From: Yang Jihong <yangjihong@bytedance.com>
To: peterz@infradead.org, mingo@redhat.com, acme@kernel.org,
namhyung@kernel.org, mark.rutland@arm.com,
alexander.shishkin@linux.intel.com, jolsa@kernel.org,
irogers@google.com, adrian.hunter@intel.com,
kan.liang@linux.intel.com, james.clark@arm.com,
linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org
Cc: yangjihong@bytedance.com
Subject: [PATCH v2 2/2] perf sched timehist: Add --prio option
Date: Mon, 19 Aug 2024 11:30:16 +0800 [thread overview]
Message-ID: <20240819033016.2427235-3-yangjihong@bytedance.com> (raw)
In-Reply-To: <20240819033016.2427235-1-yangjihong@bytedance.com>
The --prio option is used to only show events for the given task priority(ies).
The default is to show events for all priority tasks, which is consistent with
the previous behavior.
Testcase:
# perf sched record nice -n 9 perf bench sched messaging -l 10000
# Running 'sched/messaging' benchmark:
# 20 sender and receiver processes per group
# 10 groups == 400 processes run
Total time: 3.435 [sec]
[ perf record: Woken up 270 times to write data ]
[ perf record: Captured and wrote 618.688 MB perf.data (5729036 samples) ]
# perf sched timehist -h
Usage: perf sched timehist [<options>]
-C, --cpu <cpu> list of cpus to profile
-D, --dump-raw-trace dump raw trace in ASCII
-f, --force don't complain, do it
-g, --call-graph Display call chains if present (default on)
-I, --idle-hist Show idle events only
-i, --input <file> input file name
-k, --vmlinux <file> vmlinux pathname
-M, --migrations Show migration events
-n, --next Show next task
-p, --pid <pid[,pid...]>
analyze events only for given process id(s)
-s, --summary Show only syscall summary with statistics
-S, --with-summary Show all syscalls and summary with statistics
-t, --tid <tid[,tid...]>
analyze events only for given thread id(s)
-V, --cpu-visual Add CPU visual
-v, --verbose be more verbose (show symbol address, etc)
-w, --wakeups Show wakeup events
--kallsyms <file>
kallsyms pathname
--max-stack <n> Maximum number of functions to display backtrace.
--prio <prio> analyze events only for given task priority(ies)
--show-prio Show task priority
--state Show task state when sched-out
--symfs <directory>
Look for files with symbols relative to this directory
--time <str> Time span for analysis (start,stop)
# perf sched timehist --prio 140
Samples of sched_switch event do not have callchains.
Invalid prio string
# perf sched timehist --show-prio --prio 129
Samples of sched_switch event do not have callchains.
time cpu task name prio wait time sch delay run time
[tid/pid] (msec) (msec) (msec)
--------------- ------ ------------------------------ -------- --------- --------- ---------
2090450.765421 [0002] sched-messaging[1229618] 129 0.000 0.000 0.029
2090450.765445 [0007] sched-messaging[1229616] 129 0.000 0.062 0.043
2090450.765448 [0014] sched-messaging[1229619] 129 0.000 0.000 0.032
2090450.765478 [0013] sched-messaging[1229617] 129 0.000 0.065 0.048
2090450.765503 [0014] sched-messaging[1229622] 129 0.000 0.000 0.017
2090450.765550 [0002] sched-messaging[1229624] 129 0.000 0.000 0.021
2090450.765562 [0007] sched-messaging[1229621] 129 0.000 0.071 0.028
2090450.765570 [0005] sched-messaging[1229620] 129 0.000 0.064 0.066
2090450.765583 [0001] sched-messaging[1229625] 129 0.000 0.001 0.031
2090450.765595 [0013] sched-messaging[1229623] 129 0.000 0.060 0.028
2090450.765637 [0014] sched-messaging[1229628] 129 0.000 0.000 0.019
2090450.765665 [0007] sched-messaging[1229627] 129 0.000 0.038 0.030
<SNIP>
# perf sched timehist --show-prio --prio 0,120-129
Samples of sched_switch event do not have callchains.
