* [PATCH v2 1/6] perf top: Use evlist->events_stat to count events
2021-04-27 1:37 [PATCHSET v2 0/6] perf report: Make --stat output more compact Namhyung Kim
@ 2021-04-27 1:37 ` Namhyung Kim
2021-04-27 1:37 ` [PATCH v2 2/6] perf hists: Split hists_stats from events_stats Namhyung Kim
` (6 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Namhyung Kim @ 2021-04-27 1:37 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Jiri Olsa
Cc: Ingo Molnar, Peter Zijlstra, Mark Rutland, Alexander Shishkin,
LKML, Andi Kleen, Ian Rogers
It's mainly to count lost events for the warning so it should be ok
to use the evlist->stats instead. This is needed for changes in the
next commit.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
tools/perf/builtin-top.c | 20 ++++++++------------
1 file changed, 8 insertions(+), 12 deletions(-)
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 173ace43f845..69cb3635f5ef 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -328,13 +328,13 @@ static void perf_top__print_sym_table(struct perf_top *top)
printf("%-*.*s\n", win_width, win_width, graph_dotted_line);
if (!top->record_opts.overwrite &&
- (hists->stats.nr_lost_warned !=
- hists->stats.nr_events[PERF_RECORD_LOST])) {
- hists->stats.nr_lost_warned =
- hists->stats.nr_events[PERF_RECORD_LOST];
+ (top->evlist->stats.nr_lost_warned !=
+ top->evlist->stats.nr_events[PERF_RECORD_LOST])) {
+ top->evlist->stats.nr_lost_warned =
+ top->evlist->stats.nr_events[PERF_RECORD_LOST];
color_fprintf(stdout, PERF_COLOR_RED,
"WARNING: LOST %d chunks, Check IO/CPU overload",
- hists->stats.nr_lost_warned);
+ top->evlist->stats.nr_lost_warned);
++printed;
}
@@ -852,11 +852,9 @@ static void
perf_top__process_lost(struct perf_top *top, union perf_event *event,
struct evsel *evsel)
{
- struct hists *hists = evsel__hists(evsel);
-
top->lost += event->lost.lost;
top->lost_total += event->lost.lost;
- hists->stats.total_lost += event->lost.lost;
+ evsel->evlist->stats.total_lost += event->lost.lost;
}
static void
@@ -864,11 +862,9 @@ perf_top__process_lost_samples(struct perf_top *top,
union perf_event *event,
struct evsel *evsel)
{
- struct hists *hists = evsel__hists(evsel);
-
top->lost += event->lost_samples.lost;
top->lost_total += event->lost_samples.lost;
- hists->stats.total_lost_samples += event->lost_samples.lost;
+ evsel->evlist->stats.total_lost_samples += event->lost_samples.lost;
}
static u64 last_timestamp;
@@ -1205,7 +1201,7 @@ static int deliver_event(struct ordered_events *qe,
} else if (event->header.type == PERF_RECORD_LOST_SAMPLES) {
perf_top__process_lost_samples(top, event, evsel);
} else if (event->header.type < PERF_RECORD_MAX) {
- hists__inc_nr_events(evsel__hists(evsel), event->header.type);
+ events_stats__inc(&session->evlist->stats, event->header.type);
machine__process_event(machine, event, &sample);
} else
++session->evlist->stats.nr_unknown_events;
--
2.31.1.527.g47e6f16901-goog
^ permalink raw reply related [flat|nested] 9+ messages in thread* [PATCH v2 2/6] perf hists: Split hists_stats from events_stats
2021-04-27 1:37 [PATCHSET v2 0/6] perf report: Make --stat output more compact Namhyung Kim
2021-04-27 1:37 ` [PATCH v2 1/6] perf top: Use evlist->events_stat to count events Namhyung Kim
@ 2021-04-27 1:37 ` Namhyung Kim
2021-04-27 1:37 ` [PATCH v2 3/6] perf report: Show event sample counts in --stat output Namhyung Kim
` (5 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Namhyung Kim @ 2021-04-27 1:37 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Jiri Olsa
Cc: Ingo Molnar, Peter Zijlstra, Mark Rutland, Alexander Shishkin,
LKML, Andi Kleen, Ian Rogers
Each struct hists have events_stats but most of the fields were not
used. It's to count number of samples and periods whether filtered or
not. And other fields are used only by evlist.
