core.own_cpus = perf_cpu_map__get(cpus); > > - evsel->pmu_name = strdup(pmu->name); > > - evlist__add(evlist, evsel); > > - } > > - > > - return 0; > > -} > > - > > bool evlist__has_hybrid(struct evlist *evlist) > > { > > struct evsel *evsel; > > diff --git a/tools/perf/util/evlist-hybrid.h b/tools/perf/util/evlist-hybrid.h > > index 4b000eda6626..0cded76eb344 100644 > > --- a/tools/perf/util/evlist-hybrid.h > > +++ b/tools/perf/util/evlist-hybrid.h > > @@ -7,7 +7,6 @@ > > #include "evlist.h" > > #include <unistd.h> > > > > -int evlist__add_default_hybrid(struct evlist *evlist, bool precise); > > bool evlist__has_hybrid(struct evlist *evlist); > > > > #endif /* __PERF_EVLIST_HYBRID_H */ > > diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c > > index 9dfa977193b3..63f8821a5395 100644 > > --- a/tools/perf/util/evlist.c > > +++ b/tools/perf/util/evlist.c > > @@ -93,8 +93,15 @@ struct evlist *evlist__new(void) > > struct evlist *evlist__new_default(void) > > { > > struct evlist *evlist = evlist__new(); > > + bool can_profile_kernel; > > + int err; > > + > > + if (!evlist) > > + return NULL; > > > > - if (evlist && evlist__add_default(evlist)) { > > + can_profile_kernel = perf_event_paranoid_check(1); > > + err = parse_event(evlist, can_profile_kernel ? "cycles:P" : "cycles:Pu"); > > + if (err) { > > evlist__delete(evlist); > > evlist = NULL; > > } > > @@ -237,19 +244,6 @@ static void evlist__set_leader(struct evlist *evlist) > > perf_evlist__set_leader(&evlist->core); > > } > > > > -int __evlist__add_default(struct evlist *evlist, bool precise) > > -{ > > - struct evsel *evsel; > > - > > - evsel = evsel__new_cycles(precise, PERF_TYPE_HARDWARE, > > - PERF_COUNT_HW_CPU_CYCLES); > > - if (evsel == NULL) > > - return -ENOMEM; > > - > > - evlist__add(evlist, evsel); > > - return 0; > > -} > > - > > static struct evsel *evlist__dummy_event(struct evlist *evlist) > > { > > struct perf_event_attr attr = { > > diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h > > index 5e7ff44f3043..664c6bf7b3e0 100644 > > --- a/tools/perf/util/evlist.h > > +++ b/tools/perf/util/evlist.h > > @@ -100,13 +100,6 @@ void evlist__delete(struct evlist *evlist); > > void evlist__add(struct evlist *evlist, struct evsel *entry); > > void evlist__remove(struct evlist *evlist, struct evsel *evsel); > > > > -int __evlist__add_default(struct evlist *evlist, bool precise); > > - > > -static inline int evlist__add_default(struct evlist *evlist) > > -{ > > - return __evlist__add_default(evlist, true); > > -} > > - > > int evlist__add_attrs(struct evlist *evlist, struct perf_event_attr *attrs, size_t nr_attrs); > > > > int __evlist__add_default_attrs(struct evlist *evlist, > > diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c > > index 8c8f371ea2b5..1df8f967d2eb 100644 > > --- a/tools/perf/util/evsel.c > > +++ b/tools/perf/util/evsel.c > > @@ -316,48 +316,6 @@ struct evsel *evsel__new_idx(struct perf_event_attr *attr, int idx) > > return evsel; > > } > > > > -static bool perf_event_can_profile_kernel(void) > > -{ > > - return perf_event_paranoid_check(1); > > -} > > - > > -struct evsel *evsel__new_cycles(bool precise __maybe_unused, __u32 type, __u64 config) > > -{ > > - struct perf_event_attr attr = { > > - .type = type, > > - .config = config, > > - .exclude_kernel = !perf_event_can_profile_kernel(), > > - }; > > - struct evsel *evsel; > > - > > - event_attr_init(&attr); > > - > > - /* > > - * Now let the usual logic to set up the perf_event_attr defaults > > - * to kick in when we return and before perf_evsel__open() is called. > > - */ > > - evsel = evsel__new(&attr); > > - if (evsel == NULL) > > - goto out; > > - > > - arch_evsel__fixup_new_cycles(&evsel->core.attr); > > - > > - evsel->precise_max = true; > > - > > - /* use asprintf() because free(evsel) assumes name is allocated */ > > - if (asprintf(&evsel->name, "cycles%s%s%.