[deliverable/lttng-modules.git] / instrumentation / events / lttng-module / sched.h

/* SPDX-License-Identifier: GPL-2.0 */
#undef TRACE_SYSTEM
#define TRACE_SYSTEM sched

#if !defined(LTTNG_TRACE_SCHED_H) || defined(TRACE_HEADER_MULTI_READ)
#define LTTNG_TRACE_SCHED_H

#include <probes/lttng-tracepoint-event.h>
#include <linux/sched.h>
#include <linux/pid_namespace.h>
#include <linux/binfmts.h>
#include <linux/version.h>
#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,9,0))
#include <linux/sched/rt.h>
#endif
#include <wrapper/namespace.h>

#define LTTNG_MAX_PID_NS_LEVEL 32

#ifndef _TRACE_SCHED_DEF_
#define _TRACE_SCHED_DEF_

#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0))

static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
{
#ifdef CONFIG_SCHED_DEBUG
	BUG_ON(p != current);
#endif /* CONFIG_SCHED_DEBUG */
	/*
	 * Preemption ignores task state, therefore preempted tasks are always RUNNING
	 * (we will not have dequeued if state != RUNNING).
	 */
	return preempt ? TASK_RUNNING | TASK_STATE_MAX : p->state;
}

#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(3,19,0))

static inline long __trace_sched_switch_state(struct task_struct *p)
{
	long state = p->state;

#ifdef CONFIG_PREEMPT
#ifdef CONFIG_SCHED_DEBUG
	BUG_ON(p != current);
#endif /* CONFIG_SCHED_DEBUG */
	/*
	 * For all intents and purposes a preempted task is a running task.
	 */
	if (preempt_count() & PREEMPT_ACTIVE)
		state = TASK_RUNNING | TASK_STATE_MAX;
#endif	/* CONFIG_PREEMPT */

	return state;
}

#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(3,13,0))

static inline long __trace_sched_switch_state(struct task_struct *p)
{
	long state = p->state;

#ifdef CONFIG_PREEMPT
	/*
	 * For all intents and purposes a preempted task is a running task.
	 */
	if (task_preempt_count(p) & PREEMPT_ACTIVE)
		state = TASK_RUNNING | TASK_STATE_MAX;
#endif

	return state;
}

#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(3,2,0))

static inline long __trace_sched_switch_state(struct task_struct *p)
{
	long state = p->state;

#ifdef CONFIG_PREEMPT
	/*
	 * For all intents and purposes a preempted task is a running task.
	 */
	if (task_thread_info(p)->preempt_count & PREEMPT_ACTIVE)
		state = TASK_RUNNING | TASK_STATE_MAX;
#endif

	return state;
}

#else

static inline long __trace_sched_switch_state(struct task_struct *p)
{
	long state = p->state;

#ifdef CONFIG_PREEMPT
	/*
	 * For all intents and purposes a preempted task is a running task.
	 */
	if (task_thread_info(p)->preempt_count & PREEMPT_ACTIVE)
		state = TASK_RUNNING;
#endif

	return state;
}

#endif

#endif /* _TRACE_SCHED_DEF_ */

/*
 * Tracepoint for calling kthread_stop, performed to end a kthread:
 */
LTTNG_TRACEPOINT_EVENT(sched_kthread_stop,

	TP_PROTO(struct task_struct *t),

	TP_ARGS(t),

	TP_FIELDS(
		ctf_array_text(char, comm, t->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, tid, t->pid)
	)
)

/*
 * Tracepoint for the return value of the kthread stopping:
 */
LTTNG_TRACEPOINT_EVENT(sched_kthread_stop_ret,

	TP_PROTO(int ret),

	TP_ARGS(ret),

	TP_FIELDS(
		ctf_integer(int, ret, ret)
	)
)

/*
 * Tracepoint for waking up a task:
 */
#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0) || \
	LTTNG_RT_KERNEL_RANGE(4,1,10,11, 4,2,0,0) || \
	LTTNG_RT_KERNEL_RANGE(3,18,27,26, 3,19,0,0) || \
	LTTNG_RT_KERNEL_RANGE(3,14,61,63, 3,15,0,0) || \
	LTTNG_RT_KERNEL_RANGE(3,12,54,73, 3,13,0,0) || \
	LTTNG_RT_KERNEL_RANGE(3,10,97,106, 3,11,0,0) || \
	LTTNG_RT_KERNEL_RANGE(3,4,110,139, 3,5,0,0) || \
	LTTNG_RT_KERNEL_RANGE(3,2,77,111, 3,3,0,0))
LTTNG_TRACEPOINT_EVENT_CLASS(sched_wakeup_template,

