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/seccomp.h>
22 #include <linux/nodemask.h>
23 #include <linux/rcupdate.h>
24 #include <linux/resource.h>
25 #include <linux/latencytop.h>
26 #include <linux/sched/prio.h>
27 #include <linux/signal_types.h>
28 #include <linux/psi_types.h>
29 #include <linux/mm_types_task.h>
30 #include <linux/task_io_accounting.h>
31 #include <linux/rseq.h>
33 /* task_struct member predeclarations (sorted alphabetically): */
35 struct backing_dev_info;
40 struct futex_pi_state;
45 struct perf_event_context;
47 struct pipe_inode_info;
50 struct robust_list_head;
54 struct sighand_struct;
56 struct task_delay_info;
60 * Task state bitmask. NOTE! These bits are also
61 * encoded in fs/proc/array.c: get_task_state().
63 * We have two separate sets of flags: task->state
64 * is about runnability, while task->exit_state are
65 * about the task exiting. Confusing, but this way
66 * modifying one set can't modify the other one by
70 /* Used in tsk->state: */
71 #define TASK_RUNNING 0x0000
72 #define TASK_INTERRUPTIBLE 0x0001
73 #define TASK_UNINTERRUPTIBLE 0x0002
74 #define __TASK_STOPPED 0x0004
75 #define __TASK_TRACED 0x0008
76 /* Used in tsk->exit_state: */
77 #define EXIT_DEAD 0x0010
78 #define EXIT_ZOMBIE 0x0020
79 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
80 /* Used in tsk->state again: */
81 #define TASK_PARKED 0x0040
82 #define TASK_DEAD 0x0080
83 #define TASK_WAKEKILL 0x0100
84 #define TASK_WAKING 0x0200
85 #define TASK_NOLOAD 0x0400
86 #define TASK_NEW 0x0800
87 #define TASK_STATE_MAX 0x1000
89 /* Convenience macros for the sake of set_current_state: */
90 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
91 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
92 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
94 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
96 /* Convenience macros for the sake of wake_up(): */
97 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
99 /* get_task_state(): */
100 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
101 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
102 __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \
105 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
107 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
109 #define task_is_stopped_or_traced(task) ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
111 #define task_contributes_to_load(task) ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
112 (task->flags & PF_FROZEN) == 0 && \
113 (task->state & TASK_NOLOAD) == 0)
115 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
118 * Special states are those that do not use the normal wait-loop pattern. See
119 * the comment with set_special_state().
121 #define is_special_task_state(state) \
122 ((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_PARKED | TASK_DEAD))
124 #define __set_current_state(state_value) \
126 WARN_ON_ONCE(is_special_task_state(state_value));\
127 current->task_state_change = _THIS_IP_; \
128 current->state = (state_value); \
131 #define set_current_state(state_value) \
133 WARN_ON_ONCE(is_special_task_state(state_value));\
134 current->task_state_change = _THIS_IP_; \
135 smp_store_mb(current->state, (state_value)); \
138 #define set_special_state(state_value) \
140 unsigned long flags; /* may shadow */ \
141 WARN_ON_ONCE(!is_special_task_state(state_value)); \
142 raw_spin_lock_irqsave(¤t->pi_lock, flags); \
143 current->task_state_change = _THIS_IP_; \
144 current->state = (state_value); \
145 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
149 * set_current_state() includes a barrier so that the write of current->state
150 * is correctly serialised wrt the caller's subsequent test of whether to
154 * set_current_state(TASK_UNINTERRUPTIBLE);
160 * __set_current_state(TASK_RUNNING);
162 * If the caller does not need such serialisation (because, for instance, the
163 * condition test and condition change and wakeup are under the same lock) then
164 * use __set_current_state().
166 * The above is typically ordered against the wakeup, which does:
168 * need_sleep = false;
169 * wake_up_state(p, TASK_UNINTERRUPTIBLE);
171 * where wake_up_state() executes a full memory barrier before accessing the
174 * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
175 * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
176 * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
178 * However, with slightly different timing the wakeup TASK_RUNNING store can
179 * also collide with the TASK_UNINTERRUPTIBLE store. Losing that store is not
180 * a problem either because that will result in one extra go around the loop
181 * and our @cond test will save the day.
