1 /* SPDX-License-Identifier: GPL-2.0 */
6 * Define 'struct task_struct' and provide the main scheduler
7 * APIs (schedule(), wakeup variants, etc.)
10 #include <uapi/linux/sched.h>
12 #include <asm/current.h>
14 #include <linux/pid.h>
15 #include <linux/sem.h>
16 #include <linux/shm.h>
17 #include <linux/mutex.h>
18 #include <linux/plist.h>
19 #include <linux/hrtimer.h>
20 #include <linux/irqflags.h>
21 #include <linux/seccomp.h>
22 #include <linux/nodemask.h>
23 #include <linux/rcupdate.h>
24 #include <linux/refcount.h>
25 #include <linux/resource.h>
26 #include <linux/latencytop.h>
27 #include <linux/sched/prio.h>
28 #include <linux/sched/types.h>
29 #include <linux/signal_types.h>
30 #include <linux/syscall_user_dispatch.h>
31 #include <linux/mm_types_task.h>
32 #include <linux/task_io_accounting.h>
33 #include <linux/posix-timers.h>
34 #include <linux/rseq.h>
35 #include <linux/seqlock.h>
36 #include <linux/kcsan.h>
37 #include <asm/kmap_size.h>
39 /* task_struct member predeclarations (sorted alphabetically): */
41 struct backing_dev_info;
44 struct bpf_local_storage;
46 struct capture_control;
49 struct futex_pi_state;
55 struct perf_event_context;
57 struct pipe_inode_info;
60 struct robust_list_head;
66 struct sighand_struct;
68 struct task_delay_info;
72 * Task state bitmask. NOTE! These bits are also
73 * encoded in fs/proc/array.c: get_task_state().
75 * We have two separate sets of flags: task->state
76 * is about runnability, while task->exit_state are
77 * about the task exiting. Confusing, but this way
78 * modifying one set can't modify the other one by
82 /* Used in tsk->state: */
83 #define TASK_RUNNING 0x0000
84 #define TASK_INTERRUPTIBLE 0x0001
85 #define TASK_UNINTERRUPTIBLE 0x0002
86 #define __TASK_STOPPED 0x0004
87 #define __TASK_TRACED 0x0008
88 /* Used in tsk->exit_state: */
89 #define EXIT_DEAD 0x0010
90 #define EXIT_ZOMBIE 0x0020
91 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
92 /* Used in tsk->state again: */
93 #define TASK_PARKED 0x0040
94 #define TASK_DEAD 0x0080
95 #define TASK_WAKEKILL 0x0100
96 #define TASK_WAKING 0x0200
97 #define TASK_NOLOAD 0x0400
98 #define TASK_NEW 0x0800
99 /* RT specific auxilliary flag to mark RT lock waiters */
100 #define TASK_RTLOCK_WAIT 0x1000
101 #define TASK_STATE_MAX 0x2000
103 /* Convenience macros for the sake of set_current_state: */
104 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
105 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
106 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
108 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
110 /* Convenience macros for the sake of wake_up(): */
111 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
113 /* get_task_state(): */
114 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
115 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
116 __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \
119 #define task_is_running(task) (READ_ONCE((task)->__state) == TASK_RUNNING)
121 #define task_is_stopped(task) ((READ_ONCE(task->__state) & __TASK_STOPPED) != 0)
124 * Special states are those that do not use the normal wait-loop pattern. See
125 * the comment with set_special_state().
127 #define is_special_task_state(state) \
128 ((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_PARKED | TASK_DEAD))
130 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
131 # define debug_normal_state_change(state_value) \
133 WARN_ON_ONCE(is_special_task_state(state_value)); \
134 current->task_state_change = _THIS_IP_; \
137 # define debug_special_state_change(state_value) \
139 WARN_ON_ONCE(!is_special_task_state(state_value)); \
140 current->task_state_change = _THIS_IP_; \
143 # define debug_rtlock_wait_set_state() \
145 current->saved_state_change = current->task_state_change;\
146 current->task_state_change = _THIS_IP_; \
149 # define debug_rtlock_wait_restore_state() \
151 current->task_state_change = current->saved_state_change;\
155 # define debug_normal_state_change(cond) do { } while (0)
156 # define debug_special_state_change(cond) do { } while (0)
157 # define debug_rtlock_wait_set_state() do { } while (0)
158 # define debug_rtlock_wait_restore_state() do { } while (0)
162 * set_current_state() includes a barrier so that the write of current->state
163 * is correctly serialised wrt the caller's subsequent test of whether to
167 * set_current_state(TASK_UNINTERRUPTIBLE);
173 * __set_current_state(TASK_RUNNING);
175 * If the caller does not need such serialisation (because, for instance, the
176 * CONDITION test and condition change and wakeup are under the same lock) then
177 * use __set_current_state().
179 * The above is typically ordered against the wakeup, which does:
182 * wake_up_state(p, TASK_UNINTERRUPTIBLE);
184 * where wake_up_state()/try_to_wake_up() executes a full memory barrier before
185 * accessing p->state.
187 * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
188 * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
189 * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
191 * However, with slightly different timing the wakeup TASK_RUNNING store can
192 * also collide with the TASK_UNINTERRUPTIBLE store. Losing that store is not
193 * a problem either because that will result in one extra go around the loop
194 * and our @cond test will save the day.
196 * Also see the comments of try_to_wake_up().
198 #define __set_current_state(state_value) \
200 debug_normal_state_change((state_value)); \
201 WRITE_ONCE(current->__state, (state_value)); \
204 #define set_current_state(state_value) \
206 debug_normal_state_change((state_value)); \
207 smp_store_mb(current->__state, (state_value)); \
211 * set_special_state() should be used for those states when the blocking task
212 * can not use the regular condition based wait-loop. In that case we must
213 * serialize against wakeups such that any possible in-flight TASK_RUNNING
214 * stores will not collide with our state change.
216 #define set_special_state(state_value) \
218 unsigned long flags; /* may shadow */ \
220 raw_spin_lock_irqsave(¤t->pi_lock, flags); \
221 debug_special_state_change((state_value)); \
222 WRITE_ONCE(current->__state, (state_value)); \
223 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
227 * PREEMPT_RT specific variants for "sleeping" spin/rwlocks
229 * RT's spin/rwlock substitutions are state preserving. The state of the
230 * task when blocking on the lock is saved in task_struct::saved_state and
231 * restored after the lock has been acquired. These operations are
232 * serialized by task_struct::pi_lock against try_to_wake_up(). Any non RT
233 * lock related wakeups while the task is blocked on the lock are
234 * redirected to operate on task_struct::saved_state to ensure that these
235 * are not dropped. On restore task_struct::saved_state is set to
236 * TASK_RUNNING so any wakeup attempt redirected to saved_state will fail.
238 * The lock operation looks like this:
240 * current_save_and_set_rtlock_wait_state();
244 * raw_spin_unlock_irq(&lock->wait_lock);
246 * raw_spin_lock_irq(&lock->wait_lock);
247 * set_current_state(TASK_RTLOCK_WAIT);
249 * current_restore_rtlock_saved_state();
251 #define current_save_and_set_rtlock_wait_state() \
253 lockdep_assert_irqs_disabled(); \
254 raw_spin_lock(¤t->pi_lock); \
255 current->saved_state = current->__state; \
256 debug_rtlock_wait_set_state(); \
257 WRITE_ONCE(current->__state, TASK_RTLOCK_WAIT); \
258 raw_spin_unlock(¤t->pi_lock); \
261 #define current_restore_rtlock_saved_state() \
263 lockdep_assert_irqs_disabled(); \
264 raw_spin_lock(¤t->pi_lock); \
265 debug_rtlock_wait_restore_state(); \
266 WRITE_ONCE(current->__state, current->saved_state); \
267 current->saved_state = TASK_RUNNING; \
268 raw_spin_unlock(¤t->pi_lock); \
271 #define get_current_state() READ_ONCE(current->__state)
273 /* Task command name length: */
274 #define TASK_COMM_LEN 16
276 extern void scheduler_tick(void);
278 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
280 extern long schedule_timeout(long timeout);
281 extern long schedule_timeout_interruptible(long timeout);
282 extern long schedule_timeout_killable(long timeout);
283 extern long schedule_timeout_uninterruptible(long timeout);
284 extern long schedule_timeout_idle(long timeout);
285 asmlinkage void schedule(void);
286 extern void schedule_preempt_disabled(void);
287 asmlinkage void preempt_schedule_irq(void);
288 #ifdef CONFIG_PREEMPT_RT
289 extern void schedule_rtlock(void);
292 extern int __must_check io_schedule_prepare(void);
293 extern void io_schedule_finish(int token);
294 extern long io_schedule_timeout(long timeout);
295 extern void io_schedule(void);
298 * struct prev_cputime - snapshot of system and user cputime
299 * @utime: time spent in user mode
300 * @stime: time spent in system mode
301 * @lock: protects the above two fields
303 * Stores previous user/system time values such that we can guarantee
306 struct prev_cputime {
307 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
315 /* Task is sleeping or running in a CPU with VTIME inactive: */
319 /* Task runs in kernelspace in a CPU with VTIME active: */
321 /* Task runs in userspace in a CPU with VTIME active: */
323 /* Task runs as guests in a CPU with VTIME active: */
329 unsigned long long starttime;
330 enum vtime_state state;
338 * Utilization clamp constraints.
339 * @UCLAMP_MIN: Minimum utilization
340 * @UCLAMP_MAX: Maximum utilization
341 * @UCLAMP_CNT: Utilization clamp constraints count
350 extern struct root_domain def_root_domain;
351 extern struct mutex sched_domains_mutex;
355 #ifdef CONFIG_SCHED_INFO
356 /* Cumulative counters: */
358 /* # of times we have run on this CPU: */
359 unsigned long pcount;
361 /* Time spent waiting on a runqueue: */
362 unsigned long long run_delay;
366 /* When did we last run on a CPU? */
367 unsigned long long last_arrival;
369 /* When were we last queued to run? */
370 unsigned long long last_queued;
372 #endif /* CONFIG_SCHED_INFO */
376 * Integer metrics need fixed point arithmetic, e.g., sched/fair
377 * has a few: load, load_avg, util_avg, freq, and capacity.
379 * We define a basic fixed point arithmetic range, and then formalize
380 * all these metrics based on that basic range.
382 # define SCHED_FIXEDPOINT_SHIFT 10
383 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
385 /* Increase resolution of cpu_capacity calculations */
386 # define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT
387 # define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
390 unsigned long weight;
395 * struct util_est - Estimation utilization of FAIR tasks
396 * @enqueued: instantaneous estimated utilization of a task/cpu
397 * @ewma: the Exponential Weighted Moving Average (EWMA)
398 * utilization of a task
400 * Support data structure to track an Exponential Weighted Moving Average
401 * (EWMA) of a FAIR task's utilization. New samples are added to the moving
402 * average each time a task completes an activation. Sample's weight is chosen
403 * so that the EWMA will be relatively insensitive to transient changes to the
406 * The enqueued attribute has a slightly different meaning for tasks and cpus:
407 * - task: the task's util_avg at last task dequeue time
408 * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU
409 * Thus, the util_est.enqueued of a task represents the contribution on the
410 * estimated utilization of the CPU where that task is currently enqueued.
412 * Only for tasks we track a moving average of the past instantaneous
413 * estimated utilization. This allows to absorb sporadic drops in utilization
414 * of an otherwise almost periodic task.
416 * The UTIL_AVG_UNCHANGED flag is used to synchronize util_est with util_avg
417 * updates. When a task is dequeued, its util_est should not be updated if its
418 * util_avg has not been updated in the meantime.
419 * This information is mapped into the MSB bit of util_est.enqueued at dequeue
420 * time. Since max value of util_est.enqueued for a task is 1024 (PELT util_avg
421 * for a task) it is safe to use MSB.
424 unsigned int enqueued;
426 #define UTIL_EST_WEIGHT_SHIFT 2
427 #define UTIL_AVG_UNCHANGED 0x80000000
428 } __attribute__((__aligned__(sizeof(u64))));
431 * The load/runnable/util_avg accumulates an infinite geometric series
432 * (see __update_load_avg_cfs_rq() in kernel/sched/pelt.c).
434 * [load_avg definition]
436 * load_avg = runnable% * scale_load_down(load)
438 * [runnable_avg definition]
440 * runnable_avg = runnable% * SCHED_CAPACITY_SCALE
442 * [util_avg definition]
444 * util_avg = running% * SCHED_CAPACITY_SCALE
446 * where runnable% is the time ratio that a sched_entity is runnable and
447 * running% the time ratio that a sched_entity is running.
449 * For cfs_rq, they are the aggregated values of all runnable and blocked
452 * The load/runnable/util_avg doesn't directly factor frequency scaling and CPU
453 * capacity scaling. The scaling is done through the rq_clock_pelt that is used
454 * for computing those signals (see update_rq_clock_pelt())
456 * N.B., the above ratios (runnable% and running%) themselves are in the
457 * range of [0, 1]. To do fixed point arithmetics, we therefore scale them
458 * to as large a range as necessary. This is for example reflected by
459 * util_avg's SCHED_CAPACITY_SCALE.
463 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
464 * with the highest load (=88761), always runnable on a single cfs_rq,
465 * and should not overflow as the number already hits PID_MAX_LIMIT.
467 * For all other cases (including 32-bit kernels), struct load_weight's
468 * weight will overflow first before we do, because:
470 * Max(load_avg) <= Max(load.weight)
472 * Then it is the load_weight's responsibility to consider overflow
476 u64 last_update_time;
481 unsigned long load_avg;
482 unsigned long runnable_avg;
483 unsigned long util_avg;
484 struct util_est util_est;
485 } ____cacheline_aligned;
487 struct sched_statistics {
488 #ifdef CONFIG_SCHEDSTATS
498 s64 sum_sleep_runtime;
505 u64 nr_migrations_cold;
506 u64 nr_failed_migrations_affine;
507 u64 nr_failed_migrations_running;
508 u64 nr_failed_migrations_hot;
509 u64 nr_forced_migrations;
513 u64 nr_wakeups_migrate;
514 u64 nr_wakeups_local;
515 u64 nr_wakeups_remote;
516 u64 nr_wakeups_affine;
517 u64 nr_wakeups_affine_attempts;
518 u64 nr_wakeups_passive;
523 struct sched_entity {
524 /* For load-balancing: */
525 struct load_weight load;
526 struct rb_node run_node;
527 struct list_head group_node;
531 u64 sum_exec_runtime;
533 u64 prev_sum_exec_runtime;
537 struct sched_statistics statistics;
539 #ifdef CONFIG_FAIR_GROUP_SCHED
541 struct sched_entity *parent;
542 /* rq on which this entity is (to be) queued: */
543 struct cfs_rq *cfs_rq;
544 /* rq "owned" by this entity/group: */
546 /* cached value of my_q->h_nr_running */
547 unsigned long runnable_weight;
552 * Per entity load average tracking.
554 * Put into separate cache line so it does not
555 * collide with read-mostly values above.
557 struct sched_avg avg;
561 struct sched_rt_entity {
562 struct list_head run_list;
563 unsigned long timeout;
564 unsigned long watchdog_stamp;
565 unsigned int time_slice;
566 unsigned short on_rq;
567 unsigned short on_list;
569 struct sched_rt_entity *back;
570 #ifdef CONFIG_RT_GROUP_SCHED
571 struct sched_rt_entity *parent;
572 /* rq on which this entity is (to be) queued: */
574 /* rq "owned" by this entity/group: */
577 } __randomize_layout;
579 struct sched_dl_entity {
580 struct rb_node rb_node;
583 * Original scheduling parameters. Copied here from sched_attr
584 * during sched_setattr(), they will remain the same until
585 * the next sched_setattr().
587 u64 dl_runtime; /* Maximum runtime for each instance */
588 u64 dl_deadline; /* Relative deadline of each instance */
589 u64 dl_period; /* Separation of two instances (period) */
590 u64 dl_bw; /* dl_runtime / dl_period */
591 u64 dl_density; /* dl_runtime / dl_deadline */
594 * Actual scheduling parameters. Initialized with the values above,
595 * they are continuously updated during task execution. Note that
596 * the remaining runtime could be < 0 in case we are in overrun.
598 s64 runtime; /* Remaining runtime for this instance */
599 u64 deadline; /* Absolute deadline for this instance */
600 unsigned int flags; /* Specifying the scheduler behaviour */
605 * @dl_throttled tells if we exhausted the runtime. If so, the
606 * task has to wait for a replenishment to be performed at the
607 * next firing of dl_timer.
609 * @dl_boosted tells if we are boosted due to DI. If so we are
610 * outside bandwidth enforcement mechanism (but only until we
611 * exit the critical section);
613 * @dl_yielded tells if task gave up the CPU before consuming
614 * all its available runtime during the last job.
616 * @dl_non_contending tells if the task is inactive while still
617 * contributing to the active utilization. In other words, it
618 * indicates if the inactive timer has been armed and its handler
619 * has not been executed yet. This flag is useful to avoid race
620 * conditions between the inactive timer handler and the wakeup
623 * @dl_overrun tells if the task asked to be informed about runtime
626 unsigned int dl_throttled : 1;
627 unsigned int dl_yielded : 1;
628 unsigned int dl_non_contending : 1;
629 unsigned int dl_overrun : 1;
632 * Bandwidth enforcement timer. Each -deadline task has its
633 * own bandwidth to be enforced, thus we need one timer per task.
635 struct hrtimer dl_timer;
638 * Inactive timer, responsible for decreasing the active utilization
639 * at the "0-lag time". When a -deadline task blocks, it contributes
640 * to GRUB's active utilization until the "0-lag time", hence a
641 * timer is needed to decrease the active utilization at the correct
644 struct hrtimer inactive_timer;
646 #ifdef CONFIG_RT_MUTEXES
648 * Priority Inheritance. When a DEADLINE scheduling entity is boosted
649 * pi_se points to the donor, otherwise points to the dl_se it belongs
650 * to (the original one/itself).
652 struct sched_dl_entity *pi_se;
656 #ifdef CONFIG_UCLAMP_TASK
657 /* Number of utilization clamp buckets (shorter alias) */
658 #define UCLAMP_BUCKETS CONFIG_UCLAMP_BUCKETS_COUNT
661 * Utilization clamp for a scheduling entity
662 * @value: clamp value "assigned" to a se
663 * @bucket_id: bucket index corresponding to the "assigned" value
664 * @active: the se is currently refcounted in a rq's bucket
665 * @user_defined: the requested clamp value comes from user-space
667 * The bucket_id is the index of the clamp bucket matching the clamp value
668 * which is pre-computed and stored to avoid expensive integer divisions from
671 * The active bit is set whenever a task has got an "effective" value assigned,
672 * which can be different from the clamp value "requested" from user-space.
673 * This allows to know a task is refcounted in the rq's bucket corresponding
674 * to the "effective" bucket_id.
676 * The user_defined bit is set whenever a task has got a task-specific clamp
677 * value requested from userspace, i.e. the system defaults apply to this task
678 * just as a restriction. This allows to relax default clamps when a less
679 * restrictive task-specific value has been requested, thus allowing to
680 * implement a "nice" semantic. For example, a task running with a 20%
681 * default boost can still drop its own boosting to 0%.
684 unsigned int value : bits_per(SCHED_CAPACITY_SCALE);
685 unsigned int bucket_id : bits_per(UCLAMP_BUCKETS);
686 unsigned int active : 1;
687 unsigned int user_defined : 1;
689 #endif /* CONFIG_UCLAMP_TASK */
695 u8 exp_hint; /* Hint for performance. */
696 u8 need_mb; /* Readers need smp_mb(). */
698 u32 s; /* Set of bits. */
701 enum perf_event_task_context {
702 perf_invalid_context = -1,
705 perf_nr_task_contexts,
709 struct wake_q_node *next;
713 #ifdef CONFIG_KMAP_LOCAL
715 pte_t pteval[KM_MAX_IDX];
720 #ifdef CONFIG_THREAD_INFO_IN_TASK
722 * For reasons of header soup (see current_thread_info()), this
723 * must be the first element of task_struct.
725 struct thread_info thread_info;
727 unsigned int __state;
729 #ifdef CONFIG_PREEMPT_RT
730 /* saved state for "spinlock sleepers" */
731 unsigned int saved_state;
735 * This begins the randomizable portion of task_struct. Only
736 * scheduling-critical items should be added above here.
738 randomized_struct_fields_start
742 /* Per task flags (PF_*), defined further below: */
748 struct __call_single_node wake_entry;
749 #ifdef CONFIG_THREAD_INFO_IN_TASK
753 unsigned int wakee_flips;
754 unsigned long wakee_flip_decay_ts;
755 struct task_struct *last_wakee;
758 * recent_used_cpu is initially set as the last CPU used by a task
759 * that wakes affine another task. Waker/wakee relationships can
760 * push tasks around a CPU where each wakeup moves to the next one.
761 * Tracking a recently used CPU allows a quick search for a recently
762 * used CPU that may be idle.
772 unsigned int rt_priority;
774 const struct sched_class *sched_class;
775 struct sched_entity se;
776 struct sched_rt_entity rt;
777 struct sched_dl_entity dl;
779 #ifdef CONFIG_SCHED_CORE
780 struct rb_node core_node;
781 unsigned long core_cookie;
782 unsigned int core_occupation;
785 #ifdef CONFIG_CGROUP_SCHED
786 struct task_group *sched_task_group;
789 #ifdef CONFIG_UCLAMP_TASK
791 * Clamp values requested for a scheduling entity.
792 * Must be updated with task_rq_lock() held.
794 struct uclamp_se uclamp_req[UCLAMP_CNT];
796 * Effective clamp values used for a scheduling entity.
797 * Must be updated with task_rq_lock() held.
799 struct uclamp_se uclamp[UCLAMP_CNT];
802 #ifdef CONFIG_PREEMPT_NOTIFIERS
803 /* List of struct preempt_notifier: */
804 struct hlist_head preempt_notifiers;
807 #ifdef CONFIG_BLK_DEV_IO_TRACE
808 unsigned int btrace_seq;
813 const cpumask_t *cpus_ptr;
814 cpumask_t *user_cpus_ptr;
816 void *migration_pending;
818 unsigned short migration_disabled;
820 unsigned short migration_flags;
822 #ifdef CONFIG_PREEMPT_RCU
823 int rcu_read_lock_nesting;
824 union rcu_special rcu_read_unlock_special;
825 struct list_head rcu_node_entry;
826 struct rcu_node *rcu_blocked_node;
827 #endif /* #ifdef CONFIG_PREEMPT_RCU */
829 #ifdef CONFIG_TASKS_RCU
830 unsigned long rcu_tasks_nvcsw;
831 u8 rcu_tasks_holdout;
833 int rcu_tasks_idle_cpu;
834 struct list_head rcu_tasks_holdout_list;
835 #endif /* #ifdef CONFIG_TASKS_RCU */
837 #ifdef CONFIG_TASKS_TRACE_RCU
838 int trc_reader_nesting;
840 union rcu_special trc_reader_special;
841 bool trc_reader_checked;
842 struct list_head trc_holdout_list;
843 #endif /* #ifdef CONFIG_TASKS_TRACE_RCU */
845 struct sched_info sched_info;
847 struct list_head tasks;
849 struct plist_node pushable_tasks;
850 struct rb_node pushable_dl_tasks;
853 struct mm_struct *mm;
854 struct mm_struct *active_mm;
856 /* Per-thread vma caching: */
857 struct vmacache vmacache;
859 #ifdef SPLIT_RSS_COUNTING
860 struct task_rss_stat rss_stat;
865 /* The signal sent when the parent dies: */
867 /* JOBCTL_*, siglock protected: */
868 unsigned long jobctl;
870 /* Used for emulating ABI behavior of previous Linux versions: */
871 unsigned int personality;
873 /* Scheduler bits, serialized by scheduler locks: */
874 unsigned sched_reset_on_fork:1;
875 unsigned sched_contributes_to_load:1;
876 unsigned sched_migrated:1;
878 unsigned sched_psi_wake_requeue:1;
881 /* Force alignment to the next boundary: */
884 /* Unserialized, strictly 'current' */
887 * This field must not be in the scheduler word above due to wakelist
888 * queueing no longer being serialized by p->on_cpu. However:
891 * schedule() if (p->on_rq && ..) // false
892 * smp_mb__after_spinlock(); if (smp_load_acquire(&p->on_cpu) && //true
893 * deactivate_task() ttwu_queue_wakelist())
894 * p->on_rq = 0; p->sched_remote_wakeup = Y;
896 * guarantees all stores of 'current' are visible before
897 * ->sched_remote_wakeup gets used, so it can be in this word.
899 unsigned sched_remote_wakeup:1;
901 /* Bit to tell LSMs we're in execve(): */
902 unsigned in_execve:1;
903 unsigned in_iowait:1;
904 #ifndef TIF_RESTORE_SIGMASK
905 unsigned restore_sigmask:1;
908 unsigned in_user_fault:1;
910 #ifdef CONFIG_COMPAT_BRK
911 unsigned brk_randomized:1;
913 #ifdef CONFIG_CGROUPS
914 /* disallow userland-initiated cgroup migration */
915 unsigned no_cgroup_migration:1;
916 /* task is frozen/stopped (used by the cgroup freezer) */
919 #ifdef CONFIG_BLK_CGROUP
920 unsigned use_memdelay:1;
923 /* Stalled due to lack of memory */
924 unsigned in_memstall:1;
926 #ifdef CONFIG_PAGE_OWNER
927 /* Used by page_owner=on to detect recursion in page tracking. */
928 unsigned in_page_owner:1;
930 #ifdef CONFIG_EVENTFD
931 /* Recursion prevention for eventfd_signal() */
932 unsigned in_eventfd:1;
935 unsigned long atomic_flags; /* Flags requiring atomic access. */
937 struct restart_block restart_block;
942 #ifdef CONFIG_STACKPROTECTOR
943 /* Canary value for the -fstack-protector GCC feature: */
944 unsigned long stack_canary;
947 * Pointers to the (original) parent process, youngest child, younger sibling,
948 * older sibling, respectively. (p->father can be replaced with
949 * p->real_parent->pid)
952 /* Real parent process: */
953 struct task_struct __rcu *real_parent;
955 /* Recipient of SIGCHLD, wait4() reports: */
956 struct task_struct __rcu *parent;
959 * Children/sibling form the list of natural children:
961 struct list_head children;
962 struct list_head sibling;
963 struct task_struct *group_leader;
966 * 'ptraced' is the list of tasks this task is using ptrace() on.
968 * This includes both natural children and PTRACE_ATTACH targets.
969 * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
971 struct list_head ptraced;
972 struct list_head ptrace_entry;
974 /* PID/PID hash table linkage. */
975 struct pid *thread_pid;
976 struct hlist_node pid_links[PIDTYPE_MAX];
977 struct list_head thread_group;
978 struct list_head thread_node;
980 struct completion *vfork_done;
982 /* CLONE_CHILD_SETTID: */
983 int __user *set_child_tid;
985 /* CLONE_CHILD_CLEARTID: */
986 int __user *clear_child_tid;
993 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
998 struct prev_cputime prev_cputime;
999 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1003 #ifdef CONFIG_NO_HZ_FULL
1004 atomic_t tick_dep_mask;
1006 /* Context switch counts: */
1007 unsigned long nvcsw;
1008 unsigned long nivcsw;
1010 /* Monotonic time in nsecs: */
1013 /* Boot based time in nsecs: */
1016 /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
1017 unsigned long min_flt;
1018 unsigned long maj_flt;
1020 /* Empty if CONFIG_POSIX_CPUTIMERS=n */
1021 struct posix_cputimers posix_cputimers;
1023 #ifdef CONFIG_POSIX_CPU_TIMERS_TASK_WORK
1024 struct posix_cputimers_work posix_cputimers_work;
1027 /* Process credentials: */
1029 /* Tracer's credentials at attach: */
1030 const struct cred __rcu *ptracer_cred;
1032 /* Objective and real subjective task credentials (COW): */
1033 const struct cred __rcu *real_cred;
1035 /* Effective (overridable) subjective task credentials (COW): */
1036 const struct cred __rcu *cred;
1039 /* Cached requested key. */
1040 struct key *cached_requested_key;
1044 * executable name, excluding path.
1046 * - normally initialized setup_new_exec()
1047 * - access it with [gs]et_task_comm()
1048 * - lock it with task_lock()
1050 char comm[TASK_COMM_LEN];
1052 struct nameidata *nameidata;
1054 #ifdef CONFIG_SYSVIPC
1055 struct sysv_sem sysvsem;
1056 struct sysv_shm sysvshm;
1058 #ifdef CONFIG_DETECT_HUNG_TASK
1059 unsigned long last_switch_count;
1060 unsigned long last_switch_time;
1062 /* Filesystem information: */
1063 struct fs_struct *fs;
1065 /* Open file information: */
1066 struct files_struct *files;
1068 #ifdef CONFIG_IO_URING
1069 struct io_uring_task *io_uring;
1073 struct nsproxy *nsproxy;
1075 /* Signal handlers: */
1076 struct signal_struct *signal;
1077 struct sighand_struct __rcu *sighand;
1079 sigset_t real_blocked;
1080 /* Restored if set_restore_sigmask() was used: */
1081 sigset_t saved_sigmask;
1082 struct sigpending pending;
1083 #ifdef CONFIG_PREEMPT_RT
1084 /* TODO: move me into ->restart_block ? */
1085 struct kernel_siginfo forced_info;
1087 unsigned long sas_ss_sp;
1089 unsigned int sas_ss_flags;
1091 struct callback_head *task_works;
1094 #ifdef CONFIG_AUDITSYSCALL
1095 struct audit_context *audit_context;
1098 unsigned int sessionid;
1100 struct seccomp seccomp;
1101 struct syscall_user_dispatch syscall_dispatch;
1103 /* Thread group tracking: */
1107 /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
1108 spinlock_t alloc_lock;
1110 /* Protection of the PI data structures: */
1111 raw_spinlock_t pi_lock;
1113 struct wake_q_node wake_q;
1115 #ifdef CONFIG_RT_MUTEXES
1116 /* PI waiters blocked on a rt_mutex held by this task: */
1117 struct rb_root_cached pi_waiters;
1118 /* Updated under owner's pi_lock and rq lock */
1119 struct task_struct *pi_top_task;
1120 /* Deadlock detection and priority inheritance handling: */
1121 struct rt_mutex_waiter *pi_blocked_on;
1124 #ifdef CONFIG_DEBUG_MUTEXES
1125 /* Mutex deadlock detection: */
1126 struct mutex_waiter *blocked_on;
1129 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1130 int non_block_count;
1133 #ifdef CONFIG_TRACE_IRQFLAGS
1134 struct irqtrace_events irqtrace;
1135 unsigned int hardirq_threaded;
1136 u64 hardirq_chain_key;
1137 int softirqs_enabled;
1138 int softirq_context;
1141 #ifdef CONFIG_PREEMPT_RT
1142 int softirq_disable_cnt;
1145 #ifdef CONFIG_LOCKDEP
1146 # define MAX_LOCK_DEPTH 48UL
1149 unsigned int lockdep_recursion;
1150 struct held_lock held_locks[MAX_LOCK_DEPTH];
1153 #if defined(CONFIG_UBSAN) && !defined(CONFIG_UBSAN_TRAP)
1154 unsigned int in_ubsan;
1157 /* Journalling filesystem info: */
1160 /* Stacked block device info: */
1161 struct bio_list *bio_list;
1164 /* Stack plugging: */
1165 struct blk_plug *plug;
1169 struct reclaim_state *reclaim_state;
1171 struct backing_dev_info *backing_dev_info;
1173 struct io_context *io_context;
1175 #ifdef CONFIG_COMPACTION
1176 struct capture_control *capture_control;
1179 unsigned long ptrace_message;
1180 kernel_siginfo_t *last_siginfo;
1182 struct task_io_accounting ioac;
1184 /* Pressure stall state */
1185 unsigned int psi_flags;
1187 #ifdef CONFIG_TASK_XACCT
1188 /* Accumulated RSS usage: */
1190 /* Accumulated virtual memory usage: */
1192 /* stime + utime since last update: */
1195 #ifdef CONFIG_CPUSETS
1196 /* Protected by ->alloc_lock: */
1197 nodemask_t mems_allowed;
1198 /* Sequence number to catch updates: */
1199 seqcount_spinlock_t mems_allowed_seq;
1200 int cpuset_mem_spread_rotor;
1201 int cpuset_slab_spread_rotor;
1203 #ifdef CONFIG_CGROUPS
1204 /* Control Group info protected by css_set_lock: */
1205 struct css_set __rcu *cgroups;
1206 /* cg_list protected by css_set_lock and tsk->alloc_lock: */
1207 struct list_head cg_list;
1209 #ifdef CONFIG_X86_CPU_RESCTRL
1214 struct robust_list_head __user *robust_list;
1215 #ifdef CONFIG_COMPAT
1216 struct compat_robust_list_head __user *compat_robust_list;
1218 struct list_head pi_state_list;
1219 struct futex_pi_state *pi_state_cache;
1220 struct mutex futex_exit_mutex;
1221 unsigned int futex_state;
1223 #ifdef CONFIG_PERF_EVENTS
1224 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1225 struct mutex perf_event_mutex;
1226 struct list_head perf_event_list;
1228 #ifdef CONFIG_DEBUG_PREEMPT
1229 unsigned long preempt_disable_ip;
1232 /* Protected by alloc_lock: */
1233 struct mempolicy *mempolicy;
1235 short pref_node_fork;
1237 #ifdef CONFIG_NUMA_BALANCING
1239 unsigned int numa_scan_period;
1240 unsigned int numa_scan_period_max;
1241 int numa_preferred_nid;
1242 unsigned long numa_migrate_retry;
1243 /* Migration stamp: */
1245 u64 last_task_numa_placement;
1246 u64 last_sum_exec_runtime;
1247 struct callback_head numa_work;
1250 * This pointer is only modified for current in syscall and
1251 * pagefault context (and for tasks being destroyed), so it can be read
1252 * from any of the following contexts:
1253 * - RCU read-side critical section
1254 * - current->numa_group from everywhere
1255 * - task's runqueue locked, task not running
1257 struct numa_group __rcu *numa_group;
1260 * numa_faults is an array split into four regions:
1261 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1262 * in this precise order.
1264 * faults_memory: Exponential decaying average of faults on a per-node
1265 * basis. Scheduling placement decisions are made based on these
1266 * counts. The values remain static for the duration of a PTE scan.
1267 * faults_cpu: Track the nodes the process was running on when a NUMA
1268 * hinting fault was incurred.
1269 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1270 * during the current scan window. When the scan completes, the counts
1271 * in faults_memory and faults_cpu decay and these values are copied.
1273 unsigned long *numa_faults;
1274 unsigned long total_numa_faults;
1277 * numa_faults_locality tracks if faults recorded during the last
1278 * scan window were remote/local or failed to migrate. The task scan
1279 * period is adapted based on the locality of the faults with different
1280 * weights depending on whether they were shared or private faults
1282 unsigned long numa_faults_locality[3];
1284 unsigned long numa_pages_migrated;
1285 #endif /* CONFIG_NUMA_BALANCING */
1288 struct rseq __user *rseq;
1291 * RmW on rseq_event_mask must be performed atomically
1292 * with respect to preemption.
1294 unsigned long rseq_event_mask;
1297 struct tlbflush_unmap_batch tlb_ubc;
1300 refcount_t rcu_users;
1301 struct rcu_head rcu;
1304 /* Cache last used pipe for splice(): */
1305 struct pipe_inode_info *splice_pipe;
1307 struct page_frag task_frag;
1309 #ifdef CONFIG_TASK_DELAY_ACCT
1310 struct task_delay_info *delays;
1313 #ifdef CONFIG_FAULT_INJECTION
1315 unsigned int fail_nth;
1318 * When (nr_dirtied >= nr_dirtied_pause), it's time to call
1319 * balance_dirty_pages() for a dirty throttling pause:
1322 int nr_dirtied_pause;
1323 /* Start of a write-and-pause period: */
1324 unsigned long dirty_paused_when;
1326 #ifdef CONFIG_LATENCYTOP
1327 int latency_record_count;
1328 struct latency_record latency_record[LT_SAVECOUNT];
1331 * Time slack values; these are used to round up poll() and
1332 * select() etc timeout values. These are in nanoseconds.
1335 u64 default_timer_slack_ns;
1337 #if defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS)
1338 unsigned int kasan_depth;
1342 struct kcsan_ctx kcsan_ctx;
1343 #ifdef CONFIG_TRACE_IRQFLAGS
1344 struct irqtrace_events kcsan_save_irqtrace;
1348 #if IS_ENABLED(CONFIG_KUNIT)
1349 struct kunit *kunit_test;
1352 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1353 /* Index of current stored address in ret_stack: */
1357 /* Stack of return addresses for return function tracing: */
1358 struct ftrace_ret_stack *ret_stack;
1360 /* Timestamp for last schedule: */
1361 unsigned long long ftrace_timestamp;
1364 * Number of functions that haven't been traced
1365 * because of depth overrun:
1367 atomic_t trace_overrun;
1369 /* Pause tracing: */
1370 atomic_t tracing_graph_pause;
1373 #ifdef CONFIG_TRACING
1374 /* State flags for use by tracers: */
1375 unsigned long trace;
1377 /* Bitmask and counter of trace recursion: */
1378 unsigned long trace_recursion;
1379 #endif /* CONFIG_TRACING */
1382 /* See kernel/kcov.c for more details. */
1384 /* Coverage collection mode enabled for this task (0 if disabled): */
1385 unsigned int kcov_mode;
1387 /* Size of the kcov_area: */
1388 unsigned int kcov_size;
1390 /* Buffer for coverage collection: */
1393 /* KCOV descriptor wired with this task or NULL: */
1396 /* KCOV common handle for remote coverage collection: */
1399 /* KCOV sequence number: */
1402 /* Collect coverage from softirq context: */
1403 unsigned int kcov_softirq;
1407 struct mem_cgroup *memcg_in_oom;
1408 gfp_t memcg_oom_gfp_mask;
1409 int memcg_oom_order;
1411 /* Number of pages to reclaim on returning to userland: */
1412 unsigned int memcg_nr_pages_over_high;
1414 /* Used by memcontrol for targeted memcg charge: */
1415 struct mem_cgroup *active_memcg;
1418 #ifdef CONFIG_BLK_CGROUP
1419 struct request_queue *throttle_queue;
1422 #ifdef CONFIG_UPROBES
1423 struct uprobe_task *utask;
1425 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1426 unsigned int sequential_io;
1427 unsigned int sequential_io_avg;
1429 struct kmap_ctrl kmap_ctrl;
1430 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1431 unsigned long task_state_change;
1432 # ifdef CONFIG_PREEMPT_RT
1433 unsigned long saved_state_change;
1436 int pagefault_disabled;
1438 struct task_struct *oom_reaper_list;
1439 struct timer_list oom_reaper_timer;
1441 #ifdef CONFIG_VMAP_STACK
1442 struct vm_struct *stack_vm_area;
1444 #ifdef CONFIG_THREAD_INFO_IN_TASK
1445 /* A live task holds one reference: */
1446 refcount_t stack_refcount;
1448 #ifdef CONFIG_LIVEPATCH
1451 #ifdef CONFIG_SECURITY
1452 /* Used by LSM modules for access restriction: */
1455 #ifdef CONFIG_BPF_SYSCALL
1456 /* Used by BPF task local storage */
1457 struct bpf_local_storage __rcu *bpf_storage;
1458 /* Used for BPF run context */
1459 struct bpf_run_ctx *bpf_ctx;
1462 #ifdef CONFIG_GCC_PLUGIN_STACKLEAK
1463 unsigned long lowest_stack;
1464 unsigned long prev_lowest_stack;
1467 #ifdef CONFIG_X86_MCE
1468 void __user *mce_vaddr;
1473 __mce_reserved : 62;
1474 struct callback_head mce_kill_me;
1478 #ifdef CONFIG_KRETPROBES
1479 struct llist_head kretprobe_instances;
1482 #ifdef CONFIG_ARCH_HAS_PARANOID_L1D_FLUSH
1484 * If L1D flush is supported on mm context switch
1485 * then we use this callback head to queue kill work
1486 * to kill tasks that are not running on SMT disabled
1489 struct callback_head l1d_flush_kill;
1493 * New fields for task_struct should be added above here, so that
1494 * they are included in the randomized portion of task_struct.
1496 randomized_struct_fields_end
1498 /* CPU-specific state of this task: */
1499 struct thread_struct thread;
1502 * WARNING: on x86, 'thread_struct' contains a variable-sized
1503 * structure. It *MUST* be at the end of 'task_struct'.
1505 * Do not put anything below here!
1509 static inline struct pid *task_pid(struct task_struct *task)
1511 return task->thread_pid;
1515 * the helpers to get the task's different pids as they are seen
1516 * from various namespaces
1518 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1519 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1521 * task_xid_nr_ns() : id seen from the ns specified;
1523 * see also pid_nr() etc in include/linux/pid.h
1525 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns);
1527 static inline pid_t task_pid_nr(struct task_struct *tsk)
1532 static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1534 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1537 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1539 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1543 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1549 * pid_alive - check that a task structure is not stale
1550 * @p: Task structure to be checked.
1552 * Test if a process is not yet dead (at most zombie state)
1553 * If pid_alive fails, then pointers within the task structure
1554 * can be stale and must not be dereferenced.
1556 * Return: 1 if the process is alive. 0 otherwise.
1558 static inline int pid_alive(const struct task_struct *p)
1560 return p->thread_pid != NULL;
1563 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1565 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1568 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1570 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1574 static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1576 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1579 static inline pid_t task_session_vnr(struct task_struct *tsk)
1581 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1584 static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1586 return __task_pid_nr_ns(tsk, PIDTYPE_TGID, ns);
1589 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1591 return __task_pid_nr_ns(tsk, PIDTYPE_TGID, NULL);
1594 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1600 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1606 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1608 return task_ppid_nr_ns(tsk, &init_pid_ns);
1611 /* Obsolete, do not use: */
1612 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1614 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1617 #define TASK_REPORT_IDLE (TASK_REPORT + 1)
1618 #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1)
1620 static inline unsigned int task_state_index(struct task_struct *tsk)
1622 unsigned int tsk_state = READ_ONCE(tsk->__state);
1623 unsigned int state = (tsk_state | tsk->exit_state) & TASK_REPORT;
1625 BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX);
1627 if (tsk_state == TASK_IDLE)
1628 state = TASK_REPORT_IDLE;
1631 * We're lying here, but rather than expose a completely new task state
1632 * to userspace, we can make this appear as if the task has gone through
1633 * a regular rt_mutex_lock() call.
1635 if (tsk_state == TASK_RTLOCK_WAIT)
1636 state = TASK_UNINTERRUPTIBLE;
1641 static inline char task_index_to_char(unsigned int state)
1643 static const char state_char[] = "RSDTtXZPI";
1645 BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX) != sizeof(state_char) - 1);
1647 return state_char[state];
1650 static inline char task_state_to_char(struct task_struct *tsk)
1652 return task_index_to_char(task_state_index(tsk));
1656 * is_global_init - check if a task structure is init. Since init
1657 * is free to have sub-threads we need to check tgid.
1658 * @tsk: Task structure to be checked.
1660 * Check if a task structure is the first user space task the kernel created.
1662 * Return: 1 if the task structure is init. 0 otherwise.
1664 static inline int is_global_init(struct task_struct *tsk)
1666 return task_tgid_nr(tsk) == 1;
1669 extern struct pid *cad_pid;
1674 #define PF_VCPU 0x00000001 /* I'm a virtual CPU */
1675 #define PF_IDLE 0x00000002 /* I am an IDLE thread */
1676 #define PF_EXITING 0x00000004 /* Getting shut down */
1677 #define PF_IO_WORKER 0x00000010 /* Task is an IO worker */
1678 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1679 #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */
1680 #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */
1681 #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */
1682 #define PF_DUMPCORE 0x00000200 /* Dumped core */
1683 #define PF_SIGNALED 0x00000400 /* Killed by a signal */
1684 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1685 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */
1686 #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */
1687 #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */
1688 #define PF_FROZEN 0x00010000 /* Frozen for system suspend */
1689 #define PF_KSWAPD 0x00020000 /* I am kswapd */
1690 #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */
1691 #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */
1692 #define PF_LOCAL_THROTTLE 0x00100000 /* Throttle writes only against the bdi I write to,
1693 * I am cleaning dirty pages from some other bdi. */
1694 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1695 #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */
1696 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1697 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_mask */
1698 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1699 #define PF_MEMALLOC_PIN 0x10000000 /* Allocation context constrained to zones which allow long term pinning. */
1700 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1701 #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */
1704 * Only the _current_ task can read/write to tsk->flags, but other
1705 * tasks can access tsk->flags in readonly mode for example
1706 * with tsk_used_math (like during threaded core dumping).
1707 * There is however an exception to this rule during ptrace
1708 * or during fork: the ptracer task is allowed to write to the
1709 * child->flags of its traced child (same goes for fork, the parent
1710 * can write to the child->flags), because we're guaranteed the
1711 * child is not running and in turn not changing child->flags
1712 * at the same time the parent does it.
1714 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1715 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1716 #define clear_used_math() clear_stopped_child_used_math(current)
1717 #define set_used_math() set_stopped_child_used_math(current)
1719 #define conditional_stopped_child_used_math(condition, child) \
1720 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1722 #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current)
1724 #define copy_to_stopped_child_used_math(child) \
1725 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1727 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1728 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1729 #define used_math() tsk_used_math(current)
1731 static __always_inline bool is_percpu_thread(void)
1734 return (current->flags & PF_NO_SETAFFINITY) &&
1735 (current->nr_cpus_allowed == 1);
1741 /* Is the current task guaranteed to stay on its current CPU? */
1742 static inline bool is_migratable(void)
1745 return preemptible() && !current->migration_disabled;
1751 /* Per-process atomic flags. */
1752 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
1753 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
1754 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
1755 #define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */
1756 #define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/
1757 #define PFA_SPEC_IB_DISABLE 5 /* Indirect branch speculation restricted */
1758 #define PFA_SPEC_IB_FORCE_DISABLE 6 /* Indirect branch speculation permanently restricted */
1759 #define PFA_SPEC_SSB_NOEXEC 7 /* Speculative Store Bypass clear on execve() */
1761 #define TASK_PFA_TEST(name, func) \
1762 static inline bool task_##func(struct task_struct *p) \
1763 { return test_bit(PFA_##name, &p->atomic_flags); }
1765 #define TASK_PFA_SET(name, func) \
1766 static inline void task_set_##func(struct task_struct *p) \
1767 { set_bit(PFA_##name, &p->atomic_flags); }
1769 #define TASK_PFA_CLEAR(name, func) \
1770 static inline void task_clear_##func(struct task_struct *p) \
1771 { clear_bit(PFA_##name, &p->atomic_flags); }
1773 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
1774 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
1776 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
1777 TASK_PFA_SET(SPREAD_PAGE, spread_page)
1778 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
1780 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
1781 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
1782 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
1784 TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable)
1785 TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable)
1786 TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable)
1788 TASK_PFA_TEST(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1789 TASK_PFA_SET(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1790 TASK_PFA_CLEAR(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1792 TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1793 TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1795 TASK_PFA_TEST(SPEC_IB_DISABLE, spec_ib_disable)
1796 TASK_PFA_SET(SPEC_IB_DISABLE, spec_ib_disable)
1797 TASK_PFA_CLEAR(SPEC_IB_DISABLE, spec_ib_disable)
1799 TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1800 TASK_PFA_SET(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1803 current_restore_flags(unsigned long orig_flags, unsigned long flags)
1805 current->flags &= ~flags;
1806 current->flags |= orig_flags & flags;
1809 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial);
1810 extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_effective_cpus);
1812 extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask);
1813 extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask);
1814 extern int dup_user_cpus_ptr(struct task_struct *dst, struct task_struct *src, int node);
1815 extern void release_user_cpus_ptr(struct task_struct *p);
1816 extern int dl_task_check_affinity(struct task_struct *p, const struct cpumask *mask);
1817 extern void force_compatible_cpus_allowed_ptr(struct task_struct *p);
1818 extern void relax_compatible_cpus_allowed_ptr(struct task_struct *p);
1820 static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
1823 static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
1825 if (!cpumask_test_cpu(0, new_mask))
1829 static inline int dup_user_cpus_ptr(struct task_struct *dst, struct task_struct *src, int node)
1831 if (src->user_cpus_ptr)
1835 static inline void release_user_cpus_ptr(struct task_struct *p)
1837 WARN_ON(p->user_cpus_ptr);
1840 static inline int dl_task_check_affinity(struct task_struct *p, const struct cpumask *mask)
1846 extern int yield_to(struct task_struct *p, bool preempt);
1847 extern void set_user_nice(struct task_struct *p, long nice);
1848 extern int task_prio(const struct task_struct *p);
1851 * task_nice - return the nice value of a given task.
1852 * @p: the task in question.
1854 * Return: The nice value [ -20 ... 0 ... 19 ].
1856 static inline int task_nice(const struct task_struct *p)
1858 return PRIO_TO_NICE((p)->static_prio);
1861 extern int can_nice(const struct task_struct *p, const int nice);
1862 extern int task_curr(const struct task_struct *p);
1863 extern int idle_cpu(int cpu);
1864 extern int available_idle_cpu(int cpu);
1865 extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *);
1866 extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *);
1867 extern void sched_set_fifo(struct task_struct *p);
1868 extern void sched_set_fifo_low(struct task_struct *p);
1869 extern void sched_set_normal(struct task_struct *p, int nice);
1870 extern int sched_setattr(struct task_struct *, const struct sched_attr *);
1871 extern int sched_setattr_nocheck(struct task_struct *, const struct sched_attr *);
1872 extern struct task_struct *idle_task(int cpu);
1875 * is_idle_task - is the specified task an idle task?
1876 * @p: the task in question.
1878 * Return: 1 if @p is an idle task. 0 otherwise.
1880 static __always_inline bool is_idle_task(const struct task_struct *p)
1882 return !!(p->flags & PF_IDLE);
1885 extern struct task_struct *curr_task(int cpu);
1886 extern void ia64_set_curr_task(int cpu, struct task_struct *p);
1890 union thread_union {
1891 #ifndef CONFIG_ARCH_TASK_STRUCT_ON_STACK
1892 struct task_struct task;
1894 #ifndef CONFIG_THREAD_INFO_IN_TASK
1895 struct thread_info thread_info;
1897 unsigned long stack[THREAD_SIZE/sizeof(long)];
1900 #ifndef CONFIG_THREAD_INFO_IN_TASK
1901 extern struct thread_info init_thread_info;
1904 extern unsigned long init_stack[THREAD_SIZE / sizeof(unsigned long)];
1906 #ifdef CONFIG_THREAD_INFO_IN_TASK
1907 static inline struct thread_info *task_thread_info(struct task_struct *task)
1909 return &task->thread_info;
1911 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1912 # define task_thread_info(task) ((struct thread_info *)(task)->stack)
1916 * find a task by one of its numerical ids
1918 * find_task_by_pid_ns():
1919 * finds a task by its pid in the specified namespace
1920 * find_task_by_vpid():
1921 * finds a task by its virtual pid
1923 * see also find_vpid() etc in include/linux/pid.h
1926 extern struct task_struct *find_task_by_vpid(pid_t nr);
1927 extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns);
1930 * find a task by its virtual pid and get the task struct
1932 extern struct task_struct *find_get_task_by_vpid(pid_t nr);
1934 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1935 extern int wake_up_process(struct task_struct *tsk);
1936 extern void wake_up_new_task(struct task_struct *tsk);
1939 extern void kick_process(struct task_struct *tsk);
1941 static inline void kick_process(struct task_struct *tsk) { }
1944 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
1946 static inline void set_task_comm(struct task_struct *tsk, const char *from)
1948 __set_task_comm(tsk, from, false);
1951 extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk);
1952 #define get_task_comm(buf, tsk) ({ \
1953 BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \
1954 __get_task_comm(buf, sizeof(buf), tsk); \
1958 static __always_inline void scheduler_ipi(void)
1961 * Fold TIF_NEED_RESCHED into the preempt_count; anybody setting
1962 * TIF_NEED_RESCHED remotely (for the first time) will also send
1965 preempt_fold_need_resched();
1967 extern unsigned long wait_task_inactive(struct task_struct *, unsigned int match_state);
1969 static inline void scheduler_ipi(void) { }
1970 static inline unsigned long wait_task_inactive(struct task_struct *p, unsigned int match_state)
1977 * Set thread flags in other task's structures.
1978 * See asm/thread_info.h for TIF_xxxx flags available:
1980 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1982 set_ti_thread_flag(task_thread_info(tsk), flag);
1985 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1987 clear_ti_thread_flag(task_thread_info(tsk), flag);
1990 static inline void update_tsk_thread_flag(struct task_struct *tsk, int flag,
1993 update_ti_thread_flag(task_thread_info(tsk), flag, value);
1996 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1998 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2001 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2003 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2006 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2008 return test_ti_thread_flag(task_thread_info(tsk), flag);
2011 static inline void set_tsk_need_resched(struct task_struct *tsk)
2013 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2016 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2018 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2021 static inline int test_tsk_need_resched(struct task_struct *tsk)
2023 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2026 #ifdef CONFIG_PREEMPT_LAZY
2027 static inline void set_tsk_need_resched_lazy(struct task_struct *tsk)
2029 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED_LAZY);
2032 static inline void clear_tsk_need_resched_lazy(struct task_struct *tsk)
2034 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED_LAZY);
2037 static inline int test_tsk_need_resched_lazy(struct task_struct *tsk)
2039 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED_LAZY));
2042 static inline int need_resched_lazy(void)
2044 return test_thread_flag(TIF_NEED_RESCHED_LAZY);
2047 static inline int need_resched_now(void)
2049 return test_thread_flag(TIF_NEED_RESCHED);
2053 static inline void clear_tsk_need_resched_lazy(struct task_struct *tsk) { }
2054 static inline int need_resched_lazy(void) { return 0; }
2056 static inline int need_resched_now(void)
2058 return test_thread_flag(TIF_NEED_RESCHED);
2063 #ifdef CONFIG_PREEMPT_RT
2064 static inline bool task_match_saved_state(struct task_struct *p, long match_state)
2066 return p->saved_state == match_state;
2069 static inline bool task_is_traced(struct task_struct *task)
2071 bool traced = false;
2073 /* in case the task is sleeping on tasklist_lock */
2074 raw_spin_lock_irq(&task->pi_lock);
2075 if (READ_ONCE(task->__state) & __TASK_TRACED)
2077 else if (task->saved_state & __TASK_TRACED)
2079 raw_spin_unlock_irq(&task->pi_lock);
2083 static inline bool task_is_stopped_or_traced(struct task_struct *task)
2085 bool traced_stopped = false;
2086 unsigned long flags;
2088 raw_spin_lock_irqsave(&task->pi_lock, flags);
2090 if (READ_ONCE(task->__state) & (__TASK_STOPPED | __TASK_TRACED))
2091 traced_stopped = true;
2092 else if (task->saved_state & (__TASK_STOPPED | __TASK_TRACED))
2093 traced_stopped = true;
2095 raw_spin_unlock_irqrestore(&task->pi_lock, flags);
2096 return traced_stopped;
2101 static inline bool task_match_saved_state(struct task_struct *p, long match_state)
2106 static inline bool task_is_traced(struct task_struct *task)
2108 return READ_ONCE(task->__state) & __TASK_TRACED;
2111 static inline bool task_is_stopped_or_traced(struct task_struct *task)
2113 return READ_ONCE(task->__state) & (__TASK_STOPPED | __TASK_TRACED);
2117 static inline bool task_match_state_or_saved(struct task_struct *p,
2120 if (READ_ONCE(p->__state) == match_state)
2123 return task_match_saved_state(p, match_state);
2126 static inline bool task_match_state_lock(struct task_struct *p,
2131 raw_spin_lock_irq(&p->pi_lock);
2132 match = task_match_state_or_saved(p, match_state);
2133 raw_spin_unlock_irq(&p->pi_lock);
2139 * cond_resched() and cond_resched_lock(): latency reduction via
2140 * explicit rescheduling in places that are safe. The return
2141 * value indicates whether a reschedule was done in fact.
2142 * cond_resched_lock() will drop the spinlock before scheduling,
2144 #if !defined(CONFIG_PREEMPTION) || defined(CONFIG_PREEMPT_DYNAMIC)
2145 extern int __cond_resched(void);
2147 #ifdef CONFIG_PREEMPT_DYNAMIC
2149 DECLARE_STATIC_CALL(cond_resched, __cond_resched);
2151 static __always_inline int _cond_resched(void)
2153 return static_call_mod(cond_resched)();
2158 static inline int _cond_resched(void)
2160 return __cond_resched();
2163 #endif /* CONFIG_PREEMPT_DYNAMIC */
2167 static inline int _cond_resched(void) { return 0; }
2169 #endif /* !defined(CONFIG_PREEMPTION) || defined(CONFIG_PREEMPT_DYNAMIC) */
2171 #define cond_resched() ({ \
2172 __might_resched(__FILE__, __LINE__, 0); \
2176 extern int __cond_resched_lock(spinlock_t *lock);
2177 extern int __cond_resched_rwlock_read(rwlock_t *lock);
2178 extern int __cond_resched_rwlock_write(rwlock_t *lock);
2180 #define MIGHT_RESCHED_RCU_SHIFT 8
2181 #define MIGHT_RESCHED_PREEMPT_MASK ((1U << MIGHT_RESCHED_RCU_SHIFT) - 1)
2183 #ifndef CONFIG_PREEMPT_RT
2185 * Non RT kernels have an elevated preempt count due to the held lock,
2186 * but are not allowed to be inside a RCU read side critical section
2188 # define PREEMPT_LOCK_RESCHED_OFFSETS PREEMPT_LOCK_OFFSET
2191 * spin/rw_lock() on RT implies rcu_read_lock(). The might_sleep() check in
2192 * cond_resched*lock() has to take that into account because it checks for
2193 * preempt_count() and rcu_preempt_depth().
