From: Mark Rutland <mark.rutland@arm.com>
To: Ian Rogers <irogers@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>,
Ingo Molnar <mingo@redhat.com>,
Arnaldo Carvalho de Melo <acme@kernel.org>,
Namhyung Kim <namhyung@kernel.org>,
Alexander Shishkin <alexander.shishkin@linux.intel.com>,
Jiri Olsa <jolsa@kernel.org>,
Adrian Hunter <adrian.hunter@intel.com>,
Kan Liang <kan.liang@linux.intel.com>,
James Clark <james.clark@linaro.org>, Xu Yang <xu.yang_2@nxp.com>,
Thomas Falcon <thomas.falcon@intel.com>,
Andi Kleen <ak@linux.intel.com>,
linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org,
bpf@vger.kernel.org, Atish Patra <atishp@rivosinc.com>,
Beeman Strong <beeman@rivosinc.com>, Leo Yan <leo.yan@arm.com>,
Vince Weaver <vincent.weaver@maine.edu>
Subject: Re: [PATCH v4 20/21] perf parse-events: Add HW_CYCLES_STR as default cycles event string
Date: Tue, 16 Sep 2025 14:18:40 +0100 [thread overview]
Message-ID: <aMljsNZxTw5ZPfeb@J2N7QTR9R3.cambridge.arm.com> (raw)
In-Reply-To: <20250914181121.1952748-21-irogers@google.com>
On Sun, Sep 14, 2025 at 11:11:20AM -0700, Ian Rogers wrote:
> ARM managed to significantly overload the meaning of the "cycles"
> event in their PMU kernel drivers through sysfs.
Ian, please stop phrasing this as if Arm have done something wrong here.
It's true that some system PMU drivers have events named 'cycles'.
It's not true that this is "overloading" the meaning of 'cycles'; those
PMU-specific events were never intended to be conflated with the
PERF_TYPE_HARDWARE events which have the same name.
This was never a problem until the perf tool was changed to handle
events such that it blindly assumed all events were in the same
namespace. I have repeatedly explained that this was a bad idea.
There is no reason that this should be handled in an ARM-specific way;
if you want the bare 'cycles' event (withj no explcit PMU) to mean
PERF_TYPE_HARDWARE:PERF_COUNT_HW_CPU_CYCLES, then *don't* match that
with other PMU types. We cna specifically identify CPU PMUs which
support that with the extended type ID if necessary.
Mark.
> In the tool use
> "cpu-cycles" on ARM to avoid wildcard matching on different PMUS. This
> is most commonly done in test code.
>
> Signed-off-by: Ian Rogers <irogers@google.com>
> ---
> tools/perf/builtin-stat.c | 4 +-
> tools/perf/tests/code-reading.c | 4 +-
> tools/perf/tests/keep-tracking.c | 2 +-
> tools/perf/tests/parse-events.c | 100 ++++++++++++++--------------
> tools/perf/tests/perf-time-to-tsc.c | 2 +-
> tools/perf/tests/switch-tracking.c | 2 +-
> tools/perf/util/evlist.c | 2 +-
> tools/perf/util/parse-events.h | 10 +++
> tools/perf/util/perf_api_probe.c | 4 +-
> 9 files changed, 71 insertions(+), 59 deletions(-)
>
> diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> index 2c38dd98f6ca..9f522b787ad5 100644
> --- a/tools/perf/builtin-stat.c
> +++ b/tools/perf/builtin-stat.c
> @@ -1957,7 +1957,7 @@ static int add_default_events(void)
> "cpu-migrations,"
> "page-faults,"
> "instructions,"
> - "cycles,"
> + HW_CYCLES_STR ","
> "stalled-cycles-frontend,"
> "stalled-cycles-backend,"
> "branches,"
> @@ -2043,7 +2043,7 @@ static int add_default_events(void)
> * Make at least one event non-skippable so fatal errors are visible.
> * 'cycles' always used to be default and non-skippable, so use that.
