Re: [PATCH v1 4/9] perf list: Generalize limiting to a PMU name
From: Ian Rogers
Date: Mon Nov 14 2022 - 12:10:22 EST
On Mon, Nov 14, 2022 at 6:55 AM Liang, Kan <kan.liang@xxxxxxxxxxxxxxx> wrote:
>
>
>
> On 2022-11-14 9:02 a.m., Ian Rogers wrote:
> > On Mon, Nov 14, 2022 at 5:58 AM Liang, Kan <kan.liang@xxxxxxxxxxxxxxx> wrote:
> >>
> >>
> >>
> >> On 2022-11-14 2:51 a.m., Ian Rogers wrote:
> >>> Deprecate the --cputype option and add a --unit option where '--unit
> >>> cpu_atom' behaves like '--cputype atom'. The --unit option can be used
> >>> with arbitrary PMUs, for example:
> >>>
> >>> ```
> >>> $ perf list --unit msr pmu
> >>>
> >>> List of pre-defined events (to be used in -e or -M):
> >>>
> >>> msr/aperf/ [Kernel PMU event]
> >>> msr/cpu_thermal_margin/ [Kernel PMU event]
> >>> msr/mperf/ [Kernel PMU event]
> >>> msr/pperf/ [Kernel PMU event]
> >>> msr/smi/ [Kernel PMU event]
> >>> msr/tsc/ [Kernel PMU event]
> >>> ```
> >>>
> >>> Signed-off-by: Ian Rogers <irogers@xxxxxxxxxx>
> >>> ---
> >>> tools/perf/Documentation/perf-list.txt | 6 +++---
> >>> tools/perf/builtin-list.c | 18 ++++++++++++------
> >>> tools/perf/util/metricgroup.c | 3 ++-
> >>> tools/perf/util/pmu.c | 4 +---
> >>> 4 files changed, 18 insertions(+), 13 deletions(-)
> >>>
> >>> diff --git a/tools/perf/Documentation/perf-list.txt b/tools/perf/Documentation/perf-list.txt
> >>> index 57384a97c04f..44a819af573d 100644
> >>> --- a/tools/perf/Documentation/perf-list.txt
> >>> +++ b/tools/perf/Documentation/perf-list.txt
> >>> @@ -39,9 +39,9 @@ any extra expressions computed by perf stat.
> >>> --deprecated::
> >>> Print deprecated events. By default the deprecated events are hidden.
> >>>
> >>> ---cputype::
> >>> -Print events applying cpu with this type for hybrid platform
> >>> -(e.g. --cputype core or --cputype atom)
> >>
> >> The "--cputype" is removed from the documentation, but the code is still
> >> available. It sounds weird.
> >>
> >> Can we still keep the "--cputype" in the documentation? Just say that
> >> it's a deprecated option, please use the --unit cpu_atom instead. I
> >> think even better if we can throw a warning and point to --unit when the
> >> "--cputype" is used.
> >
> > I think we want to remove --cputype widely in the code and replace
> > what it does with specifying a PMU name. Advertising a flag but then
> > warning seems strange and is a behavioral change from what is
> > currently done. For raw-dump we don't document it in the man page and
> > hide the flag, this is the pattern being followed here.
>
> I see. So the --cputype is still supported, but only be hidden in the
> default. Sure, we can follow the pattern.
>
> >
> > Thanks,
> > Ian
> >
> >> Thanks,
> >> Kan
> >>> +--unit::
> >>> +Print PMU events and metrics limited to the specific PMU name.