So it'd be better to split hists_stats and events_stats to reduce
wasted memory in the struct hists. This makes the output of event
statistics in the perf report compact by skipping 0 events in each
evsel/hists.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
tools/perf/builtin-annotate.c | 2 +-
tools/perf/builtin-report.c | 4 ++--
tools/perf/tests/hists_filter.c | 14 +++++++-------
tools/perf/ui/browsers/hists.c | 17 +++++++++--------
tools/perf/util/events_stats.h | 10 +++++++---
tools/perf/util/hist.c | 20 ++++++++++++++------
tools/perf/util/hist.h | 4 ++--
7 files changed, 42 insertions(+), 29 deletions(-)
diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c
index 524e6f0dff22..717efd78eee6 100644
--- a/tools/perf/builtin-annotate.c
+++ b/tools/perf/builtin-annotate.c
@@ -418,7 +418,7 @@ static int __cmd_annotate(struct perf_annotate *ann)
total_nr_samples = 0;
evlist__for_each_entry(session->evlist, pos) {
struct hists *hists = evsel__hists(pos);
- u32 nr_samples = hists->stats.nr_events[PERF_RECORD_SAMPLE];
+ u32 nr_samples = hists->stats.nr_samples;
if (nr_samples > 0) {
total_nr_samples += nr_samples;
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index 0d65c98794a8..b0b9b60f74e5 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -436,7 +436,7 @@ static size_t hists__fprintf_nr_sample_events(struct hists *hists, struct report
{
size_t ret;
char unit;
- unsigned long nr_samples = hists->stats.nr_events[PERF_RECORD_SAMPLE];
+ unsigned long nr_samples = hists->stats.nr_samples;
u64 nr_events = hists->stats.total_period;
struct evsel *evsel = hists_to_evsel(hists);
char buf[512];
@@ -464,7 +464,7 @@ static size_t hists__fprintf_nr_sample_events(struct hists *hists, struct report
nr_samples += pos_hists->stats.nr_non_filtered_samples;
nr_events += pos_hists->stats.total_non_filtered_period;
} else {
- nr_samples += pos_hists->stats.nr_events[PERF_RECORD_SAMPLE];
+ nr_samples += pos_hists->stats.nr_samples;
nr_events += pos_hists->stats.total_period;
}
}
diff --git a/tools/perf/tests/hists_filter.c b/tools/perf/tests/hists_filter.c
index 123e07d35b55..ca6120cd1d90 100644
--- a/tools/perf/tests/hists_filter.c
+++ b/tools/perf/tests/hists_filter.c
@@ -150,13 +150,13 @@ int test__hists_filter(struct test *test __maybe_unused, int subtest __maybe_unu
}
TEST_ASSERT_VAL("Invalid nr samples",
- hists->stats.nr_events[PERF_RECORD_SAMPLE] == 10);
+ hists->stats.nr_samples == 10);
TEST_ASSERT_VAL("Invalid nr hist entries",
hists->nr_entries == 9);
TEST_ASSERT_VAL("Invalid total period",
hists->stats.total_period == 1000);
TEST_ASSERT_VAL("Unmatched nr samples",
- hists->stats.nr_events[PERF_RECORD_SAMPLE] ==
+ hists->stats.nr_samples ==
hists->stats.nr_non_filtered_samples);
TEST_ASSERT_VAL("Unmatched nr hist entries",
hists->nr_entries == hists->nr_non_filtered_entries);
@@ -175,7 +175,7 @@ int test__hists_filter(struct test *test __maybe_unused, int subtest __maybe_unu
/* normal stats should be invariant */
TEST_ASSERT_VAL("Invalid nr samples",
- hists->stats.nr_events[PERF_RECORD_SAMPLE] == 10);
+ hists->stats.nr_samples == 10);
TEST_ASSERT_VAL("Invalid nr hist entries",
hists->nr_entries == 9);
TEST_ASSERT_VAL("Invalid total period",
@@ -204,7 +204,7 @@ int test__hists_filter(struct test *test __maybe_unused, int subtest __maybe_unu
/* normal stats should be invariant */
TEST_ASSERT_VAL("Invalid nr samples",
- hists->stats.nr_events[PERF_RECORD_SAMPLE] == 10);
+ hists->stats.nr_samples == 10);
TEST_ASSERT_VAL("Invalid nr hist entries",
hists->nr_entries == 9);
TEST_ASSERT_VAL("Invalid total period",
@@ -239,7 +239,7 @@ int test__hists_filter(struct test *test __maybe_unused, int subtest __maybe_unu
/* normal stats should be invariant */
TEST_ASSERT_VAL("Invalid nr samples",
- hists->stats.nr_events[PERF_RECORD_SAMPLE] == 10);
+ hists->stats.nr_samples == 10);
TEST_ASSERT_VAL("Invalid nr hist entries",
hists->nr_entries == 9);
TEST_ASSERT_VAL("Invalid total period",
@@ -268,7 +268,7 @@ int test__hists_filter(struct test *test __maybe_unused, int subtest __maybe_unu
/* normal stats should be invariant */
TEST_ASSERT_VAL("Invalid nr samples",
- hists->stats.nr_events[PERF_RECORD_SAMPLE] == 10);
+ hists->stats.nr_samples == 10);
TEST_ASSERT_VAL("Invalid nr hist entries",
hists->nr_entries == 9);
TEST_ASSERT_VAL("Invalid total period",
@@ -299,7 +299,7 @@ int test__hists_filter(struct test *test __maybe_unused, int subtest __maybe_unu
/* normal stats should be invariant */
TEST_ASSERT_VAL("Invalid nr samples",
