Arnaldo,
I think the -n option of perf report shows bogus number of samples.
I believe it does not print the number of samples but rather the
number of events if I understand the code in hist_entry__snprintf().
I think that's useless, the number of samples is more useful.
$ perf report -h
usage: perf report [<options>] <command>
-i, --input <file> input file name
-v, --verbose be more verbose (show symbol address, etc)
-D, --dump-raw-trace dump raw trace in ASCII
-k, --vmlinux <file> vmlinux pathname
--kallsyms <file>
kallsyms pathname
-f, --force don't complain, do it
-m, --modules load module symbols - WARNING: use only with
-k and LIVE kernel
-n, --show-nr-samples
Show a column with the number of samples
$ perf record -e cycles ./repmov
$ perf report -D | fgrep RECORD_SAMPLE | wc -l
86346
$ ./perf report -n
# Events: 86K cycles
#
# Overhead Samples Command Shared Object Symbol
# ........ .......... ....... ................. .........................
#
98.92%238206388334 repmov repmov [.] main
0.08% 189506224 repmov [kernel.kallsyms] [k] perf_ctx_adjust_freq
0.06% 147582706 repmov [kernel.kallsyms] [k] perf_event_task_tick
It should be easy to reproduce with any other program.
Em Wed, Feb 09, 2011 at 12:34:44AM +0100, Stephane Eranian escreveu:
> I think the -n option of perf report shows bogus number of samples.
>
> I believe it does not print the number of samples but rather the
> number of events if I understand the code in hist_entry__snprintf().
> I think that's useless, the number of samples is more useful.
<SNIP
> $ perf record -e cycles ./repmov
> $ perf report -D | fgrep RECORD_SAMPLE | wc -l
> 86346
> $ ./perf report -n
> # Events: 86K cycles
> #
> # Overhead Samples Command Shared Object Symbol
> # ........ .......... ....... ................. .........................
> #
> 98.92%238206388334 repmov repmov [.] main
> 0.08% 189506224 repmov [kernel.kallsyms] [k] perf_ctx_adjust_freq
> 0.06% 147582706 repmov [kernel.kallsyms] [k] perf_event_task_tick
>
> It should be easy to reproduce with any other program.
Like this? Can I have your acked-by?
commit 58e57f4dd6164bdbd5bc1f7d7c1ba5eb104e9b32
Author: Arnaldo Carvalho de Melo <[email protected]>
Date: Thu Feb 17 10:37:23 2011 -0200
perf hists: Print number of samples, not the period sum
So that we match the header where we state the number of events with the
"Samples" column when using 'perf report -n/--show-nr-samples':
[root@emilia ~]# perf record -a sleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.111 MB perf.data (~4860 samples) ]
[root@emilia ~]# perf report --stdio --show-nr-samples
# Events: 11 cycles
#
# Overhead Samples Command Shared Object Symbol
# ........ .......... ........... .................. ............................
#
16.65% 1 sleep [kernel.kallsyms] [k] unmap_vmas
16.10% 1 perf libpthread-2.12.so [.] __pthread_cleanup_push_defer
15.79% 2 perf [kernel.kallsyms] [k] format_decode
12.88% 1 kworker/1:2 [kernel.kallsyms] [k] cache_reap
10.69% 1 swapper [kernel.kallsyms] [k] _raw_spin_lock
7.55% 1 sleep [kernel.kallsyms] [k] prepare_exec_creds
6.00% 1 perf [jbd2] [k] start_this_handle
5.29% 1 perf [kernel.kallsyms] [k] seq_read
4.75% 1 perf [kernel.kallsyms] [k] get_pid_task
4.30% 1 perf [kernel.kallsyms] [k] _raw_spin_unlock_irqrestore
#
# (For a higher level overview, try: perf report --sort comm,dso)
#
[root@emilia ~]#
Reported-by: Stephane Eranian <[email protected]>
Cc: Frederic Weisbecker <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Mike Galbraith <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Stephane Eranian <[email protected]>
Cc: Tom Zanussi <[email protected]>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
index 3f43723..da2899e 100644
--- a/tools/perf/util/hist.c
+++ b/tools/perf/util/hist.c
@@ -591,6 +591,7 @@ int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size,
{
struct sort_entry *se;
u64 period, total, period_sys, period_us, period_guest_sys, period_guest_us;
+ u64 nr_events;
const char *sep = symbol_conf.field_sep;
int ret;
@@ -599,6 +600,7 @@ int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size,
if (pair_hists) {
period = self->pair ? self->pair->period : 0;
+ nr_events = self->pair ? self->pair->nr_events : 0;
total = pair_hists->stats.total_period;
period_sys = self->pair ? self->pair->period_sys : 0;
period_us = self->pair ? self->pair->period_us : 0;
@@ -606,6 +608,7 @@ int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size,
period_guest_us = self->pair ? self->pair->period_guest_us : 0;
} else {
period = self->period;
+ nr_events = self->nr_events;
total = session_total;
period_sys = self->period_sys;
period_us = self->period_us;
@@ -646,9 +649,9 @@ int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size,
if (symbol_conf.show_nr_samples) {
if (sep)
- ret += snprintf(s + ret, size - ret, "%c%" PRIu64, *sep, period);
+ ret += snprintf(s + ret, size - ret, "%c%" PRIu64, *sep, nr_events);
else
- ret += snprintf(s + ret, size - ret, "%11" PRIu64, period);
+ ret += snprintf(s + ret, size - ret, "%11" PRIu64, nr_events);
}
if (pair_hists) {
On Thu, Feb 17, 2011 at 4:23 PM, Arnaldo Carvalho de Melo
<[email protected]> wrote:
> Em Wed, Feb 09, 2011 at 12:34:44AM +0100, Stephane Eranian escreveu:
>> I think the -n option of perf report shows bogus number of samples.