> */
> - if (strcmp("cycles", evsel__name(evsel)))
> + if (strcmp(HW_CYCLES_STR, evsel__name(evsel)))
> evsel->skippable = true;
> }
> }
> diff --git a/tools/perf/tests/code-reading.c b/tools/perf/tests/code-reading.c
> index 9c2091310191..baa44918f555 100644
> --- a/tools/perf/tests/code-reading.c
> +++ b/tools/perf/tests/code-reading.c
> @@ -649,7 +649,9 @@ static int do_test_code_reading(bool try_kcore)
> struct map *map;
> bool have_vmlinux, have_kcore;
> struct dso *dso;
> - const char *events[] = { "cycles", "cycles:u", "cpu-clock", "cpu-clock:u", NULL };
> + const char *events[] = {
> + HW_CYCLES_STR, HW_CYCLES_STR ":u", "cpu-clock", "cpu-clock:u", NULL
> + };
> int evidx = 0;
> struct perf_env host_env;
>
> diff --git a/tools/perf/tests/keep-tracking.c b/tools/perf/tests/keep-tracking.c
> index eafb49eb0b56..d54ddb4db47b 100644
> --- a/tools/perf/tests/keep-tracking.c
> +++ b/tools/perf/tests/keep-tracking.c
> @@ -90,7 +90,7 @@ static int test__keep_tracking(struct test_suite *test __maybe_unused, int subte
> perf_evlist__set_maps(&evlist->core, cpus, threads);
>
> CHECK__(parse_event(evlist, "dummy:u"));
> - CHECK__(parse_event(evlist, "cycles:u"));
> + CHECK__(parse_event(evlist, HW_CYCLES_STR ":u"));
>
> evlist__config(evlist, &opts, NULL);
>
> diff --git a/tools/perf/tests/parse-events.c b/tools/perf/tests/parse-events.c
> index 4e55b0d295bd..7d59648a0591 100644
> --- a/tools/perf/tests/parse-events.c
> +++ b/tools/perf/tests/parse-events.c
> @@ -198,7 +198,7 @@ static int test__checkevent_symbolic_name_config(struct evlist *evlist)
> TEST_ASSERT_VAL("wrong number of entries", 0 != evlist->core.nr_entries);
>
> perf_evlist__for_each_evsel(&evlist->core, evsel) {
> - int ret = assert_hw(evsel, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + int ret = assert_hw(evsel, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
>
> if (ret)
> return ret;
> @@ -884,7 +884,7 @@ static int test__group1(struct evlist *evlist)
>
> /* cycles:upp */
> evsel = evsel__next(evsel);
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -948,7 +948,7 @@ static int test__group2(struct evlist *evlist)
> continue;
> }
> /* cycles:k */
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -1085,7 +1085,7 @@ static int test__group4(struct evlist *evlist __maybe_unused)
>
> /* cycles:u + p */
> evsel = leader = (i == 0 ? evlist__first(evlist) : evsel__next(evsel));
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -1133,7 +1133,7 @@ static int test__group5(struct evlist *evlist __maybe_unused)
> for (int i = 0; i < num_core_entries(); i++) {
> /* cycles + G */
> evsel = leader = (i == 0 ? evlist__first(evlist) : evsel__next(evsel));
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -1168,7 +1168,7 @@ static int test__group5(struct evlist *evlist __maybe_unused)
> for (int i = 0; i < num_core_entries(); i++) {
> /* cycles:G */
> evsel = leader = evsel__next(evsel);
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -1202,7 +1202,7 @@ static int test__group5(struct evlist *evlist __maybe_unused)
> for (int i = 0; i < num_core_entries(); i++) {
> /* cycles */
> evsel = evsel__next(evsel);
