linux-kernel - Re: [RFC PATCH v6 5/5] perf sched: prefer to use prev_state_char introduced in sched

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20230803053405.4d91015d@gandalf.local.home>
Date:   Thu, 3 Aug 2023 05:34:05 -0400
From:   Steven Rostedt <rostedt@...dmis.org>
To:     Ze Gao <zegao2021@...il.com>
Cc:     Adrian Hunter <adrian.hunter@...el.com>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Ian Rogers <irogers@...gle.com>,
        Ingo Molnar <mingo@...hat.com>, Jiri Olsa <jolsa@...nel.org>,
        Mark Rutland <mark.rutland@....com>,
        Masami Hiramatsu <mhiramat@...nel.org>,
        Namhyung Kim <namhyung@...nel.org>,
        Peter Zijlstra <peterz@...radead.org>,
        linux-kernel@...r.kernel.org, linux-perf-users@...r.kernel.org,
        linux-trace-kernel@...r.kernel.org, Ze Gao <zegao@...cent.com>
Subject: Re: [RFC PATCH v6 5/5] perf sched: prefer to use prev_state_char
 introduced in sched_switch

On Thu,  3 Aug 2023 04:33:52 -0400
Ze Gao <zegao2021@...il.com> wrote:

> Since the sched_switch tracepoint introduces a new variable to
> report sched-out task state in symbolic char, we prefer to use
> it to spare from knowing internal implementations in kernel.

The above needs to be rewritten to be more comprehensive.

Feel free to reply to this thread with new versions of the change log and
such. No need to send a new patch series to fix this before you know that
the new version is acceptable or not.

Replying to a current patch series is fine, but sending out a new one
causes the spam as it's much easier to ignore a thread than to ignore a new
thread with a new series.

Thanks,

-- Steve


> 
> Also we keep the old parsing logic intact but sync the state char
> array with the latest kernel.
> 
> Signed-off-by: Ze Gao <zegao@...cent.com>
> ---
>  tools/perf/builtin-sched.c | 19 +++++++++++++++----
>  1 file changed, 15 insertions(+), 4 deletions(-)
> 
> diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
> index 5042874ba204..6f99a36206c9 100644
> --- a/tools/perf/builtin-sched.c
> +++ b/tools/perf/builtin-sched.c
> @@ -837,7 +837,7 @@ replay_wakeup_event(struct perf_sched *sched,
>  
>  static inline char task_state_char(int state)
>  {
> -	static const char state_to_char[] = "RSDTtXZPI";
> +	static const char state_to_char[] = "RSDTtXZPIp";
>  	unsigned int bit = state ? ffs(state) : 0;
>  
>  	return bit < sizeof(state_to_char) - 1 ? state_to_char[bit] : '?';
> @@ -846,9 +846,20 @@ static inline char task_state_char(int state)
>  static inline char get_task_prev_state(struct evsel *evsel,
>  				       struct perf_sample *sample)
>  {
> -	const int prev_state = evsel__intval(evsel, sample, "prev_state");
> +	char prev_state_char;
> +	int prev_state;
>  
> -	return task_state_char(prev_state);
> +	/* prefer to use prev_state_char */
> +	if (evsel__field(evsel, "prev_state_char"))
> +		prev_state_char = (char) evsel__intval(evsel,
> +				sample, "prev_state_char");
> +	else {
> +		prev_state = (int) evsel__intval(evsel,
> +				sample, "prev_state");
> +		prev_state_char = task_state_char(prev_state);
> +	}
> +
> +	return prev_state_char;
>  }
>  
>  static int replay_switch_event(struct perf_sched *sched,
> @@ -2145,7 +2156,7 @@ static void timehist_update_runtime_stats(struct thread_runtime *r,
>  		else if (r->last_time) {
>  			u64 dt_wait = tprev - r->last_time;
>  
> -			if (r->last_state == 'R')
> +			if (r->last_state == 'R' || r->last_state == 'p')
>  				r->dt_preempt = dt_wait;
>  			else if (r->last_state == 'D')
>  				r->dt_iowait = dt_wait;