Message-ID: <20230926042938.509234-4-changbin.du@huawei.com>
Date:   Tue, 26 Sep 2023 12:29:36 +0800
From:   Changbin Du <changbin.du@...wei.com>
To:     Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>,
        Arnaldo Carvalho de Melo <acme@...nel.org>
CC:     Mark Rutland <mark.rutland@....com>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Jiri Olsa <jolsa@...nel.org>,
        Namhyung Kim <namhyung@...nel.org>,
        Ian Rogers <irogers@...gle.com>,
        Adrian Hunter <adrian.hunter@...el.com>,
        <linux-perf-users@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        <changbin.du@...il.com>, Changbin Du <changbin.du@...wei.com>,
        kernel test robot <yujie.liu@...el.com>
Subject: [PATCH v5 3/5] perf: add new option '--workload-config' to set workload sched_policy/prio/cpumask

To get consistent benchmarking results, sometimes we need to set the
sched_policy/priority/cpumask of the workload to reduce system noise.

For example, CPU binding is required on big.little system.

  $ perf stat -- taskset -c 0 ls

However, the events of 'taskset' itself are also counted here. To get more
accurate result, this should be avoided.

To get away of the middleman, this adds a new option '--workload-config' to
do the same jobs for stat and record commands.

  --workload-config <[sched_policy=policy][,sched_prio=priority][,cpu-list=list]>
        setup target workload (the <command>) attributes:

		sched_policy: other|fifo|rr|batch|idle
                sched_prio: scheduling priority for fifo|rr, nice value for other
                cpu-list: CPU affinity. e.g. 1-3:5 is processors #1, #2, #3 and #5

For example,

  $ sudo perf stat --workload-config sched_policy=fifo,sched_prio=40,cpu-list=0-3:7 -- ls

Above command will make 'ls' run on CPU #0-#3 and #7 with fifo scheduler and
realtime priority is 40.

Cc: kernel test robot <yujie.liu@...el.com>
Signed-off-by: Changbin Du <changbin.du@...wei.com>

---
v2: Use cpu list spec instead of cpu mask number.
v3:
  o rename '--workload-attr' as '--workload-config'
  o transform to key-value style option
---
 tools/perf/Documentation/perf-record.txt |   7 ++
 tools/perf/Documentation/perf-stat.txt   |   6 ++
 tools/perf/builtin-record.c              |  27 ++++++
 tools/perf/builtin-stat.c                |  19 ++++
 tools/perf/util/evlist.c                 | 108 +++++++++++++++++++++++
 tools/perf/util/evlist.h                 |   3 +
 tools/perf/util/target.h                 |   9 ++
 7 files changed, 179 insertions(+)

diff --git a/tools/perf/Documentation/perf-record.txt b/tools/perf/Documentation/perf-record.txt
index d5217be012d7..da4692751e17 100644
--- a/tools/perf/Documentation/perf-record.txt
+++ b/tools/perf/Documentation/perf-record.txt
@@ -821,6 +821,13 @@ filtered through the mask provided by -C option.
 	only, as of now.  So the applications built without the frame
 	pointer might see bogus addresses.
 
+--workload-config <[sched_policy=policy][,sched_prio=priority][,cpu-list=list]>::
+	setup target workload (the <command>) attributes:
+
+	  sched_policy: other|fifo|rr|batch|idle
+	  sched_prio: scheduling priority for fifo|rr, nice value for other
+	  cpu-list: CPU affinity. e.g. 1-3:5 is processors #1, #2, #3 and #5
+
 include::intel-hybrid.txt[]
 
 SEE ALSO
diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
index 8f789fa1242e..b2038f7e236a 100644
--- a/tools/perf/Documentation/perf-stat.txt
+++ b/tools/perf/Documentation/perf-stat.txt
@@ -262,6 +262,12 @@ disable events during measurements:
  wait -n ${perf_pid}
  exit $?
 