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -1231,7 +1231,7 @@ static int test__group_gh1(struct evlist *evlist)
>
> /* cycles + :H group modifier */
> evsel = leader = (i == 0 ? evlist__first(evlist) : evsel__next(evsel));
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -1278,7 +1278,7 @@ static int test__group_gh2(struct evlist *evlist)
>
> /* cycles + :G group modifier */
> evsel = leader = (i == 0 ? evlist__first(evlist) : evsel__next(evsel));
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -1325,7 +1325,7 @@ static int test__group_gh3(struct evlist *evlist)
>
> /* cycles:G + :u group modifier */
> evsel = leader = (i == 0 ? evlist__first(evlist) : evsel__next(evsel));
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -1372,7 +1372,7 @@ static int test__group_gh4(struct evlist *evlist)
>
> /* cycles:G + :uG group modifier */
> evsel = leader = (i == 0 ? evlist__first(evlist) : evsel__next(evsel));
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -1417,7 +1417,7 @@ static int test__leader_sample1(struct evlist *evlist)
>
> /* cycles - sampling group leader */
> evsel = leader = (i == 0 ? evlist__first(evlist) : evsel__next(evsel));
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -1540,7 +1540,7 @@ static int test__pinned_group(struct evlist *evlist)
>
> /* cycles - group leader */
> evsel = leader = (i == 0 ? evlist__first(evlist) : evsel__next(evsel));
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -1594,7 +1594,7 @@ static int test__exclusive_group(struct evlist *evlist)
>
> /* cycles - group leader */
> evsel = leader = (i == 0 ? evlist__first(evlist) : evsel__next(evsel));
> - ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
> if (ret)
> return ret;
>
> @@ -1759,7 +1759,7 @@ static int test__checkevent_raw_pmu(struct evlist *evlist)
> static int test__sym_event_slash(struct evlist *evlist)
> {
> struct evsel *evsel = evlist__first(evlist);
> - int ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + int ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
>
> if (ret)
> return ret;
> @@ -1771,7 +1771,7 @@ static int test__sym_event_slash(struct evlist *evlist)
> static int test__sym_event_dc(struct evlist *evlist)
> {
> struct evsel *evsel = evlist__first(evlist);
> - int ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + int ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
>
> if (ret)
> return ret;
> @@ -1783,7 +1783,7 @@ static int test__sym_event_dc(struct evlist *evlist)
> static int test__term_equal_term(struct evlist *evlist)
> {
> struct evsel *evsel = evlist__first(evlist);
> - int ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + int ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
>
> if (ret)
> return ret;
> @@ -1795,7 +1795,7 @@ static int test__term_equal_term(struct evlist *evlist)
> static int test__term_equal_legacy(struct evlist *evlist)
> {
> struct evsel *evsel = evlist__first(evlist);
> - int ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, "cycles");
> + int ret = assert_hw(&evsel->core, PERF_COUNT_HW_CPU_CYCLES, HW_CYCLES_STR);
>
> if (ret)
> return ret;
> @@ -2006,27 +2006,27 @@ static const struct evlist_test test__events[] = {
