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

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

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

#include <lttng/tracepoint-event.h>
#include <linux/sched.h>
#include <linux/pid_namespace.h>
#include <linux/binfmts.h>
#include <lttng/kernel-version.h>
#include <linux/sched/rt.h>

#define LTTNG_MAX_PID_NS_LEVEL 32

#ifndef _TRACE_SCHED_DEF_
#define _TRACE_SCHED_DEF_

#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(5,18,0))

static inline long __trace_sched_switch_state(bool preempt,
		unsigned int prev_state,
		struct task_struct *p)
{
        unsigned int state;

#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).
         */
        if (preempt)
                return TASK_REPORT_MAX;

        /*
         * task_state_index() uses fls() and returns a value from 0-8 range.
         * Decrement it by 1 (except TASK_RUNNING state i.e 0) before using
         * it for left shift operation to get the correct task->state
         * mapping.
         */
	state = __task_state_index(prev_state, p->exit_state);

        return state ? (1 << (state - 1)) : state;
}

#elif (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,15,0))

static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
{
        unsigned int state;

#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).
         */
        if (preempt)
                return TASK_REPORT_MAX;

        /*
         * task_state_index() uses fls() and returns a value from 0-8 range.
         * Decrement it by 1 (except TASK_RUNNING state i.e 0) before using
         * it for left shift operation to get the correct task->state
         * mapping.
         */
        state = task_state_index(p);

        return state ? (1 << (state - 1)) : state;
}

#elif (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,14,0))

static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
{
        unsigned int state;

#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).
         */
        if (preempt)
                return TASK_REPORT_MAX;

        /*
         * __get_task_state() uses fls() and returns a value from 0-8 range.
         * Decrement it by 1 (except TASK_RUNNING state i.e 0) before using
         * it for left shift operation to get the correct task->state
         * mapping.
         */
        state = __get_task_state(p);

        return state ? (1 << (state - 1)) : state;
}

#else

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;
}
#endif

#endif /* _TRACE_SCHED_DEF_ */

#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
/*
 * Enumeration of the task state bitmask.
 * Only bit flags are enumerated here, not composition of states.
 */
LTTNG_TRACEPOINT_ENUM(task_state,
	TP_ENUM_VALUES(
		ctf_enum_value("TASK_RUNNING", TASK_RUNNING)
		ctf_enum_value("TASK_INTERRUPTIBLE", TASK_INTERRUPTIBLE)
		ctf_enum_value("TASK_UNINTERRUPTIBLE", TASK_UNINTERRUPTIBLE)
		ctf_enum_value("TASK_STOPPED", __TASK_STOPPED)
		ctf_enum_value("TASK_TRACED", __TASK_TRACED)
		ctf_enum_value("EXIT_DEAD", EXIT_DEAD)
		ctf_enum_value("EXIT_ZOMBIE", EXIT_ZOMBIE)
		ctf_enum_value("TASK_PARKED", TASK_PARKED)
		ctf_enum_value("TASK_DEAD", TASK_DEAD)
		ctf_enum_value("TASK_WAKEKILL", TASK_WAKEKILL)
		ctf_enum_value("TASK_WAKING", TASK_WAKING)
		ctf_enum_value("TASK_NOLOAD", TASK_NOLOAD)

#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,8,0))
		ctf_enum_value("TASK_NEW", TASK_NEW)
#endif /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,8,0)) */

		ctf_enum_value("TASK_STATE_MAX", TASK_STATE_MAX)
	)
)
#endif /* CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM */

/*
 * 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:
 */
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))
	)
)

/*
 * 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))

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

#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(5,18,0))
LTTNG_TRACEPOINT_EVENT(sched_switch,

	TP_PROTO(bool preempt,
		struct task_struct *prev,
		struct task_struct *next,
		unsigned int prev_state),