	TP_PROTO(struct task_struct *p),

	TP_ARGS(p),

	TP_FIELDS(
		ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, tid, p->pid)
		ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
		ctf_integer(int, target_cpu, task_cpu(p))
	)
)
#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0)) */
LTTNG_TRACEPOINT_EVENT_CLASS(sched_wakeup_template,

	TP_PROTO(struct task_struct *p, int success),

	TP_ARGS(p, success),

	TP_FIELDS(
		ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, tid, p->pid)
		ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
		ctf_integer(int, success, success)
		ctf_integer(int, target_cpu, task_cpu(p))
	)
)
#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0)) */

#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0) || \
	LTTNG_RT_KERNEL_RANGE(4,1,10,11, 4,2,0,0) || \
	LTTNG_RT_KERNEL_RANGE(3,18,27,26, 3,19,0,0) || \
	LTTNG_RT_KERNEL_RANGE(3,14,61,63, 3,15,0,0) || \
	LTTNG_RT_KERNEL_RANGE(3,12,54,73, 3,13,0,0) || \
	LTTNG_RT_KERNEL_RANGE(3,10,97,106, 3,11,0,0) || \
	LTTNG_RT_KERNEL_RANGE(3,4,110,139, 3,5,0,0) || \
	LTTNG_RT_KERNEL_RANGE(3,2,77,111, 3,3,0,0))

/*
 * Tracepoint called when waking a task; this tracepoint is guaranteed to be
 * called from the waking context.
 */
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_waking,
	     TP_PROTO(struct task_struct *p),
	     TP_ARGS(p))

/*
 * Tracepoint called when the task is actually woken; p->state == TASK_RUNNNG.
 * It it not always called from the waking context.
 */
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup,
	     TP_PROTO(struct task_struct *p),
	     TP_ARGS(p))

/*
 * Tracepoint for waking up a new task:
 */
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup_new,
	     TP_PROTO(struct task_struct *p),
	     TP_ARGS(p))

#else

LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup,
	     TP_PROTO(struct task_struct *p, int success),
	     TP_ARGS(p, success))

/*
 * Tracepoint for waking up a new task:
 */
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup_new,
	     TP_PROTO(struct task_struct *p, int success),
	     TP_ARGS(p, success))

#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0)) */

/*
 * Tracepoint for task switches, performed by the scheduler:
 */
LTTNG_TRACEPOINT_EVENT(sched_switch,

#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0))
	TP_PROTO(bool preempt,
		 struct task_struct *prev,
		 struct task_struct *next),

	TP_ARGS(preempt, prev, next),
#else
	TP_PROTO(struct task_struct *prev,
		 struct task_struct *next),

	TP_ARGS(prev, next),
#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0)) */

	TP_FIELDS(
		ctf_array_text(char, prev_comm,	prev->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, prev_tid, prev->pid)
		ctf_integer(int, prev_prio, prev->prio - MAX_RT_PRIO)
#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0))
		ctf_integer(long, prev_state, __trace_sched_switch_state(preempt, prev))
#else
		ctf_integer(long, prev_state, __trace_sched_switch_state(prev))
#endif
		ctf_array_text(char, next_comm, next->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, next_tid, next->pid)
		ctf_integer(int, next_prio, next->prio - MAX_RT_PRIO)
	)
)

/*
 * Tracepoint for a task being migrated:
 */
LTTNG_TRACEPOINT_EVENT(sched_migrate_task,

	TP_PROTO(struct task_struct *p, int dest_cpu),

	TP_ARGS(p, dest_cpu),

	TP_FIELDS(
		ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, tid, p->pid)
		ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
		ctf_integer(int, orig_cpu, task_cpu(p))
		ctf_integer(int, dest_cpu, dest_cpu)
	)
)

LTTNG_TRACEPOINT_EVENT_CLASS(sched_process_template,

	TP_PROTO(struct task_struct *p),

	TP_ARGS(p),

	TP_FIELDS(
		ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, tid, p->pid)
		ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	)
)

/*
 * Tracepoint for freeing a task:
 */
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_free,
	     TP_PROTO(struct task_struct *p),
	     TP_ARGS(p))


/*
 * Tracepoint for a task exiting:
 */
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_exit,
	     TP_PROTO(struct task_struct *p),
	     TP_ARGS(p))

/*
 * Tracepoint for waiting on task to unschedule:
 */
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_wait_task,
	TP_PROTO(struct task_struct *p),
	TP_ARGS(p))

/*
 * Tracepoint for a waiting task:
 */
LTTNG_TRACEPOINT_EVENT(sched_process_wait,

	TP_PROTO(struct pid *pid),

	TP_ARGS(pid),

	TP_FIELDS(
		ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, tid, pid_nr(pid))
		ctf_integer(int, prio, current->prio - MAX_RT_PRIO)
	)
)