>>
>> I believe it does not print the number of samples but rather the
>> number of events if I understand the code in hist_entry__snprintf().
>> I think that's useless, the number of samples is more useful.
>
> <SNIP
>
>> $ perf record -e cycles ./repmov
>> $ perf report -D | fgrep RECORD_SAMPLE | wc -l
>> 86346
>> $ ./perf report -n
>> # Events: 86K cycles
>> #
>> # Overhead Samples Command Shared Object Symbol
>> # ........ .......... ....... ................. .........................
>> #
>> 98.92%238206388334 repmov repmov [.] main
>> 0.08% 189506224 repmov [kernel.kallsyms] [k] perf_ctx_adjust_freq
>> 0.06% 147582706 repmov [kernel.kallsyms] [k] perf_event_task_tick
>>
>> It should be easy to reproduce with any other program.
>
> Like this? Can I have your acked-by?
>
Yes. Works for me.
Acked-by: Stephane Eranian <[email protected]>
> commit 58e57f4dd6164bdbd5bc1f7d7c1ba5eb104e9b32
> Author: Arnaldo Carvalho de Melo <[email protected]>
> Date: Thu Feb 17 10:37:23 2011 -0200
>
> perf hists: Print number of samples, not the period sum
>
> So that we match the header where we state the number of events with the
> "Samples" column when using 'perf report -n/--show-nr-samples':
>
> [root@emilia ~]# perf record -a sleep 1
> [ perf record: Woken up 1 times to write data ]
> [ perf record: Captured and wrote 0.111 MB perf.data (~4860 samples) ]
> [root@emilia ~]# perf report --stdio --show-nr-samples
> # Events: 11 cycles
> #
> # Overhead Samples Command Shared Object Symbol
> # ........ .......... ........... .................. ............................
> #
> 16.65% 1 sleep [kernel.kallsyms] [k] unmap_vmas
> 16.10% 1 perf libpthread-2.12.so [.] __pthread_cleanup_push_defer
> 15.79% 2 perf [kernel.kallsyms] [k] format_decode
> 12.88% 1 kworker/1:2 [kernel.kallsyms] [k] cache_reap
> 10.69% 1 swapper [kernel.kallsyms] [k] _raw_spin_lock
> 7.55% 1 sleep [kernel.kallsyms] [k] prepare_exec_creds
> 6.00% 1 perf [jbd2] [k] start_this_handle
> 5.29% 1 perf [kernel.kallsyms] [k] seq_read
> 4.75% 1 perf [kernel.kallsyms] [k] get_pid_task
> 4.30% 1 perf [kernel.kallsyms] [k] _raw_spin_unlock_irqrestore
>
> #
> # (For a higher level overview, try: perf report --sort comm,dso)
> #
> [root@emilia ~]#
>
> Reported-by: Stephane Eranian <[email protected]>
> Cc: Frederic Weisbecker <[email protected]>
> Cc: Ingo Molnar <[email protected]>
> Cc: Mike Galbraith <[email protected]>
> Cc: Paul Mackerras <[email protected]>
> Cc: Peter Zijlstra <[email protected]>
> Cc: Stephane Eranian <[email protected]>
> Cc: Tom Zanussi <[email protected]>
> LKML-Reference: <new-submission>
> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
>
> diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
> index 3f43723..da2899e 100644
> --- a/tools/perf/util/hist.c
> +++ b/tools/perf/util/hist.c
> @@ -591,6 +591,7 @@ int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size,
> {
> struct sort_entry *se;
> u64 period, total, period_sys, period_us, period_guest_sys, period_guest_us;
> + u64 nr_events;
> const char *sep = symbol_conf.field_sep;
> int ret;
>
> @@ -599,6 +600,7 @@ int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size,
>
> if (pair_hists) {
> period = self->pair ? self->pair->period : 0;
> + nr_events = self->pair ? self->pair->nr_events : 0;
> total = pair_hists->stats.total_period;
> period_sys = self->pair ? self->pair->period_sys : 0;
> period_us = self->pair ? self->pair->period_us : 0;
> @@ -606,6 +608,7 @@ int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size,
> period_guest_us = self->pair ? self->pair->period_guest_us : 0;
> } else {
> period = self->period;
> + nr_events = self->nr_events;
> total = session_total;
> period_sys = self->period_sys;
> period_us = self->period_us;
> @@ -646,9 +649,9 @@ int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size,
>
> if (symbol_conf.show_nr_samples) {
> if (sep)
> - ret += snprintf(s + ret, size - ret, "%c%" PRIu64, *sep, period);
> + ret += snprintf(s + ret, size - ret, "%c%" PRIu64, *sep, nr_events);
> else
> - ret += snprintf(s + ret, size - ret, "%11" PRIu64, period);
> + ret += snprintf(s + ret, size - ret, "%11" PRIu64, nr_events);
> }
>
> if (pair_hists) {
>