183 * Also see the comments of try_to_wake_up().
185 #define __set_current_state(state_value) \
186 current->state = (state_value)
188 #define set_current_state(state_value) \
189 smp_store_mb(current->state, (state_value))
192 * set_special_state() should be used for those states when the blocking task
193 * can not use the regular condition based wait-loop. In that case we must
194 * serialize against wakeups such that any possible in-flight TASK_RUNNING stores
195 * will not collide with our state change.
197 #define set_special_state(state_value) \
199 unsigned long flags; /* may shadow */ \
200 raw_spin_lock_irqsave(¤t->pi_lock, flags); \
201 current->state = (state_value); \
202 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
207 /* Task command name length: */
208 #define TASK_COMM_LEN 16
210 extern void scheduler_tick(void);
212 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
214 extern long schedule_timeout(long timeout);
215 extern long schedule_timeout_interruptible(long timeout);
216 extern long schedule_timeout_killable(long timeout);
217 extern long schedule_timeout_uninterruptible(long timeout);
218 extern long schedule_timeout_idle(long timeout);
219 asmlinkage void schedule(void);
220 extern void schedule_preempt_disabled(void);
222 extern int __must_check io_schedule_prepare(void);
223 extern void io_schedule_finish(int token);
224 extern long io_schedule_timeout(long timeout);
225 extern void io_schedule(void);
228 * struct prev_cputime - snapshot of system and user cputime
229 * @utime: time spent in user mode
230 * @stime: time spent in system mode
231 * @lock: protects the above two fields
233 * Stores previous user/system time values such that we can guarantee
236 struct prev_cputime {
237 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
245 * struct task_cputime - collected CPU time counts
246 * @utime: time spent in user mode, in nanoseconds
247 * @stime: time spent in kernel mode, in nanoseconds
248 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
250 * This structure groups together three kinds of CPU time that are tracked for
251 * threads and thread groups. Most things considering CPU time want to group
252 * these counts together and treat all three of them in parallel.
254 struct task_cputime {
257 unsigned long long sum_exec_runtime;
260 /* Alternate field names when used on cache expirations: */
261 #define virt_exp utime
262 #define prof_exp stime
263 #define sched_exp sum_exec_runtime
266 /* Task is sleeping or running in a CPU with VTIME inactive: */
268 /* Task runs in userspace in a CPU with VTIME active: */
270 /* Task runs in kernelspace in a CPU with VTIME active: */
276 unsigned long long starttime;
277 enum vtime_state state;
284 #ifdef CONFIG_SCHED_INFO
285 /* Cumulative counters: */
287 /* # of times we have run on this CPU: */
288 unsigned long pcount;
290 /* Time spent waiting on a runqueue: */
291 unsigned long long run_delay;
295 /* When did we last run on a CPU? */
296 unsigned long long last_arrival;
298 /* When were we last queued to run? */
299 unsigned long long last_queued;
301 #endif /* CONFIG_SCHED_INFO */
305 * Integer metrics need fixed point arithmetic, e.g., sched/fair
306 * has a few: load, load_avg, util_avg, freq, and capacity.
308 * We define a basic fixed point arithmetic range, and then formalize
309 * all these metrics based on that basic range.
311 # define SCHED_FIXEDPOINT_SHIFT 10
312 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
315 unsigned long weight;
320 * struct util_est - Estimation utilization of FAIR tasks
321 * @enqueued: instantaneous estimated utilization of a task/cpu
322 * @ewma: the Exponential Weighted Moving Average (EWMA)
323 * utilization of a task
325 * Support data structure to track an Exponential Weighted Moving Average
326 * (EWMA) of a FAIR task's utilization. New samples are added to the moving
327 * average each time a task completes an activation. Sample's weight is chosen
328 * so that the EWMA will be relatively insensitive to transient changes to the
331 * The enqueued attribute has a slightly different meaning for tasks and cpus:
332 * - task: the task's util_avg at last task dequeue time
333 * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU
334 * Thus, the util_est.enqueued of a task represents the contribution on the
335 * estimated utilization of the CPU where that task is currently enqueued.
337 * Only for tasks we track a moving average of the past instantaneous
338 * estimated utilization. This allows to absorb sporadic drops in utilization
339 * of an otherwise almost periodic task.
342 unsigned int enqueued;
344 #define UTIL_EST_WEIGHT_SHIFT 2
345 } __attribute__((__aligned__(sizeof(u64))));
348 * The load_avg/util_avg accumulates an infinite geometric series
349 * (see __update_load_avg() in kernel/sched/fair.c).
351 * [load_avg definition]
353 * load_avg = runnable% * scale_load_down(load)
355 * where runnable% is the time ratio that a sched_entity is runnable.
356 * For cfs_rq, it is the aggregated load_avg of all runnable and
357 * blocked sched_entities.
359 * load_avg may also take frequency scaling into account:
361 * load_avg = runnable% * scale_load_down(load) * freq%
363 * where freq% is the CPU frequency normalized to the highest frequency.
365 * [util_avg definition]
367 * util_avg = running% * SCHED_CAPACITY_SCALE
369 * where running% is the time ratio that a sched_entity is running on
370 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
371 * and blocked sched_entities.
373 * util_avg may also factor frequency scaling and CPU capacity scaling:
375 * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
377 * where freq% is the same as above, and capacity% is the CPU capacity
378 * normalized to the greatest capacity (due to uarch differences, etc).
380 * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
381 * themselves are in the range of [0, 1]. To do fixed point arithmetics,
382 * we therefore scale them to as large a range as necessary. This is for
383 * example reflected by util_avg's SCHED_CAPACITY_SCALE.
387 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
388 * with the highest load (=88761), always runnable on a single cfs_rq,
389 * and should not overflow as the number already hits PID_MAX_LIMIT.
391 * For all other cases (including 32-bit kernels), struct load_weight's
392 * weight will overflow first before we do, because:
394 * Max(load_avg) <= Max(load.weight)
396 * Then it is the load_weight's responsibility to consider overflow
400 u64 last_update_time;
402 u64 runnable_load_sum;
405 unsigned long load_avg;
406 unsigned long runnable_load_avg;
407 unsigned long util_avg;
408 struct util_est util_est;
409 } ____cacheline_aligned;
411 struct sched_statistics {
412 #ifdef CONFIG_SCHEDSTATS
422 s64 sum_sleep_runtime;
429 u64 nr_migrations_cold;
430 u64 nr_failed_migrations_affine;
431 u64 nr_failed_migrations_running;
432 u64 nr_failed_migrations_hot;
433 u64 nr_forced_migrations;
437 u64 nr_wakeups_migrate;
438 u64 nr_wakeups_local;
439 u64 nr_wakeups_remote;
440 u64 nr_wakeups_affine;
441 u64 nr_wakeups_affine_attempts;
442 u64 nr_wakeups_passive;
447 struct sched_entity {
448 /* For load-balancing: */
449 struct load_weight load;
450 unsigned long runnable_weight;
451 struct rb_node run_node;
452 struct list_head group_node;
456 u64 sum_exec_runtime;
458 u64 prev_sum_exec_runtime;
462 struct sched_statistics statistics;
464 #ifdef CONFIG_FAIR_GROUP_SCHED
466 struct sched_entity *parent;
467 /* rq on which this entity is (to be) queued: */
468 struct cfs_rq *cfs_rq;
469 /* rq "owned" by this entity/group: */
475 * Per entity load average tracking.
477 * Put into separate cache line so it does not
478 * collide with read-mostly values above.
480 struct sched_avg avg;
484 struct sched_rt_entity {
485 struct list_head run_list;
486 unsigned long timeout;
487 unsigned long watchdog_stamp;
488 unsigned int time_slice;
489 unsigned short on_rq;
490 unsigned short on_list;
492 struct sched_rt_entity *back;
493 #ifdef CONFIG_RT_GROUP_SCHED
494 struct sched_rt_entity *parent;
495 /* rq on which this entity is (to be) queued: */
497 /* rq "owned" by this entity/group: */
500 } __randomize_layout;
502 struct sched_dl_entity {
503 struct rb_node rb_node;
506 * Original scheduling parameters. Copied here from sched_attr
507 * during sched_setattr(), they will remain the same until
508 * the next sched_setattr().
510 u64 dl_runtime; /* Maximum runtime for each instance */
511 u64 dl_deadline; /* Relative deadline of each instance */
512 u64 dl_period; /* Separation of two instances (period) */
513 u64 dl_bw; /* dl_runtime / dl_period */
514 u64 dl_density; /* dl_runtime / dl_deadline */
517 * Actual scheduling parameters. Initialized with the values above,
518 * they are continuously updated during task execution. Note that
519 * the remaining runtime could be < 0 in case we are in overrun.
521 s64 runtime; /* Remaining runtime for this instance */
522 u64 deadline; /* Absolute deadline for this instance */
523 unsigned int flags; /* Specifying the scheduler behaviour */
528 * @dl_throttled tells if we exhausted the runtime. If so, the
529 * task has to wait for a replenishment to be performed at the
530 * next firing of dl_timer.
532 * @dl_boosted tells if we are boosted due to DI. If so we are
533 * outside bandwidth enforcement mechanism (but only until we
534 * exit the critical section);
536 * @dl_yielded tells if task gave up the CPU before consuming
537 * all its available runtime during the last job.
539 * @dl_non_contending tells if the task is inactive while still
540 * contributing to the active utilization. In other words, it
541 * indicates if the inactive timer has been armed and its handler
542 * has not been executed yet. This flag is useful to avoid race
543 * conditions between the inactive timer handler and the wakeup
546 * @dl_overrun tells if the task asked to be informed about runtime
549 unsigned int dl_throttled : 1;
550 unsigned int dl_boosted : 1;
551 unsigned int dl_yielded : 1;
552 unsigned int dl_non_contending : 1;
553 unsigned int dl_overrun : 1;
556 * Bandwidth enforcement timer. Each -deadline task has its
557 * own bandwidth to be enforced, thus we need one timer per task.
559 struct hrtimer dl_timer;
562 * Inactive timer, responsible for decreasing the active utilization
563 * at the "0-lag time". When a -deadline task blocks, it contributes
564 * to GRUB's active utilization until the "0-lag time", hence a
565 * timer is needed to decrease the active utilization at the correct
568 struct hrtimer inactive_timer;
575 u8 exp_hint; /* Hint for performance. */
576 u8 pad; /* No garbage from compiler! */
578 u32 s; /* Set of bits. */
581 enum perf_event_task_context {
582 perf_invalid_context = -1,
585 perf_nr_task_contexts,
589 struct wake_q_node *next;
593 #ifdef CONFIG_THREAD_INFO_IN_TASK
595 * For reasons of header soup (see current_thread_info()), this
596 * must be the first element of task_struct.
598 struct thread_info thread_info;
600 /* -1 unrunnable, 0 runnable, >0 stopped: */
604 * This begins the randomizable portion of task_struct. Only
605 * scheduling-critical items should be added above here.
607 randomized_struct_fields_start
611 /* Per task flags (PF_*), defined further below: */
616 struct llist_node wake_entry;
618 #ifdef CONFIG_THREAD_INFO_IN_TASK
622 unsigned int wakee_flips;
623 unsigned long wakee_flip_decay_ts;
624 struct task_struct *last_wakee;
627 * recent_used_cpu is initially set as the last CPU used by a task
628 * that wakes affine another task. Waker/wakee relationships can
629 * push tasks around a CPU where each wakeup moves to the next one.
630 * Tracking a recently used CPU allows a quick search for a recently
631 * used CPU that may be idle.
641 unsigned int rt_priority;
643 const struct sched_class *sched_class;
644 struct sched_entity se;
645 struct sched_rt_entity rt;
646 #ifdef CONFIG_CGROUP_SCHED
647 struct task_group *sched_task_group;
649 struct sched_dl_entity dl;
651 #ifdef CONFIG_PREEMPT_NOTIFIERS
652 /* List of struct preempt_notifier: */
653 struct hlist_head preempt_notifiers;
656 #ifdef CONFIG_BLK_DEV_IO_TRACE
657 unsigned int btrace_seq;
662 cpumask_t cpus_allowed;
664 #ifdef CONFIG_PREEMPT_RCU
665 int rcu_read_lock_nesting;
666 union rcu_special rcu_read_unlock_special;
667 struct list_head rcu_node_entry;
668 struct rcu_node *rcu_blocked_node;
669 #endif /* #ifdef CONFIG_PREEMPT_RCU */
671 #ifdef CONFIG_TASKS_RCU
672 unsigned long rcu_tasks_nvcsw;
673 u8 rcu_tasks_holdout;
675 int rcu_tasks_idle_cpu;
676 struct list_head rcu_tasks_holdout_list;
677 #endif /* #ifdef CONFIG_TASKS_RCU */
679 struct sched_info sched_info;
681 struct list_head tasks;
683 struct plist_node pushable_tasks;
684 struct rb_node pushable_dl_tasks;
687 struct mm_struct *mm;
688 struct mm_struct *active_mm;
690 /* Per-thread vma caching: */
691 struct vmacache vmacache;
693 #ifdef SPLIT_RSS_COUNTING
694 struct task_rss_stat rss_stat;
699 /* The signal sent when the parent dies: */
701 /* JOBCTL_*, siglock protected: */
702 unsigned long jobctl;
704 /* Used for emulating ABI behavior of previous Linux versions: */
705 unsigned int personality;
707 /* Scheduler bits, serialized by scheduler locks: */
708 unsigned sched_reset_on_fork:1;
709 unsigned sched_contributes_to_load:1;
710 unsigned sched_migrated:1;
711 unsigned sched_remote_wakeup:1;
713 unsigned sched_psi_wake_requeue:1;
716 /* Force alignment to the next boundary: */
719 /* Unserialized, strictly 'current' */
721 /* Bit to tell LSMs we're in execve(): */
722 unsigned in_execve:1;
723 unsigned in_iowait:1;
724 #ifndef TIF_RESTORE_SIGMASK
725 unsigned restore_sigmask:1;
728 unsigned in_user_fault:1;
730 #ifdef CONFIG_COMPAT_BRK
731 unsigned brk_randomized:1;
733 #ifdef CONFIG_CGROUPS
734 /* disallow userland-initiated cgroup migration */
735 unsigned no_cgroup_migration:1;
737 #ifdef CONFIG_BLK_CGROUP
738 /* to be used once the psi infrastructure lands upstream. */
739 unsigned use_memdelay:1;
742 unsigned long atomic_flags; /* Flags requiring atomic access. */
744 struct restart_block restart_block;
749 #ifdef CONFIG_STACKPROTECTOR
750 /* Canary value for the -fstack-protector GCC feature: */
751 unsigned long stack_canary;
754 * Pointers to the (original) parent process, youngest child, younger sibling,
755 * older sibling, respectively. (p->father can be replaced with
756 * p->real_parent->pid)
759 /* Real parent process: */
760 struct task_struct __rcu *real_parent;
762 /* Recipient of SIGCHLD, wait4() reports: */
763 struct task_struct __rcu *parent;
766 * Children/sibling form the list of natural children:
768 struct list_head children;
769 struct list_head sibling;
770 struct task_struct *group_leader;
773 * 'ptraced' is the list of tasks this task is using ptrace() on.
775 * This includes both natural children and PTRACE_ATTACH targets.