/*
 * Tracepoint for do_fork.
 * Saving both TID and PID information, especially for the child, allows
 * trace analyzers to distinguish between creation of a new process and
 * creation of a new thread. Newly created processes will have child_tid
 * == child_pid, while creation of a thread yields to child_tid !=
 * child_pid.
 */
LTTNG_TRACEPOINT_EVENT_CODE(sched_process_fork,

	TP_PROTO(struct task_struct *parent, struct task_struct *child),

	TP_ARGS(parent, child),

	TP_locvar(
		pid_t vtids[LTTNG_MAX_PID_NS_LEVEL];
		unsigned int ns_level;
	),

	TP_code_pre(
		if (child) {
			struct pid *child_pid;
			unsigned int i;

			child_pid = task_pid(child);
			tp_locvar->ns_level =
				min_t(unsigned int, child_pid->level + 1,
					LTTNG_MAX_PID_NS_LEVEL);
			for (i = 0; i < tp_locvar->ns_level; i++)
				tp_locvar->vtids[i] = child_pid->numbers[i].nr;
		}
	),

	TP_FIELDS(
		ctf_array_text(char, parent_comm, parent->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, parent_tid, parent->pid)
		ctf_integer(pid_t, parent_pid, parent->tgid)
#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0))
		ctf_integer(unsigned int, parent_ns_inum,
			({
				unsigned int parent_ns_inum = 0;

				if (parent) {
					struct pid_namespace *pid_ns;

					pid_ns = task_active_pid_ns(parent);
					if (pid_ns)
						parent_ns_inum =
							pid_ns->lttng_ns_inum;
				}
				parent_ns_inum;
			}))
#endif
		ctf_array_text(char, child_comm, child->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, child_tid, child->pid)
		ctf_sequence(pid_t, vtids, tp_locvar->vtids, u8, tp_locvar->ns_level)
		ctf_integer(pid_t, child_pid, child->tgid)
#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0))
		ctf_integer(unsigned int, child_ns_inum,
			({
				unsigned int child_ns_inum = 0;

				if (child) {
					struct pid_namespace *pid_ns;

					pid_ns = task_active_pid_ns(child);
					if (pid_ns)
						child_ns_inum =
							pid_ns->lttng_ns_inum;
				}
				child_ns_inum;
			}))
#endif
	),

	TP_code_post()
)

#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,4,0))
/*
 * Tracepoint for exec:
 */
LTTNG_TRACEPOINT_EVENT(sched_process_exec,

	TP_PROTO(struct task_struct *p, pid_t old_pid,
		 struct linux_binprm *bprm),

	TP_ARGS(p, old_pid, bprm),

	TP_FIELDS(
		ctf_string(filename, bprm->filename)
		ctf_integer(pid_t, tid, p->pid)
		ctf_integer(pid_t, old_tid, old_pid)
	)
)
#endif

/*
 * XXX the below sched_stat tracepoints only apply to SCHED_OTHER/BATCH/IDLE
 *     adding sched_stat support to SCHED_FIFO/RR would be welcome.
 */
LTTNG_TRACEPOINT_EVENT_CLASS(sched_stat_template,

	TP_PROTO(struct task_struct *tsk, u64 delay),

	TP_ARGS(tsk, delay),

	TP_FIELDS(
		ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, tid, tsk->pid)
		ctf_integer(u64, delay, delay)
	)
)


/*
 * Tracepoint for accounting wait time (time the task is runnable
 * but not actually running due to scheduler contention).
 */
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_wait,
	     TP_PROTO(struct task_struct *tsk, u64 delay),
	     TP_ARGS(tsk, delay))

/*
 * Tracepoint for accounting sleep time (time the task is not runnable,
 * including iowait, see below).
 */
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_sleep,
	     TP_PROTO(struct task_struct *tsk, u64 delay),
	     TP_ARGS(tsk, delay))

/*
 * Tracepoint for accounting iowait time (time the task is not runnable
 * due to waiting on IO to complete).
 */
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_iowait,
	     TP_PROTO(struct task_struct *tsk, u64 delay),
	     TP_ARGS(tsk, delay))

#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,3,0))
/*
 * Tracepoint for accounting blocked time (time the task is in uninterruptible).
 */
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_blocked,
	     TP_PROTO(struct task_struct *tsk, u64 delay),
	     TP_ARGS(tsk, delay))
#endif

/*
 * Tracepoint for accounting runtime (time the task is executing
 * on a CPU).
 */
LTTNG_TRACEPOINT_EVENT(sched_stat_runtime,

	TP_PROTO(struct task_struct *tsk, u64 runtime, u64 vruntime),

	TP_ARGS(tsk, runtime, vruntime),

	TP_FIELDS(
		ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, tid, tsk->pid)
		ctf_integer(u64, runtime, runtime)
		ctf_integer(u64, vruntime, vruntime)
	)
)