> /* 7 */
> },
> {
> - .name = "{instructions:k,cycles:upp}",
> + .name = "{instructions:k," HW_CYCLES_STR ":upp}",
> .check = test__group1,
> /* 8 */
> },
> {
> - .name = "{faults:k,branches}:u,cycles:k",
> + .name = "{faults:k,branches}:u," HW_CYCLES_STR ":k",
> .check = test__group2,
> /* 9 */
> },
> {
> - .name = "group1{syscalls:sys_enter_openat:H,cycles:kppp},group2{cycles,1:3}:G,instructions:u",
> + .name = "group1{syscalls:sys_enter_openat:H," HW_CYCLES_STR ":kppp},group2{" HW_CYCLES_STR ",1:3}:G,instructions:u",
> .check = test__group3,
> /* 0 */
> },
> {
> - .name = "{cycles:u,instructions:kp}:p",
> + .name = "{" HW_CYCLES_STR ":u,instructions:kp}:p",
> .check = test__group4,
> /* 1 */
> },
> {
> - .name = "{cycles,instructions}:G,{cycles:G,instructions:G},cycles",
> + .name = "{" HW_CYCLES_STR ",instructions}:G,{" HW_CYCLES_STR ":G,instructions:G}," HW_CYCLES_STR,
> .check = test__group5,
> /* 2 */
> },
> @@ -2036,27 +2036,27 @@ static const struct evlist_test test__events[] = {
> /* 3 */
> },
> {
> - .name = "{cycles,cache-misses:G}:H",
> + .name = "{" HW_CYCLES_STR ",cache-misses:G}:H",
> .check = test__group_gh1,
> /* 4 */
> },
> {
> - .name = "{cycles,cache-misses:H}:G",
> + .name = "{" HW_CYCLES_STR ",cache-misses:H}:G",
> .check = test__group_gh2,
> /* 5 */
> },
> {
> - .name = "{cycles:G,cache-misses:H}:u",
> + .name = "{" HW_CYCLES_STR ":G,cache-misses:H}:u",
> .check = test__group_gh3,
> /* 6 */
> },
> {
> - .name = "{cycles:G,cache-misses:H}:uG",
> + .name = "{" HW_CYCLES_STR ":G,cache-misses:H}:uG",
> .check = test__group_gh4,
> /* 7 */
> },
> {
> - .name = "{cycles,cache-misses,branch-misses}:S",
> + .name = "{" HW_CYCLES_STR ",cache-misses,branch-misses}:S",
> .check = test__leader_sample1,
> /* 8 */
> },
> @@ -2071,7 +2071,7 @@ static const struct evlist_test test__events[] = {
> /* 0 */
> },
> {
> - .name = "{cycles,cache-misses,branch-misses}:D",
> + .name = "{" HW_CYCLES_STR ",cache-misses,branch-misses}:D",
> .check = test__pinned_group,
> /* 1 */
> },
> @@ -2109,7 +2109,7 @@ static const struct evlist_test test__events[] = {
> /* 6 */
> },
> {
> - .name = "task-clock:P,cycles",
> + .name = "task-clock:P," HW_CYCLES_STR,
> .check = test__checkevent_precise_max_modifier,
> /* 7 */
> },
> @@ -2140,17 +2140,17 @@ static const struct evlist_test test__events[] = {
> /* 2 */
> },
> {
> - .name = "cycles/name='COMPLEX_CYCLES_NAME:orig=cycles,desc=chip-clock-ticks'/Duk",
> + .name = HW_CYCLES_STR "/name='COMPLEX_CYCLES_NAME:orig=cycles,desc=chip-clock-ticks'/Duk",
> .check = test__checkevent_complex_name,
> /* 3 */
> },
> {
> - .name = "cycles//u",
> + .name = HW_CYCLES_STR "//u",
> .check = test__sym_event_slash,
> /* 4 */
> },
> {
> - .name = "cycles:k",
> + .name = HW_CYCLES_STR ":k",
> .check = test__sym_event_dc,
> /* 5 */
> },
> @@ -2160,17 +2160,17 @@ static const struct evlist_test test__events[] = {
> /* 6 */
> },
> {
> - .name = "{cycles,cache-misses,branch-misses}:e",
> + .name = "{" HW_CYCLES_STR ",cache-misses,branch-misses}:e",
> .check = test__exclusive_group,
> /* 7 */
> },
> {
> - .name = "cycles/name=name/",
> + .name = HW_CYCLES_STR "/name=name/",
