Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230726121618.19198-1-zegao@tencent.com> <20230726121618.19198-2-zegao@tencent.com>
 <20230731113807.1a4a455c@gandalf.local.home>
In-Reply-To: <20230731113807.1a4a455c@gandalf.local.home>
From:   Ze Gao <zegao2021@gmail.com>
Date:   Tue, 1 Aug 2023 09:35:58 +0800
Message-ID: <CAD8CoPAym8fuL50vevxBxPy6-RU4kU49KV+Sz4OO0+Xh_doDCQ@mail.gmail.com>
Subject: Re: [RFC PATCH v2 1/3] sched, tracing: add to report task state in
 symbolic chars
To:     Steven Rostedt <rostedt@goodmis.org>
Cc:     Peter Zijlstra <peterz@infradead.org>,
        Namhyung Kim <namhyung@kernel.org>,
        Adrian Hunter <adrian.hunter@intel.com>,
        Alexander Shishkin <alexander.shishkin@linux.intel.com>,
        Arnaldo Carvalho de Melo <acme@kernel.org>,
        Ian Rogers <irogers@google.com>,
        Ingo Molnar <mingo@redhat.com>, Jiri Olsa <jolsa@kernel.org>,
        Mark Rutland <mark.rutland@arm.com>,
        Masami Hiramatsu <mhiramat@kernel.org>,
        linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org,
        linux-trace-kernel@vger.kernel.org,
        linux-trace-devel@vger.kernel.org, Ze Gao <zegao@tencent.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Thanks Steven,

Indeed this did not consider the struct alignment. I left the field order i=
ntact
because I was not sure whether the order here matters for reporting or pars=
ing.
Now I will fix this and send a v3, and again thanks for pointing it out.

Thanks,
Ze

On Mon, Jul 31, 2023 at 11:38=E2=80=AFPM Steven Rostedt <rostedt@goodmis.or=
g> wrote:
>
> On Wed, 26 Jul 2023 20:16:16 +0800
> Ze Gao <zegao2021@gmail.com> wrote:
>
> > @@ -231,41 +253,29 @@ TRACE_EVENT(sched_switch,
> >       TP_STRUCT__entry(
> >               __array(        char,   prev_comm,      TASK_COMM_LEN   )
> >               __field(        pid_t,  prev_pid                        )
> > -             __field(        int,    prev_prio                       )
> > -             __field(        long,   prev_state                      )
> > +             __field(        short,  prev_prio                       )
> > +             __field(        int,    prev_state                      )
> > +             __field(        char,   prev_state_char                 )
> >               __array(        char,   next_comm,      TASK_COMM_LEN   )
> >               __field(        pid_t,  next_pid                        )
> > -             __field(        int,    next_prio                       )
> > +             __field(        short,  next_prio                       )
> >       ),
>
> The above adds a bunch of holes. This needs to be reordered to condense t=
he
> event, we don't want to increase it. libtraceevent will handle reordering=
.
>
> The above produces:
>
> struct {
>         char    prev_comm[16];
>         pid_t   prev_pid;
>         short   prev_prio; <-- 2 character padding
>         int     prev_state;
>         char    prev_state_char;
>         char    next_comm[16]; <- 3 character padding
>         pid_t   next_pid;
>         short   next_prio; <- 2 char padding
> };
>
> (all events are at least 4 byte aligned, and are multiple of 4 bytes in
> size, thus that last short of next_prio did nothing)
>
> The above is a total of 56 bytes (note, that is the same as the current
> sched_switch event size);
>
> What the above should be:
>
>         TP_STRUCT__entry(
>                 __field(        pid_t,  prev_pid                        )
>                 __field(        pid_t,  next_pid                        )
>                 __field(        short,  prev_prio                       )
>                 __field(        short,  next_prio                       )
>                 __field(        int,    prev_state                      )
>                 __array(        char,   prev_comm,      TASK_COMM_LEN   )
>                 __array(        char,   next_comm,      TASK_COMM_LEN   )
>                 __field(        char,   prev_state_char                 )
>         ),
>
>
> Which would be:
>
> struct {
>         pid_t   prev_pid;
>         pid_t   next_pid;
>         short   prev_prio;
>         short   next_prio;
>         int     prev_state;
>         char    prev_comm[16];
>         char    next_comm[16];
>         char    prev_stat_char; <-- 3 characters of padding
> }
>
> which would be 52 byte. Saving us 4 bytes per event. Which is a big deal!
>
> -- Steve
>