[PATCH] trace-cmd: Add record --clock option to specify a trace clock

[Ian Webster]

This change adds a --clock option to trace-cmd record. It simply writes
trace_clock on debugfs. Examples of valid choices on most systems are: local,
global, counter, and any other choice compatible with ftrace.

This complements Yoshihiro Yunomae's change 8c9867c (trace-cmd: Add support for
extracting trace_clock in report).

Signed-off-by: Ian Webster <i...@planetaryresources.com>
---
Documentation/trace-cmd-record.1.txt | 5 +++++
trace-record.c | 19 +++++++++++++++++++
trace-usage.c | 1 +
3 files changed, 25 insertions(+)

diff --git a/Documentation/trace-cmd-record.1.txt b/Documentation/trace-cmd-record.1.txt
index 832a257..201be2b 100644
--- a/Documentation/trace-cmd-record.1.txt
+++ b/Documentation/trace-cmd-record.1.txt
@@ -240,6 +240,11 @@ OPTIONS
timestamp to gettimeofday which will allow wall time output from the
timestamps reading the created 'trace.dat' file.

+*--clock* 'clock_name'::
+ When an event is recorded into the ring buffer, a timestamp is added. This
+ stamp comes from the specified clock. Ftrace uses the "local" clock by
+ default. Other common clocks include: global, counter, uptime, perf, x86-tsc.
+
EXAMPLES
--------

diff --git a/trace-record.c b/trace-record.c
index 0199627..18de026 100644
--- a/trace-record.c
+++ b/trace-record.c
@@ -399,6 +399,20 @@ static int set_ftrace(int set, int use_proc)
return 0;
}

+static void set_trace_clock(char *clock_choice)
+{
+ FILE *fp;
+ char *path;
+
+ path = tracecmd_get_tracing_file("trace_clock");
+ fp = fopen(path, "w");
+ if (!fp)
+ die("writing to '%s'", path);
+ tracecmd_put_tracing_file(path);
+ fwrite(clock_choice, 1, strlen(clock_choice), fp);
+ fclose(fp);
+}
+
static char *
get_instance_file(struct buffer_instance *instance, const char *file)
{
@@ -2279,6 +2293,7 @@ static void record_all_events(void)
}

enum {
+ OPT_clock = 252,
OPT_nosplice = 253,
OPT_funcstack = 254,
OPT_date = 255,
@@ -2351,6 +2366,7 @@ void trace_record (int argc, char **argv)
{"func-stack", no_argument, NULL, OPT_funcstack},
{"nosplice", no_argument, NULL, OPT_nosplice},
{"help", no_argument, NULL, '?'},
+ {"clock", required_argument, NULL, OPT_clock},
{NULL, 0, NULL, 0}
};

@@ -2530,6 +2546,9 @@ void trace_record (int argc, char **argv)
case OPT_nosplice:
recorder_flags |= TRACECMD_RECORD_NOSPLICE;
break;
+ case OPT_clock:
+ set_trace_clock(optarg);
+ break;
default:
usage(argv);
}
diff --git a/trace-usage.c b/trace-usage.c
index b8f26e6..54dbb85 100644
--- a/trace-usage.c
+++ b/trace-usage.c
@@ -46,6 +46,7 @@ static struct usage_help usage_help[] = {
" -i do not fail if an event is not found\n"
" --func-stack perform a stack trace for function tracer\n"
" (use with caution)\n"
+ " --clock change the method used to produce trace timestamps\n"
},
{
"start",
--
1.7.9.5


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

[Steven Rostedt]

Should be const char *clock_choice.

> +{
> + FILE *fp;
> + char *path;
> +
> + path = tracecmd_get_tracing_file("trace_clock");
> + fp = fopen(path, "w");
> + if (!fp)
> + die("writing to '%s'", path);
> + tracecmd_put_tracing_file(path);
> + fwrite(clock_choice, 1, strlen(clock_choice), fp);
> + fclose(fp);

Whitespace damage here.

I like the change, but the patch needs to be fixed. The tabs turned to
spaces and the clock_choice above needs to be const.

Thanks,

-- Steve

[Ian Webster]

This change adds a --clock option to trace-cmd record. It simply writes
trace_clock on debugfs. Examples of valid choices on most systems are: local,
global, counter, and any other choice compatible with ftrace.

This complements Yoshihiro Yunomae's change 8c9867c (trace-cmd: Add support for
extracting trace_clock in report).

