Skip to contentSkip to conversation

perf, tools: Refactor and support interval and CSV metrics

PATCH New Andi Kleen 2 participants 11 messages Jan 16, 2016 – Jan 18, 2016 perf events

lore
Share & tools
lore

Andi Kleen

author
#0 lore raw

Rebased tree and fixed Jiri's last feedback.

[v4: Addressed all review feedback.]
[v3: Addressed all review feedback. Update manpage for CSV. Various changes
(see individual patches). Remove some more redundant code
in printout callers.]
[v2: Addressed (near) all review feedback. No manpage updates so far.
Add support for --per-core metrics. Various cleanups.]
[v3: Everything compiles again. Some more cleanups.]
[v4: Split up abstract metrics patch into two. Fix bug with earlier patch already enabling metrics for CSV/interval. Minor cleanups. Man page is included]
[v5: Fix mainly bisect problems. No regressions introduced by one patch and fixed again later. Some minor fixes in addition]
[v6: Fix running/noise printing patch.]
[v7: Reorder and merge two patches to avoid a bisect hole where unsupported was printed as 0]
[v8: Minor fixes for review feedback. See changelog in patches.]

Currently perf stat does not support printing computed metrics for interval (-I xxx) or CSV (-x,) mode. For example IPC or TSX metrics over time are quite useful to know.

This patch implements them. The main obstacle was that the
metrics printing was all open coded all over the metrics computation code. The second patch refactors the metrics printing to work through call backs that can be more easily changed. This also cleans up the metrics printing significantly. The indentation is now handled through printf, no more need to manually count spaces.

Then based on that it implements metrics printing for CSV and interval mode.

Example output:

% perf stat -I1000 -a sleep 1

#          time              counts unit events                    metric                              multiplex
     1.001301370       12020.049593      task-clock (msec)                                             (100.00%)
     1.001301370              3,952      context-switches          #    0.329 K/sec                    (100.00%)
     1.001301370                 69      cpu-migrations            #    0.006 K/sec                    (100.00%)
     1.001301370                 76      page-faults               #    0.006 K/sec                  
     1.001301370        386,582,789      cycles                    #    0.032 GHz                      (100.00%)
     1.001301370        716,441,544      stalled-cycles-frontend   #  185.33% frontend cycles idle     (100.00%)
     1.001301370    <not supported>      stalled-cycles-backend   
     1.001301370        101,751,678      instructions              #    0.26  insn per cycle         
     1.001301370                                                   #    7.04  stalled cycles per insn  (100.00%)
     1.001301370         20,914,692      branches                  #    1.740 M/sec                    (100.00%)
     1.001301370          1,943,630      branch-misses             #    9.29% of all branches        

CSV mode

% perf stat -x, -I1000 -a sleep 1

     1.000852081,12016.143006,,task-clock
     1.000852081,4457,,context-switches,12015168277,100.00,0.371,K/sec
     1.000852081,50,,cpu-migrations,12014024424,100.00,0.004,K/sec
     1.000852081,76,,page-faults,12013076716,100.00,0.006,K/sec
     1.000852081,515854373,,cycles,12011235336,100.00,0.043,GHz
     1.000852081,1030742150,,stalled-cycles-frontend,12010984057,100.00,199.81,frontend cycles idle
     1.000852081,<not supported>,,stalled-cycles-backend,0,100.00
     1.000852081,116782495,,instructions,12011130729,100.00,0.23,insn per cycle
     1.000852081,,,,12011130729,100.00,8.83,stalled cycles per insn
     1.000852081,23748237,,branches,12010745125,100.00,1.976,M/sec
     1.000852081,1976560,,branch-misses,12010501884,100.00,8.32,of all branches

Available in
git://git.kernel.org/pub/scm/linux/kernel/git/ak/linux-misc-2.6 perf/stat-metrics-12

author
#1 lore raw

PATCH 1/8 perf, tools, stat: Abstract stat metrics printing

From: Andi Kleen <[email protected]>

Abstract the printing of shadow metrics. Instead of every metric calling fprintf directly and taking care of indentation, use two call backs: one to print metrics and another to start a new line.

This will allow adding metrics to CSV mode and also using them for other purposes.

The computation of padding is now done in the central callback, instead of every metric doing it manually. This makes it easier to add new metrics.

v2: Refactor functions, printout now does more. Move shadow printing. Improve fallback callbacks. Don't
use void * callback data.
v3: Remove unnecessary hunk. Add typedef for new_line
v4: Remove unnecessary hunk. Don't print metrics for CSV/interval mode yet. Move printout change to separate patch.
v5: Fix bisect bugs. Avoid bogus frontend cycles printing. Fix indentation in different aggregation modes.