- hists->stats.nr_events[PERF_RECORD_SAMPLE] == 10);
+ hists->stats.nr_samples == 10);
TEST_ASSERT_VAL("Invalid nr hist entries",
hists->nr_entries == 9);
TEST_ASSERT_VAL("Invalid total period",
diff --git a/tools/perf/ui/browsers/hists.c b/tools/perf/ui/browsers/hists.c
index bcfd0a45953b..b72ee6822222 100644
--- a/tools/perf/ui/browsers/hists.c
+++ b/tools/perf/ui/browsers/hists.c
@@ -682,6 +682,7 @@ static int hist_browser__handle_hotkey(struct hist_browser *browser, bool warn_l
switch (key) {
case K_TIMER: {
struct hist_browser_timer *hbt = browser->hbt;
+ struct evsel *evsel = hists_to_evsel(browser->hists);
u64 nr_entries;
WARN_ON_ONCE(!hbt);
@@ -696,10 +697,10 @@ static int hist_browser__handle_hotkey(struct hist_browser *browser, bool warn_l
ui_browser__update_nr_entries(&browser->b, nr_entries);
if (warn_lost_event &&
- (browser->hists->stats.nr_lost_warned !=
- browser->hists->stats.nr_events[PERF_RECORD_LOST])) {
- browser->hists->stats.nr_lost_warned =
- browser->hists->stats.nr_events[PERF_RECORD_LOST];
+ (evsel->evlist->stats.nr_lost_warned !=
+ evsel->evlist->stats.nr_events[PERF_RECORD_LOST])) {
+ evsel->evlist->stats.nr_lost_warned =
+ evsel->evlist->stats.nr_events[PERF_RECORD_LOST];
ui_browser__warn_lost_events(&browser->b);
}
@@ -3416,7 +3417,7 @@ static void perf_evsel_menu__write(struct ui_browser *browser,
struct evsel *evsel = list_entry(entry, struct evsel, core.node);
struct hists *hists = evsel__hists(evsel);
bool current_entry = ui_browser__is_current_entry(browser, row);
- unsigned long nr_events = hists->stats.nr_events[PERF_RECORD_SAMPLE];
+ unsigned long nr_events = hists->stats.nr_samples;
const char *ev_name = evsel__name(evsel);
char bf[256], unit;
const char *warn = " ";
@@ -3432,7 +3433,7 @@ static void perf_evsel_menu__write(struct ui_browser *browser,
for_each_group_member(pos, evsel) {
struct hists *pos_hists = evsel__hists(pos);
- nr_events += pos_hists->stats.nr_events[PERF_RECORD_SAMPLE];
+ nr_events += pos_hists->stats.nr_samples;
}
}
@@ -3441,7 +3442,7 @@ static void perf_evsel_menu__write(struct ui_browser *browser,
unit, unit == ' ' ? "" : " ", ev_name);
ui_browser__printf(browser, "%s", bf);
- nr_events = hists->stats.nr_events[PERF_RECORD_LOST];
+ nr_events = evsel->evlist->stats.nr_events[PERF_RECORD_LOST];
if (nr_events != 0) {
menu->lost_events = true;
if (!current_entry)
@@ -3647,7 +3648,7 @@ static int block_hists_browser__title(struct hist_browser *browser, char *bf,
{
struct hists *hists = evsel__hists(browser->block_evsel);
const char *evname = evsel__name(browser->block_evsel);
- unsigned long nr_samples = hists->stats.nr_events[PERF_RECORD_SAMPLE];
+ unsigned long nr_samples = hists->stats.nr_samples;
int ret;
ret = scnprintf(bf, size, "# Samples: %lu", nr_samples);
diff --git a/tools/perf/util/events_stats.h b/tools/perf/util/events_stats.h
index 631a4af2ed86..e271c8004c89 100644
--- a/tools/perf/util/events_stats.h
+++ b/tools/perf/util/events_stats.h
@@ -26,15 +26,12 @@
* perf_record_sample.period and stash the result in total_period.
*/
struct events_stats {
- u64 total_period;
- u64 total_non_filtered_period;
u64 total_lost;
u64 total_lost_samples;
u64 total_aux_lost;
u64 total_aux_partial;
u64 total_invalid_chains;
u32 nr_events[PERF_RECORD_HEADER_MAX];
- u32 nr_non_filtered_samples;
u32 nr_lost_warned;
u32 nr_unknown_events;
u32 nr_invalid_chains;
@@ -44,6 +41,13 @@ struct events_stats {
u32 nr_proc_map_timeout;
};
+struct hists_stats {
+ u64 total_period;
+ u64 total_non_filtered_period;
+ u32 nr_samples;
+ u32 nr_non_filtered_samples;
+};
+
void events_stats__inc(struct events_stats *stats, u32 type);
size_t events_stats__fprintf(struct events_stats *stats, FILE *fp);
diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
index 9299ee535518..691a6a777d14 100644
--- a/tools/perf/util/hist.c
+++ b/tools/perf/util/hist.c
@@ -2325,14 +2325,19 @@ void events_stats__inc(struct events_stats *stats, u32 type)
++stats->nr_events[type];
}
-void hists__inc_nr_events(struct hists *hists, u32 type)
+static void hists_stats__inc(struct hists_stats *stats)
{
- events_stats__inc(&hists->stats, type);
+ ++stats->nr_samples;
+}
+
+void hists__inc_nr_events(struct hists *hists)
+{
+ hists_stats__inc(&hists->stats);
}
void hists__inc_nr_samples(struct hists *hists, bool filtered)
{
- events_stats__inc(&hists->stats, PERF_RECORD_SAMPLE);
+ hists_stats__inc(&hists->stats);
if (!filtered)
hists->stats.nr_non_filtered_samples++;
}