Signed-off-by: Ian Webster <i...@planetaryresources.com>

---

Changes per Steven's comments:
* add missing const
* preserve tab whitespace

Documentation/trace-cmd-record.1.txt | 5 +++++
trace-record.c | 19 +++++++++++++++++++
trace-usage.c | 1 +
3 files changed, 25 insertions(+)

diff --git a/Documentation/trace-cmd-record.1.txt b/Documentation/trace-cmd-record.1.txt

index 832a257..5e83454 100644

--- a/Documentation/trace-cmd-record.1.txt
+++ b/Documentation/trace-cmd-record.1.txt
@@ -240,6 +240,11 @@ OPTIONS
timestamp to gettimeofday which will allow wall time output from the
timestamps reading the created 'trace.dat' file.

+*--clock* 'clock_name'::
+ When an event is recorded into the ring buffer, a timestamp is added. This
+ stamp comes from the specified clock. Ftrace uses the "local" clock by
+ default. Other common clocks include: global, counter, uptime, perf, x86-tsc.
+
EXAMPLES
--------

diff --git a/trace-record.c b/trace-record.c

index 0199627..2143080 100644

--- a/trace-record.c
+++ b/trace-record.c
@@ -399,6 +399,20 @@ static int set_ftrace(int set, int use_proc)
return 0;
}

+static void set_trace_clock(const char *clock_choice)

+{
+ FILE *fp;
+ char *path;
+
+ path = tracecmd_get_tracing_file("trace_clock");
+ fp = fopen(path, "w");
+ if (!fp)
+ die("writing to '%s'", path);
+ tracecmd_put_tracing_file(path);
+ fwrite(clock_choice, 1, strlen(clock_choice), fp);
+ fclose(fp);

[Yoshihiro YUNOMAE]

Hi Ian,

(2013/12/13 8:30), Ian Webster wrote:
> This change adds a --clock option to trace-cmd record. It simply writes
> trace_clock on debugfs. Examples of valid choices on most systems are: local,
> global, counter, and any other choice compatible with ftrace.

Nice work!
Please see my following comments.

Please add an error check for fwrite(). If clock_choice is invalid,
trace-cmd would better output the error report and die.

> + fclose(fp);
> +}
> +
> static char *
> get_instance_file(struct buffer_instance *instance, const char *file)
> {
> @@ -2279,6 +2293,7 @@ static void record_all_events(void)
> }
>
> enum {
> + OPT_clock = 252,
> OPT_nosplice = 253,
> OPT_funcstack = 254,
> OPT_date = 255,
> @@ -2351,6 +2366,7 @@ void trace_record (int argc, char **argv)
> {"func-stack", no_argument, NULL, OPT_funcstack},
> {"nosplice", no_argument, NULL, OPT_nosplice},
> {"help", no_argument, NULL, '?'},
> + {"clock", required_argument, NULL, OPT_clock},
> {NULL, 0, NULL, 0}
> };
>
> @@ -2530,6 +2546,9 @@ void trace_record (int argc, char **argv)
> case OPT_nosplice:
> recorder_flags |= TRACECMD_RECORD_NOSPLICE;
> break;
> + case OPT_clock:
> + set_trace_clock(optarg);
> + break;

This implementation does not support multiple buffers.
trace_clock had better be changed after sub-buffers are created, I
think.

Example:
trace-cmd record -e sched --clock local -B foo -e block --clock x86-tsc

Thanks,
Yoshihiro YUNOMAE

--
Yoshihiro YUNOMAE
Software Platform Research Dept. Linux Technology Center
Hitachi, Ltd., Yokohama Research Laboratory
E-mail: yoshihiro....@hitachi.com

[Steven Rostedt]

On Mon, 16 Dec 2013 13:10:52 +0900
Yoshihiro YUNOMAE <yoshihiro....@hitachi.com> wrote:

> Hi Ian,
>
> (2013/12/13 8:30), Ian Webster wrote:
> > This change adds a --clock option to trace-cmd record. It simply writes
> > trace_clock on debugfs. Examples of valid choices on most systems are: local,
> > global, counter, and any other choice compatible with ftrace.
>
> Nice work!
> Please see my following comments.

Hmm, I was waiting for a reply back with an update against these
comments and never saw one. I eventually forgot about this patch :-(

I went to add it now, and noticed that I already added a -C option for
trace-clock which also works for instances.

Sorry about forgetting about this patch (just found it buried in my
inbox).

I'll be releasing 2.6 soon, with a bunch of new updates.

-- Steve