---
 tools/perf/builtin-stat.c     |  59 ++++++++++--
 tools/perf/util/stat-shadow.c | 211 +++++++++++++++++++++++-------------------
 tools/perf/util/stat.h        |  15 ++-
 3 files changed, 182 insertions(+), 103 deletions(-)
tools/perf/builtin-stat.c +53 -6
index 7f56824..cba9753 100644
@@ -735,6 +735,46 @@ static void aggr_printout(struct perf_evsel *evsel, int id, int nr)
 	}
 }
 
+struct outstate {
+	FILE *fh;
+};
+
+#define METRIC_LEN  35
+
+static void new_line_std(void *ctx)
+{
+	struct outstate *os = ctx;
+
+	fputc('\n', os->fh);
+	if (stat_config.aggr_mode == AGGR_NONE)
+		fprintf(os->fh, "        ");
+	if (stat_config.aggr_mode == AGGR_CORE)
+		fprintf(os->fh, "                  ");
+	if (stat_config.aggr_mode == AGGR_SOCKET)
+		fprintf(os->fh, "            ");
+	fprintf(os->fh, "                                                 ");
+}
+
+static void print_metric_std(void *ctx, const char *color, const char *fmt,
+			     const char *unit, double val)
+{
+	struct outstate *os = ctx;
+	FILE *out = os->fh;
+	int n;
+
+	if (unit == NULL || fmt == NULL) {
+		fprintf(out, "%-*s", METRIC_LEN, "");
+		return;
+	}
+
+	n = fprintf(out, " # ");
+	if (color)
+		n += color_fprintf(out, color, fmt, val);
+	else
+		n += fprintf(out, fmt, val);
+	fprintf(out, " %-*s", METRIC_LEN - n - 1, unit);
+}
+
 static void nsec_printout(int id, int nr, struct perf_evsel *evsel, double avg)
 {
 	FILE *output = stat_config.output;
@@ -795,20 +835,27 @@ static void abs_printout(int id, int nr, struct perf_evsel *evsel, double avg)
 
 static void printout(int id, int nr, struct perf_evsel *counter, double uval)
 {
-	int cpu = cpu_map__id_to_cpu(id);
+	struct outstate os = { .fh = stat_config.output };
+	struct perf_stat_output_ctx out;
+	print_metric_t pm = print_metric_std;
+	void (*nl)(void *);
 
-	if (stat_config.aggr_mode == AGGR_GLOBAL)
-		cpu = 0;
+	nl = new_line_std;
 
 	if (nsec_counter(counter))
 		nsec_printout(id, nr, counter, uval);
 	else
 		abs_printout(id, nr, counter, uval);
 
+	out.print_metric = pm;
+	out.new_line = nl;
+	out.ctx = &os;
+
 	if (!csv_output && !stat_config.interval)
-		perf_stat__print_shadow_stats(stat_config.output, counter,
-					      uval, cpu,
-					      stat_config.aggr_mode);
+		perf_stat__print_shadow_stats(counter, uval,
+				stat_config.aggr_mode == AGGR_GLOBAL ? 0 :
+				cpu_map__id_to_cpu(id),
+				&out);
 }
 
 static void print_aggr(char *prefix)
tools/perf/util/stat-shadow.c +116 -95

369 lines not on this page · raw message

tools/perf/util/stat.h +13 -2
index 086f4e1..f14c0f4 100644
@@ -68,11 +68,22 @@ void perf_stat_evsel_id_init(struct perf_evsel *evsel);
 
 extern struct stats walltime_nsecs_stats;
 
+typedef void (*print_metric_t)(void *ctx, const char *color, const char *unit,
+			       const char *fmt, double val);
+typedef void (*new_line_t )(void *ctx);
+
 void perf_stat__reset_shadow_stats(void);
 void perf_stat__update_shadow_stats(struct perf_evsel *counter, u64 *count,
 				    int cpu);
-void perf_stat__print_shadow_stats(FILE *out, struct perf_evsel *evsel,
-				   double avg, int cpu, enum aggr_mode aggr);
+struct perf_stat_output_ctx {
+	void *ctx;
+	print_metric_t print_metric;
+	new_line_t new_line;
+};
+
+void perf_stat__print_shadow_stats(struct perf_evsel *evsel,
+				   double avg, int cpu,
+				   struct perf_stat_output_ctx *out);
 
 void perf_evsel__reset_stat_priv(struct perf_evsel *evsel);
 int perf_evsel__alloc_stat_priv(struct perf_evsel *evsel);
Signature · 1 line
2.4.3
author
#2 lore raw

PATCH 2/8 perf, tools, stat: Add support for metrics in interval mode

In reply to Andi Kleen

From: Andi Kleen <[email protected]>

Now that we can modify the metrics printout functions easily, it's straight forward to support metric printing for interval mode. All that is needed is to print the time stamp on every new line. Pass the prefix into the context and print it out.

v2: Move wrong hunk to here.

