From: Ian Rogers <irogers@google.com>
To: Peter Zijlstra <peterz@infradead.org>,
Ingo Molnar <mingo@redhat.com>,
Arnaldo Carvalho de Melo <acme@kernel.org>,
Namhyung Kim <namhyung@kernel.org>,
Mark Rutland <mark.rutland@arm.com>,
Alexander Shishkin <alexander.shishkin@linux.intel.com>,
Jiri Olsa <jolsa@kernel.org>, Ian Rogers <irogers@google.com>,
Adrian Hunter <adrian.hunter@intel.com>,
Kan Liang <kan.liang@linux.intel.com>,
Weilin Wang <weilin.wang@intel.com>,
James Clark <james.clark@linaro.org>,
Xu Yang <xu.yang_2@nxp.com>,
John Garry <john.g.garry@oracle.com>,
Howard Chu <howardchu95@gmail.com>,
Levi Yun <yeoreum.yun@arm.com>,
Dominique Martinet <asmadeus@codewreck.org>,
linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: [PATCH v4 08/16] perf intel-tpebs: Ensure events are opened, factor out finding
Date: Tue, 8 Apr 2025 23:10:35 -0700 [thread overview]
Message-ID: <20250409061043.700792-9-irogers@google.com> (raw)
In-Reply-To: <20250409061043.700792-1-irogers@google.com>
Factor out finding an tpebs_retire_lat from an evsel. Don't blindly
return when ignoring an open request, which happens after the first
open request, ensure the event was started on a fork of perf record.
Signed-off-by: Ian Rogers <irogers@google.com>
---
tools/perf/util/intel-tpebs.c | 61 +++++++++++++++++++++--------------
1 file changed, 37 insertions(+), 24 deletions(-)
diff --git a/tools/perf/util/intel-tpebs.c b/tools/perf/util/intel-tpebs.c
index c4c818f32239..e42f3ec39a64 100644
--- a/tools/perf/util/intel-tpebs.c
+++ b/tools/perf/util/intel-tpebs.c
@@ -45,6 +45,8 @@ struct tpebs_retire_lat {
int sum;
/* Average of retire_latency, val = sum / count */
double val;
+ /* Has the event been sent to perf record? */
+ bool started;
};
static int evsel__tpebs_start_perf_record(struct evsel *evsel, int control_fd[], int ack_fd[])
@@ -94,6 +96,9 @@ static int evsel__tpebs_start_perf_record(struct evsel *evsel, int control_fd[],
tpebs_cmd.out = -1;
ret = start_command(&tpebs_cmd);
zfree(&tpebs_cmd.argv);
+ list_for_each_entry(t, &tpebs_results, nd)
+ t->started = true;
+
return ret;
}
@@ -214,6 +219,19 @@ static struct tpebs_retire_lat *tpebs_retire_lat__new(struct evsel *evsel)
return result;
}
+static struct tpebs_retire_lat *tpebs_retire_lat__find(struct evsel *evsel)
+{
+ struct tpebs_retire_lat *t;
+
+ list_for_each_entry(t, &tpebs_results, nd) {
+ if (t->tpebs_name == evsel->name ||
+ !strcmp(t->tpebs_name, evsel->name) ||
+ (evsel->metric_id && !strcmp(t->tpebs_name, evsel->metric_id)))
+ return t;
+ }
+ return NULL;
+}
+
/**
* evsel__tpebs_prepare - create tpebs data structures ready for opening.
* @evsel: retire_latency evsel, all evsels on its list will be prepared.
@@ -221,16 +239,11 @@ static struct tpebs_retire_lat *tpebs_retire_lat__new(struct evsel *evsel)
static int evsel__tpebs_prepare(struct evsel *evsel)
{
struct evsel *pos;
- struct tpebs_retire_lat *tpebs_event;
-
- list_for_each_entry(tpebs_event, &tpebs_results, nd) {
- if (!strcmp(tpebs_event->tpebs_name, evsel->name)) {
- /*
- * evsel, or an identically named one, was already
- * prepared.
- */
- return 0;
- }
+ struct tpebs_retire_lat *tpebs_event = tpebs_retire_lat__find(evsel);
+
+ if (tpebs_event) {
+ /* evsel, or an identically named one, was already prepared. */
+ return 0;
}
tpebs_event = tpebs_retire_lat__new(evsel);
if (!tpebs_event)
@@ -262,12 +275,19 @@ int evsel__tpebs_open(struct evsel *evsel)
{
int ret;
- /*
- * We should only run tpebs_start when tpebs_recording is enabled.
- * And we should only run it once with all the required events.
