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

#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

#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,19,0))
#define lttng_proc_inum ns.inum
#else
#define lttng_proc_inum proc_inum
#endif

#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;
}

#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))

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_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)
		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,

#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
	TP_PROTO(struct task_struct *p, int success),

	TP_ARGS(p, success),
#else
	TP_PROTO(struct rq *rq, struct task_struct *p, int success),

	TP_ARGS(rq, p, success),
#endif

	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)
		ctf_integer(int, success, success)
#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,32))
		ctf_integer(int, target_cpu, task_cpu(p))
#endif
	)
)
#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0)) */

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

/*
 * 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))

#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))

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))

#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */

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

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

#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */

/*
 * 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),
#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
	TP_PROTO(struct task_struct *prev,
		 struct task_struct *next),

	TP_ARGS(prev, next),
#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
	TP_PROTO(struct rq *rq, struct task_struct *prev,
		 struct task_struct *next),

	TP_ARGS(rq, prev, next),
#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */

	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))
#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
		ctf_integer(long, prev_state, __trace_sched_switch_state(prev))
#else
		ctf_integer(long, prev_state, prev->state)
#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:
 */
#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_wait_task,
	TP_PROTO(struct task_struct *p),
	TP_ARGS(p))
#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_wait_task,
	TP_PROTO(struct rq *rq, struct task_struct *p),
	TP_ARGS(rq, p))
#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */

/*
 * 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_proc_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_proc_inum;
				}
				child_ns_inum;
			}))
#endif
	),

	TP_code_post()
)

#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,33))
/*
 * Tracepoint for sending a signal:
 */
LTTNG_TRACEPOINT_EVENT(sched_signal_send,

	TP_PROTO(int sig, struct task_struct *p),

	TP_ARGS(sig, p),

	TP_FIELDS(
		ctf_integer(int, sig, sig)
		ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, tid, p->pid)
	)
)
#endif

#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

#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,32))
/*
 * 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)
	)
)
#endif

