2009-06-27 18:19:56

by Jaswinder Singh Rajput

Subject: [PATCH -tip] perf stat: memcpy is only required if no event is selected and !null_run

Set attrs and nr_counters if no event is selected and !null_run

Setting of attrs should depend on number of counters,
so we need to memcpy only for sizeof(default_attrs)

and set nr_counters as ARRAY_SIZE(default_attrs) in place of hardcoded value

Signed-off-by: Jaswinder Singh Rajput <[email protected]>
---
tools/perf/builtin-stat.c | 11 ++++++-----
1 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 3840a70..3e5ea4e 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -46,7 +46,7 @@
#include <sys/prctl.h>
#include <math.h>

-static struct perf_counter_attr default_attrs[MAX_COUNTERS] = {
+static struct perf_counter_attr default_attrs[] = {

{ .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_TASK_CLOCK },
{ .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_CONTEXT_SWITCHES},
@@ -477,16 +477,17 @@ int cmd_stat(int argc, const char **argv, const char *prefix)
{
int status;

- memcpy(attrs, default_attrs, sizeof(attrs));
-
argc = parse_options(argc, argv, options, stat_usage, 0);
if (!argc)
usage_with_options(stat_usage, options);
if (run_count <= 0 || run_count > MAX_RUN)
usage_with_options(stat_usage, options);

- if (!null_run && !nr_counters)
- nr_counters = 8;
+ /* Set attrs and nr_counters if no event is selected and !null_run */
+ if (!null_run && !nr_counters) {
+ memcpy(attrs, default_attrs, sizeof(default_attrs));
+ nr_counters = ARRAY_SIZE(default_attrs);
+ }

nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
assert(nr_cpus <= MAX_NR_CPUS);
--
1.6.0.6

2009-06-28 05:32:55

by Jaswinder Singh Rajput

Subject: Re: [PATCH -tip] perf stat: memcpy is only required if no event is selected and !null_run

If You want can also add Impact Line.

Impact: Speedup, reduce binary size.


On Sat, 2009-06-27 at 23:49 +0530, Jaswinder Singh Rajput wrote:
> Set attrs and nr_counters if no event is selected and !null_run
>
> Setting of attrs should depend on number of counters,
> so we need to memcpy only for sizeof(default_attrs)
>
> and set nr_counters as ARRAY_SIZE(default_attrs) in place of hardcoded value
>
> Signed-off-by: Jaswinder Singh Rajput <[email protected]>
> ---
> tools/perf/builtin-stat.c | 11 ++++++-----
> 1 files changed, 6 insertions(+), 5 deletions(-)
>
> diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> index 3840a70..3e5ea4e 100644
> --- a/tools/perf/builtin-stat.c
> +++ b/tools/perf/builtin-stat.c
> @@ -46,7 +46,7 @@
> #include <sys/prctl.h>
> #include <math.h>
>
> -static struct perf_counter_attr default_attrs[MAX_COUNTERS] = {
> +static struct perf_counter_attr default_attrs[] = {
>
> { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_TASK_CLOCK },
> { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_CONTEXT_SWITCHES},
> @@ -477,16 +477,17 @@ int cmd_stat(int argc, const char **argv, const char *prefix)
> {
> int status;
>
> - memcpy(attrs, default_attrs, sizeof(attrs));
> -
> argc = parse_options(argc, argv, options, stat_usage, 0);
> if (!argc)
> usage_with_options(stat_usage, options);
> if (run_count <= 0 || run_count > MAX_RUN)
> usage_with_options(stat_usage, options);
>
> - if (!null_run && !nr_counters)
> - nr_counters = 8;
> + /* Set attrs and nr_counters if no event is selected and !null_run */
> + if (!null_run && !nr_counters) {
> + memcpy(attrs, default_attrs, sizeof(default_attrs));
> + nr_counters = ARRAY_SIZE(default_attrs);
> + }
>
> nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
> assert(nr_cpus <= MAX_NR_CPUS);

2009-06-28 13:43:40

by Jaswinder Singh Rajput

Subject: [tip:perfcounters/urgent] perf stat: Micro-optimize the code: memcpy is only required if no event is selected and !null_run

Commit-ID: c3043569dc8fbe9228b76174f15d1a7152c48a20
Gitweb: http://git.kernel.org/tip/c3043569dc8fbe9228b76174f15d1a7152c48a20
Author: Jaswinder Singh Rajput <[email protected]>
AuthorDate: Sat, 27 Jun 2009 23:49:09 +0530
Committer: Ingo Molnar <[email protected]>
CommitDate: Sun, 28 Jun 2009 15:22:47 +0200

perf stat: Micro-optimize the code: memcpy is only required if no event is selected and !null_run

Set attrs and nr_counters if no event is selected and !null_run.

Setting of attrs should depend on number of counters,
so we need to memcpy only for sizeof(default_attrs)

Also set nr_counters as ARRAY_SIZE(default_attrs) in place of
hardcoded value.

Signed-off-by: Jaswinder Singh Rajput <[email protected]>
Cc: Peter Zijlstra <[email protected]>
LKML-Reference: <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>


---
tools/perf/builtin-stat.c | 11 ++++++-----
1 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 3840a70..3e5ea4e 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -46,7 +46,7 @@
#include <sys/prctl.h>
#include <math.h>

-static struct perf_counter_attr default_attrs[MAX_COUNTERS] = {
+static struct perf_counter_attr default_attrs[] = {

{ .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_TASK_CLOCK },
{ .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_CONTEXT_SWITCHES},
@@ -477,16 +477,17 @@ int cmd_stat(int argc, const char **argv, const char *prefix)
{
int status;

- memcpy(attrs, default_attrs, sizeof(attrs));
-
argc = parse_options(argc, argv, options, stat_usage, 0);
if (!argc)
usage_with_options(stat_usage, options);
if (run_count <= 0 || run_count > MAX_RUN)
usage_with_options(stat_usage, options);

- if (!null_run && !nr_counters)
- nr_counters = 8;
+ /* Set attrs and nr_counters if no event is selected and !null_run */
+ if (!null_run && !nr_counters) {
+ memcpy(attrs, default_attrs, sizeof(default_attrs));
+ nr_counters = ARRAY_SIZE(default_attrs);
+ }

nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
assert(nr_cpus <= MAX_NR_CPUS);