+--workload-config <[sched_policy=policy][,sched_prio=priority][,cpu-list=list]>::
+	setup target workload (the <command>) attributes:
+
+	  sched_policy: other|fifo|rr|batch|idle
+	  sched_prio: scheduling priority for fifo|rr, nice value for other
+	  cpu-list: CPU affinity. e.g. 1-3:5 is processors #1, #2, #3 and #5
 
 --pre::
 --post::
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 34bb31f08bb5..20799a1e60f6 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -3277,6 +3277,17 @@ static int parse_record_synth_option(const struct option *opt,
 	return 0;
 }
 
+static int record_parse_workload_attr_opt(const struct option *opt,
+					  const char *arg,
+					  int unset __maybe_unused)
+{
+	struct record_opts *opts = opt->value;
+
+	return evlist__parse_workload_config(arg, &opts->target.workload.sched_policy,
+					     &opts->target.workload.sched_priority,
+					     &opts->target.workload.cpu_map);
+}
+
 /*
  * XXX Ideally would be local to cmd_record() and passed to a record__new
  * because we need to have access to it in record__exit, that is called
@@ -3297,6 +3308,8 @@ static struct record record = {
 		.target		     = {
 			.uses_mmap   = true,
 			.default_per_cpu = true,
+			.workload.sched_policy = -1,
+			.workload.sched_priority = 0,
 		},
 		.mmap_flush          = MMAP_FLUSH_DEFAULT,
 		.nr_threads_synthesize = 1,
@@ -3321,6 +3334,12 @@ static struct record record = {
 const char record_callchain_help[] = CALLCHAIN_RECORD_HELP
 	"\n\t\t\t\tDefault: fp";
 
+const char record_workload_config_help[] =
+	"setup target workload (the <command>) attributes:\n\n"
+	HELP_PAD "sched_policy: other|fifo|rr|batch|idle\n"
+	HELP_PAD "sched_prio: scheduling priority for fifo|rr, nice value for other\n"
+	HELP_PAD "cpu-list: CPU affinity. e.g. 1-3:5 is processors #1, #2, #3 and #5";
+
 static bool dry_run;
 
 static struct parse_events_option_args parse_events_option_args = {
@@ -3535,6 +3554,10 @@ static struct option __record_options[] = {
 			    "write collected trace data into several data files using parallel threads",
 			    record__parse_threads),
 	OPT_BOOLEAN(0, "off-cpu", &record.off_cpu, "Enable off-cpu analysis"),
+	OPT_CALLBACK(0, "workload-config", &record.opts,
+		     "[sched_policy=policy][,sched_prio=priority][,cpu-list=list]",
+		     record_workload_config_help,
+		     &record_parse_workload_attr_opt),
 	OPT_END()
 };
 
@@ -4221,6 +4244,10 @@ int cmd_record(int argc, const char **argv)
 	record__free_thread_masks(rec, rec->nr_threads);
 	rec->nr_threads = 0;
 	evlist__close_control(rec->opts.ctl_fd, rec->opts.ctl_fd_ack, &rec->opts.ctl_fd_close);
+	if (rec->opts.target.workload.cpu_map) {
+		perf_cpu_map__put(rec->opts.target.workload.cpu_map);
+		rec->opts.target.workload.cpu_map = NULL;
+	}
 	return err;
 }
 
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 07b48f6df48e..a7a3a788e7d9 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -108,6 +108,8 @@ static bool all_counters_use_bpf = true;
 
 static struct target target = {
 	.uid	= UINT_MAX,
+	.workload.sched_policy = -1,
+	.workload.sched_priority = 0,
 };
 
 #define METRIC_ONLY_LEN 20
@@ -1160,6 +1162,14 @@ static int parse_cache_level(const struct option *opt,
 	return 0;
 }
 