---
 tools/perf/builtin-stat.c | 20 +++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)
tools/perf/builtin-stat.c +13 -7
index cba9753..648a8af 100644
@@ -737,6 +737,7 @@ static void aggr_printout(struct perf_evsel *evsel, int id, int nr)
 
 struct outstate {
 	FILE *fh;
+	const char *prefix;
 };
 
 #define METRIC_LEN  35
@@ -746,6 +747,7 @@ static void new_line_std(void *ctx)
 	struct outstate *os = ctx;
 
 	fputc('\n', os->fh);
+	fputs(os->prefix, os->fh);
 	if (stat_config.aggr_mode == AGGR_NONE)
 		fprintf(os->fh, "        ");
 	if (stat_config.aggr_mode == AGGR_CORE)
@@ -833,10 +835,14 @@ static void abs_printout(int id, int nr, struct perf_evsel *evsel, double avg)
 		fprintf(output, "%s%s", csv_sep, evsel->cgrp->name);
 }
 
-static void printout(int id, int nr, struct perf_evsel *counter, double uval)
+static void printout(int id, int nr, struct perf_evsel *counter, double uval,
+		     char *prefix)
 {
-	struct outstate os = { .fh = stat_config.output };
 	struct perf_stat_output_ctx out;
+	struct outstate os = {
+		.fh = stat_config.output,
+		.prefix = prefix ? prefix : ""
+	};
 	print_metric_t pm = print_metric_std;
 	void (*nl)(void *);
 
@@ -851,7 +857,7 @@ static void printout(int id, int nr, struct perf_evsel *counter, double uval)
 	out.new_line = nl;
 	out.ctx = &os;
 
-	if (!csv_output && !stat_config.interval)
+	if (!csv_output)
 		perf_stat__print_shadow_stats(counter, uval,
 				stat_config.aggr_mode == AGGR_GLOBAL ? 0 :
 				cpu_map__id_to_cpu(id),
@@ -911,7 +917,7 @@ static void print_aggr(char *prefix)
 				continue;
 			}
 			uval = val * counter->scale;
-			printout(id, nr, counter, uval);
+			printout(id, nr, counter, uval, prefix);
 			if (!csv_output)
 				print_noise(counter, 1.0);
 
@@ -942,7 +948,7 @@ static void print_aggr_thread(struct perf_evsel *counter, char *prefix)
 			fprintf(output, "%s", prefix);
 
 		uval = val * counter->scale;
-		printout(thread, 0, counter, uval);
+		printout(thread, 0, counter, uval, prefix);
 
 		if (!csv_output)
 			print_noise(counter, 1.0);
@@ -992,7 +998,7 @@ static void print_counter_aggr(struct perf_evsel *counter, char *prefix)
 	}
 
 	uval = avg * counter->scale;
-	printout(-1, 0, counter, uval);
+	printout(-1, 0, counter, uval, prefix);
 
 	print_noise(counter, avg);
 
@@ -1045,7 +1051,7 @@ static void print_counter(struct perf_evsel *counter, char *prefix)
 		}
 
 		uval = val * counter->scale;
-		printout(cpu, 0, counter, uval);
+		printout(cpu, 0, counter, uval, prefix);
 		if (!csv_output)
 			print_noise(counter, 1.0);
 		print_running(run, ena);
Signature · 1 line
2.4.3
author
#3 lore raw

PATCH 3/8 perf, tools, stat: Move noise/running printing into printout

In reply to Andi Kleen

From: Andi Kleen <[email protected]>

Move the running/noise printing into printout to avoid duplicated code in the callers.

v2: Merged with other patches. Remove unnecessary hunk. Readd hunk that ended in earlier patch.
v3: Fix noise/running output in CSV mode
v4: Merge with later patch that also moves not supported printing.