- */
- if (tpebs_cmd.pid != 0 || !tpebs_recording)
+ /* We should only run tpebs_start when tpebs_recording is enabled. */
+ if (!tpebs_recording)
return 0;
+ /* Only start the events once. */
+ if (tpebs_cmd.pid != 0) {
+ struct tpebs_retire_lat *t = tpebs_retire_lat__find(evsel);
+
+ if (!t || !t->started) {
+ /* Fail, as the event wasn't started. */
+ return -EBUSY;
+ }
+ return 0;
+ }
ret = evsel__tpebs_prepare(evsel);
if (ret)
@@ -352,7 +372,6 @@ int evsel__tpebs_open(struct evsel *evsel)
int tpebs_set_evsel(struct evsel *evsel, int cpu_map_idx, int thread)
{
__u64 val;
- bool found = false;
struct tpebs_retire_lat *t;
struct perf_counts_values *count;
@@ -367,19 +386,13 @@ int tpebs_set_evsel(struct evsel *evsel, int cpu_map_idx, int thread)
tpebs_stop();
count = perf_counts(evsel->counts, cpu_map_idx, thread);
- list_for_each_entry(t, &tpebs_results, nd) {
- if (t->tpebs_name == evsel->name ||
- (evsel->metric_id && !strcmp(t->tpebs_name, evsel->metric_id))) {
- found = true;
- break;
- }
- }
+ t = tpebs_retire_lat__find(evsel);
/* Set ena and run to non-zero */
count->ena = count->run = 1;
count->lost = 0;
- if (!found) {
+ if (!t) {
/*
* Set default value or 0 when retire_latency for this event is
* not found from sampling data (record_tpebs not set or 0
--
2.49.0.504.g3bcea36a83-goog
next prev parent reply other threads:[~2025-04-09 6:11 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-04-09 6:10 [PATCH v4 00/16] Intel TPEBS min/max/mean/last support Ian Rogers
2025-04-09 6:10 ` [PATCH v4 01/16] perf intel-tpebs: Cleanup header Ian Rogers
2025-04-09 6:10 ` [PATCH v4 02/16] perf intel-tpebs: Simplify tpebs_cmd Ian Rogers
2025-04-09 6:10 ` [PATCH v4 03/16] perf intel-tpebs: Rename tpebs_start to evsel__tpebs_open Ian Rogers
2025-04-09 6:10 ` [PATCH v4 04/16] perf intel-tpebs: Separate evsel__tpebs_prepare out of evsel__tpebs_open Ian Rogers
2025-04-09 6:10 ` [PATCH v4 05/16] perf intel-tpebs: Move cpumap_buf " Ian Rogers
2025-04-09 6:10 ` [PATCH v4 06/16] perf intel-tpebs: Reduce scope of tpebs_events_size Ian Rogers
2025-04-09 6:10 ` [PATCH v4 07/16] perf intel-tpebs: Inline get_perf_record_args Ian Rogers
2025-04-09 6:10 ` Ian Rogers [this message]
2025-04-09 6:10 ` [PATCH v4 09/16] perf intel-tpebs: Refactor tpebs_results list Ian Rogers
2025-04-09 6:10 ` [PATCH v4 10/16] perf intel-tpebs: Add support for updating counts in evsel__tpebs_read Ian Rogers
2025-04-09 6:10 ` [PATCH v4 11/16] perf intel-tpebs: Add mutex for tpebs_results Ian Rogers
2025-04-11 22:54 ` Namhyung Kim
2025-04-14 17:00 ` Ian Rogers
2025-04-09 6:10 ` [PATCH v4 12/16] perf intel-tpebs: Don't close record on read Ian Rogers
2025-04-09 6:10 ` [PATCH v4 13/16] perf intel-tpebs: Use stats for retirement latency statistics Ian Rogers
2025-04-09 6:10 ` [PATCH v4 14/16] perf stat: Add mean, min, max and last --tpebs-mode options Ian Rogers
2025-04-09 6:10 ` [PATCH v4 15/16] perf pmu-events: Add retirement latency to JSON events inside of perf Ian Rogers
2025-04-09 6:10 ` [PATCH v4 16/16] perf record: Retirement latency cleanup in evsel__config Ian Rogers
2025-04-10 3:12 ` Wang, Weilin
2025-04-11 23:09 ` [PATCH v4 00/16] Intel TPEBS min/max/mean/last support Namhyung Kim
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=20250409061043.700792-9-irogers@google.com \
--to=irogers@google.com \
--cc=acme@kernel.org \
--cc=adrian.hunter@intel.com \
--cc=alexander.shishkin@linux.intel.com \
--cc=asmadeus@codewreck.org \
--cc=howardchu95@gmail.com \
--cc=james.clark@linaro.org \
--cc=john.g.garry@oracle.com \
--cc=jolsa@kernel.org \
--cc=kan.liang@linux.intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-perf-users@vger.kernel.org \
--cc=mark.rutland@arm.com \
--cc=mingo@redhat.com \
--cc=namhyung@kernel.org \
--cc=peterz@infradead.org \
--cc=weilin.wang@intel.com \
--cc=xu.yang_2@nxp.com \
--cc=yeoreum.yun@arm.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