+static int parse_workload_attr_opt(const struct option *opt __maybe_unused, const char *arg,
+				   int unset __maybe_unused)
+{
+	return evlist__parse_workload_config(arg, &target.workload.sched_policy,
+					     &target.workload.sched_priority,
+					     &target.workload.cpu_map);
+}
+
 static struct option stat_options[] = {
 	OPT_BOOLEAN('T', "transaction", &transaction_run,
 		    "hardware transaction statistics"),
@@ -1220,6 +1230,10 @@ static struct option stat_options[] = {
 	OPT_BOOLEAN(0, "append", &append_file, "append to the output file"),
 	OPT_INTEGER(0, "log-fd", &output_fd,
 		    "log output to fd, instead of stderr"),
+	OPT_CALLBACK(0, "workload-config", &stat_config,
+		     "[sched_policy=policy][,sched_prio=priority][,cpu-list=list]",
+		     record_workload_config_help,
+		     &parse_workload_attr_opt),
 	OPT_STRING(0, "pre", &pre_cmd, "command",
 			"command to run prior to the measured command"),
 	OPT_STRING(0, "post", &post_cmd, "command",
@@ -2893,5 +2907,10 @@ int cmd_stat(int argc, const char **argv)
 	metricgroup__rblist_exit(&stat_config.metric_events);
 	evlist__close_control(stat_config.ctl_fd, stat_config.ctl_fd_ack, &stat_config.ctl_fd_close);
 
+	if (target.workload.cpu_map) {
+		perf_cpu_map__put(target.workload.cpu_map);
+		target.workload.cpu_map = NULL;
+	}
+
 	return status;
 }
diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
index 7ef43f72098e..7ad7a4fed282 100644
--- a/tools/perf/util/evlist.c
+++ b/tools/perf/util/evlist.c
@@ -33,6 +33,7 @@
 #include "util/bpf-filter.h"
 #include "util/stat.h"
 #include "util/util.h"
+#include "util/parse-sublevel-options.h"
 #include <signal.h>
 #include <unistd.h>
 #include <sched.h>
@@ -46,6 +47,7 @@
 #include <sys/mman.h>
 #include <sys/prctl.h>
 #include <sys/timerfd.h>
+#include <sys/resource.h>
 
 #include <linux/bitops.h>
 #include <linux/hash.h>
@@ -1398,6 +1400,109 @@ int evlist__open(struct evlist *evlist)
 	return err;
 }
 
+int evlist__parse_workload_config(const char *str, int *sched_policy, int *sched_priority,
+				  struct perf_cpu_map **cpu_map)
+{
+	char *policy_str = NULL;
+	int priority = -1;
+	char *cpu_list = NULL;
+	int ret;
+	struct sublevel_option workload_conf_opts[] = {
+		{ .name = "sched_policy",	.str_ptr = &policy_str},
+		{ .name = "sched_prio",		.value_ptr = &priority},
+		{ .name = "cpu-list",		.str_ptr = &cpu_list},
+		{ .name = NULL, }
+	};
+
+	ret = perf_parse_sublevel_options(str, workload_conf_opts);
+	if (ret)
+		return ret;
+
+	/* sched policy, default to 'other'. */
+	if (!policy_str || !strncmp(policy_str, "other", sizeof("other")))
+		*sched_policy = SCHED_OTHER;
+	else if (!strncmp(policy_str, "fifo", sizeof("fifo")))
+		*sched_policy = SCHED_FIFO;
+	else if (!strncmp(policy_str, "rr", sizeof("rr")))
+		*sched_policy = SCHED_RR;
+	else if (!strncmp(policy_str, "batch", sizeof("batch")))
+		*sched_policy = SCHED_BATCH;
+	else if (!strncmp(policy_str, "idle", sizeof("idle")))
+		*sched_policy = SCHED_IDLE;
+	else {
+		pr_err("workload_attr: unknown sched policy %s\n", policy_str);
+		ret = -EINVAL;
+		goto out;
+	}
+
+	/* check sched priority and set default value */
+	if (*sched_policy == SCHED_FIFO || *sched_policy == SCHED_RR) {
+		if (priority == -1)
+			priority = 99; /* default to lowest priority */
+		else if (priority < 1 || priority > 99) {
+			pr_err("workload_attr: invalid priority %d for fifo and rr, allowed [1,99]\n",
+				priority);
+			ret = -EINVAL;
+			goto out;
+		}
+	} else if (*sched_policy == SCHED_OTHER && priority == -1)
+		priority = 0;
+	*sched_priority = priority;
+
+	/* allowed cpu list */
+	*cpu_map = __perf_cpu_map__new(cpu_list, ':');
+	if (!*cpu_map) {
+		pr_err("workload_attr: failed to get cpus map from %s\n", cpu_list);
+		ret = -EINVAL;
+	}
+
+out:
+	free(policy_str);
+	free(cpu_list);
+	return ret;
+}
+
+static int configurate_workload(struct target *target)
+{
+	struct sched_param param;
+	int policy = target->workload.sched_policy;
+	int priority = target->workload.sched_priority;
+
+	if (policy >= 0) {
+		param.sched_priority = (policy == SCHED_FIFO || policy == SCHED_RR) ?
+					priority : 0;
+		if (sched_setscheduler(0, policy, &param) != 0) {
+			pr_err("failed to set the sched policy %d: %s\n", policy, strerror(errno));
+			return -1;
+		}
+
+		if (policy == SCHED_OTHER) {
+			if (setpriority(PRIO_PROCESS, 0, priority) != 0) {
+				pr_err("failed to set the nice value %d: %s\n", priority, strerror(errno));
+				return -1;
+			}
+		}
+	}
+
+	if (target->workload.cpu_map) {
+		size_t cpuset_size = -1;
+		cpu_set_t *cpu_set;
+
+		cpu_set = perf_cpu_map__2_cpuset(target->workload.cpu_map, &cpuset_size);
+		if (!cpu_set)
+			return -1;
+
+		if (sched_setaffinity(0, cpuset_size, cpu_set) != 0) {
+			pr_err("failed to set the sched affinity: %s\n", strerror(errno));
+			CPU_FREE(cpu_set);
+			return -1;
+		}
+		CPU_FREE(cpu_set);
+	}
+
+	return 0;
+}
+
 int evlist__prepare_workload(struct evlist *evlist, struct target *target, const char *argv[],
 			     bool pipe_output, void (*exec_error)(int signo, siginfo_t *info, void *ucontext))
 {
@@ -1464,6 +1569,9 @@ int evlist__prepare_workload(struct evlist *evlist, struct target *target, const
 			exit(ret);
 		}
 
