Received-SPF: pass (google.com: domain of linux-kernel+bounces-162993-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Date: Mon, 29 Apr 2024 13:02:25 -0700
In-Reply-To: <20240429200225.1271876-1-irogers@google.com>
Message-Id: <20240429200225.1271876-7-irogers@google.com>
Precedence: bulk
Mime-Version: 1.0
References: <20240429200225.1271876-1-irogers@google.com>
Subject: [PATCH v3 6/6] perf pmu: Assume sysfs events are always the same case
From: Ian Rogers <irogers@google.com>
To: Kan Liang <kan.liang@linux.intel.com>, Thomas Richter <tmricht@linux.ibm.com>, 
	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>, 
	Bjorn Helgaas <bhelgaas@google.com>, Jonathan Corbet <corbet@lwn.net>, 
	Randy Dunlap <rdunlap@infradead.org>, Jing Zhang <renyu.zj@linux.alibaba.com>, 
	James Clark <james.clark@arm.com>, Ravi Bangoria <ravi.bangoria@amd.com>, 
	linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org
Cc: Ian Rogers <irogers@google.com>
Content-Type: text/plain; charset="UTF-8"

Perf event names aren't case sensitive. For sysfs events the entire
directory of events is read then iterated comparing names in a case
insensitive way, most often to see if an event is present.

Consider:
$ perf stat -e inst_retired.any true

The event inst_retired.any may be present in any PMU, so every PMU's
sysfs events are loaded and then searched with strcasecmp to see if
any match. This event is only present on the cpu PMU as a json event
so a lot of events were loaded from sysfs unnecessarily just to prove
an event didn't exist there.

This change avoids loading all the events by assuming sysfs event
names are always either lower or uppercase. It uses file exists and
only loads the events when the desired event is present.

For the example above, the number of openat calls measured by perf
trace on a tigerlake laptop goes from 325 down to 255. The reduction
will be larger for machines with many PMUs, particularly replicated
uncore PMUs.

Ensure pmu_aliases_parse is called before all uses of the aliases
list, but remove some "pmu->sysfs_aliases_loaded" tests as they are
now part of the function.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/util/pmu.c | 31 ++++++++++++++++++++++++++-----
 1 file changed, 26 insertions(+), 5 deletions(-)

diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c
index 7849be4bfea1..7275076e6d83 100644
--- a/tools/perf/util/pmu.c
+++ b/tools/perf/util/pmu.c
@@ -425,9 +425,30 @@ static struct perf_pmu_alias *perf_pmu__find_alias(struct perf_pmu *pmu,
 {
 	struct perf_pmu_alias *alias;
 
-	if (load && !pmu->sysfs_aliases_loaded)
-		pmu_aliases_parse(pmu);
+	if (load && !pmu->sysfs_aliases_loaded) {
+		bool has_sysfs_event;
+		char event_file_name[FILENAME_MAX + 8];
 
+		/*
+		 * Test if alias/event 'name' exists in the PMU's sysfs/events
+		 * directory. If not skip parsing the sysfs aliases. Syfs event
+		 * name must be all lower or all upper case.
+		 */
+		scnprintf(event_file_name, sizeof(event_file_name), "events/%s", name);
+		for (size_t i = 7, n = 7 + strlen(name); i < n; i++)
+			event_file_name[i] = tolower(event_file_name[i]);
+
+		has_sysfs_event = perf_pmu__file_exists(pmu, event_file_name);
+		if (!has_sysfs_event) {
+			for (size_t i = 7, n = 7 + strlen(name); i < n; i++)
+				event_file_name[i] = toupper(event_file_name[i]);
+
+			has_sysfs_event = perf_pmu__file_exists(pmu, event_file_name);
+		}
+		if (has_sysfs_event)
+			pmu_aliases_parse(pmu);
+
+	}
 	list_for_each_entry(alias, &pmu->aliases, list) {
 		if (!strcasecmp(alias->name, name))
 			return alias;
@@ -1717,9 +1738,7 @@ size_t perf_pmu__num_events(struct perf_pmu *pmu)
 {
 	size_t nr;
 
-	if (!pmu->sysfs_aliases_loaded)
-		pmu_aliases_parse(pmu);
-
+	pmu_aliases_parse(pmu);
 	nr = pmu->sysfs_aliases;
 
 	if (pmu->cpu_aliases_added)
@@ -1778,6 +1797,7 @@ int perf_pmu__for_each_event(struct perf_pmu *pmu, bool skip_duplicate_pmus,
 	struct strbuf sb;
 
 	strbuf_init(&sb, /*hint=*/ 0);
+	pmu_aliases_parse(pmu);
 	pmu_add_cpu_aliases(pmu);
 	list_for_each_entry(event, &pmu->aliases, list) {
 		size_t buf_used;
@@ -2193,6 +2213,7 @@ const char *perf_pmu__name_from_config(struct perf_pmu *pmu, u64 config)
 	if (!pmu)
 		return NULL;
 
+	pmu_aliases_parse(pmu);
 	pmu_add_cpu_aliases(pmu);
 	list_for_each_entry(event, &pmu->aliases, list) {
 		struct perf_event_attr attr = {.config = 0,};
-- 
2.44.0.769.g3c40516874-goog