mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Arnaldo Carvalho de Melo <acme@kernel.org>
To: Ingo Molnar <mingo@kernel.org>
Cc: linux-kernel@vger.kernel.org, Borislav Petkov <bp@suse.de>,
	Peter Zijlstra <peterz@infradead.org>,
	Robert Richter <rric@kernel.org>, Vince Weaver <vince@deater.net>,
	Arnaldo Carvalho de Melo <acme@redhat.com>
Subject: [PATCH 02/35] perf stat: Issue a HW watchdog disable hint
Date: Mon,  6 Mar 2017 16:37:52 -0300	[thread overview]
Message-ID: <20170306193825.24011-3-acme@kernel.org> (raw)
In-Reply-To: <20170306193825.24011-1-acme@kernel.org>

From: Borislav Petkov <bp@suse.de>

When using perf stat on an AMD F15h system with the default hw events
attributes, some of the events don't get counted:

 Performance counter stats for 'sleep 1':

          0.749208      task-clock (msec)         #    0.001 CPUs utilized
                 1      context-switches          #    0.001 M/sec
                 0      cpu-migrations            #    0.000 K/sec
                54      page-faults               #    0.072 M/sec
         1,122,815      cycles                    #    1.499 GHz
           286,740      stalled-cycles-frontend   #   25.54% frontend cycles idle
     <not counted>      stalled-cycles-backend                                        (0.00%)
     ^^^^^^^^^^^^
     <not counted>      instructions                                                  (0.00%)
     ^^^^^^^^^^^^
     <not counted>      branches                                                      (0.00%)
     <not counted>      branch-misses                                                 (0.00%)

       1.001550070 seconds time elapsed

The reason is that we have the HW watchdog consuming one PMU counter and
when perf tries to schedule 6 events on 6 counters and some of those
counters are constrained to only a specific subset of PMCs by the
hardware, the event scheduling fails.

So issue a hint to disable the HW watchdog around a perf stat session.

Committer note:

Testing it...

  # perf stat -d usleep 1

   Performance counter stats for 'usleep 1':

          1.180203      task-clock (msec)         #    0.490 CPUs utilized
                 1      context-switches          #    0.847 K/sec
                 0      cpu-migrations            #    0.000 K/sec
                54      page-faults               #    0.046 M/sec
           184,754      cycles                    #    0.157 GHz
           714,553      instructions              #    3.87  insn per cycle
           154,661      branches                  #  131.046 M/sec
             7,247      branch-misses             #    4.69% of all branches
           219,984      L1-dcache-loads           #  186.395 M/sec
            17,600      L1-dcache-load-misses     #    8.00% of all L1-dcache hits    (90.16%)
     <not counted>      LLC-loads                                                     (0.00%)
     <not counted>      LLC-load-misses                                               (0.00%)

       0.002406823 seconds time elapsed

  Some events weren't counted. Try disabling the NMI watchdog:
	echo 0 > /proc/sys/kernel/nmi_watchdog
	perf stat ...
	echo 1 > /proc/sys/kernel/nmi_watchdog
  #

Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Ingo Molnar <mingo@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Robert Richter <rric@kernel.org>
Cc: Vince Weaver <vince@deater.net>
Link: http://lkml.kernel.org/r/20170211183218.ijnvb5f7ciyuunx4@pd.tnic
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/builtin-stat.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 13b54999ad79..f4f555a67e9b 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -146,6 +146,7 @@ static aggr_get_id_t		aggr_get_id;
 static bool			append_file;
 static const char		*output_name;
 static int			output_fd;
+static int			print_free_counters_hint;
 
 struct perf_stat {
 	bool			 record;
@@ -1109,6 +1110,9 @@ static void printout(int id, int nr, struct perf_evsel *counter, double uval,
 			counter->supported ? CNTR_NOT_COUNTED : CNTR_NOT_SUPPORTED,
 			csv_sep);
 