time cpu task name prio wait time sch delay run time
[tid/pid] (msec) (msec) (msec)
--------------- ------ ------------------------------ -------- --------- --------- ---------
2090450.763231 [0000] perf[1229608] 120 0.000 0.000 0.000
2090450.763235 [0000] migration/0[15] 0 0.000 0.001 0.003
2090450.763263 [0001] perf[1229608] 120 0.000 0.000 0.000
2090450.763268 [0001] migration/1[21] 0 0.000 0.001 0.004
2090450.763302 [0002] perf[1229608] 120 0.000 0.000 0.000
2090450.763309 [0002] migration/2[27] 0 0.000 0.001 0.007
2090450.763338 [0003] perf[1229608] 120 0.000 0.000 0.000
2090450.763343 [0003] migration/3[33] 0 0.000 0.001 0.004
2090450.763459 [0004] perf[1229608] 120 0.000 0.000 0.000
2090450.763469 [0004] migration/4[39] 0 0.000 0.002 0.010
2090450.763496 [0005] perf[1229608] 120 0.000 0.000 0.000
2090450.763501 [0005] migration/5[45] 0 0.000 0.001 0.004
2090450.763613 [0006] perf[1229608] 120 0.000 0.000 0.000
2090450.763622 [0006] migration/6[51] 0 0.000 0.001 0.008
2090450.763652 [0007] perf[1229608] 120 0.000 0.000 0.000
2090450.763660 [0007] migration/7[57] 0 0.000 0.001 0.008
<SNIP>
2090450.765665 [0001] <idle> 120 0.031 0.031 0.081
2090450.765665 [0007] sched-messaging[1229627] 129 0.000 0.038 0.030
2090450.765667 [0000] s1-perf[8235/7168] 120 0.008 0.000 0.004
2090450.765684 [0013] <idle> 120 0.028 0.028 0.088
2090450.765685 [0001] sched-messaging[1229630] 129 0.000 0.001 0.020
2090450.765688 [0000] <idle> 120 0.004 0.004 0.020
2090450.765689 [0002] <idle> 120 0.021 0.021 0.138
2090450.765691 [0005] sched-messaging[1229626] 129 0.000 0.085 0.029
Signed-off-by: Yang Jihong <yangjihong@bytedance.com>
---
tools/perf/Documentation/perf-sched.txt | 6 ++
tools/perf/builtin-sched.c | 74 ++++++++++++++++++++++++-
2 files changed, 79 insertions(+), 1 deletion(-)
diff --git a/tools/perf/Documentation/perf-sched.txt b/tools/perf/Documentation/perf-sched.txt
index 3efa5c58418d..3db64954a267 100644
--- a/tools/perf/Documentation/perf-sched.txt
+++ b/tools/perf/Documentation/perf-sched.txt
@@ -215,6 +215,12 @@ OPTIONS for 'perf sched timehist'
--show-prio::
Show task priority.
+--prio::
+ Only show events for given task priority(ies). Multiple priorities can be
+ provided as a comma-separated list with no spaces: 0,120. Ranges of
+ priorities are specified with -: 120-129. A combination of both can also be
+ provided: 0,120-129.
+
OPTIONS for 'perf sched replay'
------------------------------
diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
index cb61cbb72bfa..d619ae670cca 100644
--- a/tools/perf/builtin-sched.c
+++ b/tools/perf/builtin-sched.c
@@ -51,6 +51,7 @@
#define COMM_LEN 20
#define SYM_LEN 129
#define MAX_PID 1024000
+#define MAX_PRIO 140
static const char *cpu_list;
static DECLARE_BITMAP(cpu_bitmap, MAX_NR_CPUS);
@@ -234,6 +235,8 @@ struct perf_sched {
struct perf_time_interval ptime;
struct perf_time_interval hist_time;
volatile bool thread_funcs_exit;
+ const char *prio_str;
+ DECLARE_BITMAP(prio_bitmap, MAX_PRIO);
};
/* per thread run time data */
@@ -2504,12 +2507,33 @@ static bool timehist_skip_sample(struct perf_sched *sched,
struct perf_sample *sample)
{
bool rc = false;
+ int prio = -1;
+ struct thread_runtime *tr = NULL;
if (thread__is_filtered(thread)) {
rc = true;
sched->skipped_samples++;
}
+ if (sched->prio_str) {
+ /*
+ * Because priority may be changed during task execution,
+ * first read priority from prev sched_in event for current task.