> .check = test__term_equal_term,
> /* 8 */
> },
> {
> - .name = "cycles/name=l1d/",
> + .name = HW_CYCLES_STR "/name=l1d/",
> .check = test__term_equal_legacy,
> /* 9 */
> },
> @@ -2311,7 +2311,7 @@ static const struct evlist_test test__events_pmu[] = {
> /* 9 */
> },
> {
> - .name = "cpu/cycles,period=100000,config2/",
> + .name = "cpu/" HW_CYCLES_STR ",period=100000,config2/",
> .valid = test__pmu_cpu_valid,
> .check = test__checkevent_symbolic_name_config,
> /* 0 */
> @@ -2335,43 +2335,43 @@ static const struct evlist_test test__events_pmu[] = {
> /* 3 */
> },
> {
> - .name = "{cpu/instructions/k,cpu/cycles/upp}",
> + .name = "{cpu/instructions/k,cpu/" HW_CYCLES_STR "/upp}",
> .valid = test__pmu_cpu_valid,
> .check = test__group1,
> /* 4 */
> },
> {
> - .name = "{cpu/cycles/u,cpu/instructions/kp}:p",
> + .name = "{cpu/" HW_CYCLES_STR "/u,cpu/instructions/kp}:p",
> .valid = test__pmu_cpu_valid,
> .check = test__group4,
> /* 5 */
> },
> {
> - .name = "{cpu/cycles/,cpu/cache-misses/G}:H",
> + .name = "{cpu/" HW_CYCLES_STR "/,cpu/cache-misses/G}:H",
> .valid = test__pmu_cpu_valid,
> .check = test__group_gh1,
> /* 6 */
> },
> {
> - .name = "{cpu/cycles/,cpu/cache-misses/H}:G",
> + .name = "{cpu/" HW_CYCLES_STR "/,cpu/cache-misses/H}:G",
> .valid = test__pmu_cpu_valid,
> .check = test__group_gh2,
> /* 7 */
> },
> {
> - .name = "{cpu/cycles/G,cpu/cache-misses/H}:u",
> + .name = "{cpu/" HW_CYCLES_STR "/G,cpu/cache-misses/H}:u",
> .valid = test__pmu_cpu_valid,
> .check = test__group_gh3,
> /* 8 */
> },
> {
> - .name = "{cpu/cycles/G,cpu/cache-misses/H}:uG",
> + .name = "{cpu/" HW_CYCLES_STR "/G,cpu/cache-misses/H}:uG",
> .valid = test__pmu_cpu_valid,
> .check = test__group_gh4,
> /* 9 */
> },
> {
> - .name = "{cpu/cycles/,cpu/cache-misses/,cpu/branch-misses/}:S",
> + .name = "{cpu/" HW_CYCLES_STR "/,cpu/cache-misses/,cpu/branch-misses/}:S",
> .valid = test__pmu_cpu_valid,
> .check = test__leader_sample1,
> /* 0 */
> @@ -2389,7 +2389,7 @@ static const struct evlist_test test__events_pmu[] = {
> /* 2 */
> },
> {
> - .name = "{cpu/cycles/,cpu/cache-misses/,cpu/branch-misses/}:D",
> + .name = "{cpu/" HW_CYCLES_STR "/,cpu/cache-misses/,cpu/branch-misses/}:D",
> .valid = test__pmu_cpu_valid,
> .check = test__pinned_group,
> /* 3 */
> @@ -2407,13 +2407,13 @@ static const struct evlist_test test__events_pmu[] = {
> /* 5 */
> },
> {
> - .name = "cpu/cycles/u",
> + .name = "cpu/" HW_CYCLES_STR "/u",
> .valid = test__pmu_cpu_valid,
> .check = test__sym_event_slash,
> /* 6 */
> },
> {
> - .name = "cpu/cycles/k",
> + .name = "cpu/" HW_CYCLES_STR "/k",
> .valid = test__pmu_cpu_valid,
> .check = test__sym_event_dc,
> /* 7 */
> @@ -2425,19 +2425,19 @@ static const struct evlist_test test__events_pmu[] = {
> /* 8 */
> },
> {
> - .name = "{cpu/cycles/,cpu/cache-misses/,cpu/branch-misses/}:e",
> + .name = "{cpu/" HW_CYCLES_STR "/,cpu/cache-misses/,cpu/branch-misses/}:e",
> .valid = test__pmu_cpu_valid,
> .check = test__exclusive_group,
> /* 9 */
> },
> {
> - .name = "cpu/cycles,name=name/",
> + .name = "cpu/" HW_CYCLES_STR ",name=name/",
> .valid = test__pmu_cpu_valid,
> .check = test__term_equal_term,
> /* 0 */
> },
> {