> >>> +(e.g. --unit cpu, --unit msr, --unit cpu_core, --unit cpu_atom)
> >>>
> >>> [[EVENT_MODIFIERS]]
> >>> EVENT MODIFIERS
> >>> diff --git a/tools/perf/builtin-list.c b/tools/perf/builtin-list.c
> >>> index 58e1ec1654ef..cc84ced6da26 100644
> >>> --- a/tools/perf/builtin-list.c
> >>> +++ b/tools/perf/builtin-list.c
> >>> @@ -21,7 +21,6 @@
> >>>
> >>> static bool desc_flag = true;
> >>> static bool details_flag;
> >>> -static const char *hybrid_type;
> >>>
> >>> int cmd_list(int argc, const char **argv)
> >>> {
> >>> @@ -30,6 +29,8 @@ int cmd_list(int argc, const char **argv)
> >>> bool long_desc_flag = false;
> >>> bool deprecated = false;
> >>> char *pmu_name = NULL;
> >>> + const char *hybrid_name = NULL;
> >>> + const char *unit_name = NULL;
> >>> struct option list_options[] = {
> >>> OPT_BOOLEAN(0, "raw-dump", &raw_dump, "Dump raw events"),
> >>> OPT_BOOLEAN('d', "desc", &desc_flag,
> >>> @@ -40,9 +41,10 @@ int cmd_list(int argc, const char **argv)
> >>> "Print information on the perf event names and expressions used internally by events."),
> >>> OPT_BOOLEAN(0, "deprecated", &deprecated,
> >>> "Print deprecated events."),
> >>> - OPT_STRING(0, "cputype", &hybrid_type, "hybrid cpu type",
> >>> - "Print events applying cpu with this type for hybrid platform "
> >>> - "(e.g. core or atom)"),
> >>> + OPT_STRING(0, "cputype", &hybrid_name, "hybrid cpu type",
> >>> + "Limit PMU or metric printing to the given hybrid PMU (e.g. core or atom)."),
> >>> + OPT_STRING(0, "unit", &unit_name, "PMU name",
> >>> + "Limit PMU or metric printing to the specified PMU."),
> >>> OPT_INCR(0, "debug", &verbose,
> >>> "Enable debugging output"),
> >>> OPT_END()
> >>> @@ -53,6 +55,8 @@ int cmd_list(int argc, const char **argv)
> >>> };
> >>>
> >>> set_option_flag(list_options, 0, "raw-dump", PARSE_OPT_HIDDEN);
> >>> + /* Hide hybrid flag for the more generic 'unit' flag. */
> >>> + set_option_flag(list_options, 0, "cputype", PARSE_OPT_HIDDEN);
> >>>
> >>> argc = parse_options(argc, argv, list_options, list_usage,
> >>> PARSE_OPT_STOP_AT_NON_OPTION);
> >>> @@ -62,8 +66,10 @@ int cmd_list(int argc, const char **argv)
> >>> if (!raw_dump && pager_in_use())
> >>> printf("\nList of pre-defined events (to be used in -e or -M):\n\n");
> >>>
> >>> - if (hybrid_type) {
> >>> - pmu_name = perf_pmu__hybrid_type_to_pmu(hybrid_type);
> >>> + if (unit_name)
> >>> + pmu_name = strdup(unit_name);
> >>> + else if (hybrid_name) {
> >>> + pmu_name = perf_pmu__hybrid_type_to_pmu(hybrid_name);
> >>> if (!pmu_name)
> >>> pr_warning("WARNING: hybrid cputype is not supported!\n");
> >>> }
>
> Can the tool implicitly convert the --cputype to --unit at the very
> beginning? (Just need to append a prefix "cpu_".). Here we only need to
> handle the unit_name.
> The same logic may be applied for other tools if someone implements the
> --unit for the stat or record later.
>
> BTW: we may want to check the existence of a PMU here, just like what we
> did for the hybrid. If a user perf list a nonexistence PMU, we can warn
> here.
That'd be a fine follow up. I think it would be problematic as it
would warn because of a lack of permissions.
Thanks,
Ian
> Thanks,
> Kan
>
> >>> diff --git a/tools/perf/util/metricgroup.c b/tools/perf/util/metricgroup.c
> >>> index 4c98ac29ee13..1943fed9b6d9 100644
> >>> --- a/tools/perf/util/metricgroup.c
> >>> +++ b/tools/perf/util/metricgroup.c
> >>> @@ -556,11 +556,12 @@ static int metricgroup__print_callback(const struct pmu_event *pe,
> >>> void *vdata)
> >>> {
> >>> struct metricgroup_print_data *data = vdata;
> >>> + const char *pmu = pe->pmu ?: "cpu";
> >>>
> >>> if (!pe->metric_expr)
> >>> return 0;
> >>>
> >>> - if (data->pmu_name && perf_pmu__is_hybrid(pe->pmu) && strcmp(data->pmu_name, pe->pmu))
> >>> + if (data->pmu_name && strcmp(data->pmu_name, pmu))
> >>> return 0;
> >>>
> >>> return metricgroup__print_pmu_event(pe, data->metricgroups, data->filter,
> >>> diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c
> >>> index a8f9f47c6ed9..9c771f136b81 100644
> >>> --- a/tools/perf/util/pmu.c
> >>> +++ b/tools/perf/util/pmu.c
> >>> @@ -1694,10 +1694,8 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
> >>> pmu = NULL;
> >>> j = 0;
> >>> while ((pmu = perf_pmu__scan(pmu)) != NULL) {
> >>> - if (pmu_name && perf_pmu__is_hybrid(pmu->name) &&
> >>> - strcmp(pmu_name, pmu->name)) {
> >>> + if (pmu_name && pmu->name && strcmp(pmu_name, pmu->name))
> >>> continue;
> >>> - }
> >>>
> >>> list_for_each_entry(alias, &pmu->aliases, list) {
> >>> char *name = alias->desc ? alias->name :