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/kcov.h>
18 #include <linux/mutex.h>
19 #include <linux/plist.h>
20 #include <linux/hrtimer.h>
21 #include <linux/irqflags.h>
22 #include <linux/seccomp.h>
23 #include <linux/nodemask.h>
24 #include <linux/rcupdate.h>
25 #include <linux/refcount.h>
26 #include <linux/resource.h>
27 #include <linux/latencytop.h>
28 #include <linux/sched/prio.h>
29 #include <linux/sched/types.h>
30 #include <linux/signal_types.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>
38 /* task_struct member predeclarations (sorted alphabetically): */
40 struct backing_dev_info;
43 struct capture_control;
46 struct futex_pi_state;
51 struct perf_event_context;
53 struct pipe_inode_info;
56 struct robust_list_head;
62 struct sighand_struct;
64 struct task_delay_info;
69 * Task state bitmask. NOTE! These bits are also
70 * encoded in fs/proc/array.c: get_task_state().
72 * We have two separate sets of flags: task->state
73 * is about runnability, while task->exit_state are
74 * about the task exiting. Confusing, but this way
75 * modifying one set can't modify the other one by
79 /* Used in tsk->state: */
80 #define TASK_RUNNING 0x0000
81 #define TASK_INTERRUPTIBLE 0x0001
82 #define TASK_UNINTERRUPTIBLE 0x0002
83 #define __TASK_STOPPED 0x0004
84 #define __TASK_TRACED 0x0008
85 /* Used in tsk->exit_state: */
86 #define EXIT_DEAD 0x0010
87 #define EXIT_ZOMBIE 0x0020
88 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
89 /* Used in tsk->state again: */
90 #define TASK_PARKED 0x0040
91 #define TASK_DEAD 0x0080
92 #define TASK_WAKEKILL 0x0100
93 #define TASK_WAKING 0x0200
94 #define TASK_NOLOAD 0x0400
95 #define TASK_NEW 0x0800
96 #define TASK_STATE_MAX 0x1000
98 /* Convenience macros for the sake of set_current_state: */
99 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
100 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
101 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
103 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
105 /* Convenience macros for the sake of wake_up(): */
106 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
108 /* get_task_state(): */
109 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
110 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
111 __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \
114 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
116 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
118 #define task_is_stopped_or_traced(task) ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
120 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
123 * Special states are those that do not use the normal wait-loop pattern. See
124 * the comment with set_special_state().
126 #define is_special_task_state(state) \
127 ((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_PARKED | TASK_DEAD))
129 #define __set_current_state(state_value) \
131 WARN_ON_ONCE(is_special_task_state(state_value));\
132 current->task_state_change = _THIS_IP_; \
133 current->state = (state_value); \
136 #define set_current_state(state_value) \
138 WARN_ON_ONCE(is_special_task_state(state_value));\
139 current->task_state_change = _THIS_IP_; \
140 smp_store_mb(current->state, (state_value)); \
143 #define set_special_state(state_value) \
145 unsigned long flags; /* may shadow */ \
146 WARN_ON_ONCE(!is_special_task_state(state_value)); \
147 raw_spin_lock_irqsave(¤t->pi_lock, flags); \
148 current->task_state_change = _THIS_IP_; \
149 current->state = (state_value); \
150 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
154 * set_current_state() includes a barrier so that the write of current->state
155 * is correctly serialised wrt the caller's subsequent test of whether to
159 * set_current_state(TASK_UNINTERRUPTIBLE);
165 * __set_current_state(TASK_RUNNING);
167 * If the caller does not need such serialisation (because, for instance, the
168 * CONDITION test and condition change and wakeup are under the same lock) then
169 * use __set_current_state().
171 * The above is typically ordered against the wakeup, which does:
174 * wake_up_state(p, TASK_UNINTERRUPTIBLE);
176 * where wake_up_state()/try_to_wake_up() executes a full memory barrier before
177 * accessing p->state.
179 * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
180 * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
181 * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
183 * However, with slightly different timing the wakeup TASK_RUNNING store can
184 * also collide with the TASK_UNINTERRUPTIBLE store. Losing that store is not
185 * a problem either because that will result in one extra go around the loop
186 * and our @cond test will save the day.
188 * Also see the comments of try_to_wake_up().
190 #define __set_current_state(state_value) \
191 current->state = (state_value)
193 #define set_current_state(state_value) \
194 smp_store_mb(current->state, (state_value))
197 * set_special_state() should be used for those states when the blocking task
198 * can not use the regular condition based wait-loop. In that case we must
199 * serialize against wakeups such that any possible in-flight TASK_RUNNING stores
200 * will not collide with our state change.
202 #define set_special_state(state_value) \
204 unsigned long flags; /* may shadow */ \
205 raw_spin_lock_irqsave(¤t->pi_lock, flags); \
206 current->state = (state_value); \
207 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
212 /* Task command name length: */
213 #define TASK_COMM_LEN 16
215 extern void scheduler_tick(void);
217 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
219 extern long schedule_timeout(long timeout);
220 extern long schedule_timeout_interruptible(long timeout);
221 extern long schedule_timeout_killable(long timeout);
222 extern long schedule_timeout_uninterruptible(long timeout);
223 extern long schedule_timeout_idle(long timeout);
224 asmlinkage void schedule(void);
225 extern void schedule_preempt_disabled(void);
226 asmlinkage void preempt_schedule_irq(void);
228 extern int __must_check io_schedule_prepare(void);
229 extern void io_schedule_finish(int token);
230 extern long io_schedule_timeout(long timeout);
231 extern void io_schedule(void);
234 * struct prev_cputime - snapshot of system and user cputime
235 * @utime: time spent in user mode
236 * @stime: time spent in system mode
237 * @lock: protects the above two fields
239 * Stores previous user/system time values such that we can guarantee
242 struct prev_cputime {
243 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
251 /* Task is sleeping or running in a CPU with VTIME inactive: */
255 /* Task runs in kernelspace in a CPU with VTIME active: */
257 /* Task runs in userspace in a CPU with VTIME active: */
259 /* Task runs as guests in a CPU with VTIME active: */
265 unsigned long long starttime;
266 enum vtime_state state;
274 * Utilization clamp constraints.