#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,37))
/*
 * 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
f62b389e MD	1	#undef TRACE_SYSTEM
	2	#define TRACE_SYSTEM sched
	3
3bc29f0a MD	4	#if !defined(LTTNG_TRACE_SCHED_H) \|\| defined(TRACE_HEADER_MULTI_READ)
3bc29f0a MD	5	#define LTTNG_TRACE_SCHED_H
f62b389e	6
3bc29f0a	7	#include "../../../probes/lttng-tracepoint-event.h"
f62b389e	8	#include <linux/sched.h>
7bbf55ea	9	#include <linux/pid_namespace.h>
7c68b363 AG	10	#include <linux/binfmts.h>
7c68b363 AG	11	#include <linux/version.h>
090db00e MJ	12	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,9,0))
	13	#include <linux/sched/rt.h>
	14	#endif
f62b389e	15
7bbf55ea SL	16	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,19,0))
	17	#define lttng_proc_inum ns.inum
	18	#else
	19	#define lttng_proc_inum proc_inum
	20	#endif
	21
	22	#define LTTNG_MAX_PID_NS_LEVEL 32
	23
f62b389e MD	24	#ifndef _TRACE_SCHED_DEF_
	25	#define _TRACE_SCHED_DEF_
	26
3ee729fe MD	27	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0))
	28
	29	static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
	30	{
	31	#ifdef CONFIG_SCHED_DEBUG
	32	BUG_ON(p != current);
	33	#endif /* CONFIG_SCHED_DEBUG */
	34	/*
	35	* Preemption ignores task state, therefore preempted tasks are always RUNNING
	36	* (we will not have dequeued if state != RUNNING).
	37	*/
	38	return preempt ? TASK_RUNNING \| TASK_STATE_MAX : p->state;
	39	}
	40
	41	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(3,19,0))
857b6f4f MD	42
	43	static inline long __trace_sched_switch_state(struct task_struct *p)
	44	{
	45	long state = p->state;
	46
	47	#ifdef CONFIG_PREEMPT
	48	#ifdef CONFIG_SCHED_DEBUG
	49	BUG_ON(p != current);
	50	#endif /* CONFIG_SCHED_DEBUG */
	51	/*
	52	* For all intents and purposes a preempted task is a running task.
	53	*/
	54	if (preempt_count() & PREEMPT_ACTIVE)
	55	state = TASK_RUNNING \| TASK_STATE_MAX;
	56	#endif /* CONFIG_PREEMPT */
	57
	58	return state;
	59	}
	60
	61	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(3,13,0))
33673ee7 MD	62
	63	static inline long __trace_sched_switch_state(struct task_struct *p)
	64	{
	65	long state = p->state;
	66
	67	#ifdef CONFIG_PREEMPT
	68	/*
	69	* For all intents and purposes a preempted task is a running task.
	70	*/
	71	if (task_preempt_count(p) & PREEMPT_ACTIVE)
	72	state = TASK_RUNNING \| TASK_STATE_MAX;
	73	#endif
	74
	75	return state;
	76	}
	77
	78	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(3,2,0))
7c68b363	79
f62b389e MD	80	static inline long __trace_sched_switch_state(struct task_struct *p)
	81	{
	82	long state = p->state;
	83
	84	#ifdef CONFIG_PREEMPT
	85	/*
	86	* For all intents and purposes a preempted task is a running task.
	87	*/
	88	if (task_thread_info(p)->preempt_count & PREEMPT_ACTIVE)
7c68b363	89	state = TASK_RUNNING \| TASK_STATE_MAX;
7c68b363	90	#endif
33673ee7 MD	91
	92	return state;
	93	}
	94
	95	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
	96
	97	static inline long __trace_sched_switch_state(struct task_struct *p)
	98	{
	99	long state = p->state;
	100
	101	#ifdef CONFIG_PREEMPT
	102	/*
	103	* For all intents and purposes a preempted task is a running task.
	104	*/
	105	if (task_thread_info(p)->preempt_count & PREEMPT_ACTIVE)
	106	state = TASK_RUNNING;
f62b389e MD	107	#endif
	108
	109	return state;
	110	}
	111
7c68b363 AG	112	#endif
7c68b363 AG	113
f62b389e MD	114	#endif /* _TRACE_SCHED_DEF_ */
	115
	116	/*
	117	* Tracepoint for calling kthread_stop, performed to end a kthread:
	118	*/
3bc29f0a	119	LTTNG_TRACEPOINT_EVENT(sched_kthread_stop,