2195 # define PREEMPT_LOCK_RESCHED_OFFSETS \
2196 (PREEMPT_LOCK_OFFSET + (1U << MIGHT_RESCHED_RCU_SHIFT))
2199 #define cond_resched_lock(lock) ({ \
2200 __might_resched(__FILE__, __LINE__, PREEMPT_LOCK_RESCHED_OFFSETS); \
2201 __cond_resched_lock(lock); \
2204 #define cond_resched_rwlock_read(lock) ({ \
2205 __might_resched(__FILE__, __LINE__, PREEMPT_LOCK_RESCHED_OFFSETS); \
2206 __cond_resched_rwlock_read(lock); \
2209 #define cond_resched_rwlock_write(lock) ({ \
2210 __might_resched(__FILE__, __LINE__, PREEMPT_LOCK_RESCHED_OFFSETS); \
2211 __cond_resched_rwlock_write(lock); \
2214 static inline void cond_resched_rcu(void)
2216 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2224 * Does a critical section need to be broken due to another
2225 * task waiting?: (technically does not depend on CONFIG_PREEMPTION,
2226 * but a general need for low latency)
2228 static inline int spin_needbreak(spinlock_t *lock)
2230 #ifdef CONFIG_PREEMPTION
2231 return spin_is_contended(lock);
2238 * Check if a rwlock is contended.
2239 * Returns non-zero if there is another task waiting on the rwlock.
2240 * Returns zero if the lock is not contended or the system / underlying
2241 * rwlock implementation does not support contention detection.
2242 * Technically does not depend on CONFIG_PREEMPTION, but a general need
2245 static inline int rwlock_needbreak(rwlock_t *lock)
2247 #ifdef CONFIG_PREEMPTION
2248 return rwlock_is_contended(lock);
2254 static __always_inline bool need_resched(void)
2256 return unlikely(tif_need_resched());
2260 * Wrappers for p->thread_info->cpu access. No-op on UP.
2264 static inline unsigned int task_cpu(const struct task_struct *p)
2266 #ifdef CONFIG_THREAD_INFO_IN_TASK
2267 return READ_ONCE(p->cpu);
2269 return READ_ONCE(task_thread_info(p)->cpu);
2273 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2277 static inline unsigned int task_cpu(const struct task_struct *p)
2282 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2286 #endif /* CONFIG_SMP */
2288 extern bool sched_task_on_rq(struct task_struct *p);
2291 * In order to reduce various lock holder preemption latencies provide an
2292 * interface to see if a vCPU is currently running or not.
2294 * This allows us to terminate optimistic spin loops and block, analogous to
2295 * the native optimistic spin heuristic of testing if the lock owner task is
2298 #ifndef vcpu_is_preempted
2299 static inline bool vcpu_is_preempted(int cpu)
2305 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2306 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2308 #ifndef TASK_SIZE_OF
2309 #define TASK_SIZE_OF(tsk) TASK_SIZE
2313 /* Returns effective CPU energy utilization, as seen by the scheduler */
2314 unsigned long sched_cpu_util(int cpu, unsigned long max);
2315 #endif /* CONFIG_SMP */
2320 * Map the event mask on the user-space ABI enum rseq_cs_flags
2321 * for direct mask checks.
2323 enum rseq_event_mask_bits {
2324 RSEQ_EVENT_PREEMPT_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT_BIT,
2325 RSEQ_EVENT_SIGNAL_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL_BIT,
2326 RSEQ_EVENT_MIGRATE_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE_BIT,
2329 enum rseq_event_mask {
2330 RSEQ_EVENT_PREEMPT = (1U << RSEQ_EVENT_PREEMPT_BIT),
2331 RSEQ_EVENT_SIGNAL = (1U << RSEQ_EVENT_SIGNAL_BIT),
2332 RSEQ_EVENT_MIGRATE = (1U << RSEQ_EVENT_MIGRATE_BIT),
2335 static inline void rseq_set_notify_resume(struct task_struct *t)
2338 set_tsk_thread_flag(t, TIF_NOTIFY_RESUME);
2341 void __rseq_handle_notify_resume(struct ksignal *sig, struct pt_regs *regs);
2343 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
2344 struct pt_regs *regs)
2347 __rseq_handle_notify_resume(ksig, regs);
2350 static inline void rseq_signal_deliver(struct ksignal *ksig,
2351 struct pt_regs *regs)
2354 __set_bit(RSEQ_EVENT_SIGNAL_BIT, ¤t->rseq_event_mask);
2356 rseq_handle_notify_resume(ksig, regs);
2359 /* rseq_preempt() requires preemption to be disabled. */
2360 static inline void rseq_preempt(struct task_struct *t)
2362 __set_bit(RSEQ_EVENT_PREEMPT_BIT, &t->rseq_event_mask);
2363 rseq_set_notify_resume(t);
2366 /* rseq_migrate() requires preemption to be disabled. */
2367 static inline void rseq_migrate(struct task_struct *t)
2369 __set_bit(RSEQ_EVENT_MIGRATE_BIT, &t->rseq_event_mask);
2370 rseq_set_notify_resume(t);
2374 * If parent process has a registered restartable sequences area, the
2375 * child inherits. Unregister rseq for a clone with CLONE_VM set.
2377 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
2379 if (clone_flags & CLONE_VM) {
2382 t->rseq_event_mask = 0;
2384 t->rseq = current->rseq;
2385 t->rseq_sig = current->rseq_sig;
2386 t->rseq_event_mask = current->rseq_event_mask;
2390 static inline void rseq_execve(struct task_struct *t)
2394 t->rseq_event_mask = 0;
2399 static inline void rseq_set_notify_resume(struct task_struct *t)
2402 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
2403 struct pt_regs *regs)
2406 static inline void rseq_signal_deliver(struct ksignal *ksig,
2407 struct pt_regs *regs)
2410 static inline void rseq_preempt(struct task_struct *t)
2413 static inline void rseq_migrate(struct task_struct *t)
2416 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
2419 static inline void rseq_execve(struct task_struct *t)
2425 #ifdef CONFIG_DEBUG_RSEQ
2427 void rseq_syscall(struct pt_regs *regs);
2431 static inline void rseq_syscall(struct pt_regs *regs)
2437 const struct sched_avg *sched_trace_cfs_rq_avg(struct cfs_rq *cfs_rq);
2438 char *sched_trace_cfs_rq_path(struct cfs_rq *cfs_rq, char *str, int len);
2439 int sched_trace_cfs_rq_cpu(struct cfs_rq *cfs_rq);
2441 const struct sched_avg *sched_trace_rq_avg_rt(struct rq *rq);
2442 const struct sched_avg *sched_trace_rq_avg_dl(struct rq *rq);
2443 const struct sched_avg *sched_trace_rq_avg_irq(struct rq *rq);
2445 int sched_trace_rq_cpu(struct rq *rq);
2446 int sched_trace_rq_cpu_capacity(struct rq *rq);
2447 int sched_trace_rq_nr_running(struct rq *rq);
2449 const struct cpumask *sched_trace_rd_span(struct root_domain *rd);
2451 #ifdef CONFIG_SCHED_CORE
2452 extern void sched_core_free(struct task_struct *tsk);
2453 extern void sched_core_fork(struct task_struct *p);
2454 extern int sched_core_share_pid(unsigned int cmd, pid_t pid, enum pid_type type,
2455 unsigned long uaddr);
2457 static inline void sched_core_free(struct task_struct *tsk) { }
2458 static inline void sched_core_fork(struct task_struct *p) { }