+ * If prev sched_in event is not saved, then read priority from
+ * current task sched_out event.
+ */
+ tr = thread__get_runtime(thread);
+ if (tr && tr->prio != -1)
+ prio = tr->prio;
+ else if (evsel__name_is(evsel, "sched:sched_switch"))
+ prio = evsel__intval(evsel, sample, "prev_prio");
+
+ if (prio != -1 && !test_bit(prio, sched->prio_bitmap)) {
+ rc = true;
+ sched->skipped_samples++;
+ }
+ }
+
if (sched->idle_hist) {
if (!evsel__name_is(evsel, "sched:sched_switch"))
rc = true;
@@ -2723,7 +2747,7 @@ static int timehist_sched_change_event(const struct perf_tool *tool,
goto out;
}
- if (sched->show_prio)
+ if (sched->show_prio || sched->prio_str)
timehist_update_task_prio(evsel, sample, machine);
thread = timehist_get_thread(sched, sample, machine, evsel);
@@ -3142,6 +3166,47 @@ static int timehist_check_attr(struct perf_sched *sched,
return 0;
}
+static int timehist_parse_prio_str(struct perf_sched *sched)
+{
+ char *p;
+ unsigned long start_prio, end_prio;
+ const char *str = sched->prio_str;
+
+ if (!str)
+ return 0;
+
+ while (isdigit(*str)) {
+ p = NULL;
+ start_prio = strtoul(str, &p, 0);
+ if (start_prio >= MAX_PRIO || (*p != '\0' && *p != ',' && *p != '-'))
+ return -1;
+
+ if (*p == '-') {
+ str = ++p;
+ p = NULL;
+ end_prio = strtoul(str, &p, 0);
+
+ if (end_prio >= MAX_PRIO || (*p != '\0' && *p != ','))
+ return -1;
+
+ if (end_prio < start_prio)
+ return -1;
+ } else {
+ end_prio = start_prio;
+ }
+
+ for (; start_prio <= end_prio; start_prio++)
+ __set_bit(start_prio, sched->prio_bitmap);
+
+ if (*p)
+ ++p;
+
+ str = p;
+ }
+
+ return 0;
+}
+
static int perf_sched__timehist(struct perf_sched *sched)
{
struct evsel_str_handler handlers[] = {
@@ -3203,6 +3268,11 @@ static int perf_sched__timehist(struct perf_sched *sched)
if (timehist_check_attr(sched, evlist) != 0)
goto out;
+ if (timehist_parse_prio_str(sched) != 0) {
+ pr_err("Invalid prio string\n");
+ goto out;
+ }
+
setup_pager();
/* prefer sched_waking if it is captured */
@@ -3760,6 +3830,8 @@ int cmd_sched(int argc, const char **argv)
"analyze events only for given thread id(s)"),
OPT_STRING('C', "cpu", &cpu_list, "cpu", "list of cpus to profile"),
OPT_BOOLEAN(0, "show-prio", &sched.show_prio, "Show task priority"),
+ OPT_STRING(0, "prio", &sched.prio_str, "prio",
+ "analyze events only for given task priority(ies)"),
OPT_PARENT(sched_options)
};
--
2.25.1
next prev parent reply other threads:[~2024-08-19 3:30 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-08-19 3:30 [PATCH v2 0/2] perf sched timehist: Add --show-prio & --prio option Yang Jihong
2024-08-19 3:30 ` [PATCH v2 1/2] perf sched timehist: Add --show-prio option Yang Jihong
2024-08-19 3:30 ` Yang Jihong [this message]
2024-08-19 20:13 ` [PATCH v2 0/2] perf sched timehist: Add --show-prio & --prio option Namhyung Kim
2024-09-03 18:46 ` Arnaldo Carvalho de Melo
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=20240819033016.2427235-3-yangjihong@bytedance.com \
--to=yangjihong@bytedance.com \
--cc=acme@kernel.org \
--cc=adrian.hunter@intel.com \
--cc=alexander.shishkin@linux.intel.com \
--cc=irogers@google.com \
--cc=james.clark@arm.com \
--cc=jolsa@kernel.org \
--cc=kan.liang@linux.intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-perf-users@vger.kernel.org \
--cc=mark.rutland@arm.com \
--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;
as well as URLs for NNTP newsgroup(s).