f62b389e MD	120
	121	TP_PROTO(struct task_struct *t),
	122
	123	TP_ARGS(t),
	124
f127e61e MD	125	TP_FIELDS(
	126	ctf_array_text(char, comm, t->comm, TASK_COMM_LEN)
	127	ctf_integer(pid_t, tid, t->pid)
	128	)
f62b389e MD	129	)
	130
	131	/*
	132	* Tracepoint for the return value of the kthread stopping:
	133	*/
3bc29f0a	134	LTTNG_TRACEPOINT_EVENT(sched_kthread_stop_ret,
f62b389e MD	135
	136	TP_PROTO(int ret),
	137
	138	TP_ARGS(ret),
	139
f127e61e MD	140	TP_FIELDS(
	141	ctf_integer(int, ret, ret)
	142	)
f62b389e MD	143	)
	144
	145	/*
	146	* Tracepoint for waking up a task:
	147	*/
ffcf2393 MD	148	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0))
	149	LTTNG_TRACEPOINT_EVENT_CLASS(sched_wakeup_template,
	150
	151	TP_PROTO(struct task_struct *p),
	152
	153	TP_ARGS(p),
	154
	155	TP_FIELDS(
	156	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	157	ctf_integer(pid_t, tid, p->pid)
	158	ctf_integer(int, prio, p->prio)
	159	ctf_integer(int, target_cpu, task_cpu(p))
	160	)
	161	)
	162	#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0)) */
3bc29f0a	163	LTTNG_TRACEPOINT_EVENT_CLASS(sched_wakeup_template,
f62b389e	164
3a523f5b	165	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
f62b389e MD	166	TP_PROTO(struct task_struct *p, int success),
	167
	168	TP_ARGS(p, success),
3a523f5b MD	169	#else
	170	TP_PROTO(struct rq rq, struct task_struct p, int success),
	171
	172	TP_ARGS(rq, p, success),
	173	#endif
f62b389e	174
f127e61e MD	175	TP_FIELDS(
	176	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	177	ctf_integer(pid_t, tid, p->pid)
	178	ctf_integer(int, prio, p->prio)
	179	ctf_integer(int, success, success)
7c68b363	180	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,32))
f127e61e	181	ctf_integer(int, target_cpu, task_cpu(p))
7c68b363	182	#endif
7c68b363	183	)
f62b389e	184	)
ffcf2393	185	#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0)) */
f62b389e	186
ffcf2393 MD	187	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0))
	188
	189	/*
	190	* Tracepoint called when waking a task; this tracepoint is guaranteed to be
	191	* called from the waking context.
	192	*/
	193	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_waking,
	194	TP_PROTO(struct task_struct *p),
	195	TP_ARGS(p))
	196
	197	/*
	198	* Tracepoint called when the task is actually woken; p->state == TASK_RUNNNG.
	199	* It it not always called from the waking context.
	200	*/
	201	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup,
	202	TP_PROTO(struct task_struct *p),
	203	TP_ARGS(p))
	204
	205	/*
	206	* Tracepoint for waking up a new task:
	207	*/
	208	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup_new,
	209	TP_PROTO(struct task_struct *p),
	210	TP_ARGS(p))
	211
	212	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
3a523f5b	213
3bc29f0a	214	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup,
f62b389e MD	215	TP_PROTO(struct task_struct *p, int success),
	216	TP_ARGS(p, success))
	217
	218	/*
	219	* Tracepoint for waking up a new task:
	220	*/
3bc29f0a	221	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup_new,
f62b389e MD	222	TP_PROTO(struct task_struct *p, int success),
	223	TP_ARGS(p, success))
	224
3a523f5b MD	225	#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
3a523f5b MD	226
3bc29f0a	227	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup,
3a523f5b MD	228	TP_PROTO(struct rq rq, struct task_struct p, int success),
	229	TP_ARGS(rq, p, success))
	230
	231	/*
	232	* Tracepoint for waking up a new task:
	233	*/
3bc29f0a	234	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup_new,