#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,12,0) || \
	LTTNG_RT_KERNEL_RANGE(4,9,27,18, 4,10,0,0) || \
	LTTNG_RT_KERNEL_RANGE(4,11,5,1, 4,12,0,0))
/*
 * Tracepoint for showing priority inheritance modifying a tasks
 * priority.
 */
LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,

	TP_PROTO(struct task_struct *tsk, struct task_struct *pi_task),

	TP_ARGS(tsk, pi_task),

	TP_FIELDS(
		ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, tid, tsk->pid)
		ctf_integer(int, oldprio, tsk->prio - MAX_RT_PRIO)
		ctf_integer(int, newprio, pi_task ? pi_task->prio - MAX_RT_PRIO : tsk->prio - MAX_RT_PRIO)
	)
)
#else
/*
 * Tracepoint for showing priority inheritance modifying a tasks
 * priority.
 */
LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,

	TP_PROTO(struct task_struct *tsk, int newprio),

	TP_ARGS(tsk, newprio),

	TP_FIELDS(
		ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, tid, tsk->pid)
		ctf_integer(int, oldprio, tsk->prio - MAX_RT_PRIO)
		ctf_integer(int, newprio, newprio - MAX_RT_PRIO)
	)
)
#endif

#endif /* LTTNG_TRACE_SCHED_H */

/* This part must be outside protection */
#include <probes/define_trace.h>
Commit	Line	Data
9f36eaed	1	/* SPDX-License-Identifier: GPL-2.0 */
f62b389e MD	2	#undef TRACE_SYSTEM
	3	#define TRACE_SYSTEM sched
	4
3bc29f0a MD	5	#if !defined(LTTNG_TRACE_SCHED_H) \|\| defined(TRACE_HEADER_MULTI_READ)
3bc29f0a MD	6	#define LTTNG_TRACE_SCHED_H
f62b389e	7
6ec43db8	8	#include <probes/lttng-tracepoint-event.h>
f62b389e	9	#include <linux/sched.h>
7bbf55ea	10	#include <linux/pid_namespace.h>
7c68b363 AG	11	#include <linux/binfmts.h>
7c68b363 AG	12	#include <linux/version.h>
090db00e MJ	13	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,9,0))
	14	#include <linux/sched/rt.h>
	15	#endif
45b27d78	16	#include <wrapper/namespace.h>
7bbf55ea SL	17
	18	#define LTTNG_MAX_PID_NS_LEVEL 32
	19
f62b389e MD	20	#ifndef _TRACE_SCHED_DEF_
	21	#define _TRACE_SCHED_DEF_
	22
3ee729fe MD	23	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0))
	24
	25	static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
	26	{
	27	#ifdef CONFIG_SCHED_DEBUG
	28	BUG_ON(p != current);
	29	#endif /* CONFIG_SCHED_DEBUG */
	30	/*
	31	* Preemption ignores task state, therefore preempted tasks are always RUNNING
	32	* (we will not have dequeued if state != RUNNING).
	33	*/
	34	return preempt ? TASK_RUNNING \| TASK_STATE_MAX : p->state;
	35	}
	36
	37	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(3,19,0))
857b6f4f MD	38
	39	static inline long __trace_sched_switch_state(struct task_struct *p)
	40	{
	41	long state = p->state;
	42
	43	#ifdef CONFIG_PREEMPT
	44	#ifdef CONFIG_SCHED_DEBUG
	45	BUG_ON(p != current);
	46	#endif /* CONFIG_SCHED_DEBUG */
	47	/*
	48	* For all intents and purposes a preempted task is a running task.
	49	*/
	50	if (preempt_count() & PREEMPT_ACTIVE)
	51	state = TASK_RUNNING \| TASK_STATE_MAX;
	52	#endif /* CONFIG_PREEMPT */
	53
	54	return state;
	55	}
	56
	57	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(3,13,0))
33673ee7 MD	58
	59	static inline long __trace_sched_switch_state(struct task_struct *p)
	60	{
	61	long state = p->state;
	62
	63	#ifdef CONFIG_PREEMPT
	64	/*
	65	* For all intents and purposes a preempted task is a running task.
	66	*/
	67	if (task_preempt_count(p) & PREEMPT_ACTIVE)
	68	state = TASK_RUNNING \| TASK_STATE_MAX;
	69	#endif
	70
	71	return state;
	72	}
	73
	74	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(3,2,0))
7c68b363	75
f62b389e MD	76	static inline long __trace_sched_switch_state(struct task_struct *p)
	77	{
	78	long state = p->state;
	79
	80	#ifdef CONFIG_PREEMPT
	81	/*
	82	* For all intents and purposes a preempted task is a running task.
