[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_RT_KERNEL_RANGE(4,1,10,11, 4,2,0,0))
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,

#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 - MAX_RT_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) || \
	LTTNG_RT_KERNEL_RANGE(4,1,10,11, 4,2,0,0))

/*
 * 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))

#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(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)
	)
)
#elif (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
6ec43db8	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	*/
b6c74835 MJ	148	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0) \|\| \
b6c74835 MJ	149	LTTNG_RT_KERNEL_RANGE(4,1,10,11, 4,2,0,0))
ffcf2393 MD	150	LTTNG_TRACEPOINT_EVENT_CLASS(sched_wakeup_template,
	151
	152	TP_PROTO(struct task_struct *p),
	153
	154	TP_ARGS(p),
	155
	156	TP_FIELDS(
	157	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	158	ctf_integer(pid_t, tid, p->pid)
5aa835c0	159	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
ffcf2393 MD	160	ctf_integer(int, target_cpu, task_cpu(p))
	161	)
	162	)
	163	#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0)) */
3bc29f0a	164	LTTNG_TRACEPOINT_EVENT_CLASS(sched_wakeup_template,
f62b389e	165
3a523f5b	166	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
f62b389e MD	167	TP_PROTO(struct task_struct *p, int success),
	168
	169	TP_ARGS(p, success),
3a523f5b MD	170	#else
	171	TP_PROTO(struct rq rq, struct task_struct p, int success),
	172
	173	TP_ARGS(rq, p, success),
	174	#endif
f62b389e	175
f127e61e MD	176	TP_FIELDS(
	177	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	178	ctf_integer(pid_t, tid, p->pid)
5aa835c0	179	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
f127e61e	180	ctf_integer(int, success, success)
7c68b363	181	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,32))
f127e61e	182	ctf_integer(int, target_cpu, task_cpu(p))
7c68b363	183	#endif
7c68b363	184	)
f62b389e	185	)
ffcf2393	186	#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0)) */
f62b389e	187
b6c74835 MJ	188	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0) \|\| \
b6c74835 MJ	189	LTTNG_RT_KERNEL_RANGE(4,1,10,11, 4,2,0,0))
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
	214	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
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
3a523f5b MD	227	#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
3a523f5b MD	228
3bc29f0a	229	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup,
3a523f5b MD	230	TP_PROTO(struct rq rq, struct task_struct p, int success),
	231	TP_ARGS(rq, p, success))
	232
	233	/*
	234	* Tracepoint for waking up a new task:
	235	*/
3bc29f0a	236	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup_new,
3a523f5b MD	237	TP_PROTO(struct rq rq, struct task_struct p, int success),
	238	TP_ARGS(rq, p, success))
	239
	240	#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
	241
f62b389e MD	242	/*
	243	* Tracepoint for task switches, performed by the scheduler:
	244	*/
3bc29f0a	245	LTTNG_TRACEPOINT_EVENT(sched_switch,
f62b389e	246
3ee729fe MD	247	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0))
	248	TP_PROTO(bool preempt,
	249	struct task_struct *prev,
	250	struct task_struct *next),
	251
	252	TP_ARGS(preempt, prev, next),
	253	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
f62b389e MD	254	TP_PROTO(struct task_struct *prev,
	255	struct task_struct *next),
	256
	257	TP_ARGS(prev, next),
3a523f5b MD	258	#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
	259	TP_PROTO(struct rq rq, struct task_struct prev,