3a523f5b MD	235	TP_PROTO(struct rq rq, struct task_struct p, int success),
	236	TP_ARGS(rq, p, success))
	237
	238	#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
	239
f62b389e MD	240	/*
	241	* Tracepoint for task switches, performed by the scheduler:
	242	*/
3bc29f0a	243	LTTNG_TRACEPOINT_EVENT(sched_switch,
f62b389e	244
3ee729fe MD	245	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0))
	246	TP_PROTO(bool preempt,
	247	struct task_struct *prev,
	248	struct task_struct *next),
	249
	250	TP_ARGS(preempt, prev, next),
	251	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
f62b389e MD	252	TP_PROTO(struct task_struct *prev,
	253	struct task_struct *next),
	254
	255	TP_ARGS(prev, next),
3a523f5b MD	256	#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
	257	TP_PROTO(struct rq rq, struct task_struct prev,
	258	struct task_struct *next),
	259
	260	TP_ARGS(rq, prev, next),
	261	#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
f62b389e	262
f127e61e MD	263	TP_FIELDS(
	264	ctf_array_text(char, prev_comm, prev->comm, TASK_COMM_LEN)
	265	ctf_integer(pid_t, prev_tid, prev->pid)
	266	ctf_integer(int, prev_prio, prev->prio - MAX_RT_PRIO)
3ee729fe MD	267	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0))
	268	ctf_integer(long, prev_state, __trace_sched_switch_state(preempt, prev))
	269	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
f127e61e	270	ctf_integer(long, prev_state, __trace_sched_switch_state(prev))
7c68b363	271	#else
f127e61e	272	ctf_integer(long, prev_state, prev->state)
7c68b363	273	#endif
f127e61e MD	274	ctf_array_text(char, next_comm, next->comm, TASK_COMM_LEN)
	275	ctf_integer(pid_t, next_tid, next->pid)
	276	ctf_integer(int, next_prio, next->prio - MAX_RT_PRIO)
	277	)
f62b389e MD	278	)
	279
	280	/*
	281	* Tracepoint for a task being migrated:
	282	*/
3bc29f0a	283	LTTNG_TRACEPOINT_EVENT(sched_migrate_task,
f62b389e MD	284
	285	TP_PROTO(struct task_struct *p, int dest_cpu),
	286
	287	TP_ARGS(p, dest_cpu),
	288
f127e61e MD	289	TP_FIELDS(
	290	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	291	ctf_integer(pid_t, tid, p->pid)
	292	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	293	ctf_integer(int, orig_cpu, task_cpu(p))
	294	ctf_integer(int, dest_cpu, dest_cpu)
	295	)
f62b389e MD	296	)
f62b389e MD	297
3bc29f0a	298	LTTNG_TRACEPOINT_EVENT_CLASS(sched_process_template,
f62b389e MD	299
	300	TP_PROTO(struct task_struct *p),
	301
	302	TP_ARGS(p),
	303
f127e61e MD	304	TP_FIELDS(
	305	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	306	ctf_integer(pid_t, tid, p->pid)
	307	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	308	)
f62b389e MD	309	)
	310
	311	/*
	312	* Tracepoint for freeing a task:
	313	*/
3bc29f0a	314	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_free,
f62b389e MD	315	TP_PROTO(struct task_struct *p),
	316	TP_ARGS(p))
	317
	318
	319	/*
	320	* Tracepoint for a task exiting:
	321	*/
3bc29f0a	322	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_exit,
f62b389e MD	323	TP_PROTO(struct task_struct *p),
	324	TP_ARGS(p))
	325
	326	/*
	327	* Tracepoint for waiting on task to unschedule:
	328	*/
3a523f5b	329	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
3bc29f0a	330	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_wait_task,
f62b389e MD	331	TP_PROTO(struct task_struct *p),
f62b389e MD	332	TP_ARGS(p))
3a523f5b	333	#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
3bc29f0a	334	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_wait_task,
3a523f5b MD	335	TP_PROTO(struct rq rq, struct task_struct p),
	336	TP_ARGS(rq, p))