	83	*/
	84	if (task_thread_info(p)->preempt_count & PREEMPT_ACTIVE)
7c68b363	85	state = TASK_RUNNING \| TASK_STATE_MAX;
7c68b363	86	#endif
33673ee7 MD	87
	88	return state;
	89	}
	90
ead65427	91	#else
33673ee7 MD	92
	93	static inline long __trace_sched_switch_state(struct task_struct *p)
	94	{
	95	long state = p->state;
	96
	97	#ifdef CONFIG_PREEMPT
	98	/*
	99	* For all intents and purposes a preempted task is a running task.
	100	*/
	101	if (task_thread_info(p)->preempt_count & PREEMPT_ACTIVE)
	102	state = TASK_RUNNING;
f62b389e MD	103	#endif
	104
	105	return state;
	106	}
	107
7c68b363 AG	108	#endif
7c68b363 AG	109
f62b389e MD	110	#endif /* _TRACE_SCHED_DEF_ */
	111
	112	/*
	113	* Tracepoint for calling kthread_stop, performed to end a kthread:
	114	*/
3bc29f0a	115	LTTNG_TRACEPOINT_EVENT(sched_kthread_stop,
f62b389e MD	116
	117	TP_PROTO(struct task_struct *t),
	118
	119	TP_ARGS(t),
	120
f127e61e MD	121	TP_FIELDS(
	122	ctf_array_text(char, comm, t->comm, TASK_COMM_LEN)
	123	ctf_integer(pid_t, tid, t->pid)
	124	)
f62b389e MD	125	)
	126
	127	/*
	128	* Tracepoint for the return value of the kthread stopping:
	129	*/
3bc29f0a	130	LTTNG_TRACEPOINT_EVENT(sched_kthread_stop_ret,
f62b389e MD	131
	132	TP_PROTO(int ret),
	133
	134	TP_ARGS(ret),
	135
f127e61e MD	136	TP_FIELDS(
	137	ctf_integer(int, ret, ret)
	138	)
f62b389e MD	139	)
	140
	141	/*
	142	* Tracepoint for waking up a task:
	143	*/
b6c74835	144	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0) \|\| \
95ff37b6 MJ	145	LTTNG_RT_KERNEL_RANGE(4,1,10,11, 4,2,0,0) \|\| \
	146	LTTNG_RT_KERNEL_RANGE(3,18,27,26, 3,19,0,0) \|\| \
	147	LTTNG_RT_KERNEL_RANGE(3,14,61,63, 3,15,0,0) \|\| \
	148	LTTNG_RT_KERNEL_RANGE(3,12,54,73, 3,13,0,0) \|\| \
	149	LTTNG_RT_KERNEL_RANGE(3,10,97,106, 3,11,0,0) \|\| \
	150	LTTNG_RT_KERNEL_RANGE(3,4,110,139, 3,5,0,0) \|\| \
	151	LTTNG_RT_KERNEL_RANGE(3,2,77,111, 3,3,0,0))
ffcf2393 MD	152	LTTNG_TRACEPOINT_EVENT_CLASS(sched_wakeup_template,
	153
	154	TP_PROTO(struct task_struct *p),
	155
	156	TP_ARGS(p),
	157
	158	TP_FIELDS(
	159	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	160	ctf_integer(pid_t, tid, p->pid)
5aa835c0	161	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
ffcf2393 MD	162	ctf_integer(int, target_cpu, task_cpu(p))
	163	)
	164	)
	165	#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0)) */
3bc29f0a	166	LTTNG_TRACEPOINT_EVENT_CLASS(sched_wakeup_template,
f62b389e MD	167
	168	TP_PROTO(struct task_struct *p, int success),
	169
	170	TP_ARGS(p, success),
	171
f127e61e MD	172	TP_FIELDS(
	173	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	174	ctf_integer(pid_t, tid, p->pid)
5aa835c0	175	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
f127e61e	176	ctf_integer(int, success, success)
f127e61e	177	ctf_integer(int, target_cpu, task_cpu(p))
7c68b363	178	)
f62b389e	179	)
ffcf2393	180	#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0)) */
f62b389e	181
b6c74835	182	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0) \|\| \
95ff37b6 MJ	183	LTTNG_RT_KERNEL_RANGE(4,1,10,11, 4,2,0,0) \|\| \
	184	LTTNG_RT_KERNEL_RANGE(3,18,27,26, 3,19,0,0) \|\| \
	185	LTTNG_RT_KERNEL_RANGE(3,14,61,63, 3,15,0,0) \|\| \
	186	LTTNG_RT_KERNEL_RANGE(3,12,54,73, 3,13,0,0) \|\| \
	187	LTTNG_RT_KERNEL_RANGE(3,10,97,106, 3,11,0,0) \|\| \
	188	LTTNG_RT_KERNEL_RANGE(3,4,110,139, 3,5,0,0) \|\| \
	189	LTTNG_RT_KERNEL_RANGE(3,2,77,111, 3,3,0,0))
ffcf2393 MD	190
	191	/*
	192	* Tracepoint called when waking a task; this tracepoint is guaranteed to be
	193	* called from the waking context.
	194	*/
	195	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_waking,
	196	TP_PROTO(struct task_struct *p),
	197	TP_ARGS(p))