275 * @UCLAMP_MIN: Minimum utilization
276 * @UCLAMP_MAX: Maximum utilization
277 * @UCLAMP_CNT: Utilization clamp constraints count
286 extern struct root_domain def_root_domain;
287 extern struct mutex sched_domains_mutex;
291 #ifdef CONFIG_SCHED_INFO
292 /* Cumulative counters: */
294 /* # of times we have run on this CPU: */
295 unsigned long pcount;
297 /* Time spent waiting on a runqueue: */
298 unsigned long long run_delay;
302 /* When did we last run on a CPU? */
303 unsigned long long last_arrival;
305 /* When were we last queued to run? */
306 unsigned long long last_queued;
308 #endif /* CONFIG_SCHED_INFO */
312 * Integer metrics need fixed point arithmetic, e.g., sched/fair
313 * has a few: load, load_avg, util_avg, freq, and capacity.
315 * We define a basic fixed point arithmetic range, and then formalize
316 * all these metrics based on that basic range.
318 # define SCHED_FIXEDPOINT_SHIFT 10
319 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
321 /* Increase resolution of cpu_capacity calculations */
322 # define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT
323 # define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
326 unsigned long weight;
331 * struct util_est - Estimation utilization of FAIR tasks
332 * @enqueued: instantaneous estimated utilization of a task/cpu
333 * @ewma: the Exponential Weighted Moving Average (EWMA)
334 * utilization of a task
336 * Support data structure to track an Exponential Weighted Moving Average
337 * (EWMA) of a FAIR task's utilization. New samples are added to the moving
338 * average each time a task completes an activation. Sample's weight is chosen
339 * so that the EWMA will be relatively insensitive to transient changes to the
342 * The enqueued attribute has a slightly different meaning for tasks and cpus:
343 * - task: the task's util_avg at last task dequeue time
344 * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU
345 * Thus, the util_est.enqueued of a task represents the contribution on the
346 * estimated utilization of the CPU where that task is currently enqueued.
348 * Only for tasks we track a moving average of the past instantaneous
349 * estimated utilization. This allows to absorb sporadic drops in utilization
350 * of an otherwise almost periodic task.
353 unsigned int enqueued;
355 #define UTIL_EST_WEIGHT_SHIFT 2
356 } __attribute__((__aligned__(sizeof(u64))));
359 * The load/runnable/util_avg accumulates an infinite geometric series
360 * (see __update_load_avg_cfs_rq() in kernel/sched/pelt.c).
362 * [load_avg definition]
364 * load_avg = runnable% * scale_load_down(load)
366 * [runnable_avg definition]
368 * runnable_avg = runnable% * SCHED_CAPACITY_SCALE
370 * [util_avg definition]
372 * util_avg = running% * SCHED_CAPACITY_SCALE
374 * where runnable% is the time ratio that a sched_entity is runnable and
375 * running% the time ratio that a sched_entity is running.
377 * For cfs_rq, they are the aggregated values of all runnable and blocked
380 * The load/runnable/util_avg doesn't directly factor frequency scaling and CPU
381 * capacity scaling. The scaling is done through the rq_clock_pelt that is used
382 * for computing those signals (see update_rq_clock_pelt())
384 * N.B., the above ratios (runnable% and running%) themselves are in the
385 * range of [0, 1]. To do fixed point arithmetics, we therefore scale them
386 * to as large a range as necessary. This is for example reflected by
387 * util_avg's SCHED_CAPACITY_SCALE.
391 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
392 * with the highest load (=88761), always runnable on a single cfs_rq,
393 * and should not overflow as the number already hits PID_MAX_LIMIT.
395 * For all other cases (including 32-bit kernels), struct load_weight's
396 * weight will overflow first before we do, because:
398 * Max(load_avg) <= Max(load.weight)
400 * Then it is the load_weight's responsibility to consider overflow
404 u64 last_update_time;
409 unsigned long load_avg;
410 unsigned long runnable_avg;
411 unsigned long util_avg;
412 struct util_est util_est;
413 } ____cacheline_aligned;
415 struct sched_statistics {
416 #ifdef CONFIG_SCHEDSTATS
426 s64 sum_sleep_runtime;
433 u64 nr_migrations_cold;
434 u64 nr_failed_migrations_affine;
435 u64 nr_failed_migrations_running;
436 u64 nr_failed_migrations_hot;
437 u64 nr_forced_migrations;
441 u64 nr_wakeups_migrate;
442 u64 nr_wakeups_local;
443 u64 nr_wakeups_remote;
444 u64 nr_wakeups_affine;
445 u64 nr_wakeups_affine_attempts;
446 u64 nr_wakeups_passive;
451 struct sched_entity {
452 /* For load-balancing: */
453 struct load_weight load;
454 struct rb_node run_node;
455 struct list_head group_node;
459 u64 sum_exec_runtime;
461 u64 prev_sum_exec_runtime;
465 struct sched_statistics statistics;
467 #ifdef CONFIG_FAIR_GROUP_SCHED
469 struct sched_entity *parent;
470 /* rq on which this entity is (to be) queued: */
471 struct cfs_rq *cfs_rq;
472 /* rq "owned" by this entity/group: */
474 /* cached value of my_q->h_nr_running */
475 unsigned long runnable_weight;
480 * Per entity load average tracking.
482 * Put into separate cache line so it does not
483 * collide with read-mostly values above.
485 struct sched_avg avg;
489 struct sched_rt_entity {
490 struct list_head run_list;
491 unsigned long timeout;
492 unsigned long watchdog_stamp;
493 unsigned int time_slice;
494 unsigned short on_rq;
495 unsigned short on_list;
497 struct sched_rt_entity *back;
498 #ifdef CONFIG_RT_GROUP_SCHED
499 struct sched_rt_entity *parent;
500 /* rq on which this entity is (to be) queued: */
502 /* rq "owned" by this entity/group: */
505 } __randomize_layout;
507 struct sched_dl_entity {
508 struct rb_node rb_node;
511 * Original scheduling parameters. Copied here from sched_attr
512 * during sched_setattr(), they will remain the same until
513 * the next sched_setattr().
515 u64 dl_runtime; /* Maximum runtime for each instance */
516 u64 dl_deadline; /* Relative deadline of each instance */
517 u64 dl_period; /* Separation of two instances (period) */
518 u64 dl_bw; /* dl_runtime / dl_period */
519 u64 dl_density; /* dl_runtime / dl_deadline */
522 * Actual scheduling parameters. Initialized with the values above,
523 * they are continuously updated during task execution. Note that
524 * the remaining runtime could be < 0 in case we are in overrun.
526 s64 runtime; /* Remaining runtime for this instance */
527 u64 deadline; /* Absolute deadline for this instance */
528 unsigned int flags; /* Specifying the scheduler behaviour */
533 * @dl_throttled tells if we exhausted the runtime. If so, the
534 * task has to wait for a replenishment to be performed at the
535 * next firing of dl_timer.