	337	#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
f62b389e MD	338
	339	/*
	340	* Tracepoint for a waiting task:
	341	*/
3bc29f0a	342	LTTNG_TRACEPOINT_EVENT(sched_process_wait,
f62b389e MD	343
	344	TP_PROTO(struct pid *pid),
	345
	346	TP_ARGS(pid),
	347
f127e61e MD	348	TP_FIELDS(
	349	ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
	350	ctf_integer(pid_t, tid, pid_nr(pid))
	351	ctf_integer(int, prio, current->prio - MAX_RT_PRIO)
	352	)
f62b389e MD	353	)
	354
	355	/*
79b18ef7 MD	356	* Tracepoint for do_fork.
	357	* Saving both TID and PID information, especially for the child, allows
	358	* trace analyzers to distinguish between creation of a new process and
	359	* creation of a new thread. Newly created processes will have child_tid
	360	* == child_pid, while creation of a thread yields to child_tid !=
	361	* child_pid.
f62b389e	362	*/
7bbf55ea	363	LTTNG_TRACEPOINT_EVENT_CODE(sched_process_fork,
f62b389e MD	364
	365	TP_PROTO(struct task_struct parent, struct task_struct child),
	366
	367	TP_ARGS(parent, child),
	368
7bbf55ea SL	369	TP_locvar(
	370	pid_t vtids[LTTNG_MAX_PID_NS_LEVEL];
	371	unsigned int ns_level;
	372	),
	373
265822ae	374	TP_code_pre(
7bbf55ea SL	375	if (child) {
	376	struct pid *child_pid;
	377	unsigned int i;
	378
	379	child_pid = task_pid(child);
	380	tp_locvar->ns_level =
	381	min_t(unsigned int, child_pid->level + 1,
	382	LTTNG_MAX_PID_NS_LEVEL);
	383	for (i = 0; i < tp_locvar->ns_level; i++)
	384	tp_locvar->vtids[i] = child_pid->numbers[i].nr;
	385	}
	386	),
	387
f127e61e MD	388	TP_FIELDS(
	389	ctf_array_text(char, parent_comm, parent->comm, TASK_COMM_LEN)
	390	ctf_integer(pid_t, parent_tid, parent->pid)
	391	ctf_integer(pid_t, parent_pid, parent->tgid)
7bbf55ea SL	392	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0))
	393	ctf_integer(unsigned int, parent_ns_inum,
	394	({
	395	unsigned int parent_ns_inum = 0;
	396
	397	if (parent) {
	398	struct pid_namespace *pid_ns;
	399
	400	pid_ns = task_active_pid_ns(parent);
	401	if (pid_ns)
	402	parent_ns_inum =
	403	pid_ns->lttng_proc_inum;
	404	}
	405	parent_ns_inum;
	406	}))
	407	#endif
f127e61e MD	408	ctf_array_text(char, child_comm, child->comm, TASK_COMM_LEN)
f127e61e MD	409	ctf_integer(pid_t, child_tid, child->pid)
7bbf55ea	410	ctf_sequence(pid_t, vtids, tp_locvar->vtids, u8, tp_locvar->ns_level)
f127e61e	411	ctf_integer(pid_t, child_pid, child->tgid)
7bbf55ea SL	412	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0))
	413	ctf_integer(unsigned int, child_ns_inum,
	414	({
	415	unsigned int child_ns_inum = 0;
	416
	417	if (child) {
	418	struct pid_namespace *pid_ns;
	419
	420	pid_ns = task_active_pid_ns(child);
	421	if (pid_ns)
	422	child_ns_inum =
	423	pid_ns->lttng_proc_inum;
	424	}
	425	child_ns_inum;
	426	}))
	427	#endif
265822ae MD	428	),
	429
	430	TP_code_post()
f62b389e MD	431	)
f62b389e MD	432
7c68b363 AG	433	#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,33))
	434	/*
	435	* Tracepoint for sending a signal:
	436	*/
3bc29f0a	437	LTTNG_TRACEPOINT_EVENT(sched_signal_send,
7c68b363 AG	438
	439	TP_PROTO(int sig, struct task_struct *p),
	440
	441	TP_ARGS(sig, p),
	442
f127e61e MD	443	TP_FIELDS(
	444	ctf_integer(int, sig, sig)
	445	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	446	ctf_integer(pid_t, tid, p->pid)
	447	)
7c68b363 AG	448	)
	449	#endif
	450
c94b2508	451	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,4,0))
46142a81 PW	452	/*
	453	* Tracepoint for exec:
	454	*/
3bc29f0a	455	LTTNG_TRACEPOINT_EVENT(sched_process_exec,
46142a81 PW	456
	457	TP_PROTO(struct task_struct *p, pid_t old_pid,
