linux-kernel - [PATCH 3/4] perf evlist: No need to setup affinities when enabling events for pid targets

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 17 Jan 2022 13:09:30 -0300
From:   Arnaldo Carvalho de Melo <acme@...nel.org>
To:     Andi Kleen <andi@...stfloor.org>
Cc:     linux-kernel@...r.kernel.org, linux-perf-users@...r.kernel.org,
        Arnaldo Carvalho de Melo <acme@...hat.com>,
        Adrian Hunter <adrian.hunter@...el.com>,
        Ian Rogers <irogers@...gle.com>, Jiri Olsa <jolsa@...nel.org>,
        Namhyung Kim <namhyung@...nel.org>
Subject: [PATCH 3/4] perf evlist: No need to setup affinities when enabling events for pid targets

From: Arnaldo Carvalho de Melo <acme@...hat.com>

When the target is a pid, not started by 'perf stat' we need to enable
the events, and in that case there is no need to setup affinities as we
use a dummy CPU map, with just one entry set to -1.

So stop doing it to avoid this needless call to sched_getaffinity():

  # strace -ke sched_getaffinity perf stat -e cycles -p 241957 sleep 1
  <SNIP>
  sched_getaffinity(0, 512, [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31]) = 8
   > /usr/lib64/libc-2.33.so(sched_getaffinity@@GLIBC_2.3.4+0x1a) [0xe6eea]
   > /var/home/acme/bin/perf(affinity__setup+0x6a) [0x5329ca]
   > /var/home/acme/bin/perf(__evlist__enable.constprop.0+0x23) [0x4b9693]
   > /var/home/acme/bin/perf(enable_counters+0x14d) [0x42de5d]
   > /var/home/acme/bin/perf(cmd_stat+0x2358) [0x4310c8]
   > /var/home/acme/bin/perf(run_builtin+0x6a) [0x4a2cfa]
   > /var/home/acme/bin/perf(main+0x612) [0x40f8c2]
   > /usr/lib64/libc-2.33.so(__libc_start_main+0xd4) [0x27b74]
   > /var/home/acme/bin/perf(_start+0x2d) [0x40fadd]
  <SNIP>

Cc: Adrian Hunter <adrian.hunter@...el.com>
Cc: Andi Kleen <andi@...stfloor.org>
Cc: Ian Rogers <irogers@...gle.com>
Cc: Jiri Olsa <jolsa@...nel.org>
Cc: Namhyung Kim <namhyung@...nel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@...hat.com>
---
 tools/perf/util/evlist.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)

diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
index 6e88d404b5b3e96f..ae6d4363da76ec56 100644
--- a/tools/perf/util/evlist.c
+++ b/tools/perf/util/evlist.c
@@ -487,12 +487,16 @@ static void __evlist__enable(struct evlist *evlist, char *evsel_name)
 {
 	struct evsel *pos;
 	struct evlist_cpu_iterator evlist_cpu_itr;
-	struct affinity affinity;
+	struct affinity saved_affinity, *affinity = NULL;
 
-	if (affinity__setup(&affinity) < 0)
-		return;
+	// See explanation in evlist__close()
+	if (!cpu_map__is_dummy(evlist->core.cpus)) {
+		if (affinity__setup(&saved_affinity) < 0)
+			return;
+		affinity = &saved_affinity;
+	}
 
-	evlist__for_each_cpu(evlist_cpu_itr, evlist, &affinity) {
+	evlist__for_each_cpu(evlist_cpu_itr, evlist, affinity) {
 		pos = evlist_cpu_itr.evsel;
 		if (evsel__strcmp(pos, evsel_name))
 			continue;
@@ -500,7 +504,7 @@ static void __evlist__enable(struct evlist *evlist, char *evsel_name)
 			continue;
 		evsel__enable_cpu(pos, evlist_cpu_itr.cpu_map_idx);
 	}
-	affinity__cleanup(&affinity);
+	affinity__cleanup(affinity);
 	evlist__for_each_entry(evlist, pos) {
 		if (evsel__strcmp(pos, evsel_name))
 			continue;
-- 
2.34.1