	TP_ARGS(preempt, prev, next, prev_state),

	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)
#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
		ctf_enum(task_state, long, prev_state, __trace_sched_switch_state(preempt, prev_state, prev))
#else
		ctf_integer(long, prev_state, __trace_sched_switch_state(preempt, prev_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)
	)
)

#else

LTTNG_TRACEPOINT_EVENT(sched_switch,

	TP_PROTO(bool preempt,
		 struct task_struct *prev,
		 struct task_struct *next),

	TP_ARGS(preempt, prev, next),

	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)
#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
		ctf_enum(task_state, long, prev_state, __trace_sched_switch_state(preempt, prev))
#else
		ctf_integer(long, prev_state, __trace_sched_switch_state(preempt, 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)
	)
)
#endif

/*
 * 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)
		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->ns.inum;
				}
				parent_ns_inum;
			}))
		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)
		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->ns.inum;
				}
				child_ns_inum;
			}))
	),

	TP_code_post()
)

/*
 * 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)
	)
)

/*
 * 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))

/*
 * 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))

/*
 * 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 (LTTNG_LINUX_VERSION_CODE >= LTTNG_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 <lttng/define_trace.h>
Commit	Line	Data
b7cdc182	1	/* SPDX-License-Identifier: GPL-2.0-only */
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
3b4aafcb	8	#include <lttng/tracepoint-event.h>
f62b389e	9	#include <linux/sched.h>
7bbf55ea	10	#include <linux/pid_namespace.h>
7c68b363	11	#include <linux/binfmts.h>
5f4c791e	12	#include <lttng/kernel-version.h>
090db00e	13	#include <linux/sched/rt.h>
7bbf55ea SL	14
	15	#define LTTNG_MAX_PID_NS_LEVEL 32
	16
f62b389e MD	17	#ifndef _TRACE_SCHED_DEF_
	18	#define _TRACE_SCHED_DEF_
	19
8e52fd71 MJ	20	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(5,18,0))
	21
	22	static inline long __trace_sched_switch_state(bool preempt,
	23	unsigned int prev_state,
	24	struct task_struct *p)
	25	{
	26	unsigned int state;
	27
	28	#ifdef CONFIG_SCHED_DEBUG
	29	BUG_ON(p != current);
	30	#endif /* CONFIG_SCHED_DEBUG */
	31
	32	/*
	33	* Preemption ignores task state, therefore preempted tasks are always
	34	* RUNNING (we will not have dequeued if state != RUNNING).
	35	*/
	36	if (preempt)
	37	return TASK_REPORT_MAX;
	38
	39	/*
	40	* task_state_index() uses fls() and returns a value from 0-8 range.
	41	* Decrement it by 1 (except TASK_RUNNING state i.e 0) before using
	42	* it for left shift operation to get the correct task->state
	43	* mapping.
	44	*/
	45	state = __task_state_index(prev_state, p->exit_state);
	46
	47	return state ? (1 << (state - 1)) : state;
	48	}
	49
	50	#elif (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,15,0))
27e6eda7 GAPG	51
	52	static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