---
 tools/perf/builtin-stat.c | 123 ++++++++++++----------------------------------
 1 file changed, 32 insertions(+), 91 deletions(-)
tools/perf/builtin-stat.c +32 -91
index 648a8af..fe22d20 100644
@@ -836,7 +836,7 @@ static void abs_printout(int id, int nr, struct perf_evsel *evsel, double avg)
 }
 
 static void printout(int id, int nr, struct perf_evsel *counter, double uval,
-		     char *prefix)
+		     char *prefix, u64 run, u64 ena, double noise)
 {
 	struct perf_stat_output_ctx out;
 	struct outstate os = {
@@ -848,6 +848,30 @@ static void printout(int id, int nr, struct perf_evsel *counter, double uval,
 
 	nl = new_line_std;
 
+	if (run == 0 || ena == 0) {
+		aggr_printout(counter, id, nr);
+
+		fprintf(stat_config.output, "%*s%s",
+			csv_output ? 0 : 18,
+			counter->supported ? CNTR_NOT_COUNTED : CNTR_NOT_SUPPORTED,
+			csv_sep);
+
+		fprintf(stat_config.output, "%-*s%s",
+			csv_output ? 0 : unit_width,
+			counter->unit, csv_sep);
+
+		fprintf(stat_config.output, "%*s",
+			csv_output ? 0 : -25,
+			perf_evsel__name(counter));
+
+		if (counter->cgrp)
+			fprintf(stat_config.output, "%s%s",
+				csv_sep, counter->cgrp->name);
+
+		print_running(run, ena);
+		return;
+	}
+
 	if (nsec_counter(counter))
 		nsec_printout(id, nr, counter, uval);
 	else
@@ -862,6 +886,9 @@ static void printout(int id, int nr, struct perf_evsel *counter, double uval,
 				stat_config.aggr_mode == AGGR_GLOBAL ? 0 :
 				cpu_map__id_to_cpu(id),
 				&out);
+
+	print_noise(counter, noise);
+	print_running(run, ena);
 }
 
 static void print_aggr(char *prefix)
@@ -892,36 +919,8 @@ static void print_aggr(char *prefix)
 			if (prefix)
 				fprintf(output, "%s", prefix);
 
-			if (run == 0 || ena == 0) {
-				aggr_printout(counter, id, nr);
-
-				fprintf(output, "%*s%s",
-					csv_output ? 0 : 18,
-					counter->supported ? CNTR_NOT_COUNTED : CNTR_NOT_SUPPORTED,
-					csv_sep);
-
-				fprintf(output, "%-*s%s",
-					csv_output ? 0 : unit_width,
-					counter->unit, csv_sep);
-
-				fprintf(output, "%*s",
-					csv_output ? 0 : -25,
-					perf_evsel__name(counter));
-
-				if (counter->cgrp)
-					fprintf(output, "%s%s",
-						csv_sep, counter->cgrp->name);
-
-				print_running(run, ena);
-				fputc('\n', output);
-				continue;
-			}
 			uval = val * counter->scale;
-			printout(id, nr, counter, uval, prefix);
-			if (!csv_output)
-				print_noise(counter, 1.0);
-
-			print_running(run, ena);
+			printout(id, nr, counter, uval, prefix, run, ena, 1.0);
 			fputc('\n', output);
 		}
 	}
@@ -948,12 +947,7 @@ static void print_aggr_thread(struct perf_evsel *counter, char *prefix)
 			fprintf(output, "%s", prefix);
 
 		uval = val * counter->scale;
-		printout(thread, 0, counter, uval, prefix);
-
-		if (!csv_output)
-			print_noise(counter, 1.0);
-
-		print_running(run, ena);
+		printout(thread, 0, counter, uval, prefix, run, ena, 1.0);
 		fputc('\n', output);
 	}
 }
@@ -967,7 +961,6 @@ static void print_counter_aggr(struct perf_evsel *counter, char *prefix)
 	FILE *output = stat_config.output;
 	struct perf_stat_evsel *ps = counter->priv;
 	double avg = avg_stats(&ps->res_stats[0]);
-	int scaled = counter->counts->scaled;
 	double uval;
 	double avg_enabled, avg_running;
 
@@ -977,32 +970,8 @@ static void print_counter_aggr(struct perf_evsel *counter, char *prefix)
 	if (prefix)
 		fprintf(output, "%s", prefix);
 