@@ -2677,8 +2682,11 @@ size_t evlist__fprintf_nr_events(struct evlist *evlist, FILE *fp)
size_t ret = 0;
evlist__for_each_entry(evlist, pos) {
+ struct hists *hists = evsel__hists(pos);
+
ret += fprintf(fp, "%s stats:\n", evsel__name(pos));
- ret += events_stats__fprintf(&evsel__hists(pos)->stats, fp);
+ ret += fprintf(fp, "%16s events: %10d\n",
+ "SAMPLE", hists->stats.nr_samples);
}
return ret;
@@ -2698,7 +2706,7 @@ int __hists__scnprintf_title(struct hists *hists, char *bf, size_t size, bool sh
const struct dso *dso = hists->dso_filter;
struct thread *thread = hists->thread_filter;
int socket_id = hists->socket_filter;
- unsigned long nr_samples = hists->stats.nr_events[PERF_RECORD_SAMPLE];
+ unsigned long nr_samples = hists->stats.nr_samples;
u64 nr_events = hists->stats.total_period;
struct evsel *evsel = hists_to_evsel(hists);
const char *ev_name = evsel__name(evsel);
@@ -2725,7 +2733,7 @@ int __hists__scnprintf_title(struct hists *hists, char *bf, size_t size, bool sh
nr_samples += pos_hists->stats.nr_non_filtered_samples;
nr_events += pos_hists->stats.total_non_filtered_period;
} else {
- nr_samples += pos_hists->stats.nr_events[PERF_RECORD_SAMPLE];
+ nr_samples += pos_hists->stats.nr_samples;
nr_events += pos_hists->stats.total_period;
}
}
diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h
index e2faa745c8d6..6b0f708f08ac 100644
--- a/tools/perf/util/hist.h
+++ b/tools/perf/util/hist.h
@@ -96,7 +96,7 @@ struct hists {
const char *uid_filter_str;
const char *symbol_filter_str;
pthread_mutex_t lock;
- struct events_stats stats;
+ struct hists_stats stats;
u64 event_stream;
u16 col_len[HISTC_NR_COLS];
bool has_callchains;
@@ -196,7 +196,7 @@ struct hist_entry *hists__get_entry(struct hists *hists, int idx);
u64 hists__total_period(struct hists *hists);
void hists__reset_stats(struct hists *hists);
void hists__inc_stats(struct hists *hists, struct hist_entry *h);
-void hists__inc_nr_events(struct hists *hists, u32 type);
+void hists__inc_nr_events(struct hists *hists);
void hists__inc_nr_samples(struct hists *hists, bool filtered);
size_t hists__fprintf(struct hists *hists, bool show_header, int max_rows,
--
2.31.1.527.g47e6f16901-goog
^ permalink raw reply related [flat|nested] 9+ messages in thread* [PATCH v2 3/6] perf report: Show event sample counts in --stat output
2021-04-27 1:37 [PATCHSET v2 0/6] perf report: Make --stat output more compact Namhyung Kim
2021-04-27 1:37 ` [PATCH v2 1/6] perf top: Use evlist->events_stat to count events Namhyung Kim
2021-04-27 1:37 ` [PATCH v2 2/6] perf hists: Split hists_stats from events_stats Namhyung Kim
@ 2021-04-27 1:37 ` Namhyung Kim
2021-04-27 1:37 ` [PATCH v2 4/6] perf report: Add --skip-empty option to suppress 0 event stat Namhyung Kim
` (4 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Namhyung Kim @ 2021-04-27 1:37 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Jiri Olsa
Cc: Ingo Molnar, Peter Zijlstra, Mark Rutland, Alexander Shishkin,
LKML, Andi Kleen, Ian Rogers
To make the output identical with perf report -D, it needs to show
per-event sample counts along with the aggregated stat at the end.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
tools/perf/builtin-report.c | 14 ++++++++++++++
1 file changed, 14 insertions(+)
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index b0b9b60f74e5..be56f3efa413 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -708,9 +708,22 @@ static void report__output_resort(struct report *rep)
ui_progress__finish();
}
+static int count_sample_event(struct perf_tool *tool __maybe_unused,
+ union perf_event *event __maybe_unused,
+ struct perf_sample *sample __maybe_unused,
+ struct evsel *evsel,
+ struct machine *machine __maybe_unused)
+{
+ struct hists *hists = evsel__hists(evsel);
+
+ hists__inc_nr_events(hists);
+ return 0;
+}
+
static void stats_setup(struct report *rep)
{
memset(&rep->tool, 0, sizeof(rep->tool));
+ rep->tool.sample = count_sample_event;
rep->tool.no_warn = true;
}
@@ -719,6 +732,7 @@ static int stats_print(struct report *rep)
struct perf_session *session = rep->session;
perf_session__fprintf_nr_events(session, stdout);
+ perf_evlist__fprintf_nr_events(session->evlist, stdout);
return 0;
}
--
2.31.1.527.g47e6f16901-goog
^ permalink raw reply related [flat|nested] 9+ messages in thread* [PATCH v2 4/6] perf report: Add --skip-empty option to suppress 0 event stat
2021-04-27 1:37 [PATCHSET v2 0/6] perf report: Make --stat output more compact Namhyung Kim
` (2 preceding siblings ...)