	53	{
	54	unsigned int state;
	55
	56	#ifdef CONFIG_SCHED_DEBUG
	57	BUG_ON(p != current);
	58	#endif /* CONFIG_SCHED_DEBUG */
	59
	60	/*
	61	* Preemption ignores task state, therefore preempted tasks are always
	62	* RUNNING (we will not have dequeued if state != RUNNING).
	63	*/
	64	if (preempt)
	65	return TASK_REPORT_MAX;
	66
	67	/*
	68	* task_state_index() uses fls() and returns a value from 0-8 range.
	69	* Decrement it by 1 (except TASK_RUNNING state i.e 0) before using
	70	* it for left shift operation to get the correct task->state
	71	* mapping.
	72	*/
	73	state = task_state_index(p);
	74
	75	return state ? (1 << (state - 1)) : state;
	76	}
	77
5f4c791e	78	#elif (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,14,0))
27e6eda7 GAPG	79
	80	static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
	81	{
	82	unsigned int state;
	83
	84	#ifdef CONFIG_SCHED_DEBUG
	85	BUG_ON(p != current);
	86	#endif /* CONFIG_SCHED_DEBUG */
	87
	88	/*
	89	* Preemption ignores task state, therefore preempted tasks are always
	90	* RUNNING (we will not have dequeued if state != RUNNING).
	91	*/
	92	if (preempt)
	93	return TASK_REPORT_MAX;
	94
	95	/*
	96	* __get_task_state() uses fls() and returns a value from 0-8 range.
	97	* Decrement it by 1 (except TASK_RUNNING state i.e 0) before using
	98	* it for left shift operation to get the correct task->state
	99	* mapping.
	100	*/
	101	state = __get_task_state(p);
	102
	103	return state ? (1 << (state - 1)) : state;
	104	}
	105
dc818cc2	106	#else
3ee729fe MD	107
	108	static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
	109	{
	110	#ifdef CONFIG_SCHED_DEBUG
	111	BUG_ON(p != current);
	112	#endif /* CONFIG_SCHED_DEBUG */
	113	/*
	114	* Preemption ignores task state, therefore preempted tasks are always RUNNING
	115	* (we will not have dequeued if state != RUNNING).
	116	*/
	117	return preempt ? TASK_RUNNING \| TASK_STATE_MAX : p->state;
	118	}
7c68b363 AG	119	#endif
7c68b363 AG	120
f62b389e MD	121	#endif /* _TRACE_SCHED_DEF_ */
f62b389e MD	122
e54b3828	123	#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
721caea4 GB	124	/*
	125	* Enumeration of the task state bitmask.
	126	* Only bit flags are enumerated here, not composition of states.
	127	*/
	128	LTTNG_TRACEPOINT_ENUM(task_state,
	129	TP_ENUM_VALUES(
	130	ctf_enum_value("TASK_RUNNING", TASK_RUNNING)
	131	ctf_enum_value("TASK_INTERRUPTIBLE", TASK_INTERRUPTIBLE)
	132	ctf_enum_value("TASK_UNINTERRUPTIBLE", TASK_UNINTERRUPTIBLE)
	133	ctf_enum_value("TASK_STOPPED", __TASK_STOPPED)
	134	ctf_enum_value("TASK_TRACED", __TASK_TRACED)
	135	ctf_enum_value("EXIT_DEAD", EXIT_DEAD)
	136	ctf_enum_value("EXIT_ZOMBIE", EXIT_ZOMBIE)
721caea4	137	ctf_enum_value("TASK_PARKED", TASK_PARKED)
721caea4 GB	138	ctf_enum_value("TASK_DEAD", TASK_DEAD)
	139	ctf_enum_value("TASK_WAKEKILL", TASK_WAKEKILL)
	140	ctf_enum_value("TASK_WAKING", TASK_WAKING)
721caea4	141	ctf_enum_value("TASK_NOLOAD", TASK_NOLOAD)
721caea4	142
5f4c791e	143	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,8,0))
721caea4	144	ctf_enum_value("TASK_NEW", TASK_NEW)
5f4c791e	145	#endif /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,8,0)) */