	198
	199	/*
	200	* Tracepoint called when the task is actually woken; p->state == TASK_RUNNNG.
	201	* It it not always called from the waking context.
	202	*/
	203	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup,
	204	TP_PROTO(struct task_struct *p),
	205	TP_ARGS(p))
	206
	207	/*
	208	* Tracepoint for waking up a new task:
	209	*/
	210	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup_new,
	211	TP_PROTO(struct task_struct *p),
	212	TP_ARGS(p))
	213
ead65427	214	#else
3a523f5b	215
3bc29f0a	216	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup,
f62b389e MD	217	TP_PROTO(struct task_struct *p, int success),
	218	TP_ARGS(p, success))
	219
	220	/*
	221	* Tracepoint for waking up a new task:
	222	*/
3bc29f0a	223	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup_new,
f62b389e MD	224	TP_PROTO(struct task_struct *p, int success),
	225	TP_ARGS(p, success))
	226
ead65427	227	#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0)) */
3a523f5b	228
f62b389e MD	229	/*
	230	* Tracepoint for task switches, performed by the scheduler:
	231	*/
3bc29f0a	232	LTTNG_TRACEPOINT_EVENT(sched_switch,
f62b389e	233
3ee729fe MD	234	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0))
	235	TP_PROTO(bool preempt,
	236	struct task_struct *prev,
	237	struct task_struct *next),
	238
	239	TP_ARGS(preempt, prev, next),
ead65427	240	#else
f62b389e MD	241	TP_PROTO(struct task_struct *prev,
	242	struct task_struct *next),
	243
	244	TP_ARGS(prev, next),
ead65427	245	#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0)) */
f62b389e	246
f127e61e MD	247	TP_FIELDS(
	248	ctf_array_text(char, prev_comm, prev->comm, TASK_COMM_LEN)
	249	ctf_integer(pid_t, prev_tid, prev->pid)
	250	ctf_integer(int, prev_prio, prev->prio - MAX_RT_PRIO)
3ee729fe MD	251	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0))
3ee729fe MD	252	ctf_integer(long, prev_state, __trace_sched_switch_state(preempt, prev))
7c68b363	253	#else
ead65427	254	ctf_integer(long, prev_state, __trace_sched_switch_state(prev))
7c68b363	255	#endif
f127e61e MD	256	ctf_array_text(char, next_comm, next->comm, TASK_COMM_LEN)
	257	ctf_integer(pid_t, next_tid, next->pid)
	258	ctf_integer(int, next_prio, next->prio - MAX_RT_PRIO)
	259	)
f62b389e MD	260	)
	261
	262	/*
	263	* Tracepoint for a task being migrated:
	264	*/
3bc29f0a	265	LTTNG_TRACEPOINT_EVENT(sched_migrate_task,
f62b389e MD	266
	267	TP_PROTO(struct task_struct *p, int dest_cpu),
	268
	269	TP_ARGS(p, dest_cpu),
	270
f127e61e MD	271	TP_FIELDS(
	272	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	273	ctf_integer(pid_t, tid, p->pid)
	274	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	275	ctf_integer(int, orig_cpu, task_cpu(p))
	276	ctf_integer(int, dest_cpu, dest_cpu)
	277	)
f62b389e MD	278	)
f62b389e MD	279
3bc29f0a	280	LTTNG_TRACEPOINT_EVENT_CLASS(sched_process_template,
f62b389e MD	281
	282	TP_PROTO(struct task_struct *p),
	283
	284	TP_ARGS(p),
	285
f127e61e MD	286	TP_FIELDS(
	287	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	288	ctf_integer(pid_t, tid, p->pid)
	289	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	290	)
f62b389e MD	291	)
	292
	293	/*
	294	* Tracepoint for freeing a task:
	295	*/
3bc29f0a	296	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_free,
f62b389e MD	297	TP_PROTO(struct task_struct *p),
f62b389e MD	298	TP_ARGS(p))
216b6baa	299
f62b389e MD	300
	301	/*
	302	* Tracepoint for a task exiting:
	303	*/
3bc29f0a	304	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_exit,
f62b389e MD	305	TP_PROTO(struct task_struct *p),
	306	TP_ARGS(p))
	307
	308	/*
	309	* Tracepoint for waiting on task to unschedule:
	310	*/
3bc29f0a	311	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_wait_task,