2021-04-27 1:37 ` [PATCH v2 3/6] perf report: Show event sample counts in --stat output Namhyung Kim
@ 2021-04-27 1:37 ` Namhyung Kim
2021-04-27 1:37 ` [PATCH v2 5/6] perf report: Make --skip-empty as default Namhyung Kim
` (3 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Namhyung Kim @ 2021-04-27 1:37 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Jiri Olsa
Cc: Ingo Molnar, Peter Zijlstra, Mark Rutland, Alexander Shishkin,
LKML, Andi Kleen, Ian Rogers
To make the output more readable, I think it's better to remove 0's in
the output. Also the dummy event has no event stats so it just wasts
the space. Let's use the --skip-empty option to suppress it.
$ perf report --stat --skip-empty
Aggregated stats:
TOTAL events: 16530
MMAP events: 226
COMM events: 1596
EXIT events: 2
THROTTLE events: 121
UNTHROTTLE events: 117
FORK events: 1595
SAMPLE events: 719
MMAP2 events: 12147
CGROUP events: 2
FINISHED_ROUND events: 2
THREAD_MAP events: 1
CPU_MAP events: 1
TIME_CONV events: 1
cycles stats:
SAMPLE events: 719
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
tools/perf/Documentation/perf-report.txt | 3 +++
tools/perf/builtin-annotate.c | 4 ++--
tools/perf/builtin-report.c | 16 ++++++++++++----
tools/perf/ui/stdio/hist.c | 5 ++++-
tools/perf/util/events_stats.h | 3 ++-
tools/perf/util/hist.c | 6 +++++-
tools/perf/util/hist.h | 3 ++-
tools/perf/util/session.c | 5 +++--
tools/perf/util/session.h | 3 ++-
9 files changed, 35 insertions(+), 13 deletions(-)
diff --git a/tools/perf/Documentation/perf-report.txt b/tools/perf/Documentation/perf-report.txt
index f51f0000676e..24efc0583c93 100644
--- a/tools/perf/Documentation/perf-report.txt
+++ b/tools/perf/Documentation/perf-report.txt
@@ -571,6 +571,9 @@ include::itrace.txt[]
sampled cycles
'Avg Cycles' - block average sampled cycles
+--skip-empty::
+ Do not print 0 results in the --stat output.
+
include::callchain-overhead-calculation.txt[]
SEE ALSO
diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c
index 717efd78eee6..49627a7bed7c 100644
--- a/tools/perf/builtin-annotate.c
+++ b/tools/perf/builtin-annotate.c
@@ -404,8 +404,8 @@ static int __cmd_annotate(struct perf_annotate *ann)
goto out;
if (dump_trace) {
- perf_session__fprintf_nr_events(session, stdout);
- evlist__fprintf_nr_events(session->evlist, stdout);
+ perf_session__fprintf_nr_events(session, stdout, false);
+ evlist__fprintf_nr_events(session->evlist, stdout, false);
goto out;
}
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index be56f3efa413..4910194acaa6 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -85,6 +85,7 @@ struct report {
bool group_set;
bool stitch_lbr;
bool disable_order;
+ bool skip_empty;
int max_stack;
struct perf_read_values show_threads_values;
struct annotation_options annotation_opts;
@@ -530,6 +531,9 @@ static int evlist__tty_browse_hists(struct evlist *evlist, struct report *rep, c
if (symbol_conf.event_group && !evsel__is_group_leader(pos))
continue;
+ if (rep->skip_empty && !hists->stats.nr_samples)
+ continue;
+
hists__fprintf_nr_sample_events(hists, rep, evname, stdout);
if (rep->total_cycles_mode) {
@@ -731,8 +735,8 @@ static int stats_print(struct report *rep)
{
struct perf_session *session = rep->session;
- perf_session__fprintf_nr_events(session, stdout);
- perf_evlist__fprintf_nr_events(session->evlist, stdout);
+ perf_session__fprintf_nr_events(session, stdout, rep->skip_empty);
+ evlist__fprintf_nr_events(session->evlist, stdout, rep->skip_empty);
return 0;
}
@@ -944,8 +948,10 @@ static int __cmd_report(struct report *rep)
perf_session__fprintf_dsos(session, stdout);
if (dump_trace) {
- perf_session__fprintf_nr_events(session, stdout);
- evlist__fprintf_nr_events(session->evlist, stdout);
+ perf_session__fprintf_nr_events(session, stdout,
+ rep->skip_empty);
+ evlist__fprintf_nr_events(session->evlist, stdout,
+ rep->skip_empty);
return 0;
}
}
@@ -1313,6 +1319,8 @@ int cmd_report(int argc, const char **argv)
"Sort all blocks by 'Sampled Cycles%'"),
OPT_BOOLEAN(0, "disable-order", &report.disable_order,
"Disable raw trace ordering"),
+ OPT_BOOLEAN(0, "skip-empty", &report.skip_empty,
+ "Do not display empty (or dummy) events in the output"),
OPT_END()
};
struct perf_data data = {
diff --git a/tools/perf/ui/stdio/hist.c b/tools/perf/ui/stdio/hist.c
index 2ab2af4d4849..d9e634406175 100644