721caea4 GB	146
	147	ctf_enum_value("TASK_STATE_MAX", TASK_STATE_MAX)
	148	)
	149	)
e54b3828	150	#endif /* CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM */
721caea4	151
f62b389e MD	152	/*
	153	* Tracepoint for calling kthread_stop, performed to end a kthread:
	154	*/
3bc29f0a	155	LTTNG_TRACEPOINT_EVENT(sched_kthread_stop,
f62b389e MD	156
	157	TP_PROTO(struct task_struct *t),
	158
	159	TP_ARGS(t),
	160
f127e61e MD	161	TP_FIELDS(
	162	ctf_array_text(char, comm, t->comm, TASK_COMM_LEN)
	163	ctf_integer(pid_t, tid, t->pid)
	164	)
f62b389e MD	165	)
	166
	167	/*
	168	* Tracepoint for the return value of the kthread stopping:
	169	*/
3bc29f0a	170	LTTNG_TRACEPOINT_EVENT(sched_kthread_stop_ret,
f62b389e MD	171
	172	TP_PROTO(int ret),
	173
	174	TP_ARGS(ret),
	175
f127e61e MD	176	TP_FIELDS(
	177	ctf_integer(int, ret, ret)
	178	)
f62b389e MD	179	)
	180
	181	/*
	182	* Tracepoint for waking up a task:
	183	*/
ffcf2393 MD	184	LTTNG_TRACEPOINT_EVENT_CLASS(sched_wakeup_template,
	185
	186	TP_PROTO(struct task_struct *p),
	187
	188	TP_ARGS(p),
	189
	190	TP_FIELDS(
	191	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	192	ctf_integer(pid_t, tid, p->pid)
5aa835c0	193	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
ffcf2393 MD	194	ctf_integer(int, target_cpu, task_cpu(p))
	195	)
	196	)
ffcf2393 MD	197
	198	/*
	199	* Tracepoint called when waking a task; this tracepoint is guaranteed to be
	200	* called from the waking context.
	201	*/
	202	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_waking,
	203	TP_PROTO(struct task_struct *p),
	204	TP_ARGS(p))
	205
	206	/*
	207	* Tracepoint called when the task is actually woken; p->state == TASK_RUNNNG.
	208	* It it not always called from the waking context.
	209	*/
	210	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup,
	211	TP_PROTO(struct task_struct *p),
	212	TP_ARGS(p))
	213
	214	/*
	215	* Tracepoint for waking up a new task:
	216	*/
	217	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup_new,
	218	TP_PROTO(struct task_struct *p),
	219	TP_ARGS(p))
	220
f62b389e MD	221	/*
	222	* Tracepoint for task switches, performed by the scheduler:
	223	*/
8e52fd71 MJ	224
8e52fd71 MJ	225	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(5,18,0))
3bc29f0a	226	LTTNG_TRACEPOINT_EVENT(sched_switch,
f62b389e	227
3ee729fe	228	TP_PROTO(bool preempt,
8e52fd71	229	struct task_struct *prev,
9c5b8de3 MJ	230	struct task_struct *next,
9c5b8de3 MJ	231	unsigned int prev_state),
3ee729fe	232
9c5b8de3	233	TP_ARGS(preempt, prev, next, prev_state),
8e52fd71 MJ	234
	235	TP_FIELDS(
	236	ctf_array_text(char, prev_comm, prev->comm, TASK_COMM_LEN)
	237	ctf_integer(pid_t, prev_tid, prev->pid)
	238	ctf_integer(int, prev_prio, prev->prio - MAX_RT_PRIO)
	239	#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
	240	ctf_enum(task_state, long, prev_state, __trace_sched_switch_state(preempt, prev_state, prev))
ead65427	241	#else
8e52fd71 MJ	242	ctf_integer(long, prev_state, __trace_sched_switch_state(preempt, prev_state, prev))
	243	#endif
	244	ctf_array_text(char, next_comm, next->comm, TASK_COMM_LEN)
	245	ctf_integer(pid_t, next_tid, next->pid)