-	if (scaled == -1 || !counter->supported) {
-		fprintf(output, "%*s%s",
-			csv_output ? 0 : 18,
-			counter->supported ? CNTR_NOT_COUNTED : CNTR_NOT_SUPPORTED,
-			csv_sep);
-		fprintf(output, "%-*s%s",
-			csv_output ? 0 : unit_width,
-			counter->unit, csv_sep);
-		fprintf(output, "%*s",
-			csv_output ? 0 : -25,
-			perf_evsel__name(counter));
-
-		if (counter->cgrp)
-			fprintf(output, "%s%s", csv_sep, counter->cgrp->name);
-
-		print_running(avg_running, avg_enabled);
-		fputc('\n', output);
-		return;
-	}
-
 	uval = avg * counter->scale;
-	printout(-1, 0, counter, uval, prefix);
-
-	print_noise(counter, avg);
-
-	print_running(avg_running, avg_enabled);
+	printout(-1, 0, counter, uval, prefix, avg_running, avg_enabled, avg);
 	fprintf(output, "\n");
 }
 
@@ -1025,36 +994,8 @@ static void print_counter(struct perf_evsel *counter, char *prefix)
 		if (prefix)
 			fprintf(output, "%s", prefix);
 
-		if (run == 0 || ena == 0) {
-			fprintf(output, "CPU%*d%s%*s%s",
-				csv_output ? 0 : -4,
-				perf_evsel__cpus(counter)->map[cpu], csv_sep,
-				csv_output ? 0 : 18,
-				counter->supported ? CNTR_NOT_COUNTED : CNTR_NOT_SUPPORTED,
-				csv_sep);
-
-				fprintf(output, "%-*s%s",
-					csv_output ? 0 : unit_width,
-					counter->unit, csv_sep);
-
-				fprintf(output, "%*s",
-					csv_output ? 0 : -25,
-					perf_evsel__name(counter));
-
-			if (counter->cgrp)
-				fprintf(output, "%s%s",
-					csv_sep, counter->cgrp->name);
-
-			print_running(run, ena);
-			fputc('\n', output);
-			continue;
-		}
-
 		uval = val * counter->scale;
-		printout(cpu, 0, counter, uval, prefix);
-		if (!csv_output)
-			print_noise(counter, 1.0);
-		print_running(run, ena);
+		printout(cpu, 0, counter, uval, prefix, run, ena, 1.0);
 
 		fputc('\n', output);
 	}
Signature · 1 line
2.4.3
author
#4 lore raw

PATCH 4/8 perf, tools, stat: Implement CSV metrics output

In reply to Andi Kleen

From: Andi Kleen <[email protected]>

Now support CSV output for metrics. With the new output callbacks this is relatively straight forward by creating new callbacks.

The new line callback needs to know the number of fields to skip them correctly

v2: Split out function argument changes
v3: Reenable metrics for real.
v4: Fix wrong hunk from refactoring.

---
 tools/perf/builtin-stat.c | 66 +++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 64 insertions(+), 2 deletions(-)
tools/perf/builtin-stat.c +64 -2
index fe22d20..a9bdb4a 100644
@@ -738,6 +738,8 @@ static void aggr_printout(struct perf_evsel *evsel, int id, int nr)
 struct outstate {
 	FILE *fh;
 	const char *prefix;
+	int  nfields;
+	u64  run, ena;
 };
 
 #define METRIC_LEN  35
@@ -777,6 +779,49 @@ static void print_metric_std(void *ctx, const char *color, const char *fmt,
 	fprintf(out, " %-*s", METRIC_LEN - n - 1, unit);
 }
 
+static void new_line_csv(void *ctx)
+{
+	struct outstate *os = ctx;
+	int i;
+
+	fputc('\n', os->fh);
+	if (os->prefix)
+		fprintf(os->fh, "%s%s", os->prefix, csv_sep);
+	for (i = 0; i < os->nfields; i++)
+		fputs(csv_sep, os->fh);
+}
+
+static void print_metric_csv(void *ctx,
+			     const char *color __maybe_unused,
+			     const char *fmt, const char *unit, double val)
+{
+	struct outstate *os = ctx;
+	FILE *out = os->fh;
+	char buf[64], *vals, *ends;
+
+	if (unit == NULL) {
+		fprintf(out, "%s%s%s%s", csv_sep, csv_sep, csv_sep, csv_sep);
+		return;
+	}
+	fprintf(out, "%s%" PRIu64 "%s%.2f%s",
+		csv_sep,
+		os->run,
+		csv_sep,
+		os->ena ? 100.0 * os->run / os->ena : 100.0,
+		csv_sep);
+	snprintf(buf, sizeof(buf), fmt, val);
+	vals = buf;
+	while (isspace(*vals))
+		vals++;
+	ends = vals;
+	while (isdigit(*ends) || *ends == '.')
+		ends++;
+	*ends = 0;
+	while (isspace(*unit))
+		unit++;
+	fprintf(out, "%s%s%s", vals, csv_sep, unit);
+}
+
 static void nsec_printout(int id, int nr, struct perf_evsel *evsel, double avg)
 {
 	FILE *output = stat_config.output;
@@ -848,6 +893,24 @@ static void printout(int id, int nr, struct perf_evsel *counter, double uval,
 
 	nl = new_line_std;
 