	458	struct linux_binprm *bprm),
	459
	460	TP_ARGS(p, old_pid, bprm),
	461
f127e61e MD	462	TP_FIELDS(
	463	ctf_string(filename, bprm->filename)
	464	ctf_integer(pid_t, tid, p->pid)
	465	ctf_integer(pid_t, old_tid, old_pid)
	466	)
46142a81	467	)
c94b2508	468	#endif
46142a81	469
7c68b363	470	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,32))
f62b389e MD	471	/*
	472	* XXX the below sched_stat tracepoints only apply to SCHED_OTHER/BATCH/IDLE
	473	* adding sched_stat support to SCHED_FIFO/RR would be welcome.
	474	*/
3bc29f0a	475	LTTNG_TRACEPOINT_EVENT_CLASS(sched_stat_template,
f62b389e MD	476
	477	TP_PROTO(struct task_struct *tsk, u64 delay),
	478
	479	TP_ARGS(tsk, delay),
	480
f127e61e MD	481	TP_FIELDS(
	482	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	483	ctf_integer(pid_t, tid, tsk->pid)
	484	ctf_integer(u64, delay, delay)
f62b389e	485	)
f62b389e MD	486	)
	487
	488
	489	/*
	490	* Tracepoint for accounting wait time (time the task is runnable
	491	* but not actually running due to scheduler contention).
	492	*/
3bc29f0a	493	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_wait,
f62b389e MD	494	TP_PROTO(struct task_struct *tsk, u64 delay),
	495	TP_ARGS(tsk, delay))
	496
	497	/*
	498	* Tracepoint for accounting sleep time (time the task is not runnable,
	499	* including iowait, see below).
	500	*/
3bc29f0a	501	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_sleep,
f62b389e MD	502	TP_PROTO(struct task_struct *tsk, u64 delay),
	503	TP_ARGS(tsk, delay))
	504
	505	/*
	506	* Tracepoint for accounting iowait time (time the task is not runnable
	507	* due to waiting on IO to complete).
	508	*/
3bc29f0a	509	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_iowait,
f62b389e MD	510	TP_PROTO(struct task_struct *tsk, u64 delay),
	511	TP_ARGS(tsk, delay))
	512
7c68b363 AG	513	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,3,0))
	514	/*
	515	* Tracepoint for accounting blocked time (time the task is in uninterruptible).
	516	*/
3bc29f0a	517	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_blocked,
7c68b363 AG	518	TP_PROTO(struct task_struct *tsk, u64 delay),
	519	TP_ARGS(tsk, delay))
	520	#endif
	521
f62b389e MD	522	/*
	523	* Tracepoint for accounting runtime (time the task is executing
	524	* on a CPU).
	525	*/
3bc29f0a	526	LTTNG_TRACEPOINT_EVENT(sched_stat_runtime,
f62b389e MD	527
	528	TP_PROTO(struct task_struct *tsk, u64 runtime, u64 vruntime),
	529
	530	TP_ARGS(tsk, runtime, vruntime),
	531
f127e61e MD	532	TP_FIELDS(
	533	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	534	ctf_integer(pid_t, tid, tsk->pid)
	535	ctf_integer(u64, runtime, runtime)
	536	ctf_integer(u64, vruntime, vruntime)
f62b389e	537	)
f62b389e	538	)
7c68b363	539	#endif
f62b389e	540
7c68b363	541	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,37))
f62b389e MD	542	/*
	543	* Tracepoint for showing priority inheritance modifying a tasks
	544	* priority.
	545	*/
3bc29f0a	546	LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,
f62b389e MD	547
	548	TP_PROTO(struct task_struct *tsk, int newprio),
	549
	550	TP_ARGS(tsk, newprio),
	551
f127e61e MD	552	TP_FIELDS(
	553	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	554	ctf_integer(pid_t, tid, tsk->pid)
	555	ctf_integer(int, oldprio, tsk->prio - MAX_RT_PRIO)
	556	ctf_integer(int, newprio, newprio - MAX_RT_PRIO)
	557	)
f62b389e	558	)
7c68b363	559	#endif
f62b389e	560
3bc29f0a	561	#endif /* LTTNG_TRACE_SCHED_H */
f62b389e MD	562
f62b389e MD	563	/* This part must be outside protection */
5b88d86e	564	#include "../../../probes/define_trace.h"