f62b389e MD	312	TP_PROTO(struct task_struct *p),
	313	TP_ARGS(p))
	314
	315	/*
	316	* Tracepoint for a waiting task:
	317	*/
3bc29f0a	318	LTTNG_TRACEPOINT_EVENT(sched_process_wait,
f62b389e MD	319
	320	TP_PROTO(struct pid *pid),
	321
	322	TP_ARGS(pid),
	323
f127e61e MD	324	TP_FIELDS(
	325	ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
	326	ctf_integer(pid_t, tid, pid_nr(pid))
	327	ctf_integer(int, prio, current->prio - MAX_RT_PRIO)
	328	)
f62b389e MD	329	)
	330
	331	/*
79b18ef7 MD	332	* Tracepoint for do_fork.
	333	* Saving both TID and PID information, especially for the child, allows
	334	* trace analyzers to distinguish between creation of a new process and
	335	* creation of a new thread. Newly created processes will have child_tid
	336	* == child_pid, while creation of a thread yields to child_tid !=
	337	* child_pid.
f62b389e	338	*/
7bbf55ea	339	LTTNG_TRACEPOINT_EVENT_CODE(sched_process_fork,
f62b389e MD	340
	341	TP_PROTO(struct task_struct parent, struct task_struct child),
	342
	343	TP_ARGS(parent, child),
	344
7bbf55ea SL	345	TP_locvar(
	346	pid_t vtids[LTTNG_MAX_PID_NS_LEVEL];
	347	unsigned int ns_level;
	348	),
	349
265822ae	350	TP_code_pre(
7bbf55ea SL	351	if (child) {
	352	struct pid *child_pid;
	353	unsigned int i;
	354
	355	child_pid = task_pid(child);
	356	tp_locvar->ns_level =
	357	min_t(unsigned int, child_pid->level + 1,
	358	LTTNG_MAX_PID_NS_LEVEL);
	359	for (i = 0; i < tp_locvar->ns_level; i++)
	360	tp_locvar->vtids[i] = child_pid->numbers[i].nr;
	361	}
	362	),
	363
f127e61e MD	364	TP_FIELDS(
	365	ctf_array_text(char, parent_comm, parent->comm, TASK_COMM_LEN)
	366	ctf_integer(pid_t, parent_tid, parent->pid)
	367	ctf_integer(pid_t, parent_pid, parent->tgid)
7bbf55ea SL	368	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0))
	369	ctf_integer(unsigned int, parent_ns_inum,
	370	({
	371	unsigned int parent_ns_inum = 0;
	372
	373	if (parent) {
	374	struct pid_namespace *pid_ns;
	375
	376	pid_ns = task_active_pid_ns(parent);
	377	if (pid_ns)
	378	parent_ns_inum =
45b27d78	379	pid_ns->lttng_ns_inum;
7bbf55ea SL	380	}
	381	parent_ns_inum;
	382	}))
	383	#endif
f127e61e MD	384	ctf_array_text(char, child_comm, child->comm, TASK_COMM_LEN)
f127e61e MD	385	ctf_integer(pid_t, child_tid, child->pid)
7bbf55ea	386	ctf_sequence(pid_t, vtids, tp_locvar->vtids, u8, tp_locvar->ns_level)
f127e61e	387	ctf_integer(pid_t, child_pid, child->tgid)
7bbf55ea SL	388	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0))
	389	ctf_integer(unsigned int, child_ns_inum,
	390	({
	391	unsigned int child_ns_inum = 0;
	392
	393	if (child) {
	394	struct pid_namespace *pid_ns;
	395
	396	pid_ns = task_active_pid_ns(child);
	397	if (pid_ns)
	398	child_ns_inum =
45b27d78	399	pid_ns->lttng_ns_inum;
7bbf55ea SL	400	}
	401	child_ns_inum;
	402	}))
	403	#endif
265822ae MD	404	),
	405
	406	TP_code_post()
f62b389e MD	407	)
f62b389e MD	408
c94b2508	409	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,4,0))
46142a81 PW	410	/*
	411	* Tracepoint for exec:
	412	*/
3bc29f0a	413	LTTNG_TRACEPOINT_EVENT(sched_process_exec,
46142a81 PW	414
	415	TP_PROTO(struct task_struct *p, pid_t old_pid,
	416	struct linux_binprm *bprm),
	417
	418	TP_ARGS(p, old_pid, bprm),
	419
f127e61e MD	420	TP_FIELDS(
	421	ctf_string(filename, bprm->filename)
	422	ctf_integer(pid_t, tid, p->pid)
	423	ctf_integer(pid_t, old_tid, old_pid)
	424	)
46142a81	425	)
c94b2508	426	#endif
46142a81	427
f62b389e MD	428	/*
	429	* XXX the below sched_stat tracepoints only apply to SCHED_OTHER/BATCH/IDLE
	430	* adding sched_stat support to SCHED_FIFO/RR would be welcome.
	431	*/
3bc29f0a	432	LTTNG_TRACEPOINT_EVENT_CLASS(sched_stat_template,