+	if (csv_output) {
+		static int aggr_fields[] = {
+			[AGGR_GLOBAL] = 0,
+			[AGGR_THREAD] = 1,
+			[AGGR_NONE] = 1,
+			[AGGR_SOCKET] = 2,
+			[AGGR_CORE] = 2,
+		};
+
+		pm = print_metric_csv;
+		nl = new_line_csv;
+		os.nfields = 1;
+		os.nfields += aggr_fields[stat_config.aggr_mode];
+		if (counter->cgrp)
+			os.nfields++;
+		os.run = run;
+		os.ena = ena;
+	}
 	if (run == 0 || ena == 0) {
 		aggr_printout(counter, id, nr);
 
@@ -881,8 +944,7 @@ static void printout(int id, int nr, struct perf_evsel *counter, double uval,
 	out.new_line = nl;
 	out.ctx = &os;
 
-	if (!csv_output)
-		perf_stat__print_shadow_stats(counter, uval,
+	perf_stat__print_shadow_stats(counter, uval,
 				stat_config.aggr_mode == AGGR_GLOBAL ? 0 :
 				cpu_map__id_to_cpu(id),
 				&out);
Signature · 1 line
2.4.3
author
#5 lore raw

PATCH 5/8 perf, tools: Support metrics in --per-core/socket mode

In reply to Andi Kleen

From: Andi Kleen <[email protected]>

Enable metrics printing in --per-core / --per-socket mode. We need to save the shadow metrics in a unique place. Always use the first CPU in the aggregation. Then use the same CPU to retrieve the shadow value later.

Example output:

% perf stat --per-core -a ./BC1s

Performance counter stats for 'system wide':

S0-C0           2        2966.020381      task-clock (msec)         #    2.004 CPUs utilized            (100.00%)
S0-C0           2                 49      context-switches          #    0.017 K/sec                    (100.00%)
S0-C0           2                  4      cpu-migrations            #    0.001 K/sec                    (100.00%)
S0-C0           2                467      page-faults               #    0.157 K/sec
S0-C0           2      4,599,061,773      cycles                    #    1.551 GHz                      (100.00%)
S0-C0           2      9,755,886,883      instructions              #    2.12  insn per cycle           (100.00%)
S0-C0           2      1,906,272,125      branches                  #  642.704 M/sec                    (100.00%)
S0-C0           2         81,180,867      branch-misses             #    4.26% of all branches
S0-C1           2        2965.995373      task-clock (msec)         #    2.003 CPUs utilized            (100.00%)
S0-C1           2                 62      context-switches          #    0.021 K/sec                    (100.00%)
S0-C1           2                  8      cpu-migrations            #    0.003 K/sec                    (100.00%)
S0-C1           2                281      page-faults               #    0.095 K/sec
S0-C1           2          6,347,290      cycles                    #    0.002 GHz                      (100.00%)
S0-C1           2          4,654,156      instructions              #    0.73  insn per cycle           (100.00%)
S0-C1           2            947,121      branches                  #    0.319 M/sec                    (100.00%)
S0-C1           2             37,322      branch-misses             #    3.94% of all branches
       1.480409747 seconds time elapsed

v2: Rebase to older patches

---
 tools/perf/builtin-stat.c | 58 +++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 51 insertions(+), 7 deletions(-)
tools/perf/builtin-stat.c +51 -7
index a9bdb4a..304127a 100644
@@ -740,6 +740,8 @@ struct outstate {
 	const char *prefix;
 	int  nfields;
 	u64  run, ena;
+	int  id, nr;
+	struct perf_evsel *evsel;
 };
 
 #define METRIC_LEN  35
@@ -750,12 +752,9 @@ static void new_line_std(void *ctx)
 
 	fputc('\n', os->fh);
 	fputs(os->prefix, os->fh);
+	aggr_printout(os->evsel, os->id, os->nr);
 	if (stat_config.aggr_mode == AGGR_NONE)
 		fprintf(os->fh, "        ");
-	if (stat_config.aggr_mode == AGGR_CORE)
-		fprintf(os->fh, "                  ");
-	if (stat_config.aggr_mode == AGGR_SOCKET)
-		fprintf(os->fh, "            ");
 	fprintf(os->fh, "                                                 ");
 }
 
