From: Alexey Budankov <alexey.budankov@linux.intel.com>
To: Jiri Olsa <jolsa@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>,
Namhyung Kim <namhyung@kernel.org>,
Alexander Shishkin <alexander.shishkin@linux.intel.com>,
Peter Zijlstra <peterz@infradead.org>,
Ingo Molnar <mingo@redhat.com>, Andi Kleen <ak@linux.intel.com>,
linux-kernel <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v9 11/15] perf stat: implement control commands handling
Date: Tue, 7 Jul 2020 19:43:28 +0300 [thread overview]
Message-ID: <1a996f82-edce-1a27-2dda-5f162e358cfe@linux.intel.com> (raw)
In-Reply-To: <20200707160524.GA3524217@krava>
On 07.07.2020 19:05, Jiri Olsa wrote:
> On Tue, Jul 07, 2020 at 05:55:14PM +0300, Alexey Budankov wrote:
>
> SNIP
>
>> process_evlist() now looks suboptimal since record mode code directly calls evlist__ctlfd_process()
>> and then handles returned command specifically to the mode. So in v10 I replaced process_evlist()
>> call with direct evlist__ctlfd_process() call and then handling the returned command by printing
>> command msg tag and counter values in the required order. Like this:
>>
>> + clock_gettime(CLOCK_MONOTONIC, &time_start);
>> + if (!(evlist__poll(evsel_list, time_to_sleep) > 0)) { /* poll timeout or EINTR */
>> + if (timeout)
>> + break;
>> + else
>> + stop = handle_interval(interval, times);
>> + time_to_sleep = sleep_time;
>> + } else { /* fd revent */
>> + if (evlist__ctlfd_process(evsel_list, &cmd) > 0) {
>> + if (interval) {
>> + switch (cmd) {
>> + case EVLIST_CTL_CMD_ENABLE:
>> + pr_info(EVLIST_ENABLED_MSG);
>> + process_interval();
>> + break;
>> + case EVLIST_CTL_CMD_DISABLE:
>> + process_interval();
>> + pr_info(EVLIST_DISABLED_MSG);
>> + break;
>> + case EVLIST_CTL_CMD_ACK:
>> + case EVLIST_CTL_CMD_UNSUPPORTED:
>> + default:
>> + break;
>> + }
>> + }
>> + }
>> + clock_gettime(CLOCK_MONOTONIC, &time_stop);
>> + compute_tts(&time_start, &time_stop, &time_to_sleep);
>> + }
>
>
> hum, why not just get the bool from process_evlist like below?
Yes, also possible and works. However it checks twice to implement
parts of logically the same work and passes the result using extra
memory: switch/case at process_evlist(), 'if' at dispatch_events(),
dispatch_events() should also call process_interval() instead of
handle_interval() to avoid wasting of times counter for commands.
Alexey
>
> jirka
>
>
> ---
> diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> index 5021f7286422..32dd3de93f35 100644
> --- a/tools/perf/builtin-stat.c
> +++ b/tools/perf/builtin-stat.c
> @@ -485,20 +485,20 @@ static bool handle_interval(unsigned int interval, int *times)
> return false;
> }
>
> -static bool process_evlist(struct evlist *evlist, unsigned int interval, int *times)
> +static bool process_evlist(struct evlist *evlist)
> {
> - bool stop = false;
> enum evlist_ctl_cmd cmd = EVLIST_CTL_CMD_UNSUPPORTED;
> + bool display = false;
>
> if (evlist__ctlfd_process(evlist, &cmd) > 0) {
> switch (cmd) {
> case EVLIST_CTL_CMD_ENABLE:
> pr_info(EVLIST_ENABLED_MSG);
> - stop = handle_interval(interval, times);
> + display = true;
> break;
> case EVLIST_CTL_CMD_DISABLE:
> - stop = handle_interval(interval, times);
> pr_info(EVLIST_DISABLED_MSG);
> + display = true;
> break;
> case EVLIST_CTL_CMD_ACK:
> case EVLIST_CTL_CMD_UNSUPPORTED:
> @@ -507,7 +507,7 @@ static bool process_evlist(struct evlist *evlist, unsigned int interval, int *ti
> }
