linux-kernel - Re: [PATCH 1/2] perf stat: Fix segfault when counting armv8

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]

Message-ID: <20200923201900.GQ2893484@krava>
Date:   Wed, 23 Sep 2020 22:19:00 +0200
From:   Jiri Olsa <jolsa@...hat.com>
To:     Namhyung Kim <namhyung@...nel.org>
Cc:     Wei Li <liwei391@...wei.com>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Mark Rutland <mark.rutland@....com>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Andi Kleen <ak@...ux.intel.com>,
        Andi Kleen <andi@...stfloor.org>,
        Alexey Budankov <alexey.budankov@...ux.intel.com>,
        Adrian Hunter <adrian.hunter@...el.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        linux-arm-kernel@...ts.infradead.org,
        Li Bin <huawei.libin@...wei.com>
Subject: Re: [PATCH 1/2] perf stat: Fix segfault when counting armv8_pmu
 events

On Wed, Sep 23, 2020 at 11:15:06PM +0900, Namhyung Kim wrote:
> On Wed, Sep 23, 2020 at 11:08 PM Jiri Olsa <jolsa@...hat.com> wrote:
> >
> > On Wed, Sep 23, 2020 at 10:49:52PM +0900, Namhyung Kim wrote:
> > > On Wed, Sep 23, 2020 at 2:44 PM Jiri Olsa <jolsa@...hat.com> wrote:
> > > >
> > > > On Tue, Sep 22, 2020 at 11:13:45AM +0800, Wei Li wrote:
> > > > > When executing perf stat with armv8_pmu events with a workload, it will
> > > > > report a segfault as result.
> > > >
> > > > please share the perf stat command line you see that segfault for
> > >
> > > It seems the description in the patch 0/2 already has it:
> > >
> > >   [root@...alhost hulk]# tools/perf/perf stat  -e
> > > armv8_pmuv3_0/ll_cache_rd/,armv8_pmuv3_0/ll_cache_miss_rd/ ls >
> > > /dev/null
> > >   Segmentation fault
> >
> > yea I found it, but can't reproduce it.. I see the issue from
> > patch 2, but not sure what's the problem so far
> 
> I think the problem is that armv8_pmu has a cpumask,
> and the user requested per-task events.
> 
> The code tried to open the event with a dummy cpu map
> since it's not a cpu event, but the pmu has cpu map and
> it's passed to evsel.  So there's confusion somewhere
> whether it should use evsel->cpus or a dummy map.

you're right, I have following cpus file in pmu:

  # cat /sys/devices/armv8_pmuv3_0/cpus 
  0-3

covering all the cpus.. and once you have cpumask/cpus file,
you're system wide by default in current code, but we should
not crash ;-)

I tried to cover this case in patch below and I probably broke
some other use cases, but perhaps we could allow to open counters
per cpus for given workload

I'll try to look at this more tomorrow

jirka


---
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 7f8d756d9408..0c7f16a673c2 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -379,12 +379,7 @@ static int read_affinity_counters(struct timespec *rs)
 	if (affinity__setup(&affinity) < 0)
 		return -1;
 
-	ncpus = perf_cpu_map__nr(evsel_list->core.all_cpus);
-	if (!target__has_cpu(&target) || target__has_per_thread(&target))
-		ncpus = 1;
 	evlist__for_each_cpu(evsel_list, i, cpu) {
-		if (i >= ncpus)
-			break;
 		affinity__set(&affinity, cpu);
 
 		evlist__for_each_entry(evsel_list, counter) {
diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index fd865002cbbd..ef525eb2f619 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -1861,6 +1861,16 @@ void evsel__close(struct evsel *evsel)
 	perf_evsel__free_id(&evsel->core);
 }
 
+int evsel__open_threads_per_cpu(struct evsel *evsel, struct perf_thread_map *threads,
+				struct perf_cpu_map *cpus, int cpu)
+{
+	if (cpu == -1)
+		return evsel__open_cpu(evsel, cpus, threads, 0,
+					cpus ? cpus->nr : 1);
+
+	return evsel__open_cpu(evsel, cpus, threads, cpu, cpu + 1);
+}
+
 int evsel__open_per_cpu(struct evsel *evsel, struct perf_cpu_map *cpus, int cpu)
 {
 	if (cpu == -1)
diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
index 35e3f6d66085..1d055699bd1f 100644
--- a/tools/perf/util/evsel.h
+++ b/tools/perf/util/evsel.h
@@ -231,6 +231,8 @@ int evsel__enable(struct evsel *evsel);
 int evsel__disable(struct evsel *evsel);
 int evsel__disable_cpu(struct evsel *evsel, int cpu);
 
+int evsel__open_threads_per_cpu(struct evsel *evsel, struct perf_thread_map *threads,
+				struct perf_cpu_map *cpus, int cpu);
 int evsel__open_per_cpu(struct evsel *evsel, struct perf_cpu_map *cpus, int cpu);
 int evsel__open_per_thread(struct evsel *evsel, struct perf_thread_map *threads);
 int evsel__open(struct evsel *evsel, struct perf_cpu_map *cpus,
diff --git a/tools/perf/util/stat.c b/tools/perf/util/stat.c
index cdb154381a87..2b17f1315cfb 100644
--- a/tools/perf/util/stat.c
+++ b/tools/perf/util/stat.c
@@ -560,6 +560,11 @@ int create_perf_stat_counter(struct evsel *evsel,
 			attr->enable_on_exec = 1;
 	}
 
+	if (evsel->core.own_cpus && evsel->core.threads) {
+		return evsel__open_threads_per_cpu(evsel, evsel->core.threads,
+						   evsel__cpus(evsel), cpu);
+	}
+
 	if (target__has_cpu(target) && !target__has_per_thread(target))
 		return evsel__open_per_cpu(evsel, evsel__cpus(evsel), cpu);