537 * @dl_boosted tells if we are boosted due to DI. If so we are
538 * outside bandwidth enforcement mechanism (but only until we
539 * exit the critical section);
541 * @dl_yielded tells if task gave up the CPU before consuming
542 * all its available runtime during the last job.
544 * @dl_non_contending tells if the task is inactive while still
545 * contributing to the active utilization. In other words, it
546 * indicates if the inactive timer has been armed and its handler
547 * has not been executed yet. This flag is useful to avoid race
548 * conditions between the inactive timer handler and the wakeup
551 * @dl_overrun tells if the task asked to be informed about runtime
554 unsigned int dl_throttled : 1;
555 unsigned int dl_boosted : 1;
556 unsigned int dl_yielded : 1;
557 unsigned int dl_non_contending : 1;
558 unsigned int dl_overrun : 1;
561 * Bandwidth enforcement timer. Each -deadline task has its
562 * own bandwidth to be enforced, thus we need one timer per task.
564 struct hrtimer dl_timer;
567 * Inactive timer, responsible for decreasing the active utilization
568 * at the "0-lag time". When a -deadline task blocks, it contributes
569 * to GRUB's active utilization until the "0-lag time", hence a
570 * timer is needed to decrease the active utilization at the correct
573 struct hrtimer inactive_timer;
576 #ifdef CONFIG_UCLAMP_TASK
577 /* Number of utilization clamp buckets (shorter alias) */
578 #define UCLAMP_BUCKETS CONFIG_UCLAMP_BUCKETS_COUNT
581 * Utilization clamp for a scheduling entity
582 * @value: clamp value "assigned" to a se
583 * @bucket_id: bucket index corresponding to the "assigned" value
584 * @active: the se is currently refcounted in a rq's bucket
585 * @user_defined: the requested clamp value comes from user-space
587 * The bucket_id is the index of the clamp bucket matching the clamp value
588 * which is pre-computed and stored to avoid expensive integer divisions from
591 * The active bit is set whenever a task has got an "effective" value assigned,
592 * which can be different from the clamp value "requested" from user-space.
593 * This allows to know a task is refcounted in the rq's bucket corresponding
594 * to the "effective" bucket_id.
596 * The user_defined bit is set whenever a task has got a task-specific clamp
597 * value requested from userspace, i.e. the system defaults apply to this task
598 * just as a restriction. This allows to relax default clamps when a less
599 * restrictive task-specific value has been requested, thus allowing to
600 * implement a "nice" semantic. For example, a task running with a 20%
601 * default boost can still drop its own boosting to 0%.
604 unsigned int value : bits_per(SCHED_CAPACITY_SCALE);
605 unsigned int bucket_id : bits_per(UCLAMP_BUCKETS);
606 unsigned int active : 1;
607 unsigned int user_defined : 1;
609 #endif /* CONFIG_UCLAMP_TASK */
615 u8 exp_hint; /* Hint for performance. */
616 u8 need_mb; /* Readers need smp_mb(). */
618 u32 s; /* Set of bits. */
621 enum perf_event_task_context {
622 perf_invalid_context = -1,
625 perf_nr_task_contexts,
629 struct wake_q_node *next;
633 #ifdef CONFIG_THREAD_INFO_IN_TASK
635 * For reasons of header soup (see current_thread_info()), this
636 * must be the first element of task_struct.
638 struct thread_info thread_info;
640 /* -1 unrunnable, 0 runnable, >0 stopped: */
644 * This begins the randomizable portion of task_struct. Only
645 * scheduling-critical items should be added above here.
647 randomized_struct_fields_start
651 /* Per task flags (PF_*), defined further below: */
657 struct __call_single_node wake_entry;
658 #ifdef CONFIG_THREAD_INFO_IN_TASK
662 unsigned int wakee_flips;
663 unsigned long wakee_flip_decay_ts;
664 struct task_struct *last_wakee;
667 * recent_used_cpu is initially set as the last CPU used by a task
668 * that wakes affine another task. Waker/wakee relationships can
669 * push tasks around a CPU where each wakeup moves to the next one.
670 * Tracking a recently used CPU allows a quick search for a recently
671 * used CPU that may be idle.
681 unsigned int rt_priority;
683 const struct sched_class *sched_class;
684 struct sched_entity se;
685 struct sched_rt_entity rt;
686 #ifdef CONFIG_CGROUP_SCHED
687 struct task_group *sched_task_group;
689 struct sched_dl_entity dl;
691 #ifdef CONFIG_UCLAMP_TASK
693 * Clamp values requested for a scheduling entity.
694 * Must be updated with task_rq_lock() held.
696 struct uclamp_se uclamp_req[UCLAMP_CNT];
698 * Effective clamp values used for a scheduling entity.
699 * Must be updated with task_rq_lock() held.
