perf bench sched pipe: Add -G/--cgroups option

The -G/--cgroups option is to put sender and receiver in different
cgroups in order to measure cgroup context switch overheads.

Users need to make sure the cgroups exist and accessible. The following
example should the effect of this change. Please don't forget taskset
before the perf bench to measure cgroup switches properly. Otherwise
each task would run on a different CPU and generate cgroup switches
regardless of this change.

# perf stat -e context-switches,cgroup-switches \
> taskset -c 0 perf bench sched pipe -l 10000 > /dev/null

Performance counter stats for 'taskset -c 0 perf bench sched pipe -l 10000':

20,001 context-switches
2 cgroup-switches

0.053449651 seconds time elapsed

0.011286000 seconds user
0.041869000 seconds sys

# perf stat -e context-switches,cgroup-switches \
> taskset -c 0 perf bench sched pipe -l 10000 -G AAA,BBB > /dev/null

Performance counter stats for 'taskset -c 0 perf bench sched pipe -l 10000 -G AAA,BBB':

20,001 context-switches
20,001 cgroup-switches

0.052768627 seconds time elapsed

0.006284000 seconds user
0.046266000 seconds sys

Signed-off-by: Namhyung Kim <[e-mail]>
Tested-by: Arnaldo Carvalho de Melo <[e-mail]>
Link: https://lore.kernel.org/r/[e-mail]

tools/perf/Documentation/perf-bench.txt

···
 --loop=::
 Specify number of loops.
 
+-G::
+--cgroups=::
+Names of cgroups for sender and receiver, separated by a comma.
+This is useful to check cgroup context switching overhead.
+Note that perf doesn't create nor delete the cgroups, so users should
+make sure that the cgroups exist and are accessible before use.
+
+
 Example of *pipe*
 ^^^^^^^^^^^^^^^^^
 
···
         Total time:0.016 sec
                 16.948000 usecs/op
                 59004 ops/sec
+
+% perf bench sched pipe -G AAA,BBB
+(executing 1000000 pipe operations between cgroups)
+# Running 'sched/pipe' benchmark:
+# Executed 1000000 pipe operations between two processes
+
+     Total time: 6.886 [sec]
+
+       6.886208 usecs/op
+         145217 ops/sec
+
 ---------------------
 
 SUITES FOR 'syscall'

tools/perf/bench/sched-pipe.c

···
  * Ported to perf by Hitoshi Mitake <[e-mail]>
  */
 #include <subcmd/parse-options.h>
+#include <api/fs/fs.h>
 #include "bench.h"
+#include "util/cgroup.h"
 
 #include <unistd.h>
 #include <stdio.h>
···
 #include <sys/wait.h>
 #include <string.h>
 #include <errno.h>
+#include <fcntl.h>
 #include <assert.h>
 #include <sys/time.h>
 #include <sys/types.h>
···
 	int			nr;
 	int			pipe_read;
 	int			pipe_write;
+	bool			cgroup_failed;
 	pthread_t		pthread;
 };
 
···
 /* Use processes by default: */
 static bool			threaded;
 
+static char			*cgrp_names[2];
+static struct cgroup		*cgrps[2];
+
+static int parse_two_cgroups(const struct option *opt __maybe_unused,
+			     const char *str, int unset __maybe_unused)
+{
+	char *p = strdup(str);
+	char *q;
+	int ret = -1;
+
+	if (p == NULL) {
+		fprintf(stderr, "memory allocation failure\n");
+		return -1;
+	}
+
+	q = strchr(p, ',');
+	if (q == NULL) {
+		fprintf(stderr, "it should have two cgroup names: %s\n", p);
+		goto out;
+	}
+	*q = '\0';
+
+	cgrp_names[0] = strdup(p);
+	cgrp_names[1] = strdup(q + 1);
+
+	if (cgrp_names[0] == NULL || cgrp_names[1] == NULL) {
+		fprintf(stderr, "memory allocation failure\n");
+		goto out;
+	}
+	ret = 0;
+
+out:
+	free(p);
+	return ret;
+}
+
 static const struct option options[] = {
 	OPT_INTEGER('l', "loop",	&loops,		"Specify number of loops"),
 	OPT_BOOLEAN('T', "threaded",	&threaded,	"Specify threads/process based task setup"),
+	OPT_CALLBACK('G', "cgroups", NULL, "SEND,RECV",
+		     "Put sender and receivers in given cgroups",
+		     parse_two_cgroups),
 	OPT_END()
 };
 
···
 	NULL
 };
 
+static int enter_cgroup(int nr)
+{
+	char buf[32];
+	int fd, len, ret;
+	int saved_errno;
+	struct cgroup *cgrp;
+	pid_t pid;
+
+	if (cgrp_names[nr] == NULL)
+		return 0;
+
+	if (cgrps[nr] == NULL) {
+		cgrps[nr] = cgroup__new(cgrp_names[nr], /*do_open=*/true);
+		if (cgrps[nr] == NULL)
+			goto err;
+	}
+	cgrp = cgrps[nr];
+
+	if (threaded)
+		pid = syscall(__NR_gettid);
+	else
+		pid = getpid();
+
+	snprintf(buf, sizeof(buf), "%d\n", pid);
+	len = strlen(buf);
+
+	/* try cgroup v2 interface first */
+	if (threaded)
+		fd = openat(cgrp->fd, "cgroup.threads", O_WRONLY);
+	else
+		fd = openat(cgrp->fd, "cgroup.procs", O_WRONLY);
+
+	/* try cgroup v1 if failed */
+	if (fd < 0 && errno == ENOENT)
+		fd = openat(cgrp->fd, "tasks", O_WRONLY);
+
+	if (fd < 0)
+		goto err;
+
+	ret = write(fd, buf, len);
+	close(fd);
+
+	if (ret != len) {
+		printf("Cannot enter to cgroup: %s\n", cgrp->name);
+		return -1;
+	}
+	return 0;
+
+err:
+	saved_errno = errno;
+	printf("Failed to open cgroup file in %s\n", cgrp_names[nr]);
+
+	if (saved_errno == ENOENT) {
+		char mnt[PATH_MAX];
+
+		if (cgroupfs_find_mountpoint(mnt, sizeof(mnt), "perf_event") == 0)
+			printf(" Hint: create the cgroup first, like 'mkdir %s/%s'\n",
+			       mnt, cgrp_names[nr]);
+	} else if (saved_errno == EACCES && geteuid() > 0) {
+		printf(" Hint: try to run as root\n");
+	}
+
+	return -1;
+}
+
+static void exit_cgroup(int nr)
+{
+	cgroup__put(cgrps[nr]);
+	free(cgrp_names[nr]);
+}
+
 static void *worker_thread(void *__tdata)
 {
 	struct thread_data *td = __tdata;
 	int m = 0, i;
 	int ret;
+
+	ret = enter_cgroup(td->nr);
+	if (ret < 0) {
+		td->cgroup_failed = true;
+		return NULL;
+	}
 
 	for (i = 0; i < loops; i++) {
 		if (!td->nr) {
···
 
 int bench_sched_pipe(int argc, const char **argv)
 {
-	struct thread_data threads[2], *td;
+	struct thread_data threads[2] = {};
+	struct thread_data *td;
 	int pipe_1[2], pipe_2[2];
 	struct timeval start, stop, diff;
 	unsigned long long result_usec = 0;
···
 		}
 	}
 
-
 	if (threaded) {
-
 		for (t = 0; t < nr_threads; t++) {
 			td = threads + t;
 
···
 			ret = pthread_join(td->pthread, NULL);
 			BUG_ON(ret);
 		}
-
 	} else {
 		pid = fork();
 		assert(pid >= 0);
···
 
 	gettimeofday(&stop, NULL);
 	timersub(&stop, &start, &diff);
+
+	exit_cgroup(0);
+	exit_cgroup(1);
+
+	if (threads[0].cgroup_failed || threads[1].cgroup_failed)
+		return 0;
 
 	switch (bench_format) {
 	case BENCH_FORMAT_DEFAULT: