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

// SPDX-FileCopyrightText: 2011 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
//
// 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) \
	|| LTTNG_RHEL_KERNEL_RANGE(5,14,0,162,0,0, 5,15,0,0,0,0))

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_RHEL_KERNEL_RANGE(5,14,0,162,0,0, 5,15,0,0,0,0))
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))

#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(6,8,0) || \
	LTTNG_RHEL_KERNEL_RANGE(5,14,0,452,0,0, 5,15,0,0,0,0))
/*
 * 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),

	TP_ARGS(tsk, runtime),

	TP_FIELDS(
		ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, tid, tsk->pid)
		ctf_integer(u64, runtime, runtime)
	)
)
#else
/*
 * 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 (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
	1	// SPDX-FileCopyrightText: 2011 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
	2	//
	3	// SPDX-License-Identifier: GPL-2.0-only
	4
	5	#undef TRACE_SYSTEM
	6	#define TRACE_SYSTEM sched
	7
	8	#if !defined(LTTNG_TRACE_SCHED_H) \|\| defined(TRACE_HEADER_MULTI_READ)
	9	#define LTTNG_TRACE_SCHED_H
	10
	11	#include <lttng/tracepoint-event.h>
	12	#include <linux/sched.h>
	13	#include <linux/pid_namespace.h>
	14	#include <linux/binfmts.h>
	15	#include <lttng/kernel-version.h>
	16	#include <linux/sched/rt.h>
	17
	18	#define LTTNG_MAX_PID_NS_LEVEL 32
	19
	20	#ifndef _TRACE_SCHED_DEF_
	21	#define _TRACE_SCHED_DEF_
	22
	23	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(5,18,0) \
	24	\|\| LTTNG_RHEL_KERNEL_RANGE(5,14,0,162,0,0, 5,15,0,0,0,0))
	25
	26	static inline long __trace_sched_switch_state(bool preempt,
	27	unsigned int prev_state,
	28	struct task_struct *p)
	29	{
	30	unsigned int state;
	31
	32	#ifdef CONFIG_SCHED_DEBUG
	33	BUG_ON(p != current);
	34	#endif /* CONFIG_SCHED_DEBUG */
	35
	36	/*
	37	* Preemption ignores task state, therefore preempted tasks are always
	38	* RUNNING (we will not have dequeued if state != RUNNING).
	39	*/
	40	if (preempt)
	41	return TASK_REPORT_MAX;
	42
	43	/*
	44	* task_state_index() uses fls() and returns a value from 0-8 range.
	45	* Decrement it by 1 (except TASK_RUNNING state i.e 0) before using
	46	* it for left shift operation to get the correct task->state
	47	* mapping.
	48	*/
	49	state = __task_state_index(prev_state, p->exit_state);
	50
	51	return state ? (1 << (state - 1)) : state;
	52	}
	53
	54	#elif (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,15,0))
	55
	56	static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
	57	{
	58	unsigned int state;
	59
	60	#ifdef CONFIG_SCHED_DEBUG
	61	BUG_ON(p != current);
	62	#endif /* CONFIG_SCHED_DEBUG */
	63
	64	/*
	65	* Preemption ignores task state, therefore preempted tasks are always
	66	* RUNNING (we will not have dequeued if state != RUNNING).
	67	*/
	68	if (preempt)
	69	return TASK_REPORT_MAX;
	70
	71	/*
	72	* task_state_index() uses fls() and returns a value from 0-8 range.
	73	* Decrement it by 1 (except TASK_RUNNING state i.e 0) before using
	74	* it for left shift operation to get the correct task->state
	75	* mapping.
	76	*/
	77	state = task_state_index(p);
	78
	79	return state ? (1 << (state - 1)) : state;
	80	}
	81
	82	#elif (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,14,0))
	83
	84	static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
	85	{
	86	unsigned int state;
	87
	88	#ifdef CONFIG_SCHED_DEBUG
	89	BUG_ON(p != current);
	90	#endif /* CONFIG_SCHED_DEBUG */
	91
	92	/*
	93	* Preemption ignores task state, therefore preempted tasks are always
	94	* RUNNING (we will not have dequeued if state != RUNNING).
	95	*/
	96	if (preempt)
	97	return TASK_REPORT_MAX;
	98
	99	/*
	100	* __get_task_state() uses fls() and returns a value from 0-8 range.
	101	* Decrement it by 1 (except TASK_RUNNING state i.e 0) before using
	102	* it for left shift operation to get the correct task->state
	103	* mapping.
	104	*/
	105	state = __get_task_state(p);
	106
	107	return state ? (1 << (state - 1)) : state;
	108	}
	109
	110	#else
	111
	112	static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
	113	{
	114	#ifdef CONFIG_SCHED_DEBUG
	115	BUG_ON(p != current);
	116	#endif /* CONFIG_SCHED_DEBUG */
	117	/*
	118	* Preemption ignores task state, therefore preempted tasks are always RUNNING
	119	* (we will not have dequeued if state != RUNNING).
	120	*/
	121	return preempt ? TASK_RUNNING \| TASK_STATE_MAX : p->state;
	122	}
	123	#endif
	124
	125	#endif /* _TRACE_SCHED_DEF_ */
	126
	127	#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
	128	/*
	129	* Enumeration of the task state bitmask.
	130	* Only bit flags are enumerated here, not composition of states.
	131	*/
	132	LTTNG_TRACEPOINT_ENUM(task_state,
	133	TP_ENUM_VALUES(
	134	ctf_enum_value("TASK_RUNNING", TASK_RUNNING)
	135	ctf_enum_value("TASK_INTERRUPTIBLE", TASK_INTERRUPTIBLE)
	136	ctf_enum_value("TASK_UNINTERRUPTIBLE", TASK_UNINTERRUPTIBLE)
	137	ctf_enum_value("TASK_STOPPED", __TASK_STOPPED)
	138	ctf_enum_value("TASK_TRACED", __TASK_TRACED)
	139	ctf_enum_value("EXIT_DEAD", EXIT_DEAD)
	140	ctf_enum_value("EXIT_ZOMBIE", EXIT_ZOMBIE)
	141	ctf_enum_value("TASK_PARKED", TASK_PARKED)
	142	ctf_enum_value("TASK_DEAD", TASK_DEAD)
	143	ctf_enum_value("TASK_WAKEKILL", TASK_WAKEKILL)
	144	ctf_enum_value("TASK_WAKING", TASK_WAKING)
	145	ctf_enum_value("TASK_NOLOAD", TASK_NOLOAD)
	146
	147	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,8,0))
	148	ctf_enum_value("TASK_NEW", TASK_NEW)
	149	#endif /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,8,0)) */
	150
	151	ctf_enum_value("TASK_STATE_MAX", TASK_STATE_MAX)
	152	)
	153	)
	154	#endif /* CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM */
	155
	156	/*
	157	* Tracepoint for calling kthread_stop, performed to end a kthread:
	158	*/
	159	LTTNG_TRACEPOINT_EVENT(sched_kthread_stop,
	160
	161	TP_PROTO(struct task_struct *t),
	162
	163	TP_ARGS(t),
	164
	165	TP_FIELDS(
	166	ctf_array_text(char, comm, t->comm, TASK_COMM_LEN)
	167	ctf_integer(pid_t, tid, t->pid)
	168	)
	169	)
	170
	171	/*
	172	* Tracepoint for the return value of the kthread stopping:
	173	*/
	174	LTTNG_TRACEPOINT_EVENT(sched_kthread_stop_ret,
	175
	176	TP_PROTO(int ret),
	177
	178	TP_ARGS(ret),
	179
	180	TP_FIELDS(
	181	ctf_integer(int, ret, ret)
	182	)
	183	)
	184
	185	/*
	186	* Tracepoint for waking up a task:
	187	*/
	188	LTTNG_TRACEPOINT_EVENT_CLASS(sched_wakeup_template,
	189
	190	TP_PROTO(struct task_struct *p),
	191
	192	TP_ARGS(p),
	193
	194	TP_FIELDS(
	195	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	196	ctf_integer(pid_t, tid, p->pid)
	197	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	198	ctf_integer(int, target_cpu, task_cpu(p))
	199	)
	200	)
	201
	202	/*
	203	* Tracepoint called when waking a task; this tracepoint is guaranteed to be
	204	* called from the waking context.
	205	*/
	206	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_waking,
	207	TP_PROTO(struct task_struct *p),
	208	TP_ARGS(p))
	209
	210	/*
	211	* Tracepoint called when the task is actually woken; p->state == TASK_RUNNNG.
	212	* It it not always called from the waking context.
	213	*/
	214	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup,
	215	TP_PROTO(struct task_struct *p),
	216	TP_ARGS(p))
	217
	218	/*
	219	* Tracepoint for waking up a new task:
	220	*/
	221	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup_new,
	222	TP_PROTO(struct task_struct *p),
	223	TP_ARGS(p))
	224
	225	/*
	226	* Tracepoint for task switches, performed by the scheduler:
	227	*/
	228
	229	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(5,18,0) \
	230	\|\| LTTNG_RHEL_KERNEL_RANGE(5,14,0,162,0,0, 5,15,0,0,0,0))
	231	LTTNG_TRACEPOINT_EVENT(sched_switch,
	232
	233	TP_PROTO(bool preempt,
	234	struct task_struct *prev,
	235	struct task_struct *next,
	236	unsigned int prev_state),
	237
	238	TP_ARGS(preempt, prev, next, prev_state),
	239
	240	TP_FIELDS(
	241	ctf_array_text(char, prev_comm, prev->comm, TASK_COMM_LEN)
	242	ctf_integer(pid_t, prev_tid, prev->pid)
	243	ctf_integer(int, prev_prio, prev->prio - MAX_RT_PRIO)
	244	#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
	245	ctf_enum(task_state, long, prev_state, __trace_sched_switch_state(preempt, prev_state, prev))
	246	#else
	247	ctf_integer(long, prev_state, __trace_sched_switch_state(preempt, prev_state, prev))
	248	#endif
	249	ctf_array_text(char, next_comm, next->comm, TASK_COMM_LEN)
	250	ctf_integer(pid_t, next_tid, next->pid)
	251	ctf_integer(int, next_prio, next->prio - MAX_RT_PRIO)
	252	)
	253	)
	254
	255	#else
	256
	257	LTTNG_TRACEPOINT_EVENT(sched_switch,
	258
	259	TP_PROTO(bool preempt,
	260	struct task_struct *prev,
	261	struct task_struct *next),
	262
	263	TP_ARGS(preempt, prev, next),
	264
	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)
	269	#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
	270	ctf_enum(task_state, long, prev_state, __trace_sched_switch_state(preempt, prev))
	271	#else
	272	ctf_integer(long, prev_state, __trace_sched_switch_state(preempt, prev))
	273	#endif
	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	)
	278	)
	279	#endif
	280
	281	/*
	282	* Tracepoint for a task being migrated:
	283	*/
	284	LTTNG_TRACEPOINT_EVENT(sched_migrate_task,
	285
	286	TP_PROTO(struct task_struct *p, int dest_cpu),
	287
	288	TP_ARGS(p, dest_cpu),
	289
	290	TP_FIELDS(
	291	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	292	ctf_integer(pid_t, tid, p->pid)
	293	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	294	ctf_integer(int, orig_cpu, task_cpu(p))
	295	ctf_integer(int, dest_cpu, dest_cpu)
	296	)
	297	)
	298
	299	LTTNG_TRACEPOINT_EVENT_CLASS(sched_process_template,
	300
	301	TP_PROTO(struct task_struct *p),
	302
	303	TP_ARGS(p),
	304
	305	TP_FIELDS(
	306	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	307	ctf_integer(pid_t, tid, p->pid)
	308	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	309	)
	310	)
	311
	312	/*
	313	* Tracepoint for freeing a task:
	314	*/
	315	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_free,
	316	TP_PROTO(struct task_struct *p),
	317	TP_ARGS(p))
	318
	319
	320	/*
	321	* Tracepoint for a task exiting:
	322	*/
	323	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_exit,
	324	TP_PROTO(struct task_struct *p),
	325	TP_ARGS(p))
	326
	327	/*
	328	* Tracepoint for waiting on task to unschedule:
	329	*/
	330	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_wait_task,
	331	TP_PROTO(struct task_struct *p),
	332	TP_ARGS(p))
	333
	334	/*
	335	* Tracepoint for a waiting task:
	336	*/
	337	LTTNG_TRACEPOINT_EVENT(sched_process_wait,
	338
	339	TP_PROTO(struct pid *pid),
	340
	341	TP_ARGS(pid),
	342
	343	TP_FIELDS(
	344	ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
	345	ctf_integer(pid_t, tid, pid_nr(pid))
	346	ctf_integer(int, prio, current->prio - MAX_RT_PRIO)
	347	)
	348	)
	349
	350	/*
	351	* Tracepoint for do_fork.
	352	* Saving both TID and PID information, especially for the child, allows
	353	* trace analyzers to distinguish between creation of a new process and
	354	* creation of a new thread. Newly created processes will have child_tid
	355	* == child_pid, while creation of a thread yields to child_tid !=
	356	* child_pid.
	357	*/
	358	LTTNG_TRACEPOINT_EVENT_CODE(sched_process_fork,
	359
	360	TP_PROTO(struct task_struct parent, struct task_struct child),
	361
	362	TP_ARGS(parent, child),
	363
	364	TP_locvar(
	365	pid_t vtids[LTTNG_MAX_PID_NS_LEVEL];
	366	unsigned int ns_level;
	367	),
	368
	369	TP_code_pre(
	370	if (child) {
	371	struct pid *child_pid;
	372	unsigned int i;
	373
	374	child_pid = task_pid(child);
	375	tp_locvar->ns_level =
	376	min_t(unsigned int, child_pid->level + 1,
	377	LTTNG_MAX_PID_NS_LEVEL);
	378	for (i = 0; i < tp_locvar->ns_level; i++)
	379	tp_locvar->vtids[i] = child_pid->numbers[i].nr;
	380	}
	381	),
	382
	383	TP_FIELDS(
	384	ctf_array_text(char, parent_comm, parent->comm, TASK_COMM_LEN)
	385	ctf_integer(pid_t, parent_tid, parent->pid)
	386	ctf_integer(pid_t, parent_pid, parent->tgid)
	387	ctf_integer(unsigned int, parent_ns_inum,
	388	({
	389	unsigned int parent_ns_inum = 0;
	390
	391	if (parent) {
	392	struct pid_namespace *pid_ns;
	393
	394	pid_ns = task_active_pid_ns(parent);
	395	if (pid_ns)
	396	parent_ns_inum =
	397	pid_ns->ns.inum;
	398	}
	399	parent_ns_inum;
	400	}))
	401	ctf_array_text(char, child_comm, child->comm, TASK_COMM_LEN)
	402	ctf_integer(pid_t, child_tid, child->pid)
	403	ctf_sequence(pid_t, vtids, tp_locvar->vtids, u8, tp_locvar->ns_level)
	404	ctf_integer(pid_t, child_pid, child->tgid)
	405	ctf_integer(unsigned int, child_ns_inum,
	406	({
	407	unsigned int child_ns_inum = 0;
	408
	409	if (child) {
	410	struct pid_namespace *pid_ns;
	411
	412	pid_ns = task_active_pid_ns(child);
	413	if (pid_ns)
	414	child_ns_inum =
	415	pid_ns->ns.inum;
	416	}
	417	child_ns_inum;
	418	}))
	419	),
	420
	421	TP_code_post()
	422	)
	423
	424	/*
	425	* Tracepoint for exec:
	426	*/
	427	LTTNG_TRACEPOINT_EVENT(sched_process_exec,
	428
	429	TP_PROTO(struct task_struct *p, pid_t old_pid,
	430	struct linux_binprm *bprm),
	431
	432	TP_ARGS(p, old_pid, bprm),
	433
	434	TP_FIELDS(
	435	ctf_string(filename, bprm->filename)
	436	ctf_integer(pid_t, tid, p->pid)
	437	ctf_integer(pid_t, old_tid, old_pid)
	438	)
	439	)
	440
	441	/*
	442	* XXX the below sched_stat tracepoints only apply to SCHED_OTHER/BATCH/IDLE
	443	* adding sched_stat support to SCHED_FIFO/RR would be welcome.
	444	*/
	445	LTTNG_TRACEPOINT_EVENT_CLASS(sched_stat_template,
	446
	447	TP_PROTO(struct task_struct *tsk, u64 delay),
	448
	449	TP_ARGS(tsk, delay),
	450
	451	TP_FIELDS(
	452	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	453	ctf_integer(pid_t, tid, tsk->pid)
	454	ctf_integer(u64, delay, delay)
	455	)
	456	)
	457
	458
	459	/*
	460	* Tracepoint for accounting wait time (time the task is runnable
	461	* but not actually running due to scheduler contention).
	462	*/
	463	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_wait,
	464	TP_PROTO(struct task_struct *tsk, u64 delay),
	465	TP_ARGS(tsk, delay))
	466
	467	/*
	468	* Tracepoint for accounting sleep time (time the task is not runnable,
	469	* including iowait, see below).
	470	*/
	471	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_sleep,
	472	TP_PROTO(struct task_struct *tsk, u64 delay),
	473	TP_ARGS(tsk, delay))
	474
	475	/*
	476	* Tracepoint for accounting iowait time (time the task is not runnable
	477	* due to waiting on IO to complete).
	478	*/
	479	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_iowait,
	480	TP_PROTO(struct task_struct *tsk, u64 delay),
	481	TP_ARGS(tsk, delay))
	482
	483	/*
	484	* Tracepoint for accounting blocked time (time the task is in uninterruptible).
	485	*/
	486	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_blocked,
	487	TP_PROTO(struct task_struct *tsk, u64 delay),
	488	TP_ARGS(tsk, delay))
	489
	490	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(6,8,0) \|\| \
	491	LTTNG_RHEL_KERNEL_RANGE(5,14,0,452,0,0, 5,15,0,0,0,0))
	492	/*
	493	* Tracepoint for accounting runtime (time the task is executing
	494	* on a CPU).
	495	*/
	496	LTTNG_TRACEPOINT_EVENT(sched_stat_runtime,
	497
	498	TP_PROTO(struct task_struct *tsk, u64 runtime),
	499
	500	TP_ARGS(tsk, runtime),
	501
	502	TP_FIELDS(
	503	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	504	ctf_integer(pid_t, tid, tsk->pid)
	505	ctf_integer(u64, runtime, runtime)
	506	)
	507	)
	508	#else
	509	/*
	510	* Tracepoint for accounting runtime (time the task is executing
	511	* on a CPU).
	512	*/
	513	LTTNG_TRACEPOINT_EVENT(sched_stat_runtime,
	514
	515	TP_PROTO(struct task_struct *tsk, u64 runtime, u64 vruntime),
	516
	517	TP_ARGS(tsk, runtime, vruntime),
	518
	519	TP_FIELDS(
	520	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	521	ctf_integer(pid_t, tid, tsk->pid)
	522	ctf_integer(u64, runtime, runtime)
	523	ctf_integer(u64, vruntime, vruntime)
	524	)
	525	)
	526	#endif
	527
	528	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,12,0) \|\| \
	529	LTTNG_RT_KERNEL_RANGE(4,9,27,18, 4,10,0,0) \|\| \
	530	LTTNG_RT_KERNEL_RANGE(4,11,5,1, 4,12,0,0))
	531	/*
	532	* Tracepoint for showing priority inheritance modifying a tasks
	533	* priority.
	534	*/
	535	LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,
	536
	537	TP_PROTO(struct task_struct tsk, struct task_struct pi_task),
	538
	539	TP_ARGS(tsk, pi_task),
	540
	541	TP_FIELDS(
	542	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	543	ctf_integer(pid_t, tid, tsk->pid)
	544	ctf_integer(int, oldprio, tsk->prio - MAX_RT_PRIO)
	545	ctf_integer(int, newprio, pi_task ? pi_task->prio - MAX_RT_PRIO : tsk->prio - MAX_RT_PRIO)
	546	)
	547	)
	548	#else
	549	/*
	550	* Tracepoint for showing priority inheritance modifying a tasks
	551	* priority.
	552	*/
	553	LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,
	554
	555	TP_PROTO(struct task_struct *tsk, int newprio),
	556
	557	TP_ARGS(tsk, newprio),
	558
	559	TP_FIELDS(
	560	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	561	ctf_integer(pid_t, tid, tsk->pid)
	562	ctf_integer(int, oldprio, tsk->prio - MAX_RT_PRIO)
	563	ctf_integer(int, newprio, newprio - MAX_RT_PRIO)
	564	)
	565	)
	566	#endif
	567
	568	#endif /* LTTNG_TRACE_SCHED_H */
	569
	570	/* This part must be outside protection */
	571	#include <lttng/define_trace.h>