701 struct uclamp_se uclamp[UCLAMP_CNT];
704 #ifdef CONFIG_PREEMPT_NOTIFIERS
705 /* List of struct preempt_notifier: */
706 struct hlist_head preempt_notifiers;
709 #ifdef CONFIG_BLK_DEV_IO_TRACE
710 unsigned int btrace_seq;
715 const cpumask_t *cpus_ptr;
718 #ifdef CONFIG_PREEMPT_RCU
719 int rcu_read_lock_nesting;
720 union rcu_special rcu_read_unlock_special;
721 struct list_head rcu_node_entry;
722 struct rcu_node *rcu_blocked_node;
723 #endif /* #ifdef CONFIG_PREEMPT_RCU */
725 #ifdef CONFIG_TASKS_RCU
726 unsigned long rcu_tasks_nvcsw;
727 u8 rcu_tasks_holdout;
729 int rcu_tasks_idle_cpu;
730 struct list_head rcu_tasks_holdout_list;
731 #endif /* #ifdef CONFIG_TASKS_RCU */
733 #ifdef CONFIG_TASKS_TRACE_RCU
734 int trc_reader_nesting;
736 union rcu_special trc_reader_special;
737 bool trc_reader_checked;
738 struct list_head trc_holdout_list;
739 #endif /* #ifdef CONFIG_TASKS_TRACE_RCU */
741 struct sched_info sched_info;
743 struct list_head tasks;
745 struct plist_node pushable_tasks;
746 struct rb_node pushable_dl_tasks;
749 struct mm_struct *mm;
750 struct mm_struct *active_mm;
752 /* Per-thread vma caching: */
753 struct vmacache vmacache;
755 #ifdef SPLIT_RSS_COUNTING
756 struct task_rss_stat rss_stat;
761 /* The signal sent when the parent dies: */
763 /* JOBCTL_*, siglock protected: */
764 unsigned long jobctl;
766 /* Used for emulating ABI behavior of previous Linux versions: */
767 unsigned int personality;
769 /* Scheduler bits, serialized by scheduler locks: */
770 unsigned sched_reset_on_fork:1;
771 unsigned sched_contributes_to_load:1;
772 unsigned sched_migrated:1;
773 unsigned sched_remote_wakeup:1;
775 unsigned sched_psi_wake_requeue:1;
778 /* Force alignment to the next boundary: */
781 /* Unserialized, strictly 'current' */
783 /* Bit to tell LSMs we're in execve(): */
784 unsigned in_execve:1;
785 unsigned in_iowait:1;
786 #ifndef TIF_RESTORE_SIGMASK
787 unsigned restore_sigmask:1;
790 unsigned in_user_fault:1;
792 #ifdef CONFIG_COMPAT_BRK
793 unsigned brk_randomized:1;
795 #ifdef CONFIG_CGROUPS
796 /* disallow userland-initiated cgroup migration */
797 unsigned no_cgroup_migration:1;
798 /* task is frozen/stopped (used by the cgroup freezer) */
801 #ifdef CONFIG_BLK_CGROUP
802 unsigned use_memdelay:1;
805 /* Stalled due to lack of memory */
806 unsigned in_memstall:1;
809 unsigned long atomic_flags; /* Flags requiring atomic access. */
811 struct restart_block restart_block;
816 #ifdef CONFIG_STACKPROTECTOR
817 /* Canary value for the -fstack-protector GCC feature: */
818 unsigned long stack_canary;
821 * Pointers to the (original) parent process, youngest child, younger sibling,
822 * older sibling, respectively. (p->father can be replaced with
823 * p->real_parent->pid)
826 /* Real parent process: */
827 struct task_struct __rcu *real_parent;
829 /* Recipient of SIGCHLD, wait4() reports: */
830 struct task_struct __rcu *parent;
833 * Children/sibling form the list of natural children:
835 struct list_head children;
836 struct list_head sibling;
837 struct task_struct *group_leader;
840 * 'ptraced' is the list of tasks this task is using ptrace() on.
842 * This includes both natural children and PTRACE_ATTACH targets.
843 * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
845 struct list_head ptraced;
846 struct list_head ptrace_entry;
848 /* PID/PID hash table linkage. */
849 struct pid *thread_pid;
850 struct hlist_node pid_links[PIDTYPE_MAX];
851 struct list_head thread_group;
852 struct list_head thread_node;
854 struct completion *vfork_done;
856 /* CLONE_CHILD_SETTID: */
857 int __user *set_child_tid;
859 /* CLONE_CHILD_CLEARTID: */
860 int __user *clear_child_tid;
864 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
869 struct prev_cputime prev_cputime;
870 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
874 #ifdef CONFIG_NO_HZ_FULL
875 atomic_t tick_dep_mask;
877 /* Context switch counts: */
879 unsigned long nivcsw;
881 /* Monotonic time in nsecs: */
884 /* Boot based time in nsecs: */
887 /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
888 unsigned long min_flt;
889 unsigned long maj_flt;
891 /* Empty if CONFIG_POSIX_CPUTIMERS=n */
892 struct posix_cputimers posix_cputimers;
894 #ifdef CONFIG_POSIX_CPU_TIMERS_TASK_WORK
895 struct posix_cputimers_work posix_cputimers_work;
898 /* Process credentials: */
900 /* Tracer's credentials at attach: */
901 const struct cred __rcu *ptracer_cred;
903 /* Objective and real subjective task credentials (COW): */
904 const struct cred __rcu *real_cred;
906 /* Effective (overridable) subjective task credentials (COW): */
907 const struct cred __rcu *cred;
910 /* Cached requested key. */
911 struct key *cached_requested_key;
915 * executable name, excluding path.
917 * - normally initialized setup_new_exec()
918 * - access it with [gs]et_task_comm()
919 * - lock it with task_lock()
921 char comm[TASK_COMM_LEN];
923 struct nameidata *nameidata;
925 #ifdef CONFIG_SYSVIPC
926 struct sysv_sem sysvsem;
927 struct sysv_shm sysvshm;
929 #ifdef CONFIG_DETECT_HUNG_TASK
930 unsigned long last_switch_count;
931 unsigned long last_switch_time;
933 /* Filesystem information: */
934 struct fs_struct *fs;
936 /* Open file information: */
937 struct files_struct *files;
939 #ifdef CONFIG_IO_URING
940 struct io_uring_task *io_uring;
944 struct nsproxy *nsproxy;
946 /* Signal handlers: */
947 struct signal_struct *signal;
948 struct sighand_struct __rcu *sighand;
950 sigset_t real_blocked;
951 /* Restored if set_restore_sigmask() was used: */
952 sigset_t saved_sigmask;
953 struct sigpending pending;
954 unsigned long sas_ss_sp;
956 unsigned int sas_ss_flags;
958 struct callback_head *task_works;
961 #ifdef CONFIG_AUDITSYSCALL
962 struct audit_context *audit_context;
965 unsigned int sessionid;
967 struct seccomp seccomp;
969 /* Thread group tracking: */
973 /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
974 spinlock_t alloc_lock;
976 /* Protection of the PI data structures: */
977 raw_spinlock_t pi_lock;
979 struct wake_q_node wake_q;
981 #ifdef CONFIG_RT_MUTEXES
982 /* PI waiters blocked on a rt_mutex held by this task: */
983 struct rb_root_cached pi_waiters;
984 /* Updated under owner's pi_lock and rq lock */
985 struct task_struct *pi_top_task;
986 /* Deadlock detection and priority inheritance handling: */
987 struct rt_mutex_waiter *pi_blocked_on;
990 #ifdef CONFIG_DEBUG_MUTEXES
991 /* Mutex deadlock detection: */
992 struct mutex_waiter *blocked_on;
995 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
999 #ifdef CONFIG_TRACE_IRQFLAGS
1000 struct irqtrace_events irqtrace;
1001 unsigned int hardirq_threaded;
1002 u64 hardirq_chain_key;
1003 int softirqs_enabled;
1004 int softirq_context;
1008 #ifdef CONFIG_LOCKDEP
1009 # define MAX_LOCK_DEPTH 48UL
1012 unsigned int lockdep_recursion;
1013 struct held_lock held_locks[MAX_LOCK_DEPTH];
1016 #if defined(CONFIG_UBSAN) && !defined(CONFIG_UBSAN_TRAP)
1017 unsigned int in_ubsan;
1020 /* Journalling filesystem info: */
1023 /* Stacked block device info: */
1024 struct bio_list *bio_list;
1027 /* Stack plugging: */
1028 struct blk_plug *plug;
1032 struct reclaim_state *reclaim_state;
1034 struct backing_dev_info *backing_dev_info;
1036 struct io_context *io_context;
1038 #ifdef CONFIG_COMPACTION
1039 struct capture_control *capture_control;
1042 unsigned long ptrace_message;
1043 kernel_siginfo_t *last_siginfo;
1045 struct task_io_accounting ioac;
1047 /* Pressure stall state */
1048 unsigned int psi_flags;
1050 #ifdef CONFIG_TASK_XACCT
1051 /* Accumulated RSS usage: */
1053 /* Accumulated virtual memory usage: */
1055 /* stime + utime since last update: */
1058 #ifdef CONFIG_CPUSETS
1059 /* Protected by ->alloc_lock: */
1060 nodemask_t mems_allowed;
1061 /* Seqence number to catch updates: */
1062 seqcount_spinlock_t mems_allowed_seq;
1063 int cpuset_mem_spread_rotor;
1064 int cpuset_slab_spread_rotor;
1066 #ifdef CONFIG_CGROUPS
1067 /* Control Group info protected by css_set_lock: */
1068 struct css_set __rcu *cgroups;
1069 /* cg_list protected by css_set_lock and tsk->alloc_lock: */
1070 struct list_head cg_list;
1072 #ifdef CONFIG_X86_CPU_RESCTRL
1077 struct robust_list_head __user *robust_list;
1078 #ifdef CONFIG_COMPAT
1079 struct compat_robust_list_head __user *compat_robust_list;
1081 struct list_head pi_state_list;
1082 struct futex_pi_state *pi_state_cache;
1083 struct mutex futex_exit_mutex;
1084 unsigned int futex_state;
1086 #ifdef CONFIG_PERF_EVENTS
1087 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1088 struct mutex perf_event_mutex;
1089 struct list_head perf_event_list;
1091 #ifdef CONFIG_DEBUG_PREEMPT
1092 unsigned long preempt_disable_ip;
1095 /* Protected by alloc_lock: */
1096 struct mempolicy *mempolicy;
1098 short pref_node_fork;
1100 #ifdef CONFIG_NUMA_BALANCING
1102 unsigned int numa_scan_period;
1103 unsigned int numa_scan_period_max;
1104 int numa_preferred_nid;
1105 unsigned long numa_migrate_retry;
1106 /* Migration stamp: */
1108 u64 last_task_numa_placement;
1109 u64 last_sum_exec_runtime;
1110 struct callback_head numa_work;
1113 * This pointer is only modified for current in syscall and
1114 * pagefault context (and for tasks being destroyed), so it can be read
1115 * from any of the following contexts:
1116 * - RCU read-side critical section
1117 * - current->numa_group from everywhere
1118 * - task's runqueue locked, task not running
1120 struct numa_group __rcu *numa_group;
1123 * numa_faults is an array split into four regions:
1124 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1125 * in this precise order.
1127 * faults_memory: Exponential decaying average of faults on a per-node
1128 * basis. Scheduling placement decisions are made based on these
1129 * counts. The values remain static for the duration of a PTE scan.
1130 * faults_cpu: Track the nodes the process was running on when a NUMA
1131 * hinting fault was incurred.
1132 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1133 * during the current scan window. When the scan completes, the counts
1134 * in faults_memory and faults_cpu decay and these values are copied.
1136 unsigned long *numa_faults;
1137 unsigned long total_numa_faults;
1140 * numa_faults_locality tracks if faults recorded during the last
1141 * scan window were remote/local or failed to migrate. The task scan
1142 * period is adapted based on the locality of the faults with different
1143 * weights depending on whether they were shared or private faults
1145 unsigned long numa_faults_locality[3];
1147 unsigned long numa_pages_migrated;
1148 #endif /* CONFIG_NUMA_BALANCING */
1151 struct rseq __user *rseq;
1154 * RmW on rseq_event_mask must be performed atomically
1155 * with respect to preemption.
1157 unsigned long rseq_event_mask;
1160 struct tlbflush_unmap_batch tlb_ubc;
1163 refcount_t rcu_users;
1164 struct rcu_head rcu;
1167 /* Cache last used pipe for splice(): */
1168 struct pipe_inode_info *splice_pipe;
1170 struct page_frag task_frag;
1172 #ifdef CONFIG_TASK_DELAY_ACCT
1173 struct task_delay_info *delays;
1176 #ifdef CONFIG_FAULT_INJECTION
1178 unsigned int fail_nth;
1181 * When (nr_dirtied >= nr_dirtied_pause), it's time to call
1182 * balance_dirty_pages() for a dirty throttling pause:
1185 int nr_dirtied_pause;
1186 /* Start of a write-and-pause period: */
1187 unsigned long dirty_paused_when;
1189 #ifdef CONFIG_LATENCYTOP
1190 int latency_record_count;
1191 struct latency_record latency_record[LT_SAVECOUNT];
1194 * Time slack values; these are used to round up poll() and
1195 * select() etc timeout values. These are in nanoseconds.
1198 u64 default_timer_slack_ns;
1201 unsigned int kasan_depth;
1205 struct kcsan_ctx kcsan_ctx;
1206 #ifdef CONFIG_TRACE_IRQFLAGS
1207 struct irqtrace_events kcsan_save_irqtrace;
1211 #if IS_ENABLED(CONFIG_KUNIT)
1212 struct kunit *kunit_test;
1215 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1216 /* Index of current stored address in ret_stack: */
1220 /* Stack of return addresses for return function tracing: */
1221 struct ftrace_ret_stack *ret_stack;
1223 /* Timestamp for last schedule: */
1224 unsigned long long ftrace_timestamp;
1227 * Number of functions that haven't been traced
1228 * because of depth overrun:
1230 atomic_t trace_overrun;
1232 /* Pause tracing: */
1233 atomic_t tracing_graph_pause;
1236 #ifdef CONFIG_TRACING
1237 /* State flags for use by tracers: */
1238 unsigned long trace;
1240 /* Bitmask and counter of trace recursion: */
1241 unsigned long trace_recursion;
1242 #endif /* CONFIG_TRACING */
1245 /* See kernel/kcov.c for more details. */
1247 /* Coverage collection mode enabled for this task (0 if disabled): */
1248 unsigned int kcov_mode;
1250 /* Size of the kcov_area: */
1251 unsigned int kcov_size;
1253 /* Buffer for coverage collection: */
1256 /* KCOV descriptor wired with this task or NULL: */
1259 /* KCOV common handle for remote coverage collection: */
1262 /* KCOV sequence number: */
1265 /* Collect coverage from softirq context: */
1266 unsigned int kcov_softirq;
1270 struct mem_cgroup *memcg_in_oom;
1271 gfp_t memcg_oom_gfp_mask;
1272 int memcg_oom_order;
1274 /* Number of pages to reclaim on returning to userland: */
1275 unsigned int memcg_nr_pages_over_high;
1277 /* Used by memcontrol for targeted memcg charge: */
1278 struct mem_cgroup *active_memcg;
1281 #ifdef CONFIG_BLK_CGROUP
1282 struct request_queue *throttle_queue;
1285 #ifdef CONFIG_UPROBES
1286 struct uprobe_task *utask;
1288 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1289 unsigned int sequential_io;
1290 unsigned int sequential_io_avg;
1292 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1293 unsigned long task_state_change;
1295 int pagefault_disabled;
1297 struct task_struct *oom_reaper_list;
1299 #ifdef CONFIG_VMAP_STACK
1300 struct vm_struct *stack_vm_area;
1302 #ifdef CONFIG_THREAD_INFO_IN_TASK
1303 /* A live task holds one reference: */
1304 refcount_t stack_refcount;
1306 #ifdef CONFIG_LIVEPATCH
1309 #ifdef CONFIG_SECURITY
1310 /* Used by LSM modules for access restriction: */
1314 #ifdef CONFIG_GCC_PLUGIN_STACKLEAK
1315 unsigned long lowest_stack;
1316 unsigned long prev_lowest_stack;
1319 #ifdef CONFIG_X86_MCE
1320 void __user *mce_vaddr;
1325 __mce_reserved : 62;
1326 struct callback_head mce_kill_me;
1330 * New fields for task_struct should be added above here, so that
1331 * they are included in the randomized portion of task_struct.
1333 randomized_struct_fields_end
1335 /* CPU-specific state of this task: */
1336 struct thread_struct thread;
1339 * WARNING: on x86, 'thread_struct' contains a variable-sized
1340 * structure. It *MUST* be at the end of 'task_struct'.
1342 * Do not put anything below here!
1346 static inline struct pid *task_pid(struct task_struct *task)
1348 return task->thread_pid;
1352 * the helpers to get the task's different pids as they are seen
1353 * from various namespaces
1355 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1356 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1358 * task_xid_nr_ns() : id seen from the ns specified;
1360 * see also pid_nr() etc in include/linux/pid.h
1362 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns);
1364 static inline pid_t task_pid_nr(struct task_struct *tsk)
1369 static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1371 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1374 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1376 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1380 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1386 * pid_alive - check that a task structure is not stale
1387 * @p: Task structure to be checked.
1389 * Test if a process is not yet dead (at most zombie state)
1390 * If pid_alive fails, then pointers within the task structure
1391 * can be stale and must not be dereferenced.
1393 * Return: 1 if the process is alive. 0 otherwise.
1395 static inline int pid_alive(const struct task_struct *p)
1397 return p->thread_pid != NULL;
1400 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1402 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1405 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1407 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1411 static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1413 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1416 static inline pid_t task_session_vnr(struct task_struct *tsk)
1418 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1421 static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1423 return __task_pid_nr_ns(tsk, PIDTYPE_TGID, ns);
1426 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1428 return __task_pid_nr_ns(tsk, PIDTYPE_TGID, NULL);
1431 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1437 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1443 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1445 return task_ppid_nr_ns(tsk, &init_pid_ns);
1448 /* Obsolete, do not use: */
1449 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1451 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1454 #define TASK_REPORT_IDLE (TASK_REPORT + 1)
1455 #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1)
1457 static inline unsigned int task_state_index(struct task_struct *tsk)
1459 unsigned int tsk_state = READ_ONCE(tsk->state);
1460 unsigned int state = (tsk_state | tsk->exit_state) & TASK_REPORT;
1462 BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX);
1464 if (tsk_state == TASK_IDLE)
1465 state = TASK_REPORT_IDLE;
1470 static inline char task_index_to_char(unsigned int state)
1472 static const char state_char[] = "RSDTtXZPI";
1474 BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX) != sizeof(state_char) - 1);
1476 return state_char[state];
1479 static inline char task_state_to_char(struct task_struct *tsk)
1481 return task_index_to_char(task_state_index(tsk));
1485 * is_global_init - check if a task structure is init. Since init
1486 * is free to have sub-threads we need to check tgid.
1487 * @tsk: Task structure to be checked.
1489 * Check if a task structure is the first user space task the kernel created.
1491 * Return: 1 if the task structure is init. 0 otherwise.
1493 static inline int is_global_init(struct task_struct *tsk)
1495 return task_tgid_nr(tsk) == 1;
1498 extern struct pid *cad_pid;
1503 #define PF_VCPU 0x00000001 /* I'm a virtual CPU */
1504 #define PF_IDLE 0x00000002 /* I am an IDLE thread */
1505 #define PF_EXITING 0x00000004 /* Getting shut down */
1506 #define PF_IO_WORKER 0x00000010 /* Task is an IO worker */
1507 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1508 #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */
1509 #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */
1510 #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */
1511 #define PF_DUMPCORE 0x00000200 /* Dumped core */
1512 #define PF_SIGNALED 0x00000400 /* Killed by a signal */
1513 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1514 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */
1515 #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */
1516 #define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */
1517 #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */
1518 #define PF_FROZEN 0x00010000 /* Frozen for system suspend */
1519 #define PF_KSWAPD 0x00020000 /* I am kswapd */
1520 #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */
1521 #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */
1522 #define PF_LOCAL_THROTTLE 0x00100000 /* Throttle writes only against the bdi I write to,
1523 * I am cleaning dirty pages from some other bdi. */
1524 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1525 #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */
1526 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1527 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_mask */
1528 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1529 #define PF_MEMALLOC_NOCMA 0x10000000 /* All allocation request will have _GFP_MOVABLE cleared */
1530 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1531 #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */
1534 * Only the _current_ task can read/write to tsk->flags, but other
1535 * tasks can access tsk->flags in readonly mode for example
1536 * with tsk_used_math (like during threaded core dumping).
1537 * There is however an exception to this rule during ptrace
1538 * or during fork: the ptracer task is allowed to write to the
1539 * child->flags of its traced child (same goes for fork, the parent
1540 * can write to the child->flags), because we're guaranteed the
1541 * child is not running and in turn not changing child->flags
1542 * at the same time the parent does it.