> - .name = "cpu/cycles,name=l1d/",
> + .name = "cpu/" HW_CYCLES_STR ",name=l1d/",
> .valid = test__pmu_cpu_valid,
> .check = test__term_equal_legacy,
> /* 1 */
> diff --git a/tools/perf/tests/perf-time-to-tsc.c b/tools/perf/tests/perf-time-to-tsc.c
> index d4437410c99f..7ebcb1f004b2 100644
> --- a/tools/perf/tests/perf-time-to-tsc.c
> +++ b/tools/perf/tests/perf-time-to-tsc.c
> @@ -101,7 +101,7 @@ static int test__perf_time_to_tsc(struct test_suite *test __maybe_unused, int su
>
> perf_evlist__set_maps(&evlist->core, cpus, threads);
>
> - CHECK__(parse_event(evlist, "cycles:u"));
> + CHECK__(parse_event(evlist, HW_CYCLES_STR ":u"));
>
> evlist__config(evlist, &opts, NULL);
>
> diff --git a/tools/perf/tests/switch-tracking.c b/tools/perf/tests/switch-tracking.c
> index 5be294014d3b..ad3a87978c0d 100644
> --- a/tools/perf/tests/switch-tracking.c
> +++ b/tools/perf/tests/switch-tracking.c
> @@ -332,7 +332,7 @@ static int process_events(struct evlist *evlist,
> static int test__switch_tracking(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
> {
> const char *sched_switch = "sched:sched_switch";
> - const char *cycles = "cycles:u";
> + const char *cycles = HW_CYCLES_STR ":u";
> struct switch_tracking switch_tracking = { .tids = NULL, };
> struct record_opts opts = {
> .mmap_pages = UINT_MAX,
> diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
> index e8217efdda53..d7e935faeda0 100644
> --- a/tools/perf/util/evlist.c
> +++ b/tools/perf/util/evlist.c
> @@ -112,7 +112,7 @@ struct evlist *evlist__new_default(void)
> char buf[256];
> int err;
>
> - snprintf(buf, sizeof(buf), "%s/cycles/%s", pmu->name,
> + snprintf(buf, sizeof(buf), "%s/%s/%s", pmu->name, HW_CYCLES_STR,
> can_profile_kernel ? "P" : "Pu");
> err = parse_event(evlist, buf);
> if (err) {
> diff --git a/tools/perf/util/parse-events.h b/tools/perf/util/parse-events.h
> index db92cd67bc0f..304676bf32dd 100644
> --- a/tools/perf/util/parse-events.h
> +++ b/tools/perf/util/parse-events.h
> @@ -20,6 +20,16 @@ struct option;
> struct perf_pmu;
> struct strbuf;
>
> +/*
> + * The name used for the "cycles" event. A different event name is used on ARM
> + * as many ARM PMUs define a "cycles" event.
> + */
> +#if defined(__aarch64__) || defined(__arm__)
> +#define HW_CYCLES_STR "cpu-cycles"
> +#else
> +#define HW_CYCLES_STR "cycles"
> +#endif
> +
> const char *event_type(size_t type);
>
> /* Arguments encoded in opt->value. */
> diff --git a/tools/perf/util/perf_api_probe.c b/tools/perf/util/perf_api_probe.c
> index 6ecf38314f01..693bb5891bc4 100644
> --- a/tools/perf/util/perf_api_probe.c
> +++ b/tools/perf/util/perf_api_probe.c
> @@ -74,9 +74,9 @@ static bool perf_probe_api(setup_probe_fn_t fn)
> if (!ret)
> return true;
>
> - pmu = perf_pmus__scan_core(/*pmu=*/NULL);
> + pmu = perf_pmus__find_core_pmu();
> if (pmu) {
> - const char *try[] = {"cycles", "instructions", NULL};
> + const char *try[] = {HW_CYCLES_STR, "instructions", NULL};
> char buf[256];
> int i = 0;
>
> --
> 2.51.0.384.g4c02a37b29-goog
>
next prev parent reply other threads:[~2025-09-16 13:18 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-09-14 18:11 [PATCH v4 00/21] Legacy hardware/cache events as json Ian Rogers