*s", > > - (attr.precise_ip || attr.exclude_kernel) ? ":" : "", > > - attr.exclude_kernel ? "u" : "", > > - attr.precise_ip ? attr.precise_ip + 1 : 0, "ppp") < 0) > > - goto error_free; > > -out: > > - return evsel; > > -error_free: > > - evsel__delete(evsel); > > - evsel = NULL; > > - goto out; > > -} > > - > > int copy_config_terms(struct list_head *dst, struct list_head *src) > > { > > struct evsel_config_term *pos, *tmp; > > @@ -1131,10 +1089,6 @@ void __weak arch_evsel__set_sample_weight(struct evsel *evsel) > > evsel__set_sample_bit(evsel, WEIGHT); > > } > > > > -void __weak arch_evsel__fixup_new_cycles(struct perf_event_attr *attr __maybe_unused) > > -{ > > -} > > - > > void __weak arch__post_evsel_config(struct evsel *evsel __maybe_unused, > > struct perf_event_attr *attr __maybe_unused) > > { > > diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h > > index df8928745fc6..429b172cc94d 100644 > > --- a/tools/perf/util/evsel.h > > +++ b/tools/perf/util/evsel.h > > @@ -243,8 +243,6 @@ static inline struct evsel *evsel__newtp(const char *sys, const char *name) > > } > > #endif > > > > -struct evsel *evsel__new_cycles(bool precise, __u32 type, __u64 config); > > - > > #ifdef HAVE_LIBTRACEEVENT > > struct tep_event *event_format__new(const char *sys, const char *name); > > #endif > > @@ -312,7 +310,6 @@ void __evsel__reset_sample_bit(struct evsel *evsel, enum perf_event_sample_forma > > void evsel__set_sample_id(struct evsel *evsel, bool use_sample_identifier); > > > > void arch_evsel__set_sample_weight(struct evsel *evsel); > > -void arch_evsel__fixup_new_cycles(struct perf_event_attr *attr); > > void arch__post_evsel_config(struct evsel *evsel, struct perf_event_attr *attr); > > > > int evsel__set_filter(struct evsel *evsel, const char *filter); > > -- > > 2.41.0.rc0.172.g3f132b7071-goog > > > > -- > > - Arnaldo
diff --git a/tools/perf/arch/x86/util/evsel.c b/tools/perf/arch/x86/util/evsel.c index ea3972d785d1..153cdca94cd4 100644 --- a/tools/perf/arch/x86/util/evsel.c +++ b/tools/perf/arch/x86/util/evsel.c @@ -16,26 +16,6 @@ void arch_evsel__set_sample_weight(struct evsel *evsel) evsel__set_sample_bit(evsel, WEIGHT_STRUCT); } -void arch_evsel__fixup_new_cycles(struct perf_event_attr *attr) -{ - struct perf_env env = { .total_mem = 0, } ; - - if (!perf_env__cpuid(&env)) - return; - - /* - * On AMD, precise cycles event sampling internally uses IBS pmu. - * But IBS does not have filtering capabilities and perf by default - * sets exclude_guest = 1. This makes IBS pmu event init fail and - * thus perf ends up doing non-precise sampling. Avoid it by clearing - * exclude_guest. - */ - if (env.cpuid && strstarts(env.cpuid, "AuthenticAMD")) - attr->exclude_guest = 0; - - free(env.cpuid); -} - /* Check whether the evsel's PMU supports the perf metrics */ bool evsel__sys_has_perf_metrics(const struct evsel *evsel) { diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 88f7b4241153..d80b54a6f450 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -4161,18 +4161,11 @@ int cmd_record(int argc, const