	260	struct task_struct *next),
	261
	262	TP_ARGS(rq, prev, next),
	263	#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
f62b389e	264
f127e61e MD	265	TP_FIELDS(
	266	ctf_array_text(char, prev_comm, prev->comm, TASK_COMM_LEN)
	267	ctf_integer(pid_t, prev_tid, prev->pid)
	268	ctf_integer(int, prev_prio, prev->prio - MAX_RT_PRIO)
3ee729fe MD	269	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,4,0))
	270	ctf_integer(long, prev_state, __trace_sched_switch_state(preempt, prev))
	271	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
f127e61e	272	ctf_integer(long, prev_state, __trace_sched_switch_state(prev))
7c68b363	273	#else
f127e61e	274	ctf_integer(long, prev_state, prev->state)
7c68b363	275	#endif
f127e61e MD	276	ctf_array_text(char, next_comm, next->comm, TASK_COMM_LEN)
	277	ctf_integer(pid_t, next_tid, next->pid)
	278	ctf_integer(int, next_prio, next->prio - MAX_RT_PRIO)
	279	)
f62b389e MD	280	)
	281
	282	/*
	283	* Tracepoint for a task being migrated:
	284	*/
3bc29f0a	285	LTTNG_TRACEPOINT_EVENT(sched_migrate_task,
f62b389e MD	286
	287	TP_PROTO(struct task_struct *p, int dest_cpu),
	288
	289	TP_ARGS(p, dest_cpu),
	290
f127e61e MD	291	TP_FIELDS(
	292	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	293	ctf_integer(pid_t, tid, p->pid)
	294	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	295	ctf_integer(int, orig_cpu, task_cpu(p))
	296	ctf_integer(int, dest_cpu, dest_cpu)
	297	)
f62b389e MD	298	)
f62b389e MD	299
3bc29f0a	300	LTTNG_TRACEPOINT_EVENT_CLASS(sched_process_template,
f62b389e MD	301
	302	TP_PROTO(struct task_struct *p),
	303
	304	TP_ARGS(p),
	305
f127e61e MD	306	TP_FIELDS(
	307	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	308	ctf_integer(pid_t, tid, p->pid)
	309	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	310	)
f62b389e MD	311	)
	312
	313	/*
	314	* Tracepoint for freeing a task:
	315	*/
3bc29f0a	316	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_free,
f62b389e MD	317	TP_PROTO(struct task_struct *p),
f62b389e MD	318	TP_ARGS(p))
216b6baa	319
f62b389e MD	320
	321	/*
	322	* Tracepoint for a task exiting:
	323	*/
3bc29f0a	324	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_exit,
f62b389e MD	325	TP_PROTO(struct task_struct *p),
	326	TP_ARGS(p))
	327
	328	/*
	329	* Tracepoint for waiting on task to unschedule:
	330	*/
3a523f5b	331	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35))
3bc29f0a	332	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_wait_task,
f62b389e MD	333	TP_PROTO(struct task_struct *p),
f62b389e MD	334	TP_ARGS(p))
3a523f5b	335	#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
3bc29f0a	336	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_wait_task,
3a523f5b MD	337	TP_PROTO(struct rq rq, struct task_struct p),
	338	TP_ARGS(rq, p))
	339	#endif /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35)) */
f62b389e MD	340
	341	/*
	342	* Tracepoint for a waiting task:
	343	*/
3bc29f0a	344	LTTNG_TRACEPOINT_EVENT(sched_process_wait,
f62b389e MD	345
	346	TP_PROTO(struct pid *pid),
	347
	348	TP_ARGS(pid),
	349
f127e61e MD	350	TP_FIELDS(
	351	ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
	352	ctf_integer(pid_t, tid, pid_nr(pid))
	353	ctf_integer(int, prio, current->prio - MAX_RT_PRIO)
	354	)
f62b389e MD	355	)
	356
	357	/*
79b18ef7 MD	358	* Tracepoint for do_fork.
	359	* Saving both TID and PID information, especially for the child, allows
	360	* trace analyzers to distinguish between creation of a new process and
	361	* creation of a new thread. Newly created processes will have child_tid
	362	* == child_pid, while creation of a thread yields to child_tid !=
	363	* child_pid.
f62b389e	364	*/
7bbf55ea	365	LTTNG_TRACEPOINT_EVENT_CODE(sched_process_fork,
f62b389e MD	366
	367	TP_PROTO(struct task_struct parent, struct task_struct child),
	368
	369	TP_ARGS(parent, child),
	370
7bbf55ea SL	371	TP_locvar(
	372	pid_t vtids[LTTNG_MAX_PID_NS_LEVEL];
	373	unsigned int ns_level;
	374	),
	375
265822ae	376	TP_code_pre(
7bbf55ea SL	377	if (child) {
	378	struct pid *child_pid;
	379	unsigned int i;
	380
	381	child_pid = task_pid(child);
	382	tp_locvar->ns_level =
	383	min_t(unsigned int, child_pid->level + 1,
	384	LTTNG_MAX_PID_NS_LEVEL);
	385	for (i = 0; i < tp_locvar->ns_level; i++)
	386	tp_locvar->vtids[i] = child_pid->numbers[i].nr;
	387	}
	388	),
	389
f127e61e MD	390	TP_FIELDS(
	391	ctf_array_text(char, parent_comm, parent->comm, TASK_COMM_LEN)
	392	ctf_integer(pid_t, parent_tid, parent->pid)
	393	ctf_integer(pid_t, parent_pid, parent->tgid)
7bbf55ea SL	394	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0))
	395	ctf_integer(unsigned int, parent_ns_inum,
	396	({
	397	unsigned int parent_ns_inum = 0;
	398
	399	if (parent) {
	400	struct pid_namespace *pid_ns;
	401
	402	pid_ns = task_active_pid_ns(parent);
	403	if (pid_ns)
	404	parent_ns_inum =
	405	pid_ns->lttng_proc_inum;
	406	}
	407	parent_ns_inum;
	408	}))
	409	#endif
f127e61e MD	410	ctf_array_text(char, child_comm, child->comm, TASK_COMM_LEN)
f127e61e MD	411	ctf_integer(pid_t, child_tid, child->pid)
7bbf55ea	412	ctf_sequence(pid_t, vtids, tp_locvar->vtids, u8, tp_locvar->ns_level)
f127e61e	413	ctf_integer(pid_t, child_pid, child->tgid)
7bbf55ea SL	414	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,8,0))
	415	ctf_integer(unsigned int, child_ns_inum,
	416	({
	417	unsigned int child_ns_inum = 0;
	418
	419	if (child) {
	420	struct pid_namespace *pid_ns;
	421
	422	pid_ns = task_active_pid_ns(child);
	423	if (pid_ns)
	424	child_ns_inum =
	425	pid_ns->lttng_proc_inum;
	426	}
	427	child_ns_inum;
	428	}))
	429	#endif
265822ae MD	430	),
	431
	432	TP_code_post()
f62b389e MD	433	)
f62b389e MD	434
7c68b363 AG	435	#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,33))
	436	/*
	437	* Tracepoint for sending a signal:
	438	*/
3bc29f0a	439	LTTNG_TRACEPOINT_EVENT(sched_signal_send,
7c68b363 AG	440
	441	TP_PROTO(int sig, struct task_struct *p),
	442
	443	TP_ARGS(sig, p),
	444
f127e61e MD	445	TP_FIELDS(
	446	ctf_integer(int, sig, sig)
	447	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	448	ctf_integer(pid_t, tid, p->pid)
	449	)
7c68b363 AG	450	)
	451	#endif
	452
c94b2508	453	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,4,0))
46142a81 PW	454	/*
	455	* Tracepoint for exec:
	456	*/
3bc29f0a	457	LTTNG_TRACEPOINT_EVENT(sched_process_exec,
46142a81 PW	458
	459	TP_PROTO(struct task_struct *p, pid_t old_pid,
	460	struct linux_binprm *bprm),
	461
	462	TP_ARGS(p, old_pid, bprm),
	463
f127e61e MD	464	TP_FIELDS(
	465	ctf_string(filename, bprm->filename)
	466	ctf_integer(pid_t, tid, p->pid)
	467	ctf_integer(pid_t, old_tid, old_pid)
	468	)
46142a81	469	)
c94b2508	470	#endif
46142a81	471
7c68b363	472	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,32))
f62b389e MD	473	/*
	474	* XXX the below sched_stat tracepoints only apply to SCHED_OTHER/BATCH/IDLE
	475	* adding sched_stat support to SCHED_FIFO/RR would be welcome.
	476	*/
3bc29f0a	477	LTTNG_TRACEPOINT_EVENT_CLASS(sched_stat_template,
f62b389e MD	478
	479	TP_PROTO(struct task_struct *tsk, u64 delay),
	480
	481	TP_ARGS(tsk, delay),
	482
f127e61e MD	483	TP_FIELDS(
	484	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	485	ctf_integer(pid_t, tid, tsk->pid)
	486	ctf_integer(u64, delay, delay)
f62b389e	487	)
f62b389e MD	488	)
	489
	490
	491	/*
	492	* Tracepoint for accounting wait time (time the task is runnable
	493	* but not actually running due to scheduler contention).