--- a/tools/perf/ui/stdio/hist.c
+++ b/tools/perf/ui/stdio/hist.c
@@ -897,7 +897,8 @@ size_t hists__fprintf(struct hists *hists, bool show_header, int max_rows,
return ret;
}
-size_t events_stats__fprintf(struct events_stats *stats, FILE *fp)
+size_t events_stats__fprintf(struct events_stats *stats, FILE *fp,
+ bool skip_empty)
{
int i;
size_t ret = 0;
@@ -908,6 +909,8 @@ size_t events_stats__fprintf(struct events_stats *stats, FILE *fp)
name = perf_event__name(i);
if (!strcmp(name, "UNKNOWN"))
continue;
+ if (skip_empty && !stats->nr_events[i])
+ continue;
ret += fprintf(fp, "%16s events: %10d\n", name, stats->nr_events[i]);
}
diff --git a/tools/perf/util/events_stats.h b/tools/perf/util/events_stats.h
index e271c8004c89..3480bafd414b 100644
--- a/tools/perf/util/events_stats.h
+++ b/tools/perf/util/events_stats.h
@@ -50,6 +50,7 @@ struct hists_stats {
void events_stats__inc(struct events_stats *stats, u32 type);
-size_t events_stats__fprintf(struct events_stats *stats, FILE *fp);
+size_t events_stats__fprintf(struct events_stats *stats, FILE *fp,
+ bool skip_empty);
#endif /* __PERF_EVENTS_STATS_ */
diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
index 691a6a777d14..65fe65ba03c2 100644
--- a/tools/perf/util/hist.c
+++ b/tools/perf/util/hist.c
@@ -2676,7 +2676,8 @@ void hist__account_cycles(struct branch_stack *bs, struct addr_location *al,
}
}
-size_t evlist__fprintf_nr_events(struct evlist *evlist, FILE *fp)
+size_t evlist__fprintf_nr_events(struct evlist *evlist, FILE *fp,
+ bool skip_empty)
{
struct evsel *pos;
size_t ret = 0;
@@ -2684,6 +2685,9 @@ size_t evlist__fprintf_nr_events(struct evlist *evlist, FILE *fp)
evlist__for_each_entry(evlist, pos) {
struct hists *hists = evsel__hists(pos);
+ if (skip_empty && !hists->stats.nr_samples)
+ continue;
+
ret += fprintf(fp, "%s stats:\n", evsel__name(pos));
ret += fprintf(fp, "%16s events: %10d\n",
"SAMPLE", hists->stats.nr_samples);
diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h
index 6b0f708f08ac..5343b62476e6 100644
--- a/tools/perf/util/hist.h
+++ b/tools/perf/util/hist.h
@@ -202,7 +202,8 @@ void hists__inc_nr_samples(struct hists *hists, bool filtered);
size_t hists__fprintf(struct hists *hists, bool show_header, int max_rows,
int max_cols, float min_pcnt, FILE *fp,
bool ignore_callchains);
-size_t evlist__fprintf_nr_events(struct evlist *evlist, FILE *fp);
+size_t evlist__fprintf_nr_events(struct evlist *evlist, FILE *fp,
+ bool skip_empty);
void hists__filter_by_dso(struct hists *hists);
void hists__filter_by_thread(struct hists *hists);
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
index eba3769be3f1..a6659b616e6d 100644
--- a/tools/perf/util/session.c
+++ b/tools/perf/util/session.c
@@ -2352,7 +2352,8 @@ size_t perf_session__fprintf_dsos_buildid(struct perf_session *session, FILE *fp
return machines__fprintf_dsos_buildid(&session->machines, fp, skip, parm);
}
-size_t perf_session__fprintf_nr_events(struct perf_session *session, FILE *fp)
+size_t perf_session__fprintf_nr_events(struct perf_session *session, FILE *fp,
+ bool skip_empty)
{
size_t ret;
const char *msg = "";
@@ -2362,7 +2363,7 @@ size_t perf_session__fprintf_nr_events(struct perf_session *session, FILE *fp)
ret = fprintf(fp, "\nAggregated stats:%s\n", msg);
- ret += events_stats__fprintf(&session->evlist->stats, fp);
+ ret += events_stats__fprintf(&session->evlist->stats, fp, skip_empty);
return ret;
}
diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h
index f76480166d38..e31ba4c92a6c 100644
--- a/tools/perf/util/session.h
+++ b/tools/perf/util/session.h
@@ -113,7 +113,8 @@ size_t perf_session__fprintf_dsos(struct perf_session *session, FILE *fp);
size_t perf_session__fprintf_dsos_buildid(struct perf_session *session, FILE *fp,
bool (fn)(struct dso *dso, int parm), int parm);
-size_t perf_session__fprintf_nr_events(struct perf_session *session, FILE *fp);
+size_t perf_session__fprintf_nr_events(struct perf_session *session, FILE *fp,
+ bool skip_empty);
struct evsel *perf_session__find_first_evtype(struct perf_session *session,
unsigned int type);
--
2.31.1.527.g47e6f16901-goog
^ permalink raw reply related [flat|nested] 9+ messages in thread* [PATCH v2 5/6] perf report: Make --skip-empty as default
2021-04-27 1:37 [PATCHSET v2 0/6] perf report: Make --stat output more compact Namhyung Kim
` (3 preceding siblings ...)