2025-09-14 18:11 ` [PATCH v4 01/21] perf parse-events: Fix legacy cache events if event is duplicated in a PMU Ian Rogers
2025-09-14 18:11 ` [PATCH v4 02/21] perf perf_api_probe: Avoid scanning all PMUs, try software PMU first Ian Rogers
2025-09-14 18:11 ` [PATCH v4 03/21] perf record: Skip don't fail for events that don't open Ian Rogers
2025-09-14 18:11 ` [PATCH v4 04/21] perf jevents: Support copying the source json files to OUTPUT Ian Rogers
2025-09-14 18:11 ` [PATCH v4 05/21] perf pmu: Don't eagerly parse event terms Ian Rogers
2025-09-14 18:11 ` [PATCH v4 06/21] perf parse-events: Remove unused FILE input argument to scanner Ian Rogers
2025-09-14 18:11 ` [PATCH v4 07/21] perf pmu: Use fd rather than FILE from new_alias Ian Rogers
2025-09-14 18:11 ` [PATCH v4 08/21] perf pmu: Factor term parsing into a perf_event_attr into a helper Ian Rogers
2025-09-14 18:11 ` [PATCH v4 09/21] perf parse-events: Add terms for legacy hardware and cache config values Ian Rogers
2025-09-14 18:11 ` [PATCH v4 10/21] perf jevents: Add legacy json terms and default_core event table helper Ian Rogers
2025-09-14 18:11 ` [PATCH v4 11/21] perf pmu: Add and use legacy_terms in alias information Ian Rogers
2025-09-14 18:11 ` [PATCH v4 12/21] perf jevents: Add legacy-hardware and legacy-cache json Ian Rogers
2025-09-14 18:11 ` [PATCH v4 13/21] perf print-events: Remove print_hwcache_events Ian Rogers
2025-09-14 18:11 ` [PATCH v4 14/21] perf print-events: Remove print_symbol_events Ian Rogers
2025-09-14 18:11 ` [PATCH v4 15/21] perf parse-events: Remove hard coded legacy hardware and cache parsing Ian Rogers
2025-09-14 18:11 ` [PATCH v4 16/21] perf record: Use evlist__new_default when no events specified Ian Rogers
2025-09-14 18:11 ` [PATCH v4 17/21] perf top: " Ian Rogers
2025-09-14 18:11 ` [PATCH v4 18/21] perf evlist: Avoid scanning all PMUs for evlist__new_default Ian Rogers
2025-09-14 18:11 ` [PATCH v4 19/21] perf evsel: Improvements to __evsel__match Ian Rogers
2025-09-14 18:11 ` [PATCH v4 20/21] perf parse-events: Add HW_CYCLES_STR as default cycles event string Ian Rogers
2025-09-16 13:18 ` Mark Rutland [this message]
2025-09-16 15:49 ` Ian Rogers
2025-09-16 17:56 ` Ian Rogers
2025-09-17 6:23 ` Namhyung Kim
2025-09-20 15:16 ` Ian Rogers
2025-09-14 18:11 ` [PATCH v4 21/21] perf test: Make stat grep more robust Ian Rogers
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=aMljsNZxTw5ZPfeb@J2N7QTR9R3.cambridge.arm.com \
--to=mark.rutland@arm.com \
--cc=acme@kernel.org \
--cc=adrian.hunter@intel.com \
--cc=ak@linux.intel.com \
--cc=alexander.shishkin@linux.intel.com \
--cc=atishp@rivosinc.com \
--cc=beeman@rivosinc.com \
--cc=bpf@vger.kernel.org \
--cc=irogers@google.com \
--cc=james.clark@linaro.org \
--cc=jolsa@kernel.org \
--cc=kan.liang@linux.intel.com \
--cc=leo.yan@arm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-perf-users@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=namhyung@kernel.org \
--cc=peterz@infradead.org \
--cc=thomas.falcon@intel.com \
--cc=vincent.weaver@maine.edu \
--cc=xu.yang_2@nxp.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®