mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Wei Li <liwei391@huawei.com>
To: Arnaldo Carvalho de Melo <acme@kernel.org>,
	Mark Rutland <mark.rutland@arm.com>,
	Alexander Shishkin <alexander.shishkin@linux.intel.com>,
	Jiri Olsa <jolsa@redhat.com>,
	"Namhyung Kim" <namhyung@kernel.org>,
	Andi Kleen <ak@linux.intel.com>,
	Alexey Budankov <alexey.budankov@linux.intel.com>,
	Adrian Hunter <adrian.hunter@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@redhat.com>, <linux-kernel@vger.kernel.org>,
	<linux-arm-kernel@lists.infradead.org>, <huawei.libin@huawei.com>
Subject: [PATCH 1/2] perf stat: Fix segfault when counting armv8_pmu events
Date: Tue, 22 Sep 2020 11:13:45 +0800	[thread overview]
Message-ID: <20200922031346.15051-2-liwei391@huawei.com> (raw)
In-Reply-To: <20200922031346.15051-1-liwei391@huawei.com>

When executing perf stat with armv8_pmu events with a workload, it will
report a segfault as result.

(gdb) bt
#0  0x0000000000603fc8 in perf_evsel__close_fd_cpu (evsel=<optimized out>,
    cpu=<optimized out>) at evsel.c:122
#1  perf_evsel__close_cpu (evsel=evsel@entry=0x716e950, cpu=7) at evsel.c:156
#2  0x00000000004d4718 in evlist__close (evlist=0x70a7cb0) at util/evlist.c:1242
#3  0x0000000000453404 in __run_perf_stat (argc=3, argc@entry=1, argv=0x30,
    argv@entry=0xfffffaea2f90, run_idx=119, run_idx@entry=1701998435)
    at builtin-stat.c:929
#4  0x0000000000455058 in run_perf_stat (run_idx=1701998435, argv=0xfffffaea2f90,
    argc=1) at builtin-stat.c:947
#5  cmd_stat (argc=1, argv=0xfffffaea2f90) at builtin-stat.c:2357
#6  0x00000000004bb888 in run_builtin (p=p@entry=0x9764b8 <commands+288>,
    argc=argc@entry=4, argv=argv@entry=0xfffffaea2f90) at perf.c:312
#7  0x00000000004bbb54 in handle_internal_command (argc=argc@entry=4,
    argv=argv@entry=0xfffffaea2f90) at perf.c:364
#8  0x0000000000435378 in run_argv (argcp=<synthetic pointer>,
    argv=<synthetic pointer>) at perf.c:408
#9  main (argc=4, argv=0xfffffaea2f90) at perf.c:538

After debugging, i found the root reason is that the xyarray fd is created
by evsel__open_per_thread() ignoring the cpu passed in
create_perf_stat_counter(), while the evsel' cpumap is assigned as the
corresponding PMU's cpumap in __add_event(). Thus, the xyarray fd is created
with ncpus of dummy cpumap and an out of bounds 'cpu' index will be used in
perf_evsel__close_fd_cpu().

To address this, add a flag to mark this situation and avoid using the
affinity technique when closing/enabling/disabling events.

Fixes: 7736627b865d ("perf stat: Use affinity for closing file descriptors")
Fixes: 704e2f5b700d ("perf stat: Use affinity for enabling/disabling events")
Signed-off-by: Wei Li <liwei391@huawei.com>
---
 tools/lib/perf/include/internal/evlist.h |  1 +
 tools/perf/builtin-stat.c                |  3 +++
 tools/perf/util/evlist.c                 | 23 ++++++++++++++++++++++-
 3 files changed, 26 insertions(+), 1 deletion(-)

diff --git a/tools/lib/perf/include/internal/evlist.h b/tools/lib/perf/include/internal/evlist.h
index 2d0fa02b036f..c02d7e583846 100644
--- a/tools/lib/perf/include/internal/evlist.h
+++ b/tools/lib/perf/include/internal/evlist.h
@@ -17,6 +17,7 @@ struct perf_evlist {
 	struct list_head	 entries;
 	int			 nr_entries;
 	bool			 has_user_cpus;
+	bool			 open_per_thread;
 	struct perf_cpu_map	*cpus;
 	struct perf_cpu_map	*all_cpus;
 	struct perf_thread_map	*threads;
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index fddc97cac984..6e6ceacce634 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -725,6 +725,9 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)
 	if (group)
 		perf_evlist__set_leader(evsel_list);
 