2021-04-27 1:37 ` [PATCH v2 4/6] perf report: Add --skip-empty option to suppress 0 event stat Namhyung Kim
@ 2021-04-27 1:37 ` Namhyung Kim
2021-04-27 1:37 ` [PATCH v2 6/6] perf report: Print percentage of each event statistics Namhyung Kim
` (2 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Namhyung Kim @ 2021-04-27 1:37 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Jiri Olsa
Cc: Ingo Molnar, Peter Zijlstra, Mark Rutland, Alexander Shishkin,
LKML, Andi Kleen, Ian Rogers
so that the compact output is shown by default. Also add
'report.skip-empty' config option to override the default.
Users can also use --no-skip-empty command line option to
change the behavior anytime.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
tools/perf/Documentation/perf-config.txt | 5 +++++
tools/perf/builtin-report.c | 6 ++++++
2 files changed, 11 insertions(+)
diff --git a/tools/perf/Documentation/perf-config.txt b/tools/perf/Documentation/perf-config.txt
index 154a1ced72b2..b0872c801866 100644
--- a/tools/perf/Documentation/perf-config.txt
+++ b/tools/perf/Documentation/perf-config.txt
@@ -123,6 +123,7 @@ Example
queue-size = 0
children = true
group = true
+ skip-empty = true
[llvm]
dump-obj = true
@@ -531,6 +532,10 @@ Variables
0.07% 0.00% noploop ld-2.15.so [.] strcmp
0.03% 0.00% noploop [kernel.kallsyms] [k] timerqueue_del
+ report.skip-empty::
+ This option can change default stat behavior with empty results.
+ If it's set true, 'perf report --stat' will not show 0 stats.
+
top.*::
top.children::
Same as 'report.children'. So if it is enabled, the output of 'top'
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index 4910194acaa6..36f9ccfeb38a 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -136,6 +136,11 @@ static int report__config(const char *var, const char *value, void *cb)
return 0;
}
+ if (!strcmp(var, "report.skip-empty")) {
+ rep->skip_empty = perf_config_bool(var, value);
+ return 0;
+ }
+
return 0;
}
@@ -1160,6 +1165,7 @@ int cmd_report(int argc, const char **argv)
.pretty_printing_style = "normal",
.socket_filter = -1,
.annotation_opts = annotation__default_options,
+ .skip_empty = true,
};
const struct option options[] = {
OPT_STRING('i', "input", &input_name, "file",
--
2.31.1.527.g47e6f16901-goog
^ permalink raw reply related [flat|nested] 9+ messages in thread* [PATCH v2 6/6] perf report: Print percentage of each event statistics
2021-04-27 1:37 [PATCHSET v2 0/6] perf report: Make --stat output more compact Namhyung Kim
` (4 preceding siblings ...)
2021-04-27 1:37 ` [PATCH v2 5/6] perf report: Make --skip-empty as default Namhyung Kim
@ 2021-04-27 1:37 ` Namhyung Kim
2021-04-27 5:05 ` [PATCHSET v2 0/6] perf report: Make --stat output more compact Andi Kleen
2021-04-27 13:15 ` Jiri Olsa
7 siblings, 0 replies; 9+ messages in thread
From: Namhyung Kim @ 2021-04-27 1:37 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Jiri Olsa
Cc: Ingo Molnar, Peter Zijlstra, Mark Rutland, Alexander Shishkin,
LKML, Andi Kleen, Ian Rogers
It's sometimes useful to see how many samples vs other events in the
data file with percent values.
$ perf report --stat
Aggregated stats:
TOTAL events: 20064
MMAP events: 239 ( 1.2%)
COMM events: 1518 ( 7.6%)
EXIT events: 1 ( 0.0%)
FORK events: 1517 ( 7.6%)
SAMPLE events: 4015 (20.0%)
MMAP2 events: 12769 (63.6%)
FINISHED_ROUND events: 2 ( 0.0%)
THREAD_MAP events: 1 ( 0.0%)
CPU_MAP events: 1 ( 0.0%)
TIME_CONV events: 1 ( 0.0%)
cycles stats:
SAMPLE events: 2475
instructions stats:
SAMPLE events: 1540
Suggested-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
tools/perf/ui/stdio/hist.c | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
diff --git a/tools/perf/ui/stdio/hist.c b/tools/perf/ui/stdio/hist.c
index d9e634406175..f36270485168 100644
--- a/tools/perf/ui/stdio/hist.c
+++ b/tools/perf/ui/stdio/hist.c
@@ -902,6 +902,7 @@ size_t events_stats__fprintf(struct events_stats *stats, FILE *fp,
{
int i;
size_t ret = 0;
+ u32 total = stats->nr_events[0];
for (i = 0; i < PERF_RECORD_HEADER_MAX; ++i) {
const char *name;
@@ -912,7 +913,14 @@ size_t events_stats__fprintf(struct events_stats *stats, FILE *fp,
if (skip_empty && !stats->nr_events[i])
continue;
- ret += fprintf(fp, "%16s events: %10d\n", name, stats->nr_events[i]);
+ if (i && total) {
+ ret += fprintf(fp, "%16s events: %10d (%4.1f%%)\n",
+ name, stats->nr_events[i],
+ 100.0 * stats->nr_events[i] / total);
+ } else {
+ ret += fprintf(fp, "%16s events: %10d\n",
+ name, stats->nr_events[i]);
+ }
}
return ret;
--
2.31.1.527.g47e6f16901-goog
^ permalink raw reply related [flat|nested] 9+ messages in thread* Re: [PATCHSET v2 0/6] perf report: Make --stat output more compact
2021-04-27 1:37 [PATCHSET v2 0/6] perf report: Make --stat output more compact Namhyung Kim
` (5 preceding siblings ...)