1544 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1545 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1546 #define clear_used_math() clear_stopped_child_used_math(current)
1547 #define set_used_math() set_stopped_child_used_math(current)
1549 #define conditional_stopped_child_used_math(condition, child) \
1550 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1552 #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current)
1554 #define copy_to_stopped_child_used_math(child) \
1555 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1557 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1558 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1559 #define used_math() tsk_used_math(current)
1561 static inline bool is_percpu_thread(void)
1564 return (current->flags & PF_NO_SETAFFINITY) &&
1565 (current->nr_cpus_allowed == 1);
1571 /* Per-process atomic flags. */
1572 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
1573 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
1574 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
1575 #define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */
1576 #define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/
1577 #define PFA_SPEC_IB_DISABLE 5 /* Indirect branch speculation restricted */
1578 #define PFA_SPEC_IB_FORCE_DISABLE 6 /* Indirect branch speculation permanently restricted */
1579 #define PFA_SPEC_SSB_NOEXEC 7 /* Speculative Store Bypass clear on execve() */
1581 #define TASK_PFA_TEST(name, func) \
1582 static inline bool task_##func(struct task_struct *p) \
1583 { return test_bit(PFA_##name, &p->atomic_flags); }
1585 #define TASK_PFA_SET(name, func) \
1586 static inline void task_set_##func(struct task_struct *p) \
1587 { set_bit(PFA_##name, &p->atomic_flags); }
1589 #define TASK_PFA_CLEAR(name, func) \
1590 static inline void task_clear_##func(struct task_struct *p) \
1591 { clear_bit(PFA_##name, &p->atomic_flags); }
1593 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
1594 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
1596 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
1597 TASK_PFA_SET(SPREAD_PAGE, spread_page)
1598 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
1600 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
1601 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
1602 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
1604 TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable)
1605 TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable)
1606 TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable)
1608 TASK_PFA_TEST(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1609 TASK_PFA_SET(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1610 TASK_PFA_CLEAR(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1612 TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1613 TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1615 TASK_PFA_TEST(SPEC_IB_DISABLE, spec_ib_disable)
1616 TASK_PFA_SET(SPEC_IB_DISABLE, spec_ib_disable)
1617 TASK_PFA_CLEAR(SPEC_IB_DISABLE, spec_ib_disable)
1619 TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1620 TASK_PFA_SET(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1623 current_restore_flags(unsigned long orig_flags, unsigned long flags)
1625 current->flags &= ~flags;
1626 current->flags |= orig_flags & flags;
1629 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial);
1630 extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed);
1632 extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask);
1633 extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask);
1635 static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
1638 static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
1640 if (!cpumask_test_cpu(0, new_mask))
1646 extern int yield_to(struct task_struct *p, bool preempt);
1647 extern void set_user_nice(struct task_struct *p, long nice);
1648 extern int task_prio(const struct task_struct *p);
1651 * task_nice - return the nice value of a given task.
1652 * @p: the task in question.
1654 * Return: The nice value [ -20 ... 0 ... 19 ].
1656 static inline int task_nice(const struct task_struct *p)
1658 return PRIO_TO_NICE((p)->static_prio);
1661 extern int can_nice(const struct task_struct *p, const int nice);
1662 extern int task_curr(const struct task_struct *p);
1663 extern int idle_cpu(int cpu);
1664 extern int available_idle_cpu(int cpu);
1665 extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *);
1666 extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *);
1667 extern void sched_set_fifo(struct task_struct *p);
1668 extern void sched_set_fifo_low(struct task_struct *p);
1669 extern void sched_set_normal(struct task_struct *p, int nice);
1670 extern int sched_setattr(struct task_struct *, const struct sched_attr *);
1671 extern int sched_setattr_nocheck(struct task_struct *, const struct sched_attr *);
1672 extern struct task_struct *idle_task(int cpu);
1675 * is_idle_task - is the specified task an idle task?
1676 * @p: the task in question.
1678 * Return: 1 if @p is an idle task. 0 otherwise.
1680 static __always_inline bool is_idle_task(const struct task_struct *p)
1682 return !!(p->flags & PF_IDLE);
1685 extern struct task_struct *curr_task(int cpu);
1686 extern void ia64_set_curr_task(int cpu, struct task_struct *p);
1690 union thread_union {
1691 #ifndef CONFIG_ARCH_TASK_STRUCT_ON_STACK
1692 struct task_struct task;
1694 #ifndef CONFIG_THREAD_INFO_IN_TASK
1695 struct thread_info thread_info;
1697 unsigned long stack[THREAD_SIZE/sizeof(long)];
1700 #ifndef CONFIG_THREAD_INFO_IN_TASK
1701 extern struct thread_info init_thread_info;
1704 extern unsigned long init_stack[THREAD_SIZE / sizeof(unsigned long)];
1706 #ifdef CONFIG_THREAD_INFO_IN_TASK
1707 static inline struct thread_info *task_thread_info(struct task_struct *task)
1709 return &task->thread_info;
1711 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1712 # define task_thread_info(task) ((struct thread_info *)(task)->stack)
1716 * find a task by one of its numerical ids
1718 * find_task_by_pid_ns():
1719 * finds a task by its pid in the specified namespace
1720 * find_task_by_vpid():
1721 * finds a task by its virtual pid
1723 * see also find_vpid() etc in include/linux/pid.h
1726 extern struct task_struct *find_task_by_vpid(pid_t nr);
1727 extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns);
1730 * find a task by its virtual pid and get the task struct
1732 extern struct task_struct *find_get_task_by_vpid(pid_t nr);
1734 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1735 extern int wake_up_process(struct task_struct *tsk);
1736 extern void wake_up_new_task(struct task_struct *tsk);
1739 extern void kick_process(struct task_struct *tsk);
1741 static inline void kick_process(struct task_struct *tsk) { }
1744 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
1746 static inline void set_task_comm(struct task_struct *tsk, const char *from)
1748 __set_task_comm(tsk, from, false);
1751 extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk);
1752 #define get_task_comm(buf, tsk) ({ \
1753 BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \
1754 __get_task_comm(buf, sizeof(buf), tsk); \
1758 static __always_inline void scheduler_ipi(void)
1761 * Fold TIF_NEED_RESCHED into the preempt_count; anybody setting
1762 * TIF_NEED_RESCHED remotely (for the first time) will also send
1765 preempt_fold_need_resched();
1767 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1769 static inline void scheduler_ipi(void) { }
1770 static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state)
1777 * Set thread flags in other task's structures.
