Release 4.10 tools/perf/builtin-sched.c

#include "builtin.h"
#include "perf.h"

#include "util/util.h"
#include "util/evlist.h"
#include "util/cache.h"
#include "util/evsel.h"
#include "util/symbol.h"
#include "util/thread.h"
#include "util/header.h"
#include "util/session.h"
#include "util/tool.h"
#include "util/cloexec.h"
#include "util/thread_map.h"
#include "util/color.h"
#include "util/stat.h"
#include "util/callchain.h"
#include "util/time-utils.h"

#include <subcmd/parse-options.h>
#include "util/trace-event.h"

#include "util/debug.h"

#include <linux/log2.h>
#include <sys/prctl.h>
#include <sys/resource.h>

#include <semaphore.h>
#include <pthread.h>
#include <math.h>
#include <api/fs/fs.h>
#include <linux/time64.h>


#define PR_SET_NAME		15               
/* Set process name */

#define MAX_CPUS		4096

#define COMM_LEN		20

#define SYM_LEN			129

#define MAX_PID			1024000

struct sched_atom;


struct task_desc {
	
unsigned long		nr;
	
unsigned long		pid;
	
char			comm[COMM_LEN];

	
unsigned long		nr_events;
	
unsigned long		curr_event;
	
struct sched_atom	**atoms;

	
pthread_t		thread;
	
sem_t			sleep_sem;

	
sem_t			ready_for_work;
	
sem_t			work_done_sem;

	
u64			cpu_usage;
};


enum sched_event_type {
	
SCHED_EVENT_RUN,
	
SCHED_EVENT_SLEEP,
	
SCHED_EVENT_WAKEUP,
	
SCHED_EVENT_MIGRATION,
};


struct sched_atom {
	
enum sched_event_type	type;
	
int			specific_wait;
	
u64			timestamp;
	
u64			duration;
	
unsigned long		nr;
	
sem_t			*wait_sem;
	
struct task_desc	*wakee;
};


#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"


enum thread_state {
	
THREAD_SLEEPING = 0,
	
THREAD_WAIT_CPU,
	
THREAD_SCHED_IN,
	
THREAD_IGNORE
};


struct work_atom {
	
struct list_head	list;
	
enum thread_state	state;
	
u64			sched_out_time;
	
u64			wake_up_time;
	
u64			sched_in_time;
	
u64			runtime;
};


struct work_atoms {
	
struct list_head	work_list;
	
struct thread		*thread;
	
struct rb_node		node;
	
u64			max_lat;
	
u64			max_lat_at;
	
u64			total_lat;
	
u64			nb_atoms;
	
u64			total_runtime;
	
int			num_merged;
};


typedef int (*sort_fn_t)(struct work_atoms *, struct work_atoms *);

struct perf_sched;


struct trace_sched_handler {
	
int (*switch_event)(struct perf_sched *sched, struct perf_evsel *evsel,
			    struct perf_sample *sample, struct machine *machine);

	
int (*runtime_event)(struct perf_sched *sched, struct perf_evsel *evsel,
			     struct perf_sample *sample, struct machine *machine);

	
int (*wakeup_event)(struct perf_sched *sched, struct perf_evsel *evsel,
			    struct perf_sample *sample, struct machine *machine);

	/* PERF_RECORD_FORK event, not sched_process_fork tracepoint */
	
int (*fork_event)(struct perf_sched *sched, union perf_event *event,
			  struct machine *machine);

	
int (*migrate_task_event)(struct perf_sched *sched,
				  struct perf_evsel *evsel,
				  struct perf_sample *sample,
				  struct machine *machine);
};


#define COLOR_PIDS PERF_COLOR_BLUE

#define COLOR_CPUS PERF_COLOR_BG_RED


struct perf_sched_map {
	DECLARE_BITMAP(comp_cpus_mask, MAX_CPUS);
	
int			*comp_cpus;
	
bool			 comp;
	
struct thread_map	*color_pids;
	
const char		*color_pids_str;
	
struct cpu_map		*color_cpus;
	
const char		*color_cpus_str;
	
struct cpu_map		*cpus;
	
const char		*cpus_str;
};


struct perf_sched {
	
struct perf_tool tool;
	
const char	 *sort_order;
	
unsigned long	 nr_tasks;
	
struct task_desc **pid_to_task;
	
struct task_desc **tasks;
	
const struct trace_sched_handler *tp_handler;
	
pthread_mutex_t	 start_work_mutex;
	
pthread_mutex_t	 work_done_wait_mutex;
	
int		 profile_cpu;
/*
 * Track the current task - that way we can know whether there's any
 * weird events, such as a task being switched away that is not current.
 */
	
int		 max_cpu;
	
u32		 curr_pid[MAX_CPUS];
	
struct thread	 *curr_thread[MAX_CPUS];
	
char		 next_shortname1;
	
char		 next_shortname2;
	
unsigned int	 replay_repeat;
	
unsigned long	 nr_run_events;
	
unsigned long	 nr_sleep_events;
	
unsigned long	 nr_wakeup_events;
	
unsigned long	 nr_sleep_corrections;
	
unsigned long	 nr_run_events_optimized;
	
unsigned long	 targetless_wakeups;
	
unsigned long	 multitarget_wakeups;
	
unsigned long	 nr_runs;
	
unsigned long	 nr_timestamps;
	
unsigned long	 nr_unordered_timestamps;
	
unsigned long	 nr_context_switch_bugs;
	
unsigned long	 nr_events;
	
unsigned long	 nr_lost_chunks;
	
unsigned long	 nr_lost_events;
	
u64		 run_measurement_overhead;
	
u64		 sleep_measurement_overhead;
	
u64		 start_time;
	
u64		 cpu_usage;
	
u64		 runavg_cpu_usage;
	
u64		 parent_cpu_usage;
	
u64		 runavg_parent_cpu_usage;
	
u64		 sum_runtime;
	
u64		 sum_fluct;
	
u64		 run_avg;
	
u64		 all_runtime;
	
u64		 all_count;
	
u64		 cpu_last_switched[MAX_CPUS];
	


struct rb_root	 atom_root, sorted_atom_root, merged_atom_root;
	

struct list_head sort_list, cmp_pid;
	
bool force;
	
bool skip_merge;
	
struct perf_sched_map map;

	/* options for timehist command */
	
bool		summary;
	
bool		summary_only;
	
bool		idle_hist;
	
bool		show_callchain;
	
unsigned int	max_stack;
	
bool		show_cpu_visual;
	
bool		show_wakeups;
	
bool		show_migrations;
	
u64		skipped_samples;
	
const char	*time_str;
	
struct perf_time_interval ptime;
	
struct perf_time_interval hist_time;
};

/* per thread run time data */

struct thread_runtime {
	
u64 last_time;      /* time of previous sched in/out event */
	
u64 dt_run;         /* run time */
	
u64 dt_wait;        /* time between CPU access (off cpu) */
	
u64 dt_delay;       /* time between wakeup and sched-in */
	
u64 ready_to_run;   /* time of wakeup */

	
struct stats run_stats;
	
u64 total_run_time;

	
u64 migrations;
};

/* per event run time data */

struct evsel_runtime {
	
u64 *last_time; /* time this event was last seen per cpu */
	
u32 ncpu;       /* highest cpu slot allocated */
};

/* per cpu idle time data */

struct idle_thread_runtime {
	
struct thread_runtime	tr;
	
struct thread		*last_thread;
	
struct rb_root		sorted_root;
	
struct callchain_root	callchain;
	
struct callchain_cursor	cursor;
};

/* track idle times per cpu */

static struct thread **idle_threads;

static int idle_max_cpu;

static char idle_comm[] = "<idle>";



static u64 get_nsecs(void)
{
	struct timespec ts;

	clock_gettime(CLOCK_MONOTONIC, &ts);

	return ts.tv_sec * NSEC_PER_SEC + ts.tv_nsec;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
24
77.42%
2
66.67%
ingo molnar
ingo molnar
7
22.58%
1
33.33%
Total
31
100.00%
3
100.00%



static void burn_nsecs(struct perf_sched *sched, u64 nsecs)
{
	u64 T0 = get_nsecs(), T1;

	do {
		T1 = get_nsecs();
	} while (T1 + sched->run_measurement_overhead < T0 + nsecs);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
37
86.05%
1
50.00%
ingo molnar
ingo molnar
6
13.95%
1
50.00%
Total
43
100.00%
2
100.00%



static void sleep_nsecs(u64 nsecs)
{
	struct timespec ts;

	ts.tv_nsec = nsecs % 999999999;
	ts.tv_sec = nsecs / 999999999;

	nanosleep(&ts, NULL);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
37
100.00%
1
100.00%
Total
37
100.00%
1
100.00%



static void calibrate_run_measurement_overhead(struct perf_sched *sched)
{
	u64 T0, T1, delta, min_delta = NSEC_PER_SEC;
	int i;

	for (i = 0; i < 10; i++) {
		T0 = get_nsecs();
		burn_nsecs(sched, 0);
		T1 = get_nsecs();
		delta = T1-T0;
		min_delta = min(min_delta, delta);
	}
	sched->run_measurement_overhead = min_delta;

	printf("run measurement overhead: %" PRIu64 " nsecs\n", min_delta);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
62
71.26%
2
66.67%
ingo molnar
ingo molnar
25
28.74%
1
33.33%
Total
87
100.00%
3
100.00%



static void calibrate_sleep_measurement_overhead(struct perf_sched *sched)
{
	u64 T0, T1, delta, min_delta = NSEC_PER_SEC;
	int i;

	for (i = 0; i < 10; i++) {
		T0 = get_nsecs();
		sleep_nsecs(10000);
		T1 = get_nsecs();
		delta = T1-T0;
		min_delta = min(min_delta, delta);
	}
	min_delta -= 10000;
	sched->sleep_measurement_overhead = min_delta;

	printf("sleep measurement overhead: %" PRIu64 " nsecs\n", min_delta);
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
54
60.67%
2
50.00%
ingo molnar
ingo molnar
35
39.33%
2
50.00%
Total
89
100.00%
4
100.00%


static struct sched_atom *

get_new_event(struct task_desc *task, u64 timestamp)
{
	struct sched_atom *event = zalloc(sizeof(*event));
	unsigned long idx = task->nr_events;
	size_t size;

	event->timestamp = timestamp;
	event->nr = idx;

	task->nr_events++;
	size = sizeof(struct sched_atom *) * task->nr_events;
	task->atoms = realloc(task->atoms, size);
	BUG_ON(!task->atoms);

	task->atoms[idx] = event;

	return event;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
71
68.27%
1
25.00%
ingo molnar
ingo molnar
33
31.73%
3
75.00%
Total
104
100.00%
4
100.00%



static struct sched_atom *last_event(struct task_desc *task)
{
	if (!task->nr_events)
		return NULL;

	return task->atoms[task->nr_events - 1];
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
28
80.00%
1
50.00%
ingo molnar
ingo molnar
7
20.00%
1
50.00%
Total
35
100.00%
2
100.00%



static void add_sched_event_run(struct perf_sched *sched, struct task_desc *task,
				u64 timestamp, u64 duration)
{
	struct sched_atom *event, *curr_event = last_event(task);

	/*
         * optimize an existing RUN event by merging this one
         * to it:
         */
	if (curr_event && curr_event->type == SCHED_EVENT_RUN) {
		sched->nr_run_events_optimized++;
		curr_event->duration += duration;
		return;
	}

	event = get_new_event(task, timestamp);

	event->type = SCHED_EVENT_RUN;
	event->duration = duration;

	sched->nr_run_events++;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
69
80.23%
1
33.33%
ingo molnar
ingo molnar
15
17.44%
1
33.33%
xiao guangrong
xiao guangrong
2
2.33%
1
33.33%
Total
86
100.00%
3
100.00%



static void add_sched_event_wakeup(struct perf_sched *sched, struct task_desc *task,
				   u64 timestamp, struct task_desc *wakee)
{
	struct sched_atom *event, *wakee_event;

	event = get_new_event(task, timestamp);
	event->type = SCHED_EVENT_WAKEUP;
	event->wakee = wakee;

	wakee_event = last_event(wakee);
	if (!wakee_event || wakee_event->type != SCHED_EVENT_SLEEP) {
		sched->targetless_wakeups++;
		return;
	}
	if (wakee_event->wait_sem) {
		sched->multitarget_wakeups++;
		return;
	}

	wakee_event->wait_sem = zalloc(sizeof(*wakee_event->wait_sem));
	sem_init(wakee_event->wait_sem, 0, 0);
	wakee_event->specific_wait = 1;
	event->wait_sem = wakee_event->wait_sem;

	sched->nr_wakeup_events++;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
120
86.96%
1
33.33%
xiao guangrong
xiao guangrong
11
7.97%
1
33.33%
ingo molnar
ingo molnar
7
5.07%
1
33.33%
Total
138
100.00%
3
100.00%



static void add_sched_event_sleep(struct perf_sched *sched, struct task_desc *task,
				  u64 timestamp, u64 task_state __maybe_unused)
{
	struct sched_atom *event = get_new_event(task, timestamp);

	event->type = SCHED_EVENT_SLEEP;

	sched->nr_sleep_events++;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
37
80.43%
1
33.33%
ingo molnar
ingo molnar
9
19.57%
2
66.67%
Total
46
100.00%
3
100.00%



static struct task_desc *register_pid(struct perf_sched *sched,
				      unsigned long pid, const char *comm)
{
	struct task_desc *task;
	static int pid_max;

	if (sched->pid_to_task == NULL) {
		if (sysctl__read_int("kernel/pid_max", &pid_max) < 0)
			pid_max = MAX_PID;
		BUG_ON((sched->pid_to_task = calloc(pid_max, sizeof(struct task_desc *))) == NULL);
	}
	if (pid >= (unsigned long)pid_max) {
		BUG_ON((sched->pid_to_task = realloc(sched->pid_to_task, (pid + 1) *
			sizeof(struct task_desc *))) == NULL);
		while (pid >= (unsigned long)pid_max)
			sched->pid_to_task[pid_max++] = NULL;
	}

	task = sched->pid_to_task[pid];

	if (task)
		return task;

	task = zalloc(sizeof(*task));
	task->pid = pid;
	task->nr = sched->nr_tasks;
	strcpy(task->comm, comm);
	/*
         * every task starts in sleeping state - this gets ignored
         * if there's no wakeup pointing to this sleep state:
         */
	add_sched_event_sleep(sched, task, 0, 0);

	sched->pid_to_task[pid] = task;
	sched->nr_tasks++;
	sched->tasks = realloc(sched->tasks, sched->nr_tasks * sizeof(struct task_desc *));
	BUG_ON(!sched->tasks);
	sched->tasks[task->nr] = task;

	if (verbose)
		printf("registered task #%ld, PID %ld (%s)\n", sched->nr_tasks, pid, comm);

	return task;
}


Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
136
48.57%
2
28.57%
yunlong song
yunlong song
112
40.00%
3
42.86%
ingo molnar
ingo molnar
28
10.00%
1
14.29%
xiao guangrong
xiao guangrong
4
1.43%
1
14.29%
Total
280
100.00%
7
100.00%



static void print_task_traces(struct perf_sched *sched)
{
	struct task_desc *task;
	unsigned long i;

	for (i = 0; i < sched->nr_tasks; i++) {
		task = sched->tasks[i];
		printf("task %6ld (%20s:%10ld), nr_events: %ld\n",
			task->nr, task->comm, task->pid, task->nr_events);
	}
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
41
61.19%
1
33.33%
ingo molnar
ingo molnar
25
37.31%
1
33.33%
xiao guangrong
xiao guangrong
1
1.49%
1
33.33%
Total
67
100.00%
3
100.00%



static void add_cross_task_wakeups(struct perf_sched *sched)
{
	struct task_desc *task1, *task2;
	unsigned long i, j;

	for (i = 0; i < sched->nr_tasks; i++) {
		task1 = sched->tasks[i];
		j = i + 1;
		if (j == sched->nr_tasks)
			j = 0;
		task2 = sched->tasks[j];
		add_sched_event_wakeup(sched, task1, 0, task2);
	}
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
ingo molnar
ingo molnar
47
52.81%
1
50.00%
arnaldo carvalho de melo
arnaldo carvalho de melo
42
47.19%
1
50.00%
Total
89
100.00%
2
100.00%



static void perf_sched__process_event(struct perf_sched *sched,
				      struct sched_atom *atom)
{
	int ret = 0;

	switch (atom->type) {
		case SCHED_EVENT_RUN:
			burn_nsecs(sched, atom->duration);
			break;
		case SCHED_EVENT_SLEEP:
			if (atom->wait_sem)
				ret = sem_wait(atom->wait_sem);
			BUG_ON(ret);
			break;
		case SCHED_EVENT_WAKEUP:
			if (atom->wait_sem)
				ret = sem_post(atom->wait_sem);
			BUG_ON(ret);
			break;
		case SCHED_EVENT_MIGRATION:
			break;
		default:
			BUG_ON(1);
	}
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
64
64.00%
1
50.00%
ingo molnar
ingo molnar
36
36.00%
1
50.00%
Total
100
100.00%
2
100.00%



static u64 get_cpu_usage_nsec_parent(void)
{
	struct rusage ru;
	u64 sum;
	int err;

	err = getrusage(RUSAGE_SELF, &ru);
	BUG_ON(err);

	sum =  ru.ru_utime.tv_sec * NSEC_PER_SEC + ru.ru_utime.tv_usec * NSEC_PER_USEC;
	sum += ru.ru_stime.tv_sec * NSEC_PER_SEC + ru.ru_stime.tv_usec * NSEC_PER_USEC;

	return sum;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
62
86.11%
2
66.67%
ingo molnar
ingo molnar
10
13.89%
1
33.33%
Total
72
100.00%
3
100.00%



static int self_open_counters(struct perf_sched *sched, unsigned long cur_task)
{
	struct perf_event_attr attr;
	char sbuf[STRERR_BUFSIZE], info[STRERR_BUFSIZE];
	int fd;
	struct rlimit limit;
	bool need_privilege = false;

	memset(&attr, 0, sizeof(attr));

	attr.type = PERF_TYPE_SOFTWARE;
	attr.config = PERF_COUNT_SW_TASK_CLOCK;

force_again:
	fd = sys_perf_event_open(&attr, 0, -1, -1,
				 perf_event_open_cloexec_flag());

	if (fd < 0) {
		if (errno == EMFILE) {
			if (sched->force) {
				BUG_ON(getrlimit(RLIMIT_NOFILE, &limit) == -1);
				limit.rlim_cur += sched->nr_tasks - cur_task;
				if (limit.rlim_cur > limit.rlim_max) {
					limit.rlim_max = limit.rlim_cur;
					need_privilege = true;
				}
				if (setrlimit(RLIMIT_NOFILE, &limit) == -1) {
					if (need_privilege && errno == EPERM)
						strcpy(info, "Need privilege\n");
				} else
					goto force_again;
			} else
				strcpy(info, "Have a try with -f option\n");
		}
		pr_err("Error: sys_perf_event_open() syscall returned "
		       "with %d (%s)\n%s", fd,
		       str_error_r(errno, sbuf, sizeof(sbuf)), info);
		exit(EXIT_FAILURE);
	}
	return fd;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
yunlong song
yunlong song
140
60.34%
2
25.00%
arnaldo carvalho de melo
arnaldo carvalho de melo
61
26.29%
2
25.00%
ingo molnar
ingo molnar
14
6.03%
1
12.50%
masami hiramatsu
masami hiramatsu
13
5.60%
1
12.50%
namhyung kim
namhyung kim
2
0.86%
1
12.50%
yann droneaud
yann droneaud
2
0.86%
1
12.50%
Total
232
100.00%
8
100.00%



static u64 get_cpu_usage_nsec_self(int fd)
{
	u64 runtime;
	int ret;

	ret = read(fd, &runtime, sizeof(runtime));
	BUG_ON(ret != sizeof(runtime));

	return runtime;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
42
97.67%
1
50.00%
ingo molnar
ingo molnar
1
2.33%
1
50.00%
Total
43
100.00%
2
100.00%


struct sched_thread_parms {
	
struct task_desc  *task;
	
struct perf_sched *sched;
	
int fd;
};



static void *thread_func(void *ctx)
{
	struct sched_thread_parms *parms = ctx;
	struct task_desc *this_task = parms->task;
	struct perf_sched *sched = parms->sched;
	u64 cpu_usage_0, cpu_usage_1;
	unsigned long i, ret;
	char comm2[22];
	int fd = parms->fd;

	zfree(&parms);

	sprintf(comm2, ":%s", this_task->comm);
	prctl(PR_SET_NAME, comm2);
	if (fd < 0)
		return NULL;
again:
	ret = sem_post(&this_task->ready_for_work);
	BUG_ON(ret);
	ret = pthread_mutex_lock(&sched->start_work_mutex);
	BUG_ON(ret);
	ret = pthread_mutex_unlock(&sched->start_work_mutex);
	BUG_ON(ret);

	cpu_usage_0 = get_cpu_usage_nsec_self(fd);

	for (i = 0; i < this_task->nr_events; i++) {
		this_task->curr_event = i;
		perf_sched__process_event(sched, this_task->atoms[i]);
	}

	cpu_usage_1 = get_cpu_usage_nsec_self(fd);
	this_task->cpu_usage = cpu_usage_1 - cpu_usage_0;
	ret = sem_post(&this_task->work_done_sem);
	BUG_ON(ret);

	ret = pthread_mutex_lock(&sched->work_done_wait_mutex);
	BUG_ON(ret);
	ret = pthread_mutex_unlock(&sched->work_done_wait_mutex);
	BUG_ON(ret);

	goto again;
}

Contributors
Person
Tokens
Prop
Commits
CommitProp
arnaldo carvalho de melo
arnaldo carvalho de melo
177
72.24%
2
50.00%
ingo molnar
ingo molnar
64
26.12%
1
25.00%
yunlong song
yunlong song
4
1.63%
1
25.00%
Total
245
100.00%
4
100.00%



static void create_tasks(struct perf_sched *sched)
{
	struct task_desc *task;
	pthread_attr_t attr;
	unsigned long i;
	int err;

	err = pthread_attr_init(&attr);
	BUG_ON(err);
	err = pthread_attr_setstacksize(&attr,
			(size_t) max(16 * 1024, PTHREAD_STACK_MIN));
	BUG_ON(err);
	err = pthread_mutex_lock(&sched->start_work_mutex);
	BUG_ON(err);
	err = pthread_mutex_lock(&sched->work_done_wait_mutex);
	BUG_ON(err);
	for (i = 0; i < sched->nr_tasks; i++) {
		struct sched_thread_parms *parms = malloc(sizeof(*parms));