[PATCH] perf tools: Handle sparse CPU numbers correctly

From: Paul Mackerras
Date: Wed Mar 10 2010 - 04:36:21 EST


At present, the perf subcommands that do system-wide monitoring
(perf stat, perf record and perf top) don't work properly unless
the online cpus are numbered 0, 1, ..., N-1. These tools ask for
the number of online cpus with sysconf(_SC_NPROCESSORS_ONLN) and
then try to create events for cpus 0, 1, ..., N-1.

This creates problems for systems where the online cpus are
numbered sparsely. For example, a POWER6 system in single-threaded
mode (i.e. only running 1 hardware thread per core) will have only
even-numbered cpus online.

This fixes the problem by reading the /sys/devices/system/cpu/online
file to find out which cpus are online. The code that does that is
in tools/perf/util/cpumap.[ch], and consists of a read_cpu_map()
function that sets up a cpumap[] array and returns the number of
online cpus. If /sys/devices/system/cpu/online can't be read or
can't be parsed successfully, it falls back to using sysconf to ask
how many cpus are online and sets up an identity map in cpumap[].

The perf record, perf stat and perf top code then calls read_cpu_map()
in the system-wide monitoring case (instead of sysconf) and uses
cpumap[] to get the cpu numbers to pass to perf_event_open.

Signed-off-by: Paul Mackerras <paulus@xxxxxxxxx>
---
I think this approach is a bit better than my previous approach of
just keeping on incrementing the cpu number.

tools/perf/Makefile | 2 +
tools/perf/builtin-record.c | 7 ++---
tools/perf/builtin-stat.c | 10 ++++---
tools/perf/builtin-top.c | 9 +++---
tools/perf/util/cpumap.c | 59 +++++++++++++++++++++++++++++++++++++++++++
tools/perf/util/cpumap.h | 7 +++++
6 files changed, 81 insertions(+), 13 deletions(-)
create mode 100644 tools/perf/util/cpumap.c
create mode 100644 tools/perf/util/cpumap.h

diff --git a/tools/perf/Makefile b/tools/perf/Makefile
index 54a5b50..aeb9aac 100644
--- a/tools/perf/Makefile
+++ b/tools/perf/Makefile
@@ -387,6 +387,7 @@ LIB_H += util/thread.h
LIB_H += util/trace-event.h
LIB_H += util/probe-finder.h
LIB_H += util/probe-event.h
+LIB_H += util/cpumap.h

LIB_OBJS += util/abspath.o
LIB_OBJS += util/alias.o
@@ -433,6 +434,7 @@ LIB_OBJS += util/sort.o
LIB_OBJS += util/hist.o
LIB_OBJS += util/probe-event.o
LIB_OBJS += util/util.o
+LIB_OBJS += util/cpumap.o

BUILTIN_OBJS += builtin-annotate.o

diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 771533c..cb28224 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -22,6 +22,7 @@
#include "util/debug.h"
#include "util/session.h"
#include "util/symbol.h"
+#include "util/cpumap.h"

#include <unistd.h>
#include <sched.h>
@@ -418,9 +419,6 @@ static int __cmd_record(int argc, const char **argv)
char buf;

page_size = sysconf(_SC_PAGE_SIZE);
- nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
- assert(nr_cpus <= MAX_NR_CPUS);
- assert(nr_cpus >= 0);

atexit(sig_atexit);
signal(SIGCHLD, sig_handler);
@@ -544,8 +542,9 @@ static int __cmd_record(int argc, const char **argv)
if ((!system_wide && !inherit) || profile_cpu != -1) {
open_counters(profile_cpu, target_pid);
} else {
+ nr_cpus = read_cpu_map();
for (i = 0; i < nr_cpus; i++)
- open_counters(i, target_pid);
+ open_counters(cpumap[i], target_pid);
}

if (file_new) {
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index e8c85d5..95db31c 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -45,6 +45,7 @@
#include "util/event.h"
#include "util/debug.h"
#include "util/header.h"
+#include "util/cpumap.h"

#include <sys/prctl.h>
#include <math.h>
@@ -151,7 +152,7 @@ static void create_perf_stat_counter(int counter, int pid)
unsigned int cpu;

for (cpu = 0; cpu < nr_cpus; cpu++) {
- fd[cpu][counter] = sys_perf_event_open(attr, -1, cpu, -1, 0);
+ fd[cpu][counter] = sys_perf_event_open(attr, -1, cpumap[cpu], -1, 0);
if (fd[cpu][counter] < 0 && verbose)
fprintf(stderr, ERR_PERF_OPEN, counter,
fd[cpu][counter], strerror(errno));
@@ -519,9 +520,10 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used)
nr_counters = ARRAY_SIZE(default_attrs);
}

- nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
- assert(nr_cpus <= MAX_NR_CPUS);
- assert((int)nr_cpus >= 0);
+ if (system_wide)
+ nr_cpus = read_cpu_map();
+ else
+ nr_cpus = 1;

/*
* We dont want to block the signals - that would cause
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 31f2e59..0b719e3 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -28,6 +28,7 @@
#include <linux/rbtree.h>
#include "util/parse-options.h"
#include "util/parse-events.h"
+#include "util/cpumap.h"

#include "util/debug.h"

@@ -1123,7 +1124,7 @@ static void start_counter(int i, int counter)

cpu = profile_cpu;
if (target_pid == -1 && profile_cpu == -1)
- cpu = i;
+ cpu = cpumap[i];

attr = attrs + counter;

@@ -1347,12 +1348,10 @@ int cmd_top(int argc, const char **argv, const char *prefix __used)
attrs[counter].sample_period = default_interval;
}

- nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
- assert(nr_cpus <= MAX_NR_CPUS);
- assert(nr_cpus >= 0);
-
if (target_pid != -1 || profile_cpu != -1)
nr_cpus = 1;
+ else
+ nr_cpus = read_cpu_map();

get_term_dimensions(&winsize);
if (print_entries == 0) {
diff --git a/tools/perf/util/cpumap.c b/tools/perf/util/cpumap.c
new file mode 100644
index 0000000..4e01490
--- /dev/null
+++ b/tools/perf/util/cpumap.c
@@ -0,0 +1,59 @@
+#include "util.h"
+#include "../perf.h"
+#include "cpumap.h"
+#include <assert.h>
+#include <stdio.h>
+
+int cpumap[MAX_NR_CPUS];
+
+static int default_cpu_map(void)
+{
+ int nr_cpus, i;
+
+ nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
+ assert(nr_cpus <= MAX_NR_CPUS);
+ assert((int)nr_cpus >= 0);
+
+ for (i = 0; i < nr_cpus; ++i)
+ cpumap[i] = i;
+
+ return nr_cpus;
+}
+
+int read_cpu_map(void)
+{
+ FILE *onlnf;
+ int nr_cpus = 0;
+ int n, cpu, prev;
+ char sep;
+
+ onlnf = fopen("/sys/devices/system/cpu/online", "r");
+ if (!onlnf)
+ return default_cpu_map();
+
+ sep = 0;
+ prev = -1;
+ for (;;) {
+ n = fscanf(onlnf, "%u%c", &cpu, &sep);
+ if (n <= 0)
+ break;
+ if (prev >= 0) {
+ assert(nr_cpus + cpu - prev - 1 < MAX_NR_CPUS);
+ while (++prev < cpu)
+ cpumap[nr_cpus++] = prev;
+ }
+ assert (nr_cpus < MAX_NR_CPUS);
+ cpumap[nr_cpus++] = cpu;
+ if (n == 2 && sep == '-')
+ prev = cpu;
+ else
+ prev = -1;
+ if (n == 1 || sep == '\n')
+ break;
+ }
+ fclose(onlnf);
+ if (nr_cpus > 0)
+ return nr_cpus;
+
+ return default_cpu_map();
+}
diff --git a/tools/perf/util/cpumap.h b/tools/perf/util/cpumap.h
new file mode 100644
index 0000000..86c78bb
--- /dev/null
+++ b/tools/perf/util/cpumap.h
@@ -0,0 +1,7 @@
+#ifndef __PERF_CPUMAP_H
+#define __PERF_CPUMAP_H
+
+extern int read_cpu_map(void);
+extern int cpumap[];
+
+#endif /* __PERF_CPUMAP_H */
--
1.6.5.3.171.ge36e

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/