@@ -787,6 +786,7 @@ static void new_line_csv(void *ctx)
 	fputc('\n', os->fh);
 	if (os->prefix)
 		fprintf(os->fh, "%s%s", os->prefix, csv_sep);
+	aggr_printout(os->evsel, os->id, os->nr);
 	for (i = 0; i < os->nfields; i++)
 		fputs(csv_sep, os->fh);
 }
@@ -850,6 +850,22 @@ static void nsec_printout(int id, int nr, struct perf_evsel *evsel, double avg)
 		fprintf(output, "%s%s", csv_sep, evsel->cgrp->name);
 }
 
+static int first_shadow_cpu(struct perf_evsel *evsel, int id)
+{
+	int i;
+
+	if (aggr_get_id == NULL)
+		return 0;
+
+	for (i = 0; i < perf_evsel__nr_cpus(evsel); i++) {
+		int cpu2 = perf_evsel__cpus(evsel)->map[i];
+
+		if (aggr_get_id(evsel_list->cpus, cpu2) == id)
+			return cpu2;
+	}
+	return 0;
+}
+
 static void abs_printout(int id, int nr, struct perf_evsel *evsel, double avg)
 {
 	FILE *output = stat_config.output;
@@ -886,7 +902,10 @@ static void printout(int id, int nr, struct perf_evsel *counter, double uval,
 	struct perf_stat_output_ctx out;
 	struct outstate os = {
 		.fh = stat_config.output,
-		.prefix = prefix ? prefix : ""
+		.prefix = prefix ? prefix : "",
+		.id = id,
+		.nr = nr,
+		.evsel = counter,
 	};
 	print_metric_t pm = print_metric_std;
 	void (*nl)(void *);
@@ -946,13 +965,36 @@ static void printout(int id, int nr, struct perf_evsel *counter, double uval,
 
 	perf_stat__print_shadow_stats(counter, uval,
 				stat_config.aggr_mode == AGGR_GLOBAL ? 0 :
-				cpu_map__id_to_cpu(id),
+				first_shadow_cpu(counter, id),
 				&out);
-
 	print_noise(counter, noise);
 	print_running(run, ena);
 }
 
+static void aggr_update_shadow(void)
+{
+	int cpu, cpu2, s2, id, s;
+	u64 val;
+	struct perf_evsel *counter;
+
+	for (s = 0; s < aggr_map->nr; s++) {
+		id = aggr_map->map[s];
+		evlist__for_each(evsel_list, counter) {
+			val = 0;
+			for (cpu = 0; cpu < perf_evsel__nr_cpus(counter); cpu++) {
+				cpu2 = perf_evsel__cpus(counter)->map[cpu];
+				s2 = aggr_get_id(evsel_list->cpus, cpu2);
+				if (s2 != id)
+					continue;
+				val += perf_counts(counter->counts, cpu, 0)->val;
+			}
+			val = val * counter->scale;
+			perf_stat__update_shadow_stats(counter, &val,
+						       first_shadow_cpu(counter, id));
+		}
+	}
+}
+
 static void print_aggr(char *prefix)
 {
 	FILE *output = stat_config.output;
@@ -964,6 +1006,8 @@ static void print_aggr(char *prefix)
 	if (!(aggr_map || aggr_get_id))
 		return;
 
+	aggr_update_shadow();
+
 	for (s = 0; s < aggr_map->nr; s++) {
 		id = aggr_map->map[s];
 		evlist__for_each(evsel_list, counter) {
Signature · 1 line
2.4.3
author
#6 lore raw

PATCH 6/8 perf, tools, stat: Document CSV format in manpage

In reply to Andi Kleen

From: Andi Kleen <[email protected]>

With all the recently added fields in the perf stat CSV output we should finally document them in the man page. Do this here.

---
 tools/perf/Documentation/perf-stat.txt | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)
tools/perf/Documentation/perf-stat.txt +19 -0
index 52ef7a9..6629a8e 100644
@@ -211,6 +211,25 @@ $ perf stat -- make -j
 
  Wall-clock time elapsed:   719.554352 msecs
 
+CSV FORMAT
+----------
+
+With -x, perf stat is able to output a not-quite-CSV format output
+(commas in the output are not put into "")
+
+The fields are in this order:
+
+	- optional usec time stamp in fractions of second (with -I xxx)
+	- unit of the counter value or empty
+	- event name
+	- counter value
+	- run time of counter
+	- percentage of measurement time the counter was running
+	- optional metric value
+	- optional unit of metric
+
+Additional metrics may be printed with all earlier fields being empty.
+
 SEE ALSO
 --------
 linkperf:perf-top[1], linkperf:perf-list[1]
Signature · 1 line
2.4.3
author
#7 lore raw

PATCH 7/8 perf, tools, stat: Implement --metric-only mode

In reply to Andi Kleen

From: Andi Kleen <[email protected]>

Add a new mode to only print metrics. Sometimes we don't care about the raw values, just want the computed metrics. This allows more compact printing, so with -I each sample is only a single line. This also allows easier plotting and processing with other tools.