	494	*/
3bc29f0a	495	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_wait,
f62b389e MD	496	TP_PROTO(struct task_struct *tsk, u64 delay),
	497	TP_ARGS(tsk, delay))
	498
	499	/*
	500	* Tracepoint for accounting sleep time (time the task is not runnable,
	501	* including iowait, see below).
	502	*/
3bc29f0a	503	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_sleep,
f62b389e MD	504	TP_PROTO(struct task_struct *tsk, u64 delay),
	505	TP_ARGS(tsk, delay))
	506
	507	/*
	508	* Tracepoint for accounting iowait time (time the task is not runnable
	509	* due to waiting on IO to complete).
	510	*/
3bc29f0a	511	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_iowait,
f62b389e MD	512	TP_PROTO(struct task_struct *tsk, u64 delay),
	513	TP_ARGS(tsk, delay))
	514
7c68b363 AG	515	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,3,0))
	516	/*
	517	* Tracepoint for accounting blocked time (time the task is in uninterruptible).
	518	*/
3bc29f0a	519	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_blocked,
7c68b363 AG	520	TP_PROTO(struct task_struct *tsk, u64 delay),
	521	TP_ARGS(tsk, delay))
	522	#endif
	523
f62b389e MD	524	/*
	525	* Tracepoint for accounting runtime (time the task is executing
	526	* on a CPU).
	527	*/
3bc29f0a	528	LTTNG_TRACEPOINT_EVENT(sched_stat_runtime,
f62b389e MD	529
	530	TP_PROTO(struct task_struct *tsk, u64 runtime, u64 vruntime),
	531
	532	TP_ARGS(tsk, runtime, vruntime),
	533
f127e61e MD	534	TP_FIELDS(
	535	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	536	ctf_integer(pid_t, tid, tsk->pid)
	537	ctf_integer(u64, runtime, runtime)
	538	ctf_integer(u64, vruntime, vruntime)
f62b389e	539	)
f62b389e	540	)
7c68b363	541	#endif
f62b389e	542
0395b92c	543	#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,12,0) \|\| \
959a7d80 MJ	544	LTTNG_RT_KERNEL_RANGE(4,9,27,18, 4,10,0,0) \|\| \
959a7d80 MJ	545	LTTNG_RT_KERNEL_RANGE(4,11,5,1, 4,12,0,0))
35970083 MJ	546	/*
	547	* Tracepoint for showing priority inheritance modifying a tasks
	548	* priority.
	549	*/
	550	LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,
	551
	552	TP_PROTO(struct task_struct tsk, struct task_struct pi_task),
	553
	554	TP_ARGS(tsk, pi_task),
	555
	556	TP_FIELDS(
	557	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	558	ctf_integer(pid_t, tid, tsk->pid)
	559	ctf_integer(int, oldprio, tsk->prio - MAX_RT_PRIO)
	560	ctf_integer(int, newprio, pi_task ? pi_task->prio - MAX_RT_PRIO : tsk->prio - MAX_RT_PRIO)
	561	)
	562	)
	563	#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,37))
f62b389e MD	564	/*
	565	* Tracepoint for showing priority inheritance modifying a tasks
	566	* priority.
	567	*/
3bc29f0a	568	LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,
f62b389e MD	569
	570	TP_PROTO(struct task_struct *tsk, int newprio),
	571
	572	TP_ARGS(tsk, newprio),
	573
f127e61e MD	574	TP_FIELDS(
	575	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	576	ctf_integer(pid_t, tid, tsk->pid)
	577	ctf_integer(int, oldprio, tsk->prio - MAX_RT_PRIO)
	578	ctf_integer(int, newprio, newprio - MAX_RT_PRIO)
	579	)
f62b389e	580	)
7c68b363	581	#endif
f62b389e	582
3bc29f0a	583	#endif /* LTTNG_TRACE_SCHED_H */
f62b389e MD	584
f62b389e MD	585	/* This part must be outside protection */
6ec43db8	586	#include <probes/define_trace.h>