2021-04-27 1:37 ` [PATCH v2 6/6] perf report: Print percentage of each event statistics Namhyung Kim
@ 2021-04-27 5:05 ` Andi Kleen
2021-04-27 13:15 ` Jiri Olsa
7 siblings, 0 replies; 9+ messages in thread
From: Andi Kleen @ 2021-04-27 5:05 UTC (permalink / raw)
To: Namhyung Kim
Cc: Arnaldo Carvalho de Melo, Jiri Olsa, Ingo Molnar, Peter Zijlstra,
Mark Rutland, Alexander Shishkin, LKML, Ian Rogers
On Mon, Apr 26, 2021 at 06:37:11PM -0700, Namhyung Kim wrote:
> Hello,
>
> This patchset changes the output of perf report --stat.
>
> Changes from v1)
> * fix build error in TUI (Jiri)
> * print percentage of each event (Andi)
Looks all good to me.
Reviewed-by: Andi Kleen <ak@linux.intel.com>
^ permalink raw reply [flat|nested] 9+ messages in thread* Re: [PATCHSET v2 0/6] perf report: Make --stat output more compact
2021-04-27 1:37 [PATCHSET v2 0/6] perf report: Make --stat output more compact Namhyung Kim
` (6 preceding siblings ...)
2021-04-27 5:05 ` [PATCHSET v2 0/6] perf report: Make --stat output more compact Andi Kleen
@ 2021-04-27 13:15 ` Jiri Olsa
7 siblings, 0 replies; 9+ messages in thread
From: Jiri Olsa @ 2021-04-27 13:15 UTC (permalink / raw)
To: Namhyung Kim
Cc: Arnaldo Carvalho de Melo, Ingo Molnar, Peter Zijlstra,
Mark Rutland, Alexander Shishkin, LKML, Andi Kleen, Ian Rogers
On Mon, Apr 26, 2021 at 06:37:11PM -0700, Namhyung Kim wrote:
> Hello,
>
> This patchset changes the output of perf report --stat.
>
> Changes from v1)
> * fix build error in TUI (Jiri)
> * print percentage of each event (Andi)
>
>
> The perf report --stat shows event statistics like below:
>
> $ perf report --stat
>
> Aggregated stats:
> TOTAL events: 20064
> MMAP events: 239
> LOST events: 0
> COMM events: 1518
> EXIT events: 1
> THROTTLE events: 0
> UNTHROTTLE events: 0
> FORK events: 1517
> READ events: 0
> SAMPLE events: 4015
> MMAP2 events: 12769
> AUX events: 0
> ITRACE_START events: 0
> LOST_SAMPLES events: 0
> SWITCH events: 0
> SWITCH_CPU_WIDE events: 0
> NAMESPACES events: 0
> KSYMBOL events: 0
> BPF_EVENT events: 0
> CGROUP events: 0
> TEXT_POKE events: 0
> ATTR events: 0
> EVENT_TYPE events: 0
> TRACING_DATA events: 0
> BUILD_ID events: 0
> FINISHED_ROUND events: 2
> ID_INDEX events: 0
> AUXTRACE_INFO events: 0
> AUXTRACE events: 0
> AUXTRACE_ERROR events: 0
> THREAD_MAP events: 1
> CPU_MAP events: 1
> STAT_CONFIG events: 0
> STAT events: 0
> STAT_ROUND events: 0
> EVENT_UPDATE events: 0
> TIME_CONV events: 1
> FEATURE events: 0
> COMPRESSED events: 0
>
> But it's too long and mostly 0 so we can make it more compact. Also
> perf report -D has similar output at the end with each event's sample
> count. It'd be better if we can have the same output in both case.
>
> So I added --skip-empty (and --no-skip-empty automatically) to suppres
> the 0 output and add the event stats like below.
>
> $ perf report --stat --skip-empty
>
> Aggregated stats:
> TOTAL events: 20064
> MMAP events: 239 ( 1.2%)
> COMM events: 1518 ( 7.6%)
> EXIT events: 1 ( 0.0%)
> FORK events: 1517 ( 7.6%)
> SAMPLE events: 4015 (20.0%)
> MMAP2 events: 12769 (63.6%)
> FINISHED_ROUND events: 2 ( 0.0%)
> THREAD_MAP events: 1 ( 0.0%)
> CPU_MAP events: 1 ( 0.0%)
> TIME_CONV events: 1 ( 0.0%)
> cycles stats:
> SAMPLE events: 2475
> instructions stats:
> SAMPLE events: 1540
>
>
> And I make it enabled by default with a new config option
> report.skip-empty to change the behavior if needed.
>
> Thanks,
> Namhyung
>
>
> Namhyung Kim (6):
> perf top: Use evlist->events_stat to count events
> perf hists: Split hists_stats from events_stats
> perf report: Show event sample counts in --stat output
> perf report: Add --skip-empty option to suppress 0 event stat
> perf report: Make --skip-empty as default
> perf report: Print percentage of each event statistics
Acked-by: Jiri Olsa <jolsa@redhat.com>
thanks,
jirka
^ permalink raw reply [flat|nested] 9+ messages in thread