[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>
#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,9,0))
#include <linux/sched/rt.h>
#endif
#include <wrapper/namespace.h>

#define LTTNG_MAX_PID_NS_LEVEL 32

#ifndef _TRACE_SCHED_DEF_
#define _TRACE_SCHED_DEF_

#if (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;
}

#elif (LTTNG_LINUX_VERSION_CODE >= LTTNG_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 (LTTNG_LINUX_VERSION_CODE >= LTTNG_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 (LTTNG_LINUX_VERSION_CODE >= LTTNG_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 (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,2,0))

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

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

	return state;
}

#else

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

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

	return state;
}

#endif

#endif /* _TRACE_SCHED_DEF_ */

#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)

#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,9,0))
		ctf_enum_value("TASK_PARKED", TASK_PARKED)
#endif /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,9,0)) */

		ctf_enum_value("TASK_DEAD", TASK_DEAD)
		ctf_enum_value("TASK_WAKEKILL", TASK_WAKEKILL)
		ctf_enum_value("TASK_WAKING", TASK_WAKING)

#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,2,0))
		ctf_enum_value("TASK_NOLOAD", TASK_NOLOAD)
#endif /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,2,0)) */

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

	TP_PROTO(struct task_struct *p),

	TP_ARGS(p),

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

	TP_PROTO(struct task_struct *p, int success),

	TP_ARGS(p, success),

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

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

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

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

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

#else

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

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

#endif /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,3,0)) */

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

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

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

	TP_ARGS(prev, next),
#endif /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,4,0)) */

	TP_FIELDS(
		ctf_array_text(char, prev_comm,	prev->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, prev_tid, prev->pid)
		ctf_integer(int, prev_prio, prev->prio - MAX_RT_PRIO)
#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,4,0))
#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
#else
#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
		ctf_enum(task_state, long, prev_state, __trace_sched_switch_state(prev))
#else
		ctf_integer(long, prev_state, __trace_sched_switch_state(prev))
#endif
#endif
		ctf_array_text(char, next_comm, next->comm, TASK_COMM_LEN)
		ctf_integer(pid_t, next_tid, next->pid)
		ctf_integer(int, next_prio, next->prio - MAX_RT_PRIO)
	)
)

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

	TP_PROTO(struct task_struct *p, int dest_cpu),

	TP_ARGS(p, dest_cpu),

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

LTTNG_TRACEPOINT_EVENT_CLASS(sched_process_template,

	TP_PROTO(struct task_struct *p),

	TP_ARGS(p),

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

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


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

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

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

	TP_PROTO(struct pid *pid),

	TP_ARGS(pid),

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

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

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

	TP_ARGS(parent, child),

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

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

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

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

				if (parent) {
					struct pid_namespace *pid_ns;

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

				if (child) {
					struct pid_namespace *pid_ns;

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

	TP_code_post()
)

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

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

	TP_ARGS(p, old_pid, bprm),

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

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

	TP_PROTO(struct task_struct *tsk, u64 delay),

	TP_ARGS(tsk, delay),

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


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

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

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

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

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

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

	TP_ARGS(tsk, runtime, vruntime),

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

#if (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 MJ	12	#include <lttng/kernel-version.h>
5f4c791e MJ	13	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,9,0))
090db00e MJ	14	#include <linux/sched/rt.h>
090db00e MJ	15	#endif
a6cf40a4	16	#include <wrapper/namespace.h>
7bbf55ea SL	17
	18	#define LTTNG_MAX_PID_NS_LEVEL 32
	19
f62b389e MD	20	#ifndef _TRACE_SCHED_DEF_
	21	#define _TRACE_SCHED_DEF_
	22
5f4c791e	23	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,15,0))
27e6eda7 GAPG	24
	25	static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
	26	{
	27	unsigned int state;
	28
	29	#ifdef CONFIG_SCHED_DEBUG
	30	BUG_ON(p != current);
	31	#endif /* CONFIG_SCHED_DEBUG */
	32
	33	/*
	34	* Preemption ignores task state, therefore preempted tasks are always
	35	* RUNNING (we will not have dequeued if state != RUNNING).
	36	*/
	37	if (preempt)
	38	return TASK_REPORT_MAX;
	39
	40	/*
	41	* task_state_index() uses fls() and returns a value from 0-8 range.
	42	* Decrement it by 1 (except TASK_RUNNING state i.e 0) before using
	43	* it for left shift operation to get the correct task->state
	44	* mapping.
	45	*/
	46	state = task_state_index(p);
	47
	48	return state ? (1 << (state - 1)) : state;
	49	}
	50
5f4c791e	51	#elif (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,14,0))
27e6eda7 GAPG	52
	53	static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
	54	{
	55	unsigned int state;
	56
	57	#ifdef CONFIG_SCHED_DEBUG
	58	BUG_ON(p != current);
	59	#endif /* CONFIG_SCHED_DEBUG */
	60
	61	/*
	62	* Preemption ignores task state, therefore preempted tasks are always
	63	* RUNNING (we will not have dequeued if state != RUNNING).
	64	*/
	65	if (preempt)
	66	return TASK_REPORT_MAX;
	67
	68	/*
	69	* __get_task_state() uses fls() and returns a value from 0-8 range.
	70	* Decrement it by 1 (except TASK_RUNNING state i.e 0) before using
	71	* it for left shift operation to get the correct task->state
	72	* mapping.
	73	*/
	74	state = __get_task_state(p);
	75
	76	return state ? (1 << (state - 1)) : state;
	77	}
	78
5f4c791e	79	#elif (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,4,0))
3ee729fe MD	80
	81	static inline long __trace_sched_switch_state(bool preempt, struct task_struct *p)
	82	{
	83	#ifdef CONFIG_SCHED_DEBUG
	84	BUG_ON(p != current);
	85	#endif /* CONFIG_SCHED_DEBUG */
	86	/*
	87	* Preemption ignores task state, therefore preempted tasks are always RUNNING
	88	* (we will not have dequeued if state != RUNNING).
	89	*/
	90	return preempt ? TASK_RUNNING \| TASK_STATE_MAX : p->state;
	91	}
	92
5f4c791e	93	#elif (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,19,0))
857b6f4f MD	94
	95	static inline long __trace_sched_switch_state(struct task_struct *p)
	96	{
	97	long state = p->state;
	98
	99	#ifdef CONFIG_PREEMPT
	100	#ifdef CONFIG_SCHED_DEBUG
	101	BUG_ON(p != current);
	102	#endif /* CONFIG_SCHED_DEBUG */
	103	/*
	104	* For all intents and purposes a preempted task is a running task.
	105	*/
	106	if (preempt_count() & PREEMPT_ACTIVE)
	107	state = TASK_RUNNING \| TASK_STATE_MAX;
	108	#endif /* CONFIG_PREEMPT */
	109
	110	return state;
	111	}
	112
5f4c791e	113	#elif (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,13,0))
33673ee7 MD	114
	115	static inline long __trace_sched_switch_state(struct task_struct *p)
	116	{
	117	long state = p->state;
	118
	119	#ifdef CONFIG_PREEMPT
	120	/*
	121	* For all intents and purposes a preempted task is a running task.
	122	*/
	123	if (task_preempt_count(p) & PREEMPT_ACTIVE)
	124	state = TASK_RUNNING \| TASK_STATE_MAX;
	125	#endif
	126
	127	return state;
	128	}
	129
5f4c791e	130	#elif (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,2,0))
7c68b363	131
f62b389e MD	132	static inline long __trace_sched_switch_state(struct task_struct *p)
	133	{
	134	long state = p->state;
	135
	136	#ifdef CONFIG_PREEMPT
	137	/*
	138	* For all intents and purposes a preempted task is a running task.
	139	*/
	140	if (task_thread_info(p)->preempt_count & PREEMPT_ACTIVE)
7c68b363	141	state = TASK_RUNNING \| TASK_STATE_MAX;
7c68b363	142	#endif
33673ee7 MD	143
	144	return state;
	145	}
	146
ead65427	147	#else
33673ee7 MD	148
	149	static inline long __trace_sched_switch_state(struct task_struct *p)
	150	{
	151	long state = p->state;
	152
	153	#ifdef CONFIG_PREEMPT
	154	/*
	155	* For all intents and purposes a preempted task is a running task.
	156	*/
	157	if (task_thread_info(p)->preempt_count & PREEMPT_ACTIVE)
	158	state = TASK_RUNNING;
f62b389e MD	159	#endif
	160
	161	return state;
	162	}
	163
7c68b363 AG	164	#endif
7c68b363 AG	165
f62b389e MD	166	#endif /* _TRACE_SCHED_DEF_ */
f62b389e MD	167
e54b3828	168	#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
721caea4 GB	169	/*
	170	* Enumeration of the task state bitmask.
	171	* Only bit flags are enumerated here, not composition of states.
	172	*/
	173	LTTNG_TRACEPOINT_ENUM(task_state,
	174	TP_ENUM_VALUES(
	175	ctf_enum_value("TASK_RUNNING", TASK_RUNNING)
	176	ctf_enum_value("TASK_INTERRUPTIBLE", TASK_INTERRUPTIBLE)
	177	ctf_enum_value("TASK_UNINTERRUPTIBLE", TASK_UNINTERRUPTIBLE)
	178	ctf_enum_value("TASK_STOPPED", __TASK_STOPPED)
	179	ctf_enum_value("TASK_TRACED", __TASK_TRACED)
	180	ctf_enum_value("EXIT_DEAD", EXIT_DEAD)
	181	ctf_enum_value("EXIT_ZOMBIE", EXIT_ZOMBIE)
	182
5f4c791e	183	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,9,0))
721caea4	184	ctf_enum_value("TASK_PARKED", TASK_PARKED)
5f4c791e	185	#endif /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,9,0)) */
721caea4 GB	186
	187	ctf_enum_value("TASK_DEAD", TASK_DEAD)
	188	ctf_enum_value("TASK_WAKEKILL", TASK_WAKEKILL)
	189	ctf_enum_value("TASK_WAKING", TASK_WAKING)
	190
5f4c791e	191	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,2,0))
721caea4	192	ctf_enum_value("TASK_NOLOAD", TASK_NOLOAD)
5f4c791e	193	#endif /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,2,0)) */
721caea4	194
5f4c791e	195	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,8,0))
721caea4	196	ctf_enum_value("TASK_NEW", TASK_NEW)
5f4c791e	197	#endif /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,8,0)) */
721caea4 GB	198
	199	ctf_enum_value("TASK_STATE_MAX", TASK_STATE_MAX)
	200	)
	201	)
e54b3828	202	#endif /* CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM */
721caea4	203
f62b389e MD	204	/*
	205	* Tracepoint for calling kthread_stop, performed to end a kthread:
	206	*/
3bc29f0a	207	LTTNG_TRACEPOINT_EVENT(sched_kthread_stop,
f62b389e MD	208
	209	TP_PROTO(struct task_struct *t),
	210
	211	TP_ARGS(t),
	212
f127e61e MD	213	TP_FIELDS(
	214	ctf_array_text(char, comm, t->comm, TASK_COMM_LEN)
	215	ctf_integer(pid_t, tid, t->pid)
	216	)
f62b389e MD	217	)
	218
	219	/*
	220	* Tracepoint for the return value of the kthread stopping:
	221	*/
3bc29f0a	222	LTTNG_TRACEPOINT_EVENT(sched_kthread_stop_ret,
f62b389e MD	223
	224	TP_PROTO(int ret),
	225
	226	TP_ARGS(ret),
	227
f127e61e MD	228	TP_FIELDS(
	229	ctf_integer(int, ret, ret)
	230	)
f62b389e MD	231	)
	232
	233	/*
	234	* Tracepoint for waking up a task:
	235	*/
5f4c791e	236	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,3,0) \|\| \
95ff37b6 MJ	237	LTTNG_RT_KERNEL_RANGE(4,1,10,11, 4,2,0,0) \|\| \
	238	LTTNG_RT_KERNEL_RANGE(3,18,27,26, 3,19,0,0) \|\| \
	239	LTTNG_RT_KERNEL_RANGE(3,14,61,63, 3,15,0,0) \|\| \
	240	LTTNG_RT_KERNEL_RANGE(3,12,54,73, 3,13,0,0) \|\| \
	241	LTTNG_RT_KERNEL_RANGE(3,10,97,106, 3,11,0,0) \|\| \
	242	LTTNG_RT_KERNEL_RANGE(3,4,110,139, 3,5,0,0) \|\| \
	243	LTTNG_RT_KERNEL_RANGE(3,2,77,111, 3,3,0,0))
ffcf2393 MD	244	LTTNG_TRACEPOINT_EVENT_CLASS(sched_wakeup_template,
	245
	246	TP_PROTO(struct task_struct *p),
	247
	248	TP_ARGS(p),
	249
	250	TP_FIELDS(
	251	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	252	ctf_integer(pid_t, tid, p->pid)
5aa835c0	253	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
ffcf2393 MD	254	ctf_integer(int, target_cpu, task_cpu(p))
	255	)
	256	)
5f4c791e	257	#else /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,3,0)) */
3bc29f0a	258	LTTNG_TRACEPOINT_EVENT_CLASS(sched_wakeup_template,
f62b389e MD	259
	260	TP_PROTO(struct task_struct *p, int success),
	261
	262	TP_ARGS(p, success),
	263
f127e61e MD	264	TP_FIELDS(
	265	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	266	ctf_integer(pid_t, tid, p->pid)
5aa835c0	267	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
f127e61e	268	ctf_integer(int, success, success)
f127e61e	269	ctf_integer(int, target_cpu, task_cpu(p))
7c68b363	270	)
f62b389e	271	)
5f4c791e	272	#endif /* #else #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,3,0)) */
f62b389e	273
5f4c791e	274	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,3,0) \|\| \
95ff37b6 MJ	275	LTTNG_RT_KERNEL_RANGE(4,1,10,11, 4,2,0,0) \|\| \
	276	LTTNG_RT_KERNEL_RANGE(3,18,27,26, 3,19,0,0) \|\| \
	277	LTTNG_RT_KERNEL_RANGE(3,14,61,63, 3,15,0,0) \|\| \
	278	LTTNG_RT_KERNEL_RANGE(3,12,54,73, 3,13,0,0) \|\| \
	279	LTTNG_RT_KERNEL_RANGE(3,10,97,106, 3,11,0,0) \|\| \
	280	LTTNG_RT_KERNEL_RANGE(3,4,110,139, 3,5,0,0) \|\| \
	281	LTTNG_RT_KERNEL_RANGE(3,2,77,111, 3,3,0,0))
ffcf2393 MD	282
	283	/*
	284	* Tracepoint called when waking a task; this tracepoint is guaranteed to be
	285	* called from the waking context.
	286	*/
	287	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_waking,
	288	TP_PROTO(struct task_struct *p),
	289	TP_ARGS(p))
	290
	291	/*
	292	* Tracepoint called when the task is actually woken; p->state == TASK_RUNNNG.
	293	* It it not always called from the waking context.
	294	*/
	295	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup,
	296	TP_PROTO(struct task_struct *p),
	297	TP_ARGS(p))
	298
	299	/*
	300	* Tracepoint for waking up a new task:
	301	*/
	302	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup_new,
	303	TP_PROTO(struct task_struct *p),
	304	TP_ARGS(p))
	305
ead65427	306	#else
3a523f5b	307
3bc29f0a	308	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup,
f62b389e MD	309	TP_PROTO(struct task_struct *p, int success),
	310	TP_ARGS(p, success))
	311
	312	/*
	313	* Tracepoint for waking up a new task:
	314	*/
3bc29f0a	315	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_wakeup_template, sched_wakeup_new,
f62b389e MD	316	TP_PROTO(struct task_struct *p, int success),
	317	TP_ARGS(p, success))
	318
5f4c791e	319	#endif /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,3,0)) */
3a523f5b	320
f62b389e MD	321	/*
	322	* Tracepoint for task switches, performed by the scheduler:
	323	*/
3bc29f0a	324	LTTNG_TRACEPOINT_EVENT(sched_switch,
f62b389e	325
5f4c791e	326	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,4,0))
3ee729fe MD	327	TP_PROTO(bool preempt,
	328	struct task_struct *prev,
	329	struct task_struct *next),
	330
	331	TP_ARGS(preempt, prev, next),
ead65427	332	#else
f62b389e MD	333	TP_PROTO(struct task_struct *prev,
	334	struct task_struct *next),
	335
	336	TP_ARGS(prev, next),
5f4c791e	337	#endif /* #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,4,0)) */
f62b389e	338
f127e61e MD	339	TP_FIELDS(
	340	ctf_array_text(char, prev_comm, prev->comm, TASK_COMM_LEN)
	341	ctf_integer(pid_t, prev_tid, prev->pid)
	342	ctf_integer(int, prev_prio, prev->prio - MAX_RT_PRIO)
5f4c791e	343	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,4,0))
e54b3828	344	#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
721caea4	345	ctf_enum(task_state, long, prev_state, __trace_sched_switch_state(preempt, prev))
7c68b363	346	#else
e54b3828 MJ	347	ctf_integer(long, prev_state, __trace_sched_switch_state(preempt, prev))
	348	#endif
	349	#else
	350	#ifdef CONFIG_LTTNG_EXPERIMENTAL_BITWISE_ENUM
721caea4	351	ctf_enum(task_state, long, prev_state, __trace_sched_switch_state(prev))
e54b3828 MJ	352	#else
	353	ctf_integer(long, prev_state, __trace_sched_switch_state(prev))
	354	#endif
7c68b363	355	#endif
f127e61e MD	356	ctf_array_text(char, next_comm, next->comm, TASK_COMM_LEN)
	357	ctf_integer(pid_t, next_tid, next->pid)
	358	ctf_integer(int, next_prio, next->prio - MAX_RT_PRIO)
	359	)
f62b389e MD	360	)
	361
	362	/*
	363	* Tracepoint for a task being migrated:
	364	*/
3bc29f0a	365	LTTNG_TRACEPOINT_EVENT(sched_migrate_task,
f62b389e MD	366
	367	TP_PROTO(struct task_struct *p, int dest_cpu),
	368
	369	TP_ARGS(p, dest_cpu),
	370
f127e61e MD	371	TP_FIELDS(
	372	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	373	ctf_integer(pid_t, tid, p->pid)
	374	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	375	ctf_integer(int, orig_cpu, task_cpu(p))
	376	ctf_integer(int, dest_cpu, dest_cpu)
	377	)
f62b389e MD	378	)
f62b389e MD	379
3bc29f0a	380	LTTNG_TRACEPOINT_EVENT_CLASS(sched_process_template,
f62b389e MD	381
	382	TP_PROTO(struct task_struct *p),
	383
	384	TP_ARGS(p),
	385
f127e61e MD	386	TP_FIELDS(
	387	ctf_array_text(char, comm, p->comm, TASK_COMM_LEN)
	388	ctf_integer(pid_t, tid, p->pid)
	389	ctf_integer(int, prio, p->prio - MAX_RT_PRIO)
	390	)
f62b389e MD	391	)
	392
	393	/*
	394	* Tracepoint for freeing a task:
	395	*/
3bc29f0a	396	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_free,
f62b389e MD	397	TP_PROTO(struct task_struct *p),
f62b389e MD	398	TP_ARGS(p))
216b6baa	399
f62b389e MD	400
	401	/*
	402	* Tracepoint for a task exiting:
	403	*/
3bc29f0a	404	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_process_exit,
f62b389e MD	405	TP_PROTO(struct task_struct *p),
	406	TP_ARGS(p))
	407
	408	/*
	409	* Tracepoint for waiting on task to unschedule:
	410	*/
3bc29f0a	411	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_process_template, sched_wait_task,
f62b389e MD	412	TP_PROTO(struct task_struct *p),
	413	TP_ARGS(p))
	414
	415	/*
	416	* Tracepoint for a waiting task:
	417	*/
3bc29f0a	418	LTTNG_TRACEPOINT_EVENT(sched_process_wait,
f62b389e MD	419
	420	TP_PROTO(struct pid *pid),
	421
	422	TP_ARGS(pid),
	423
f127e61e MD	424	TP_FIELDS(
	425	ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
	426	ctf_integer(pid_t, tid, pid_nr(pid))
	427	ctf_integer(int, prio, current->prio - MAX_RT_PRIO)
	428	)
f62b389e MD	429	)
	430
	431	/*
79b18ef7 MD	432	* Tracepoint for do_fork.
	433	* Saving both TID and PID information, especially for the child, allows
	434	* trace analyzers to distinguish between creation of a new process and
	435	* creation of a new thread. Newly created processes will have child_tid
	436	* == child_pid, while creation of a thread yields to child_tid !=
	437	* child_pid.
f62b389e	438	*/
7bbf55ea	439	LTTNG_TRACEPOINT_EVENT_CODE(sched_process_fork,
f62b389e MD	440
	441	TP_PROTO(struct task_struct parent, struct task_struct child),
	442
	443	TP_ARGS(parent, child),
	444
7bbf55ea SL	445	TP_locvar(
	446	pid_t vtids[LTTNG_MAX_PID_NS_LEVEL];
	447	unsigned int ns_level;
	448	),
	449
265822ae	450	TP_code_pre(
7bbf55ea SL	451	if (child) {
	452	struct pid *child_pid;
	453	unsigned int i;
	454
	455	child_pid = task_pid(child);
	456	tp_locvar->ns_level =
	457	min_t(unsigned int, child_pid->level + 1,
	458	LTTNG_MAX_PID_NS_LEVEL);
	459	for (i = 0; i < tp_locvar->ns_level; i++)
	460	tp_locvar->vtids[i] = child_pid->numbers[i].nr;
	461	}
	462	),
	463
f127e61e MD	464	TP_FIELDS(
	465	ctf_array_text(char, parent_comm, parent->comm, TASK_COMM_LEN)
	466	ctf_integer(pid_t, parent_tid, parent->pid)
	467	ctf_integer(pid_t, parent_pid, parent->tgid)
5f4c791e	468	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,8,0))
7bbf55ea SL	469	ctf_integer(unsigned int, parent_ns_inum,
	470	({
	471	unsigned int parent_ns_inum = 0;
	472
	473	if (parent) {
	474	struct pid_namespace *pid_ns;
	475
	476	pid_ns = task_active_pid_ns(parent);
	477	if (pid_ns)
	478	parent_ns_inum =
a6cf40a4	479	pid_ns->lttng_ns_inum;
7bbf55ea SL	480	}
	481	parent_ns_inum;
	482	}))
	483	#endif
f127e61e MD	484	ctf_array_text(char, child_comm, child->comm, TASK_COMM_LEN)
f127e61e MD	485	ctf_integer(pid_t, child_tid, child->pid)
7bbf55ea	486	ctf_sequence(pid_t, vtids, tp_locvar->vtids, u8, tp_locvar->ns_level)
f127e61e	487	ctf_integer(pid_t, child_pid, child->tgid)
5f4c791e	488	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,8,0))
7bbf55ea SL	489	ctf_integer(unsigned int, child_ns_inum,
	490	({
	491	unsigned int child_ns_inum = 0;
	492
	493	if (child) {
	494	struct pid_namespace *pid_ns;
	495
	496	pid_ns = task_active_pid_ns(child);
	497	if (pid_ns)
	498	child_ns_inum =
a6cf40a4	499	pid_ns->lttng_ns_inum;
7bbf55ea SL	500	}
	501	child_ns_inum;
	502	}))
	503	#endif
265822ae MD	504	),
	505
	506	TP_code_post()
f62b389e MD	507	)
f62b389e MD	508
5f4c791e	509	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,4,0))
46142a81 PW	510	/*
	511	* Tracepoint for exec:
	512	*/
3bc29f0a	513	LTTNG_TRACEPOINT_EVENT(sched_process_exec,
46142a81 PW	514
	515	TP_PROTO(struct task_struct *p, pid_t old_pid,
	516	struct linux_binprm *bprm),
	517
	518	TP_ARGS(p, old_pid, bprm),
	519
f127e61e MD	520	TP_FIELDS(
	521	ctf_string(filename, bprm->filename)
	522	ctf_integer(pid_t, tid, p->pid)
	523	ctf_integer(pid_t, old_tid, old_pid)
	524	)
46142a81	525	)
c94b2508	526	#endif
46142a81	527
f62b389e MD	528	/*
	529	* XXX the below sched_stat tracepoints only apply to SCHED_OTHER/BATCH/IDLE
	530	* adding sched_stat support to SCHED_FIFO/RR would be welcome.
	531	*/
3bc29f0a	532	LTTNG_TRACEPOINT_EVENT_CLASS(sched_stat_template,
f62b389e MD	533
	534	TP_PROTO(struct task_struct *tsk, u64 delay),
	535
	536	TP_ARGS(tsk, delay),
	537
f127e61e MD	538	TP_FIELDS(
	539	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	540	ctf_integer(pid_t, tid, tsk->pid)
	541	ctf_integer(u64, delay, delay)
f62b389e	542	)
f62b389e MD	543	)
	544
	545
	546	/*
	547	* Tracepoint for accounting wait time (time the task is runnable
	548	* but not actually running due to scheduler contention).
	549	*/
3bc29f0a	550	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_wait,
f62b389e MD	551	TP_PROTO(struct task_struct *tsk, u64 delay),
	552	TP_ARGS(tsk, delay))
	553
	554	/*
	555	* Tracepoint for accounting sleep time (time the task is not runnable,
	556	* including iowait, see below).
	557	*/
3bc29f0a	558	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_sleep,
f62b389e MD	559	TP_PROTO(struct task_struct *tsk, u64 delay),
	560	TP_ARGS(tsk, delay))
	561
	562	/*
	563	* Tracepoint for accounting iowait time (time the task is not runnable
	564	* due to waiting on IO to complete).
	565	*/
3bc29f0a	566	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_iowait,
f62b389e MD	567	TP_PROTO(struct task_struct *tsk, u64 delay),
	568	TP_ARGS(tsk, delay))
	569
5f4c791e	570	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(3,3,0))
7c68b363 AG	571	/*
	572	* Tracepoint for accounting blocked time (time the task is in uninterruptible).
	573	*/
3bc29f0a	574	LTTNG_TRACEPOINT_EVENT_INSTANCE(sched_stat_template, sched_stat_blocked,
7c68b363 AG	575	TP_PROTO(struct task_struct *tsk, u64 delay),
	576	TP_ARGS(tsk, delay))
	577	#endif
	578
f62b389e MD	579	/*
	580	* Tracepoint for accounting runtime (time the task is executing
	581	* on a CPU).
	582	*/
3bc29f0a	583	LTTNG_TRACEPOINT_EVENT(sched_stat_runtime,
f62b389e MD	584
	585	TP_PROTO(struct task_struct *tsk, u64 runtime, u64 vruntime),
	586
	587	TP_ARGS(tsk, runtime, vruntime),
	588
f127e61e MD	589	TP_FIELDS(
	590	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	591	ctf_integer(pid_t, tid, tsk->pid)
	592	ctf_integer(u64, runtime, runtime)
	593	ctf_integer(u64, vruntime, vruntime)
f62b389e	594	)
f62b389e MD	595	)
f62b389e MD	596
5f4c791e	597	#if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(4,12,0) \|\| \
46dded2d MJ	598	LTTNG_RT_KERNEL_RANGE(4,9,27,18, 4,10,0,0) \|\| \
46dded2d MJ	599	LTTNG_RT_KERNEL_RANGE(4,11,5,1, 4,12,0,0))
673e9a03 MJ	600	/*
	601	* Tracepoint for showing priority inheritance modifying a tasks
	602	* priority.
	603	*/
	604	LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,
	605
	606	TP_PROTO(struct task_struct tsk, struct task_struct pi_task),
	607
	608	TP_ARGS(tsk, pi_task),
	609
	610	TP_FIELDS(
	611	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	612	ctf_integer(pid_t, tid, tsk->pid)
	613	ctf_integer(int, oldprio, tsk->prio - MAX_RT_PRIO)
	614	ctf_integer(int, newprio, pi_task ? pi_task->prio - MAX_RT_PRIO : tsk->prio - MAX_RT_PRIO)
	615	)
	616	)
ead65427	617	#else
f62b389e MD	618	/*
	619	* Tracepoint for showing priority inheritance modifying a tasks
	620	* priority.
	621	*/
3bc29f0a	622	LTTNG_TRACEPOINT_EVENT(sched_pi_setprio,
f62b389e MD	623
	624	TP_PROTO(struct task_struct *tsk, int newprio),
	625
	626	TP_ARGS(tsk, newprio),
	627
f127e61e MD	628	TP_FIELDS(
	629	ctf_array_text(char, comm, tsk->comm, TASK_COMM_LEN)
	630	ctf_integer(pid_t, tid, tsk->pid)
	631	ctf_integer(int, oldprio, tsk->prio - MAX_RT_PRIO)
	632	ctf_integer(int, newprio, newprio - MAX_RT_PRIO)
	633	)
f62b389e	634	)
7c68b363	635	#endif
f62b389e	636
3bc29f0a	637	#endif /* LTTNG_TRACE_SCHED_H */
f62b389e MD	638
f62b389e MD	639	/* This part must be outside protection */
3b4aafcb	640	#include <lttng/define_trace.h>