	246	ctf_integer(int, next_prio, next->prio - MAX_RT_PRIO)
	247	)
	248	)
	249
dc818cc2	250	#else
8e52fd71 MJ	251
	252	LTTNG_TRACEPOINT_EVENT(sched_switch,
	253
	254	TP_PROTO(bool preempt,
	255	struct task_struct *prev,
f62b389e MD	256	struct task_struct *next),
f62b389e MD	257
8e52fd71	258	TP_ARGS(preempt, prev, next),
f62b389e	259
f127e61e MD	260	TP_FIELDS(
	261	ctf_array_text(char, prev_comm, prev->comm, TASK_COMM_LEN)
	262	ctf_integer(pid_t, prev_tid, prev->pid)
	263	ctf_integer(int, prev_prio, prev->prio - MAX_RT_PRIO)
e54b3828	264	#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
721caea4	265	ctf_enum(task_state, long, prev_state, __trace_sched_switch_state(preempt, prev))
7c68b363	266	#else
e54b3828 MJ	267	ctf_integer(long, prev_state, __trace_sched_switch_state(preempt, prev))
e54b3828 MJ	268	#endif
8e52fd71 MJ	269	ctf_array_text(char, next_comm, next->comm, TASK_COMM_LEN)
	270	ctf_integer(pid_t, next_tid, next->pid)
	271	ctf_integer(int, next_prio, next->prio - MAX_RT_PRIO)
	272	)
	273	)
8e52fd71	274	#endif
f62b389e MD	275
	276	/*
	277	* Tracepoint for a task being migrated:
	278	*/
3bc29f0a	279	LTTNG_TRACEPOINT_EVENT(sched_migrate_task,
f62b389e MD	280
	281	TP_PROTO(struct task_struct *p, int dest_cpu),
	282
	283	TP_ARGS(p, dest_cpu),
	284
f127e61e MD	285	TP_FIELDS(
	286	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	287	ctf_integer(pid_t, tid, p->pid)
	288	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	289	ctf_integer(int, orig_cpu, task_cpu(p))
	290	ctf_integer(int, dest_cpu, dest_cpu)
	291	)
f62b389e MD	292	)
f62b389e MD	293
3bc29f0a	294	LTTNG_TRACEPOINT_EVENT_CLASS(sched_process_template,
f62b389e MD	295
	296	TP_PROTO(struct task_struct *p),
	297
	298	TP_ARGS(p),
	299
f127e61e MD	300	TP_FIELDS(
	301	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	302	ctf_integer(pid_t, tid, p->pid)
	303	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	304	)
f62b389e MD	305	)
	306
	307	/*
	308	* Tracepoint for freeing a task:
	309	*/
3bc29f0a	310	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_free,
f62b389e MD	311	TP_PROTO(struct task_struct *p),
f62b389e MD	312	TP_ARGS(p))
216b6baa	313
f62b389e MD	314
	315	/*
	316	* Tracepoint for a task exiting:
	317	*/
3bc29f0a	318	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_exit,
f62b389e MD	319	TP_PROTO(struct task_struct *p),
	320	TP_ARGS(p))
	321
	322	/*
	323	* Tracepoint for waiting on task to unschedule:
	324	*/
3bc29f0a	325	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_wait_task,
f62b389e MD	326	TP_PROTO(struct task_struct *p),
	327	TP_ARGS(p))
	328
	329	/*
	330	* Tracepoint for a waiting task:
	331	*/
3bc29f0a	332	LTTNG_TRACEPOINT_EVENT(sched_process_wait,
f62b389e MD	333
	334	TP_PROTO(struct pid *pid),
	335
	336	TP_ARGS(pid),
	337
f127e61e MD	338	TP_FIELDS(
	339	ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
	340	ctf_integer(pid_t, tid, pid_nr(pid))
	341	ctf_integer(int, prio, current->prio - MAX_RT_PRIO)
	342	)
f62b389e MD	343	)
	344
	345	/*
79b18ef7 MD	346	* Tracepoint for do_fork.
	347	* Saving both TID and PID information, especially for the child, allows