+	if (!(target__has_cpu(&target) && !target__has_per_thread(&target)))
+		evsel_list->core.open_per_thread = true;
+
 	if (affinity__setup(&affinity) < 0)
 		return -1;
 
diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
index e3fa3bf7498a..bf8a3ccc599f 100644
--- a/tools/perf/util/evlist.c
+++ b/tools/perf/util/evlist.c
@@ -383,6 +383,15 @@ void evlist__disable(struct evlist *evlist)
 	int cpu, i, imm = 0;
 	bool has_imm = false;
 
+	if (evlist->core.open_per_thread) {
+		evlist__for_each_entry(evlist, pos) {
+			if (pos->disabled || !evsel__is_group_leader(pos) || !pos->core.fd)
+				continue;
+			evsel__disable(pos);
+		}
+		goto out;
+	}
+
 	if (affinity__setup(&affinity) < 0)
 		return;
 
@@ -414,6 +423,7 @@ void evlist__disable(struct evlist *evlist)
 		pos->disabled = true;
 	}
 
+out:
 	evlist->enabled = false;
 }
 
@@ -423,6 +433,15 @@ void evlist__enable(struct evlist *evlist)
 	struct affinity affinity;
 	int cpu, i;
 
+	if (evlist->core.open_per_thread) {
+		evlist__for_each_entry(evlist, pos) {
+			if (!evsel__is_group_leader(pos) || !pos->core.fd)
+				continue;
+			evsel__enable(pos);
+		}
+		goto out;
+	}
+
 	if (affinity__setup(&affinity) < 0)
 		return;
 
@@ -444,6 +463,7 @@ void evlist__enable(struct evlist *evlist)
 		pos->disabled = false;
 	}
 
+out:
 	evlist->enabled = true;
 }
 
@@ -1223,9 +1243,10 @@ void evlist__close(struct evlist *evlist)
 
 	/*
 	 * With perf record core.cpus is usually NULL.
+	 * Or perf stat may open events per-thread.
 	 * Use the old method to handle this for now.
 	 */
-	if (!evlist->core.cpus) {
+	if (evlist->core.open_per_thread || !evlist->core.cpus) {
 		evlist__for_each_entry_reverse(evlist, evsel)
 			evsel__close(evsel);
 		return;
-- 
2.17.1


  reply	other threads:[~2020-09-22  3:14 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-09-22  3:13 [PATCH 0/2] perf stat: Unbreak perf stat with ARMv8 PMU events Wei Li
2020-09-22  3:13 ` Wei Li [this message]
2020-09-22 19:23   ` [PATCH 1/2] perf stat: Fix segfault when counting armv8_pmu events Andi Kleen
2020-09-22 19:50     ` Andi Kleen
2020-09-24 14:14       ` liwei (GF)
2020-09-23  5:44   ` Jiri Olsa
2020-09-23 13:49     ` Namhyung Kim
2020-09-23 14:07       ` Jiri Olsa
2020-09-23 14:15         ` Namhyung Kim
2020-09-23 20:19           ` Jiri Olsa
2020-09-24 14:36             ` Namhyung Kim
2020-09-25 21:01               ` Jiri Olsa
2020-10-02  8:59               ` Jiri Olsa
2020-10-06  6:51                 ` Song Bao Hua (Barry Song)
2020-09-22  3:13 ` [PATCH 2/2] perf stat: Unbreak perf stat with " Wei Li

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=20200922031346.15051-2-liwei391@huawei.com \
    --to=liwei391@huawei.com \
    --cc=acme@kernel.org \
    --cc=adrian.hunter@intel.com \
    --cc=ak@linux.intel.com \
    --cc=alexander.shishkin@linux.intel.com \
    --cc=alexey.budankov@linux.intel.com \
    --cc=huawei.libin@huawei.com \
    --cc=jolsa@redhat.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mark.rutland@arm.com \
    --cc=mingo@redhat.com \
    --cc=namhyung@kernel.org \
    --cc=peterz@infradead.org \
    /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®