> }
>
> - return stop;
> + return display;
> }
>
> static void enable_counters(void)
> @@ -618,7 +618,8 @@ static int dispatch_events(bool forks, int timeout, int interval, int *times)
> stop = handle_interval(interval, times);
> time_to_sleep = sleep_time;
> } else { /* fd revent */
> - stop = process_evlist(evsel_list, interval, times);
> + if (process_evlist(evsel_list))
> + stop = handle_interval(interval, times);
> clock_gettime(CLOCK_MONOTONIC, &time_stop);
> diff_timespec(&time_diff, &time_stop, &time_start);
> time_to_sleep -= time_diff.tv_sec * MSEC_PER_SEC +
>
next prev parent reply other threads:[~2020-07-07 16:43 UTC|newest]
Thread overview: 33+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-07-03 7:38 [PATCH v9 00/15] perf: support enable and disable commands in stat and record modes Alexey Budankov
2020-07-03 7:41 ` [PATCH v9 01/15] tools/libperf: avoid fds moving by fdarray__filter() Alexey Budankov
2020-07-03 7:41 ` [PATCH v9 02/15] tools/libperf: add properties to struct pollfd *entries objects Alexey Budankov
2020-07-06 12:24 ` Jiri Olsa
2020-07-06 13:05 ` Alexey Budankov
2020-07-03 7:42 ` [PATCH v9 03/15] tools/libperf: don't count nonfilterable fds by fdarray__filter() Alexey Budankov
2020-07-03 7:43 ` [PATCH v9 04/15] perf evlist: introduce control file descriptors Alexey Budankov
2020-07-03 7:43 ` [PATCH v9 05/15] perf evlist: implement control command handling functions Alexey Budankov
2020-07-03 7:44 ` [PATCH v9 06/15] perf stat: factor out body of event handling loop for system wide Alexey Budankov
2020-07-03 7:45 ` [PATCH v9 07/15] perf stat: move target check to loop control statement Alexey Budankov
2020-07-03 7:45 ` [PATCH v9 08/15] perf stat: factor out body of event handling loop for fork case Alexey Budankov
2020-07-03 7:46 ` [PATCH v9 09/15] perf stat: factor out event handling loop into dispatch_events() Alexey Budankov
2020-07-06 12:27 ` Jiri Olsa
2020-07-06 13:07 ` Alexey Budankov
2020-07-03 7:46 ` [PATCH v9 10/15] perf stat: extend -D,--delay option with -1 value Alexey Budankov
2020-07-03 7:47 ` [PATCH v9 11/15] perf stat: implement control commands handling Alexey Budankov
2020-07-06 12:34 ` Jiri Olsa
2020-07-06 14:47 ` Alexey Budankov
2020-07-06 19:34 ` Jiri Olsa
2020-07-07 13:07 ` Alexey Budankov
2020-07-07 13:14 ` Jiri Olsa
2020-07-07 13:24 ` Alexey Budankov
2020-07-07 14:23 ` Jiri Olsa
2020-07-07 14:55 ` Alexey Budankov
2020-07-07 16:05 ` Jiri Olsa
2020-07-07 16:43 ` Alexey Budankov [this message]
2020-07-07 17:19 ` Alexey Budankov
2020-07-06 12:37 ` Jiri Olsa
2020-07-06 13:10 ` Alexey Budankov
2020-07-03 7:47 ` [PATCH v9 12/15] perf stat: introduce --control fd:ctl-fd[,ack-fd] options Alexey Budankov
2020-07-03 7:48 ` [PATCH v9 13/15] perf record: extend -D,--delay option with -1 value Alexey Budankov
2020-07-03 7:49 ` [PATCH v9 14/15] perf record: implement control commands handling Alexey Budankov
2020-07-03 7:49 ` [PATCH v9 15/15] perf record: introduce --control fd:ctl-fd[,ack-fd] options Alexey Budankov
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1a996f82-edce-1a27-2dda-5f162e358cfe@linux.intel.com \
--to=alexey.budankov@linux.intel.com \
--cc=acme@kernel.org \
--cc=ak@linux.intel.com \
--cc=alexander.shishkin@linux.intel.com \
--cc=jolsa@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=namhyung@kernel.org \
--cc=peterz@infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox