Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit 39aeb52f authored by mingo's avatar mingo Committed by Ingo Molnar
Browse files

perf sched: Add support for sched:sched_stat_runtime events 



This allows more precise 'perf sched latency' output:

 ---------------------------------------------------------------------------------------
  Task                  |  Runtime ms | Switches | Average delay ms | Maximum delay ms |
 ---------------------------------------------------------------------------------------
  ksoftirqd/0-4         |    0.010 ms |        2 | avg:    2.476 ms | max:    2.977 ms |
  perf-12328            |   15.844 ms |       66 | avg:    1.118 ms | max:    9.979 ms |
  bdi-default-235       |    0.009 ms |        1 | avg:    0.998 ms | max:    0.998 ms |
  events/1-8            |    0.020 ms |        2 | avg:    0.998 ms | max:    0.998 ms |
  events/0-7            |    0.018 ms |        2 | avg:    0.992 ms | max:    0.996 ms |
  sleep-12329           |    0.742 ms |        3 | avg:    0.906 ms | max:    2.289 ms |
  sshd-12122            |    0.163 ms |        2 | avg:    0.283 ms | max:    0.562 ms |
  loop-getpid-lon-12322 | 1023.636 ms |       69 | avg:    0.208 ms | max:    5.996 ms |
  loop-getpid-lon-12321 | 1038.638 ms |        5 | avg:    0.073 ms | max:    0.171 ms |
  migration/1-5         |    0.000 ms |        1 | avg:    0.006 ms | max:    0.006 ms |
 ---------------------------------------------------------------------------------------
  TOTAL:                | 2079.078 ms |      153 |
 -------------------------------------------------

Also, streamline the code a bit more, add asserts for various state
machine failures (they should be debugged if they occur) and fix
a few odd ends.

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
parent 08f69e6c

tools/perf/builtin-sched.c

+173 −91
Original line number Original line Diff line number Diff line
@@ -50,7 +50,7 @@ static u64 sleep_measurement_overhead; 




static unsigned long		nr_tasks;

static unsigned long		nr_tasks;





struct sched_event;

struct sched_atom;





struct task_desc {

struct task_desc {

	unsigned long		nr;

	unsigned long		nr;
@@ -59,7 +59,7 @@ struct task_desc { 




	unsigned long		nr_events;

	unsigned long		nr_events;

	unsigned long		curr_event;

	unsigned long		curr_event;

	struct sched_event	**events;

	struct sched_atom	**atoms;





	pthread_t		thread;

	pthread_t		thread;

	sem_t			sleep_sem;

	sem_t			sleep_sem;
@@ -76,7 +76,7 @@ enum sched_event_type { 
	SCHED_EVENT_WAKEUP,

	SCHED_EVENT_WAKEUP,

};

};





struct sched_event {

struct sched_atom {

	enum sched_event_type	type;

	enum sched_event_type	type;

	u64			timestamp;

	u64			timestamp;

	u64			duration;

	u64			duration;
@@ -137,8 +137,8 @@ struct work_atom { 
	u64			runtime;

	u64			runtime;

};

};





struct task_atoms {

struct work_atoms {

	struct list_head	atom_list;

	struct list_head	work_list;

	struct thread		*thread;

	struct thread		*thread;

	struct rb_node		node;

	struct rb_node		node;

	u64			max_lat;

	u64			max_lat;
@@ -147,7 +147,7 @@ struct task_atoms { 
	u64			total_runtime;

	u64			total_runtime;

};

};





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

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





static struct rb_root		atom_root, sorted_atom_root;

static struct rb_root		atom_root, sorted_atom_root;
@@ -220,10 +220,10 @@ static void calibrate_sleep_measurement_overhead(void) 
	printf("sleep measurement overhead: %Ld nsecs\n", min_delta);

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

}

}





static struct sched_event *

static struct sched_atom *

get_new_event(struct task_desc *task, u64 timestamp)

get_new_event(struct task_desc *task, u64 timestamp)

{

{

	struct sched_event *event = calloc(1, sizeof(*event));

	struct sched_atom *event = calloc(1, sizeof(*event));

	unsigned long idx = task->nr_events;

	unsigned long idx = task->nr_events;

	size_t size;

	size_t size;
@@ -231,27 +231,27 @@ get_new_event(struct task_desc *task, u64 timestamp) 
	event->nr = idx;

	event->nr = idx;





	task->nr_events++;

	task->nr_events++;

	size = sizeof(struct sched_event *) * task->nr_events;

	size = sizeof(struct sched_atom *) * task->nr_events;

	task->events = realloc(task->events, size);

	task->atoms = realloc(task->atoms, size);

	BUG_ON(!task->events);

	BUG_ON(!task->atoms);





	task->events[idx] = event;

	task->atoms[idx] = event;





	return event;

	return event;

}

}





static struct sched_event *last_event(struct task_desc *task)

static struct sched_atom *last_event(struct task_desc *task)

{

{

	if (!task->nr_events)

	if (!task->nr_events)

		return NULL;

		return NULL;





	return task->events[task->nr_events - 1];

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

}

}





static void

static void

add_sched_event_run(struct task_desc *task, u64 timestamp, u64 duration)

add_sched_event_run(struct task_desc *task, u64 timestamp, u64 duration)

{

{

	struct sched_event *event, *curr_event = last_event(task);

	struct sched_atom *event, *curr_event = last_event(task);





	/*

	/*

	 * optimize an existing RUN event by merging this one

	 * optimize an existing RUN event by merging this one
@@ -275,7 +275,7 @@ static void 
add_sched_event_wakeup(struct task_desc *task, u64 timestamp,

add_sched_event_wakeup(struct task_desc *task, u64 timestamp,

		       struct task_desc *wakee)

		       struct task_desc *wakee)

{

{

	struct sched_event *event, *wakee_event;

	struct sched_atom *event, *wakee_event;





	event = get_new_event(task, timestamp);

	event = get_new_event(task, timestamp);

	event->type = SCHED_EVENT_WAKEUP;

	event->type = SCHED_EVENT_WAKEUP;
@@ -303,7 +303,7 @@ static void 
add_sched_event_sleep(struct task_desc *task, u64 timestamp,

add_sched_event_sleep(struct task_desc *task, u64 timestamp,

		      u64 task_state __used)

		      u64 task_state __used)

{

{

	struct sched_event *event = get_new_event(task, timestamp);

	struct sched_atom *event = get_new_event(task, timestamp);





	event->type = SCHED_EVENT_SLEEP;

	event->type = SCHED_EVENT_SLEEP;
@@ -372,27 +372,27 @@ static void add_cross_task_wakeups(void) 
}

}





static void

static void

process_sched_event(struct task_desc *this_task __used, struct sched_event *event)

process_sched_event(struct task_desc *this_task __used, struct sched_atom *atom)

{

{

	int ret = 0;

	int ret = 0;

	u64 now;

	u64 now;

	long long delta;

	long long delta;





	now = get_nsecs();

	now = get_nsecs();

	delta = start_time + event->timestamp - now;

	delta = start_time + atom->timestamp - now;





	switch (event->type) {

	switch (atom->type) {

		case SCHED_EVENT_RUN:

		case SCHED_EVENT_RUN:

			burn_nsecs(event->duration);

			burn_nsecs(atom->duration);

			break;

			break;

		case SCHED_EVENT_SLEEP:

		case SCHED_EVENT_SLEEP:

			if (event->wait_sem)

			if (atom->wait_sem)

				ret = sem_wait(event->wait_sem);

				ret = sem_wait(atom->wait_sem);

			BUG_ON(ret);

			BUG_ON(ret);

			break;

			break;

		case SCHED_EVENT_WAKEUP:

		case SCHED_EVENT_WAKEUP:

			if (event->wait_sem)

			if (atom->wait_sem)

				ret = sem_post(event->wait_sem);

				ret = sem_post(atom->wait_sem);

			BUG_ON(ret);

			BUG_ON(ret);

			break;

			break;

		default:

		default:
@@ -467,7 +467,7 @@ static void *thread_func(void *ctx) 




	for (i = 0; i < this_task->nr_events; i++) {

	for (i = 0; i < this_task->nr_events; i++) {

		this_task->curr_event = i;

		this_task->curr_event = i;

		process_sched_event(this_task, this_task->events[i]);

		process_sched_event(this_task, this_task->atoms[i]);

	}

	}





	cpu_usage_1 = get_cpu_usage_nsec_self();

	cpu_usage_1 = get_cpu_usage_nsec_self();
@@ -649,7 +649,7 @@ static void __cmd_replay(void) 
	if (multitarget_wakeups)

	if (multitarget_wakeups)

		printf("multi-target wakeups: %ld\n", multitarget_wakeups);

		printf("multi-target wakeups: %ld\n", multitarget_wakeups);

	if (nr_run_events_optimized)

	if (nr_run_events_optimized)

		printf("run events optimized: %ld\n",

		printf("run atoms optimized: %ld\n",

			nr_run_events_optimized);

			nr_run_events_optimized);





	print_task_traces();

	print_task_traces();
@@ -727,6 +727,20 @@ struct trace_switch_event { 
	u32 next_prio;

	u32 next_prio;

};

};





struct trace_runtime_event {

	u32 size;



	u16 common_type;

	u8 common_flags;

	u8 common_preempt_count;

	u32 common_pid;

	u32 common_tgid;



	char comm[16];

	u32 pid;

	u64 runtime;

	u64 vruntime;

};





struct trace_wakeup_event {

struct trace_wakeup_event {

	u32 size;

	u32 size;
@@ -767,6 +781,12 @@ struct trace_sched_handler { 
			     u64 timestamp,

			     u64 timestamp,

			     struct thread *thread);

			     struct thread *thread);





	void (*runtime_event)(struct trace_runtime_event *,

			      struct event *,

			      int cpu,

			      u64 timestamp,

			      struct thread *thread);



	void (*wakeup_event)(struct trace_wakeup_event *,

	void (*wakeup_event)(struct trace_wakeup_event *,

			     struct event *,

			     struct event *,

			     int cpu,

			     int cpu,
@@ -881,7 +901,7 @@ struct sort_dimension { 
static LIST_HEAD(cmp_pid);

static LIST_HEAD(cmp_pid);





static int

static int

thread_lat_cmp(struct list_head *list, struct task_atoms *l, struct task_atoms *r)

thread_lat_cmp(struct list_head *list, struct work_atoms *l, struct work_atoms *r)

{

{

	struct sort_dimension *sort;

	struct sort_dimension *sort;

	int ret = 0;

	int ret = 0;
@@ -897,18 +917,18 @@ thread_lat_cmp(struct list_head *list, struct task_atoms *l, struct task_atoms * 
	return ret;

	return ret;

}

}





static struct task_atoms *

static struct work_atoms *

thread_atoms_search(struct rb_root *root, struct thread *thread,

thread_atoms_search(struct rb_root *root, struct thread *thread,

			 struct list_head *sort_list)

			 struct list_head *sort_list)

{

{

	struct rb_node *node = root->rb_node;

	struct rb_node *node = root->rb_node;

	struct task_atoms key = { .thread = thread };

	struct work_atoms key = { .thread = thread };





	while (node) {

	while (node) {

		struct task_atoms *atoms;

		struct work_atoms *atoms;

		int cmp;

		int cmp;





		atoms = container_of(node, struct task_atoms, node);

		atoms = container_of(node, struct work_atoms, node);





		cmp = thread_lat_cmp(sort_list, &key, atoms);

		cmp = thread_lat_cmp(sort_list, &key, atoms);

		if (cmp > 0)

		if (cmp > 0)
@@ -924,16 +944,16 @@ thread_atoms_search(struct rb_root *root, struct thread *thread, 
}

}





static void

static void

__thread_latency_insert(struct rb_root *root, struct task_atoms *data,

__thread_latency_insert(struct rb_root *root, struct work_atoms *data,

			 struct list_head *sort_list)

			 struct list_head *sort_list)

{

{

	struct rb_node **new = &(root->rb_node), *parent = NULL;

	struct rb_node **new = &(root->rb_node), *parent = NULL;





	while (*new) {

	while (*new) {

		struct task_atoms *this;

		struct work_atoms *this;

		int cmp;

		int cmp;





		this = container_of(*new, struct task_atoms, node);

		this = container_of(*new, struct work_atoms, node);

		parent = *new;

		parent = *new;





		cmp = thread_lat_cmp(sort_list, data, this);

		cmp = thread_lat_cmp(sort_list, data, this);
@@ -950,14 +970,14 @@ __thread_latency_insert(struct rb_root *root, struct task_atoms *data, 




static void thread_atoms_insert(struct thread *thread)

static void thread_atoms_insert(struct thread *thread)

{

{

	struct task_atoms *atoms;

	struct work_atoms *atoms;





	atoms = calloc(sizeof(*atoms), 1);

	atoms = calloc(sizeof(*atoms), 1);

	if (!atoms)

	if (!atoms)

		die("No memory");

		die("No memory");





	atoms->thread = thread;

	atoms->thread = thread;

	INIT_LIST_HEAD(&atoms->atom_list);

	INIT_LIST_HEAD(&atoms->work_list);

	__thread_latency_insert(&atom_root, atoms, &cmp_pid);

	__thread_latency_insert(&atom_root, atoms, &cmp_pid);

}

}
@@ -980,9 +1000,8 @@ static char sched_out_state(struct trace_switch_event *switch_event) 
}

}





static void

static void

lat_sched_out(struct task_atoms *atoms,

add_sched_out_event(struct work_atoms *atoms,

	      struct trace_switch_event *switch_event __used,

		    char run_state,

	      u64 delta,

		    u64 timestamp)

		    u64 timestamp)

{

{

	struct work_atom *atom;

	struct work_atom *atom;
@@ -993,25 +1012,37 @@ lat_sched_out(struct task_atoms *atoms, 




	atom->sched_out_time = timestamp;

	atom->sched_out_time = timestamp;





	if (sched_out_state(switch_event) == 'R') {

	if (run_state == 'R') {

		atom->state = THREAD_WAIT_CPU;

		atom->state = THREAD_WAIT_CPU;

		atom->wake_up_time = atom->sched_out_time;

		atom->wake_up_time = atom->sched_out_time;

	}

	}





	atom->runtime = delta;

	list_add_tail(&atom->list, &atoms->work_list);

	list_add_tail(&atom->list, &atoms->atom_list);

}

}





static void

static void

lat_sched_in(struct task_atoms *atoms, u64 timestamp)

add_runtime_event(struct work_atoms *atoms, u64 delta, u64 timestamp __used)

{

	struct work_atom *atom;



	BUG_ON(list_empty(&atoms->work_list));



	atom = list_entry(atoms->work_list.prev, struct work_atom, list);



	atom->runtime += delta;

	atoms->total_runtime += delta;

}



static void

add_sched_in_event(struct work_atoms *atoms, u64 timestamp)

{

{

	struct work_atom *atom;

	struct work_atom *atom;

	u64 delta;

	u64 delta;





	if (list_empty(&atoms->atom_list))

	if (list_empty(&atoms->work_list))

		return;

		return;





	atom = list_entry(atoms->atom_list.prev, struct work_atom, list);

	atom = list_entry(atoms->work_list.prev, struct work_atom, list);





	if (atom->state != THREAD_WAIT_CPU)

	if (atom->state != THREAD_WAIT_CPU)

		return;

		return;
@@ -1029,7 +1060,6 @@ lat_sched_in(struct task_atoms *atoms, u64 timestamp) 
	if (delta > atoms->max_lat)

	if (delta > atoms->max_lat)

		atoms->max_lat = delta;

		atoms->max_lat = delta;

	atoms->nb_atoms++;

	atoms->nb_atoms++;

	atoms->total_runtime += atom->runtime;

}

}





static void

static void
@@ -1039,13 +1069,12 @@ latency_switch_event(struct trace_switch_event *switch_event, 
		     u64 timestamp,

		     u64 timestamp,

		     struct thread *thread __used)

		     struct thread *thread __used)

{

{

	struct task_atoms *out_atoms, *in_atoms;

	struct work_atoms *out_events, *in_events;

	struct thread *sched_out, *sched_in;

	struct thread *sched_out, *sched_in;

	u64 timestamp0;

	u64 timestamp0;

	s64 delta;

	s64 delta;





	if (cpu >= MAX_CPUS || cpu < 0)

	BUG_ON(cpu >= MAX_CPUS || cpu < 0);

		return;





	timestamp0 = cpu_last_switched[cpu];

	timestamp0 = cpu_last_switched[cpu];

	cpu_last_switched[cpu] = timestamp;

	cpu_last_switched[cpu] = timestamp;
@@ -1061,34 +1090,63 @@ latency_switch_event(struct trace_switch_event *switch_event, 
	sched_out = threads__findnew(switch_event->prev_pid, &threads, &last_match);

	sched_out = threads__findnew(switch_event->prev_pid, &threads, &last_match);

	sched_in = threads__findnew(switch_event->next_pid, &threads, &last_match);

	sched_in = threads__findnew(switch_event->next_pid, &threads, &last_match);





	in_atoms = thread_atoms_search(&atom_root, sched_in, &cmp_pid);

	out_events = thread_atoms_search(&atom_root, sched_out, &cmp_pid);

	if (!in_atoms) {

	if (!out_events) {

		thread_atoms_insert(sched_out);

		out_events = thread_atoms_search(&atom_root, sched_out, &cmp_pid);

		if (!out_events)

			die("out-event: Internal tree error");

	}

	add_sched_out_event(out_events, sched_out_state(switch_event), timestamp);



	in_events = thread_atoms_search(&atom_root, sched_in, &cmp_pid);

	if (!in_events) {

		thread_atoms_insert(sched_in);

		thread_atoms_insert(sched_in);

		in_atoms = thread_atoms_search(&atom_root, sched_in, &cmp_pid);

		in_events = thread_atoms_search(&atom_root, sched_in, &cmp_pid);

		if (!in_atoms)

		if (!in_events)

			die("in-atom: Internal tree error");

			die("in-event: Internal tree error");

		/*

		 * Take came in we have not heard about yet,

		 * add in an initial atom in runnable state:

		 */

		add_sched_out_event(in_events, 'R', timestamp);

	}

	add_sched_in_event(in_events, timestamp);

}

}





	out_atoms = thread_atoms_search(&atom_root, sched_out, &cmp_pid);

static void

	if (!out_atoms) {

latency_runtime_event(struct trace_runtime_event *runtime_event,

		thread_atoms_insert(sched_out);

		     struct event *event __used,

		out_atoms = thread_atoms_search(&atom_root, sched_out, &cmp_pid);

		     int cpu,

		if (!out_atoms)

		     u64 timestamp,

			die("out-atom: Internal tree error");

		     struct thread *this_thread __used)

{

	struct work_atoms *atoms;

	struct thread *thread;



	BUG_ON(cpu >= MAX_CPUS || cpu < 0);



	thread = threads__findnew(runtime_event->pid, &threads, &last_match);

	atoms = thread_atoms_search(&atom_root, thread, &cmp_pid);

	if (!atoms) {

		thread_atoms_insert(thread);

		atoms = thread_atoms_search(&atom_root, thread, &cmp_pid);

		if (!atoms)

			die("in-event: Internal tree error");

		add_sched_out_event(atoms, 'R', timestamp);

	}

	}





	lat_sched_in(in_atoms, timestamp);

	add_runtime_event(atoms, runtime_event->runtime, timestamp);

	lat_sched_out(out_atoms, switch_event, delta, timestamp);

}

}





static void

static void

latency_wakeup_event(struct trace_wakeup_event *wakeup_event,

latency_wakeup_event(struct trace_wakeup_event *wakeup_event,

		     struct event *event __used,

		     struct event *__event __used,

		     int cpu __used,

		     int cpu __used,

		     u64 timestamp,

		     u64 timestamp,

		     struct thread *thread __used)

		     struct thread *thread __used)

{

{

	struct task_atoms *atoms;

	struct work_atoms *atoms;

	struct work_atom *atom;

	struct work_atom *atom;

	struct thread *wakee;

	struct thread *wakee;
@@ -1100,16 +1158,20 @@ latency_wakeup_event(struct trace_wakeup_event *wakeup_event, 
	atoms = thread_atoms_search(&atom_root, wakee, &cmp_pid);

	atoms = thread_atoms_search(&atom_root, wakee, &cmp_pid);

	if (!atoms) {

	if (!atoms) {

		thread_atoms_insert(wakee);

		thread_atoms_insert(wakee);

		return;

		atoms = thread_atoms_search(&atom_root, wakee, &cmp_pid);

		if (!atoms)

			die("wakeup-event: Internal tree error");

		add_sched_out_event(atoms, 'S', timestamp);

	}

	}





	if (list_empty(&atoms->atom_list))

	BUG_ON(list_empty(&atoms->work_list));

		return;





	atom = list_entry(atoms->atom_list.prev, struct work_atom, list);

	atom = list_entry(atoms->work_list.prev, struct work_atom, list);





	if (atom->state != THREAD_SLEEPING)

	if (atom->state != THREAD_SLEEPING) {

		printf("boo2\n");

		return;

		return;

	}





	nr_timestamps++;

	nr_timestamps++;

	if (atom->sched_out_time > timestamp) {

	if (atom->sched_out_time > timestamp) {
@@ -1124,40 +1186,41 @@ latency_wakeup_event(struct trace_wakeup_event *wakeup_event, 
static struct trace_sched_handler lat_ops  = {

static struct trace_sched_handler lat_ops  = {

	.wakeup_event		= latency_wakeup_event,

	.wakeup_event		= latency_wakeup_event,

	.switch_event		= latency_switch_event,

	.switch_event		= latency_switch_event,

	.runtime_event		= latency_runtime_event,

	.fork_event		= latency_fork_event,

	.fork_event		= latency_fork_event,

};

};





static void output_lat_thread(struct task_atoms *atom_list)

static void output_lat_thread(struct work_atoms *work_list)

{

{

	int i;

	int i;

	int ret;

	int ret;

	u64 avg;

	u64 avg;





	if (!atom_list->nb_atoms)

	if (!work_list->nb_atoms)

		return;

		return;

	/*

	/*

	 * Ignore idle threads:

	 * Ignore idle threads:

	 */

	 */

	if (!atom_list->thread->pid)

	if (!work_list->thread->pid)

		return;

		return;





	all_runtime += atom_list->total_runtime;

	all_runtime += work_list->total_runtime;

	all_count += atom_list->nb_atoms;

	all_count += work_list->nb_atoms;





	ret = printf("  %s-%d ", atom_list->thread->comm, atom_list->thread->pid);

	ret = printf("  %s-%d ", work_list->thread->comm, work_list->thread->pid);





	for (i = 0; i < 24 - ret; i++)

	for (i = 0; i < 24 - ret; i++)

		printf(" ");

		printf(" ");





	avg = atom_list->total_lat / atom_list->nb_atoms;

	avg = work_list->total_lat / work_list->nb_atoms;





	printf("|%9.3f ms |%9llu | avg:%9.3f ms | max:%9.3f ms |\n",

	printf("|%9.3f ms |%9llu | avg:%9.3f ms | max:%9.3f ms |\n",

	      (double)atom_list->total_runtime / 1e6,

	      (double)work_list->total_runtime / 1e6,

		 atom_list->nb_atoms, (double)avg / 1e6,

		 work_list->nb_atoms, (double)avg / 1e6,

		 (double)atom_list->max_lat / 1e6);

		 (double)work_list->max_lat / 1e6);

}

}





static int pid_cmp(struct task_atoms *l, struct task_atoms *r)

static int pid_cmp(struct work_atoms *l, struct work_atoms *r)

{

{

	if (l->thread->pid < r->thread->pid)

	if (l->thread->pid < r->thread->pid)

		return -1;

		return -1;
@@ -1172,7 +1235,7 @@ static struct sort_dimension pid_sort_dimension = { 
	.cmp			= pid_cmp,

	.cmp			= pid_cmp,

};

};





static int avg_cmp(struct task_atoms *l, struct task_atoms *r)

static int avg_cmp(struct work_atoms *l, struct work_atoms *r)

{

{

	u64 avgl, avgr;

	u64 avgl, avgr;
@@ -1198,7 +1261,7 @@ static struct sort_dimension avg_sort_dimension = { 
	.cmp			= avg_cmp,

	.cmp			= avg_cmp,

};

};





static int max_cmp(struct task_atoms *l, struct task_atoms *r)

static int max_cmp(struct work_atoms *l, struct work_atoms *r)

{

{

	if (l->max_lat < r->max_lat)

	if (l->max_lat < r->max_lat)

		return -1;

		return -1;
@@ -1213,7 +1276,7 @@ static struct sort_dimension max_sort_dimension = { 
	.cmp			= max_cmp,

	.cmp			= max_cmp,

};

};





static int switch_cmp(struct task_atoms *l, struct task_atoms *r)

static int switch_cmp(struct work_atoms *l, struct work_atoms *r)

{

{

	if (l->nb_atoms < r->nb_atoms)

	if (l->nb_atoms < r->nb_atoms)

		return -1;

		return -1;
@@ -1228,7 +1291,7 @@ static struct sort_dimension switch_sort_dimension = { 
	.cmp			= switch_cmp,

	.cmp			= switch_cmp,

};

};





static int runtime_cmp(struct task_atoms *l, struct task_atoms *r)

static int runtime_cmp(struct work_atoms *l, struct work_atoms *r)

{

{

	if (l->total_runtime < r->total_runtime)

	if (l->total_runtime < r->total_runtime)

		return -1;

		return -1;
@@ -1277,13 +1340,13 @@ static void sort_lat(void) 
	struct rb_node *node;

	struct rb_node *node;





	for (;;) {

	for (;;) {

		struct task_atoms *data;

		struct work_atoms *data;

		node = rb_first(&atom_root);

		node = rb_first(&atom_root);

		if (!node)

		if (!node)

			break;

			break;





		rb_erase(node, &atom_root);

		rb_erase(node, &atom_root);

		data = rb_entry(node, struct task_atoms, node);

		data = rb_entry(node, struct work_atoms, node);

		__thread_latency_insert(&sorted_atom_root, data, &sort_list);

		__thread_latency_insert(&sorted_atom_root, data, &sort_list);

	}

	}

}

}
@@ -1303,10 +1366,10 @@ static void __cmd_lat(void) 
	next = rb_first(&sorted_atom_root);

	next = rb_first(&sorted_atom_root);





	while (next) {

	while (next) {

		struct task_atoms *atom_list;

		struct work_atoms *work_list;





		atom_list = rb_entry(next, struct task_atoms, node);

		work_list = rb_entry(next, struct work_atoms, node);

		output_lat_thread(atom_list);

		output_lat_thread(work_list);

		next = rb_next(next);

		next = rb_next(next);

	}

	}
@@ -1368,6 +1431,23 @@ process_sched_switch_event(struct raw_event_sample *raw, 
	trace_handler->switch_event(&switch_event, event, cpu, timestamp, thread);

	trace_handler->switch_event(&switch_event, event, cpu, timestamp, thread);

}

}





static void

process_sched_runtime_event(struct raw_event_sample *raw,

			   struct event *event,

			   int cpu __used,

			   u64 timestamp __used,

			   struct thread *thread __used)

{

	struct trace_runtime_event runtime_event;



	FILL_ARRAY(runtime_event, comm, event, raw->data);

	FILL_FIELD(runtime_event, pid, event, raw->data);

	FILL_FIELD(runtime_event, runtime, event, raw->data);

	FILL_FIELD(runtime_event, vruntime, event, raw->data);



	trace_handler->runtime_event(&runtime_event, event, cpu, timestamp, thread);

}



static void

static void

process_sched_fork_event(struct raw_event_sample *raw,

process_sched_fork_event(struct raw_event_sample *raw,

			 struct event *event,

			 struct event *event,
@@ -1410,6 +1490,8 @@ process_raw_event(event_t *raw_event __used, void *more_data, 




	if (!strcmp(event->name, "sched_switch"))

	if (!strcmp(event->name, "sched_switch"))

		process_sched_switch_event(raw, event, cpu, timestamp, thread);

		process_sched_switch_event(raw, event, cpu, timestamp, thread);

	if (!strcmp(event->name, "sched_stat_runtime"))

		process_sched_runtime_event(raw, event, cpu, timestamp, thread);

	if (!strcmp(event->name, "sched_wakeup"))

	if (!strcmp(event->name, "sched_wakeup"))

		process_sched_wakeup_event(raw, event, cpu, timestamp, thread);

		process_sched_wakeup_event(raw, event, cpu, timestamp, thread);

	if (!strcmp(event->name, "sched_wakeup_new"))

	if (!strcmp(event->name, "sched_wakeup_new"))