+		if (counter->supported)
+			print_free_counters_hint = 1;
+
 		fprintf(stat_config.output, "%-*s%s",
 			csv_output ? 0 : unit_width,
 			counter->unit, csv_sep);
@@ -1477,6 +1481,13 @@ static void print_footer(void)
 				avg_stats(&walltime_nsecs_stats));
 	}
 	fprintf(output, "\n\n");
+
+	if (print_free_counters_hint)
+		fprintf(output,
+"Some events weren't counted. Try disabling the NMI watchdog:\n"
+"	echo 0 > /proc/sys/kernel/nmi_watchdog\n"
+"	perf stat ...\n"
+"	echo 1 > /proc/sys/kernel/nmi_watchdog\n");
 }
 
 static void print_counters(struct timespec *ts, int argc, const char **argv)
-- 
2.9.3

  parent reply	other threads:[~2017-03-06 19:47 UTC|newest]

Thread overview: 39+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-03-06 19:37 [GIT PULL 00/35] perf/core improvements and fixes Arnaldo Carvalho de Melo
2017-03-06 19:37 ` [PATCH 01/35] perf vendor events: Add mapping for KnightsMill PMU events Arnaldo Carvalho de Melo
2017-03-06 19:37 ` Arnaldo Carvalho de Melo [this message]
2017-03-06 19:37 ` [PATCH 03/35] tools include: Adopt __compiletime_error Arnaldo Carvalho de Melo
2017-03-06 19:37 ` [PATCH 04/35] tools arch x86: Include asm/cmpxchg.h Arnaldo Carvalho de Melo
2017-03-06 19:37 ` [PATCH 05/35] tools arch x86: Introduce atomic_cmpxchg() Arnaldo Carvalho de Melo
2017-03-06 19:37 ` [PATCH 06/35] tools include: Introduce atomic_cmpxchg_{relaxed,release}() Arnaldo Carvalho de Melo
2017-03-06 19:37 ` [PATCH 07/35] tools include: Provide gcc based cmpxchg fallback for !x86 Arnaldo Carvalho de Melo
2017-03-06 19:37 ` [PATCH 08/35] tools include: Add UINT_MAX def to kernel.h Arnaldo Carvalho de Melo
2017-03-06 19:37 ` [PATCH 09/35] tools include: Adopt kernel's refcount.h Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 10/35] perf cgroup: Convert cgroup_sel.refcnt from atomic_t to refcount_t Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 11/35] perf cpumap: Convert cpu_map.refcnt " Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 12/35] perf comm: Convert comm_str.refcnt " Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 13/35] perf dso: Convert dso.refcnt " Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 14/35] perf map: Convert map.refcnt " Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 15/35] perf map: Convert map_groups.refcnt " Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 16/35] perf evlist: Convert perf_map.refcnt " Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 17/35] perf thread: convert thread.refcnt " Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 18/35] perf thread_map: Convert thread_map.refcnt " Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 19/35] perf evlist: Clarify a bit the use of perf_mmap->refcnt Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 20/35] perf tools: Allow sorting by symbol size Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 21/35] perf ftrace: Add support for --pid option Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 22/35] perf cpumap: Introduce cpu_map__snprint_mask() Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 23/35] perf ftrace: Add support for -a and -C option Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 24/35] perf ftrace: Use pager for displaying result Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 25/35] kretprobes: Ensure probe location is at function entry Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 26/35] trace/kprobes: Allow return probes with offsets and absolute addresses Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 27/35] perf probe: Generalize probe event file open routine Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 28/35] perf intel-PT/BTS: Add missing initialization Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 29/35] trace/kprobes: Add back warning about offset in return probes Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 30/35] perf tools: Force uncore events to system wide monitoring Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 31/35] tools build: Add test for sched_getcpu() Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 32/35] perf bench futex: Use __maybe_unused Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 33/35] perf bench futex: Fix build on musl + clang Arnaldo Carvalho de Melo
2017-09-08  8:04   ` Jörg Krause
2017-09-08 13:47     ` Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 34/35] tools build: Use the same CC for feature detection and actual build Arnaldo Carvalho de Melo
2017-03-06 19:38 ` [PATCH 35/35] perf bench numa: Add more comment for -c option Arnaldo Carvalho de Melo
2017-03-07  7:17 ` [GIT PULL 00/35] perf/core improvements and fixes Ingo Molnar

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=20170306193825.24011-3-acme@kernel.org \
    --to=acme@kernel.org \
    --cc=acme@redhat.com \
    --cc=bp@suse.de \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@kernel.org \
    --cc=peterz@infradead.org \
    --cc=rric@kernel.org \
    --cc=vince@deater.net \
    /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

Powered by JetHome