+		if (configurate_workload(target) != 0)
+			exit(-1);
+
 		execvp(argv[0], (char **)argv);
 
 		if (exec_error) {
diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h
index 664c6bf7b3e0..540e17d0d9fe 100644
--- a/tools/perf/util/evlist.h
+++ b/tools/perf/util/evlist.h
@@ -15,6 +15,7 @@
 #include <pthread.h>
 #include <signal.h>
 #include <unistd.h>
+#include <sched.h>
 
 struct pollfd;
 struct thread_map;
@@ -180,6 +181,8 @@ void evlist__set_id_pos(struct evlist *evlist);
 void evlist__config(struct evlist *evlist, struct record_opts *opts, struct callchain_param *callchain);
 int record_opts__config(struct record_opts *opts);
 
+int evlist__parse_workload_config(const char *str, int *sched_policy, int *sched_priority,
+				  struct perf_cpu_map **cpu_set);
 int evlist__prepare_workload(struct evlist *evlist, struct target *target,
 			     const char *argv[], bool pipe_output,
 			     void (*exec_error)(int signo, siginfo_t *info, void *ucontext));
diff --git a/tools/perf/util/target.h b/tools/perf/util/target.h
index d582cae8e105..78b7e7ab1c7b 100644
--- a/tools/perf/util/target.h
+++ b/tools/perf/util/target.h
@@ -4,6 +4,7 @@
 
 #include <stdbool.h>
 #include <sys/types.h>
+#include <sched.h>
 
 struct target {
 	const char   *pid;
@@ -19,6 +20,12 @@ struct target {
 	bool	     use_bpf;
 	int	     initial_delay;
 	const char   *attr_map;
+
+	struct {
+		int	 sched_policy;
+		int	 sched_priority;
+		struct perf_cpu_map *cpu_map;
+	} workload;
 };
 
 enum target_errno {
@@ -103,4 +110,6 @@ static inline bool target__uses_dummy_map(struct target *target)
 	return use_dummy;
 }
 
+extern const char record_workload_config_help[];
+
 #endif /* _PERF_TARGET_H */
-- 
2.25.1

Powered by blists - more mailing lists