	348	* trace analyzers to distinguish between creation of a new process and
	349	* creation of a new thread. Newly created processes will have child_tid
	350	* == child_pid, while creation of a thread yields to child_tid !=
	351	* child_pid.
f62b389e	352	*/
7bbf55ea	353	LTTNG_TRACEPOINT_EVENT_CODE(sched_process_fork,
f62b389e MD	354
	355	TP_PROTO(struct task_struct parent, struct task_struct child),
	356
	357	TP_ARGS(parent, child),
	358
7bbf55ea SL	359	TP_locvar(
	360	pid_t vtids[LTTNG_MAX_PID_NS_LEVEL];
	361	unsigned int ns_level;
	362	),
	363
265822ae	364	TP_code_pre(
7bbf55ea SL	365	if (child) {
	366	struct pid *child_pid;
	367	unsigned int i;
	368
	369	child_pid = task_pid(child);
	370	tp_locvar->ns_level =
	371	min_t(unsigned int, child_pid->level + 1,
	372	LTTNG_MAX_PID_NS_LEVEL);
	373	for (i = 0; i < tp_locvar->ns_level; i++)
	374	tp_locvar->vtids[i] = child_pid->numbers[i].nr;
	375	}
	376	),
	377
f127e61e MD	378	TP_FIELDS(
	379	ctf_array_text(char, parent_comm, parent->comm, TASK_COMM_LEN)
	380	ctf_integer(pid_t, parent_tid, parent->pid)
	381	ctf_integer(pid_t, parent_pid, parent->tgid)
7bbf55ea SL	382	ctf_integer(unsigned int, parent_ns_inum,
	383	({
	384	unsigned int parent_ns_inum = 0;
	385
	386	if (parent) {
	387	struct pid_namespace *pid_ns;
	388
	389	pid_ns = task_active_pid_ns(parent);
	390	if (pid_ns)
	391	parent_ns_inum =
93e5bcec	392	pid_ns->ns.inum;
7bbf55ea SL	393	}
	394	parent_ns_inum;
	395	}))
f127e61e MD	396	ctf_array_text(char, child_comm, child->comm, TASK_COMM_LEN)
f127e61e MD	397	ctf_integer(pid_t, child_tid, child->pid)
7bbf55ea	398	ctf_sequence(pid_t, vtids, tp_locvar->vtids, u8, tp_locvar->ns_level)
f127e61e	399	ctf_integer(pid_t, child_pid, child->tgid)
7bbf55ea SL	400	ctf_integer(unsigned int, child_ns_inum,
	401	({
	402	unsigned int child_ns_inum = 0;
	403
	404	if (child) {
	405	struct pid_namespace *pid_ns;
	406
	407	pid_ns = task_active_pid_ns(child);
	408	if (pid_ns)
	409	child_ns_inum =
93e5bcec	410	pid_ns->ns.inum;
7bbf55ea SL	411	}
	412	child_ns_inum;
	413	}))
265822ae MD	414	),
	415
	416	TP_code_post()
f62b389e MD	417	)
f62b389e MD	418
46142a81 PW	419	/*
	420	* Tracepoint for exec:
	421	*/
3bc29f0a	422	LTTNG_TRACEPOINT_EVENT(sched_process_exec,
46142a81 PW	423
	424	TP_PROTO(struct task_struct *p, pid_t old_pid,
	425	struct linux_binprm *bprm),
	426
	427	TP_ARGS(p, old_pid, bprm),
	428
f127e61e MD	429	TP_FIELDS(
	430	ctf_string(filename, bprm->filename)
	431	ctf_integer(pid_t, tid, p->pid)
	432	ctf_integer(pid_t, old_tid, old_pid)
	433	)
46142a81 PW	434	)
46142a81 PW	435
f62b389e MD	436	/*
	437	* XXX the below sched_stat tracepoints only apply to SCHED_OTHER/BATCH/IDLE
	438	* adding sched_stat support to SCHED_FIFO/RR would be welcome.
	439	*/
3bc29f0a	440	LTTNG_TRACEPOINT_EVENT_CLASS(sched_stat_template,
f62b389e MD	441
	442	TP_PROTO(struct task_struct *tsk, u64 delay),
	443
	444	TP_ARGS(tsk, delay),
	445
f127e61e MD	446	TP_FIELDS(
	447	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	448	ctf_integer(pid_t, tid, tsk->pid)
	449	ctf_integer(u64, delay, delay)
f62b389e	450	)
f62b389e MD	451	)
	452
	453
	454	/*
	455	* Tracepoint for accounting wait time (time the task is runnable
	456	* but not actually running due to scheduler contention).
	457	*/
3bc29f0a	458	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_wait,
f62b389e MD	459	TP_PROTO(struct task_struct *tsk, u64 delay),
	460	TP_ARGS(tsk, delay))
	461
	462	/*
	463	* Tracepoint for accounting sleep time (time the task is not runnable,
	464	* including iowait, see below).
	465	*/
3bc29f0a	466	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_sleep,
f62b389e MD	467	TP_PROTO(struct task_struct *tsk, u64 delay),
	468	TP_ARGS(tsk, delay))
	469
	470	/*
	471	* Tracepoint for accounting iowait time (time the task is not runnable
	472	* due to waiting on IO to complete).
	473	*/
3bc29f0a	474	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_iowait,
f62b389e MD	475	TP_PROTO(struct task_struct *tsk, u64 delay),
	476	TP_ARGS(tsk, delay))
	477
7c68b363 AG	478	/*
	479	* Tracepoint for accounting blocked time (time the task is in uninterruptible).
	480	*/
3bc29f0a	481	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_blocked,
7c68b363 AG	482	TP_PROTO(struct task_struct *tsk, u64 delay),
7c68b363 AG	483	TP_ARGS(tsk, delay))
7c68b363	484
f62b389e MD	485	/*
	486	* Tracepoint for accounting runtime (time the task is executing
	487	* on a CPU).
	488	*/
3bc29f0a	489	LTTNG_TRACEPOINT_EVENT(sched_stat_runtime,
f62b389e MD	490
	491	TP_PROTO(struct task_struct *tsk, u64 runtime, u64 vruntime),
	492
	493	TP_ARGS(tsk, runtime, vruntime),
	494
f127e61e MD	495	TP_FIELDS(
	496	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	497	ctf_integer(pid_t, tid, tsk->pid)
	498	ctf_integer(u64, runtime, runtime)
	499	ctf_integer(u64, vruntime, vruntime)
f62b389e	500	)
f62b389e MD	501	)
f62b389e MD	502
5f4c791e	503	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,12,0) \|\| \
46dded2d MJ	504	LTTNG_RT_KERNEL_RANGE(4,9,27,18, 4,10,0,0) \|\| \
46dded2d MJ	505	LTTNG_RT_KERNEL_RANGE(4,11,5,1, 4,12,0,0))
673e9a03 MJ	506	/*
	507	* Tracepoint for showing priority inheritance modifying a tasks
	508	* priority.
	509	*/
	510	LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,
	511
	512	TP_PROTO(struct task_struct tsk, struct task_struct pi_task),
	513
	514	TP_ARGS(tsk, pi_task),
	515
	516	TP_FIELDS(
	517	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	518	ctf_integer(pid_t, tid, tsk->pid)
	519	ctf_integer(int, oldprio, tsk->prio - MAX_RT_PRIO)
	520	ctf_integer(int, newprio, pi_task ? pi_task->prio - MAX_RT_PRIO : tsk->prio - MAX_RT_PRIO)
	521	)
	522	)
ead65427	523	#else
f62b389e MD	524	/*
	525	* Tracepoint for showing priority inheritance modifying a tasks
	526	* priority.
	527	*/
3bc29f0a	528	LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,
f62b389e MD	529
	530	TP_PROTO(struct task_struct *tsk, int newprio),
	531
	532	TP_ARGS(tsk, newprio),
	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(int, oldprio, tsk->prio - MAX_RT_PRIO)
	538	ctf_integer(int, newprio, newprio - MAX_RT_PRIO)
	539	)
f62b389e	540	)
7c68b363	541	#endif
f62b389e	542
3bc29f0a	543	#endif /* LTTNG_TRACE_SCHED_H */
f62b389e MD	544
f62b389e MD	545	/* This part must be outside protection */
3b4aafcb	546	#include <lttng/define_trace.h>