The main target is with using --topdown, but it also works with -T and standard perf stat. A few metrics are not supported.

To avoiding having to hardcode all the metrics in the code it uses a two pass approach: first compute dummy metrics and only print the headers in the print_metric callback. Then use the callback to print the actual values. There are some additional changes in the stat printout code to handle all metrics being on a single line.

Example:

% perf stat -a -I 1000 --metric-only

     1.000604977 frontend cycles idle     backend cycles idle      insn per cycle           stalled cycles per insn  branch-misses of all branches
     1.000604977                                                    0.76                                             2.35%
     2.000924680                                                    0.72                                             2.34%
     3.001139592                                                    0.76                                             2.57%
     4.001358452                                                    0.73                                             2.44%

The output is fairly wide, but that's a trade off for the concise format.

v2: Lots of updates.

---
 tools/perf/Documentation/perf-stat.txt |   4 +
 tools/perf/builtin-stat.c              | 210 +++++++++++++++++++++++++++++++--
 2 files changed, 204 insertions(+), 10 deletions(-)
tools/perf/Documentation/perf-stat.txt +4 -0
index 6629a8e..51e3c5a 100644
@@ -139,6 +139,10 @@ Print count deltas every N milliseconds (minimum: 10ms)
 The overhead percentage could be high in some cases, for instance with small, sub 100ms intervals.  Use with caution.
 	example: 'perf stat -I 1000 -e cycles -a sleep 5'
 
+--metric-only::
+Only print computed metrics. Print them in a single line.
+Don't show any raw values. Not supported with -A or --per-thread.
+
 --per-socket::
 Aggregate counts per processor socket for system-wide mode measurements.  This
 is a useful mode to detect imbalance between sockets.  To enable this mode,
tools/perf/builtin-stat.c +200 -10

317 lines not on this page · raw message

Signature · 1 line
2.4.3
author
#8 lore raw

PATCH 8/8 perf, tools, stat: Fix indentation of counter running printout

In reply to Andi Kleen

From: Andi Kleen <[email protected]>

Fix the indentation of the counter running printout for non supported counters.

---
 tools/perf/builtin-stat.c | 2 ++
 1 file changed, 2 insertions(+)
tools/perf/builtin-stat.c +2 -0
index 7f2ed13..5a02870 100644
@@ -1055,6 +1055,8 @@ static void printout(int id, int nr, struct perf_evsel *counter, double uval,
 			fprintf(stat_config.output, "%s%s",
 				csv_sep, counter->cgrp->name);
 
+		pm(&os, NULL, NULL, "", 0);
+
 		print_running(run, ena);
 		return;
 	}
Signature · 1 line
2.4.3
#9 lore raw
In reply to Andi Kleen

On Fri, Jan 15, 2016 at 05:11:07PM -0800, Andi Kleen wrote:

SNIP

Quoted 4 lines
     1.000852081,1976560,,branch-misses,12010501884,100.00,8.32,of all branches

Available in
git://git.kernel.org/pub/scm/linux/kernel/git/ak/linux-misc-2.6 perf/stat-metrics-12

getting little tired of this:
http://marc.info/?l=linux-kernel&m=145076648115589&w=2

first patch in your perf/stat-metrics-12:

[jolsa@krava perf]$ ./perf stat -e instructions,cycles kill
kill: not enough arguments

Performance counter stats for 'kill':

           772,743      instructions              #    0.82  insn per cycle         
                                                                                    
           947,484      cycles                                                      

       0.002016698 seconds time elapsed

jirka

author
#10 lore raw
Quoted 11 lines · Jiri Olsa

On Fri, Jan 15, 2016 at 05:11:07PM -0800, Andi Kleen wrote:

SNIP

Quoted 4 lines
     1.000852081,1976560,,branch-misses,12010501884,100.00,8.32,of all branches

Available in
git://git.kernel.org/pub/scm/linux/kernel/git/ak/linux-misc-2.6 perf/stat-metrics-12

getting little tired of this:
http://marc.info/?l=linux-kernel&m=145076648115589&w=2

Sorry I thought I had fixed it, but didn't. Next time for real.

-Andi