f62b389e MD	433
	434	TP_PROTO(struct task_struct *tsk, u64 delay),
	435
	436	TP_ARGS(tsk, delay),
	437
f127e61e MD	438	TP_FIELDS(
	439	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	440	ctf_integer(pid_t, tid, tsk->pid)
	441	ctf_integer(u64, delay, delay)
f62b389e	442	)
f62b389e MD	443	)
	444
	445
	446	/*
	447	* Tracepoint for accounting wait time (time the task is runnable
	448	* but not actually running due to scheduler contention).
	449	*/
3bc29f0a	450	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_wait,
f62b389e MD	451	TP_PROTO(struct task_struct *tsk, u64 delay),
	452	TP_ARGS(tsk, delay))
	453
	454	/*
	455	* Tracepoint for accounting sleep time (time the task is not runnable,
	456	* including iowait, see below).
	457	*/
3bc29f0a	458	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_sleep,
f62b389e MD	459	TP_PROTO(struct task_struct *tsk, u64 delay),
	460	TP_ARGS(tsk, delay))
	461
	462	/*
	463	* Tracepoint for accounting iowait time (time the task is not runnable
	464	* due to waiting on IO to complete).
	465	*/
3bc29f0a	466	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_iowait,
f62b389e MD	467	TP_PROTO(struct task_struct *tsk, u64 delay),
	468	TP_ARGS(tsk, delay))
	469
7c68b363 AG	470	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,3,0))
	471	/*
	472	* Tracepoint for accounting blocked time (time the task is in uninterruptible).
	473	*/
3bc29f0a	474	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_blocked,
7c68b363 AG	475	TP_PROTO(struct task_struct *tsk, u64 delay),
	476	TP_ARGS(tsk, delay))
	477	#endif
	478
f62b389e MD	479	/*
	480	* Tracepoint for accounting runtime (time the task is executing
	481	* on a CPU).
	482	*/
3bc29f0a	483	LTTNG_TRACEPOINT_EVENT(sched_stat_runtime,
f62b389e MD	484
	485	TP_PROTO(struct task_struct *tsk, u64 runtime, u64 vruntime),
	486
	487	TP_ARGS(tsk, runtime, vruntime),
	488
f127e61e MD	489	TP_FIELDS(
	490	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	491	ctf_integer(pid_t, tid, tsk->pid)
	492	ctf_integer(u64, runtime, runtime)
	493	ctf_integer(u64, vruntime, vruntime)
f62b389e	494	)
f62b389e MD	495	)
f62b389e MD	496
262ab189	497	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,12,0) \|\| \
46dded2d MJ	498	LTTNG_RT_KERNEL_RANGE(4,9,27,18, 4,10,0,0) \|\| \
46dded2d MJ	499	LTTNG_RT_KERNEL_RANGE(4,11,5,1, 4,12,0,0))
673e9a03 MJ	500	/*
	501	* Tracepoint for showing priority inheritance modifying a tasks
	502	* priority.
	503	*/
	504	LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,
	505
	506	TP_PROTO(struct task_struct tsk, struct task_struct pi_task),
	507
	508	TP_ARGS(tsk, pi_task),
	509
	510	TP_FIELDS(
	511	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	512	ctf_integer(pid_t, tid, tsk->pid)
	513	ctf_integer(int, oldprio, tsk->prio - MAX_RT_PRIO)
	514	ctf_integer(int, newprio, pi_task ? pi_task->prio - MAX_RT_PRIO : tsk->prio - MAX_RT_PRIO)
	515	)
	516	)
ead65427	517	#else
f62b389e MD	518	/*
	519	* Tracepoint for showing priority inheritance modifying a tasks
	520	* priority.
	521	*/
3bc29f0a	522	LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,
f62b389e MD	523
	524	TP_PROTO(struct task_struct *tsk, int newprio),
	525
	526	TP_ARGS(tsk, newprio),
	527
f127e61e MD	528	TP_FIELDS(
	529	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	530	ctf_integer(pid_t, tid, tsk->pid)
	531	ctf_integer(int, oldprio, tsk->prio - MAX_RT_PRIO)
	532	ctf_integer(int, newprio, newprio - MAX_RT_PRIO)
	533	)
f62b389e	534	)
7c68b363	535	#endif
f62b389e	536
3bc29f0a	537	#endif /* LTTNG_TRACE_SCHED_H */
f62b389e MD	538
f62b389e MD	539	/* This part must be outside protection */
6ec43db8	540	#include <probes/define_trace.h>