char **argv) record.opts.tail_synthesize = true; if (rec->evlist->core.nr_entries == 0) { - if (perf_pmu__has_hybrid()) { - err = evlist__add_default_hybrid(rec->evlist, - !record.opts.no_samples); - } else { - err = __evlist__add_default(rec->evlist, - !record.opts.no_samples); - } + bool can_profile_kernel = perf_event_paranoid_check(1); - if (err < 0) { - pr_err("Not enough memory for event selector list\n"); + err = parse_event(rec->evlist, can_profile_kernel ? "cycles:P" : "cycles:Pu"); + if (err) goto out; - } } if (rec->opts.target.tid && !rec->opts.no_inherit_set) diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 48ee49e95c5e..27a7f068207d 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -1653,10 +1653,12 @@ int cmd_top(int argc, const char **argv) if (annotate_check_args(&top.annotation_opts) < 0) goto out_delete_evlist; - if (!top.evlist->core.nr_entries && - evlist__add_default(top.evlist) < 0) { - pr_err("Not enough memory for event selector list\n"); - goto out_delete_evlist; + if (!top.evlist->core.nr_entries) { + bool can_profile_kernel = perf_event_paranoid_check(1); + int err = parse_event(top.evlist, can_profile_kernel ? "cycles:P" : "cycles:Pu"); + + if (err) + goto out_delete_evlist; } status = evswitch__init(&top.evswitch, top.evlist, stderr); diff --git a/tools/perf/util/evlist-hybrid.c b/tools/perf/util/evlist-hybrid.c index 0f59c80f27b2..64f78d06fe19 100644 --- a/tools/perf/util/evlist-hybrid.c +++ b/tools/perf/util/evlist-hybrid.c @@ -16,31 +16,6 @@ #include <perf/evsel.h> #include <perf/cpumap.h> -int evlist__add_default_hybrid(struct evlist *evlist, bool precise) -{ - struct evsel *evsel; - struct perf_pmu *pmu; - __u64 config; - struct perf_cpu_map *cpus; - - perf_pmu__for_each_hybrid_pmu(pmu) { - config = PERF_COUNT_HW_CPU_CYCLES | - ((__u64)pmu->type << PERF_PMU_TYPE_SHIFT); - evsel = evsel__new_cycles(precise, PERF_TYPE_HARDWARE, - config); - if (!evsel) - return -ENOMEM; - - cpus = perf_cpu_map__get(pmu->cpus); - evsel->core.cpus = cpus; - evsel->core.own_cpus = perf_cpu_map__get(cpus); - evsel->pmu_name = strdup(pmu->name); - evlist__add(evlist, evsel); - } - - return 0; -} - bool evlist__has_hybrid(struct evlist *evlist) { struct evsel *evsel; diff --git a/tools/perf/util/evlist-hybrid.h b/tools/perf/util/evlist-hybrid.h index 4b000eda6626..0cded76eb344 100644 --- a/tools/perf/util/evlist-hybrid.h +++ b/tools/perf/util/evlist-hybrid.h @@ -7,7 +7,6 @@ #include "evlist.h" #include <unistd.h> -int evlist__add_default_hybrid(struct evlist *evlist, bool precise); bool evlist__has_hybrid(struct evlist *evlist); #endif /* __PERF_EVLIST_HYBRID_H */ diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 9dfa977193b3..63f8821a5395 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -93,8 +93,15 @@ struct evlist *evlist__new(void) struct evlist *evlist__new_default(void) { struct evlist *evlist = evlist__new(); + bool can_profile_kernel; + int err; + + if (!evlist) + return NULL; - if (evlist && evlist__add_default(evlist)) { + can_profile_kernel = perf_event_paranoid_check(1); + err = parse_event(evlist, can_profile_kernel ? "cycles:P" : "cycles:Pu"); + if (err) { evlist__delete(evlist); evlist = NULL; } @@ -237,19 +244,6 @@ static void evlist__set_leader(struct evlist *evlist) perf_evlist__set_leader(&evlist->core); } -int __evlist__add_default(struct evlist *evlist, bool precise) -{ - struct evsel *evsel; - - evsel = evsel__new_cycles(precise, PERF_TYPE_HARDWARE, - PERF_COUNT_HW_CPU_CYCLES); - if (evsel == NULL) - return -ENOMEM; - - evlist__add(evlist, evsel); - return 0; -} - static struct evsel *evlist__dummy_event(struct evlist *evlist) { struct perf_event_attr attr = { diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 5e7ff44f3043..664c6bf7b3e0 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -100,13 +100,6 @@ void evlist__delete(struct evlist *evlist); void evlist__add(struct evlist *evlist, struct evsel *entry); void evlist__remove(struct evlist *evlist, struct evsel *evsel); -int __evlist__add_default(struct evlist *evlist, bool precise); - -static inline int evlist__add_default(struct evlist *evlist) -{ - return __evlist__add_default(evlist, true); -} - int evlist__add_attrs(struct evlist *evlist, struct perf_event_attr *attrs, size_t nr_attrs); int __evlist__add_default_attrs(struct evlist *evlist, diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index 8c8f371ea2b5..1df8f967d2eb 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -316,48 +316,6 @@ struct evsel *evsel__new_idx(struct perf_event_attr *attr, int idx) return evsel; } -static bool perf_event_can_profile_kernel(void) -{ - return perf_event_paranoid_check(1); -} - -struct evsel *evsel__new_cycles(bool precise __maybe_unused, __u32 type, __u64 config) -{ - struct perf_event_attr attr = { - .type = type, - .config = config, - .exclude_kernel = !perf_event_can_profile_kernel(), - }; - struct evsel *evsel; - - event_attr_init(&attr); - - /* - * Now let the usual logic to set up the perf_event_attr defaults - * to kick in when we return and before perf_evsel__open() is called. - */ - evsel = evsel__new(&attr); - if (evsel == NULL) - goto out; - - arch_evsel__fixup_new_cycles(&evsel->core.attr); - - evsel->precise_max = true; - - /* use asprintf() because free(evsel) assumes name is allocated */ - if (asprintf(&evsel->name, "cycles%s%s%.*s", - (attr.precise_ip || attr.exclude_kernel) ? ":" : "", - attr.exclude_kernel ? "u" : "", - attr.precise_ip ? attr.precise_ip + 1 : 0, "ppp") < 0) - goto error_free; -out: - return evsel; -error_free: - evsel__delete(evsel); - evsel = NULL; - goto out; -} - int copy_config_terms(struct list_head *dst, struct list_head *src) { struct evsel_config_term *pos, *tmp; @@ -1131,10 +1089,6 @@ void __weak arch_evsel__set_sample_weight(struct evsel *evsel) evsel__set_sample_bit(evsel, WEIGHT); } -void __weak arch_evsel__fixup_new_cycles(struct perf_event_attr *attr __maybe_unused) -{ -} - void __weak arch__post_evsel_config(struct evsel *evsel __maybe_unused, struct perf_event_attr *attr __maybe_unused) { diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index df8928745fc6..429b172cc94d 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -243,8 +243,6 @@ static inline struct evsel *evsel__newtp(const char *sys, const char *name) } #endif -struct evsel *evsel__new_cycles(bool precise, __u32 type, __u64 config); - #ifdef HAVE_LIBTRACEEVENT struct tep_event *event_format__new(const char *sys, const char *name); #endif @@ -312,7 +310,6 @@ void __evsel__reset_sample_bit(struct evsel *evsel, enum perf_event_sample_forma void evsel__set_sample_id(struct evsel *evsel, bool use_sample_identifier); void arch_evsel__set_sample_weight(struct evsel *evsel); -void arch_evsel__fixup_new_cycles(struct perf_event_attr *attr); void arch__post_evsel_config(struct evsel *evsel, struct perf_event_attr *attr); int evsel__set_filter(struct evsel *evsel, const char *filter);