1778 * See asm/thread_info.h for TIF_xxxx flags available:
1780 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1782 set_ti_thread_flag(task_thread_info(tsk), flag);
1785 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1787 clear_ti_thread_flag(task_thread_info(tsk), flag);
1790 static inline void update_tsk_thread_flag(struct task_struct *tsk, int flag,
1793 update_ti_thread_flag(task_thread_info(tsk), flag, value);
1796 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1798 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1801 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1803 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1806 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1808 return test_ti_thread_flag(task_thread_info(tsk), flag);
1811 static inline void set_tsk_need_resched(struct task_struct *tsk)
1813 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1816 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1818 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1821 static inline int test_tsk_need_resched(struct task_struct *tsk)
1823 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
1827 * cond_resched() and cond_resched_lock(): latency reduction via
1828 * explicit rescheduling in places that are safe. The return
1829 * value indicates whether a reschedule was done in fact.
1830 * cond_resched_lock() will drop the spinlock before scheduling,
1832 #ifndef CONFIG_PREEMPTION
1833 extern int _cond_resched(void);
1835 static inline int _cond_resched(void) { return 0; }
1838 #define cond_resched() ({ \
1839 ___might_sleep(__FILE__, __LINE__, 0); \
1843 extern int __cond_resched_lock(spinlock_t *lock);
1845 #define cond_resched_lock(lock) ({ \
1846 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
1847 __cond_resched_lock(lock); \
1850 static inline void cond_resched_rcu(void)
1852 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
1860 * Does a critical section need to be broken due to another
1861 * task waiting?: (technically does not depend on CONFIG_PREEMPTION,
1862 * but a general need for low latency)
1864 static inline int spin_needbreak(spinlock_t *lock)
1866 #ifdef CONFIG_PREEMPTION
1867 return spin_is_contended(lock);
1873 static __always_inline bool need_resched(void)
1875 return unlikely(tif_need_resched());
1879 * Wrappers for p->thread_info->cpu access. No-op on UP.
1883 static inline unsigned int task_cpu(const struct task_struct *p)
1885 #ifdef CONFIG_THREAD_INFO_IN_TASK
1886 return READ_ONCE(p->cpu);
1888 return READ_ONCE(task_thread_info(p)->cpu);
1892 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1896 static inline unsigned int task_cpu(const struct task_struct *p)
1901 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1905 #endif /* CONFIG_SMP */
1908 * In order to reduce various lock holder preemption latencies provide an
1909 * interface to see if a vCPU is currently running or not.
1911 * This allows us to terminate optimistic spin loops and block, analogous to
1912 * the native optimistic spin heuristic of testing if the lock owner task is
1915 #ifndef vcpu_is_preempted
1916 static inline bool vcpu_is_preempted(int cpu)
1922 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
1923 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
1925 #ifndef TASK_SIZE_OF
1926 #define TASK_SIZE_OF(tsk) TASK_SIZE
1932 * Map the event mask on the user-space ABI enum rseq_cs_flags
1933 * for direct mask checks.
1935 enum rseq_event_mask_bits {
1936 RSEQ_EVENT_PREEMPT_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT_BIT,
1937 RSEQ_EVENT_SIGNAL_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL_BIT,
1938 RSEQ_EVENT_MIGRATE_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE_BIT,
1941 enum rseq_event_mask {
1942 RSEQ_EVENT_PREEMPT = (1U << RSEQ_EVENT_PREEMPT_BIT),
1943 RSEQ_EVENT_SIGNAL = (1U << RSEQ_EVENT_SIGNAL_BIT),
1944 RSEQ_EVENT_MIGRATE = (1U << RSEQ_EVENT_MIGRATE_BIT),
1947 static inline void rseq_set_notify_resume(struct task_struct *t)
1950 set_tsk_thread_flag(t, TIF_NOTIFY_RESUME);
1953 void __rseq_handle_notify_resume(struct ksignal *sig, struct pt_regs *regs);
1955 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
1956 struct pt_regs *regs)
1959 __rseq_handle_notify_resume(ksig, regs);
1962 static inline void rseq_signal_deliver(struct ksignal *ksig,
1963 struct pt_regs *regs)
1966 __set_bit(RSEQ_EVENT_SIGNAL_BIT, ¤t->rseq_event_mask);
1968 rseq_handle_notify_resume(ksig, regs);
1971 /* rseq_preempt() requires preemption to be disabled. */
1972 static inline void rseq_preempt(struct task_struct *t)
1974 __set_bit(RSEQ_EVENT_PREEMPT_BIT, &t->rseq_event_mask);
1975 rseq_set_notify_resume(t);
1978 /* rseq_migrate() requires preemption to be disabled. */
1979 static inline void rseq_migrate(struct task_struct *t)
1981 __set_bit(RSEQ_EVENT_MIGRATE_BIT, &t->rseq_event_mask);
1982 rseq_set_notify_resume(t);
1986 * If parent process has a registered restartable sequences area, the
1987 * child inherits. Unregister rseq for a clone with CLONE_VM set.
1989 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
1991 if (clone_flags & CLONE_VM) {
1994 t->rseq_event_mask = 0;
1996 t->rseq = current->rseq;
1997 t->rseq_sig = current->rseq_sig;
1998 t->rseq_event_mask = current->rseq_event_mask;
2002 static inline void rseq_execve(struct task_struct *t)
2006 t->rseq_event_mask = 0;
2011 static inline void rseq_set_notify_resume(struct task_struct *t)
2014 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
2015 struct pt_regs *regs)
2018 static inline void rseq_signal_deliver(struct ksignal *ksig,
2019 struct pt_regs *regs)
2022 static inline void rseq_preempt(struct task_struct *t)
2025 static inline void rseq_migrate(struct task_struct *t)
2028 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
2031 static inline void rseq_execve(struct task_struct *t)
2037 #ifdef CONFIG_DEBUG_RSEQ
2039 void rseq_syscall(struct pt_regs *regs);
2043 static inline void rseq_syscall(struct pt_regs *regs)
2049 const struct sched_avg *sched_trace_cfs_rq_avg(struct cfs_rq *cfs_rq);
2050 char *sched_trace_cfs_rq_path(struct cfs_rq *cfs_rq, char *str, int len);
2051 int sched_trace_cfs_rq_cpu(struct cfs_rq *cfs_rq);
2053 const struct sched_avg *sched_trace_rq_avg_rt(struct rq *rq);
2054 const struct sched_avg *sched_trace_rq_avg_dl(struct rq *rq);
2055 const struct sched_avg *sched_trace_rq_avg_irq(struct rq *rq);
2057 int sched_trace_rq_cpu(struct rq *rq);
2058 int sched_trace_rq_cpu_capacity(struct rq *rq);
2059 int sched_trace_rq_nr_running(struct rq *rq);
2061 const struct cpumask *sched_trace_rd_span(struct root_domain *rd);