776 * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
778 struct list_head ptraced;
779 struct list_head ptrace_entry;
781 /* PID/PID hash table linkage. */
782 struct pid *thread_pid;
783 struct hlist_node pid_links[PIDTYPE_MAX];
784 struct list_head thread_group;
785 struct list_head thread_node;
787 struct completion *vfork_done;
789 /* CLONE_CHILD_SETTID: */
790 int __user *set_child_tid;
792 /* CLONE_CHILD_CLEARTID: */
793 int __user *clear_child_tid;
797 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
802 struct prev_cputime prev_cputime;
803 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
807 #ifdef CONFIG_NO_HZ_FULL
808 atomic_t tick_dep_mask;
810 /* Context switch counts: */
812 unsigned long nivcsw;
814 /* Monotonic time in nsecs: */
817 /* Boot based time in nsecs: */
820 /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
821 unsigned long min_flt;
822 unsigned long maj_flt;
824 #ifdef CONFIG_POSIX_TIMERS
825 struct task_cputime cputime_expires;
826 struct list_head cpu_timers[3];
829 /* Process credentials: */
831 /* Tracer's credentials at attach: */
832 const struct cred __rcu *ptracer_cred;
834 /* Objective and real subjective task credentials (COW): */
835 const struct cred __rcu *real_cred;
837 /* Effective (overridable) subjective task credentials (COW): */
838 const struct cred __rcu *cred;
841 * executable name, excluding path.
843 * - normally initialized setup_new_exec()
844 * - access it with [gs]et_task_comm()
845 * - lock it with task_lock()
847 char comm[TASK_COMM_LEN];
849 struct nameidata *nameidata;
851 #ifdef CONFIG_SYSVIPC
852 struct sysv_sem sysvsem;
853 struct sysv_shm sysvshm;
855 #ifdef CONFIG_DETECT_HUNG_TASK
856 unsigned long last_switch_count;
857 unsigned long last_switch_time;
859 /* Filesystem information: */
860 struct fs_struct *fs;
862 /* Open file information: */
863 struct files_struct *files;
866 struct nsproxy *nsproxy;
868 /* Signal handlers: */
869 struct signal_struct *signal;
870 struct sighand_struct *sighand;
872 sigset_t real_blocked;
873 /* Restored if set_restore_sigmask() was used: */
874 sigset_t saved_sigmask;
875 struct sigpending pending;
876 unsigned long sas_ss_sp;
878 unsigned int sas_ss_flags;
880 struct callback_head *task_works;
882 struct audit_context *audit_context;
883 #ifdef CONFIG_AUDITSYSCALL
885 unsigned int sessionid;
887 struct seccomp seccomp;
889 /* Thread group tracking: */
893 /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
894 spinlock_t alloc_lock;
896 /* Protection of the PI data structures: */
897 raw_spinlock_t pi_lock;
899 struct wake_q_node wake_q;
901 #ifdef CONFIG_RT_MUTEXES
902 /* PI waiters blocked on a rt_mutex held by this task: */
903 struct rb_root_cached pi_waiters;
904 /* Updated under owner's pi_lock and rq lock */
905 struct task_struct *pi_top_task;
906 /* Deadlock detection and priority inheritance handling: */
907 struct rt_mutex_waiter *pi_blocked_on;
910 #ifdef CONFIG_DEBUG_MUTEXES
911 /* Mutex deadlock detection: */
912 struct mutex_waiter *blocked_on;
915 #ifdef CONFIG_TRACE_IRQFLAGS
916 unsigned int irq_events;
917 unsigned long hardirq_enable_ip;
918 unsigned long hardirq_disable_ip;
919 unsigned int hardirq_enable_event;
920 unsigned int hardirq_disable_event;
921 int hardirqs_enabled;
923 unsigned long softirq_disable_ip;
924 unsigned long softirq_enable_ip;
925 unsigned int softirq_disable_event;
926 unsigned int softirq_enable_event;
927 int softirqs_enabled;
931 #ifdef CONFIG_LOCKDEP
932 # define MAX_LOCK_DEPTH 48UL
935 unsigned int lockdep_recursion;
936 struct held_lock held_locks[MAX_LOCK_DEPTH];
940 unsigned int in_ubsan;
943 /* Journalling filesystem info: */
946 /* Stacked block device info: */
947 struct bio_list *bio_list;
950 /* Stack plugging: */
951 struct blk_plug *plug;
955 struct reclaim_state *reclaim_state;
957 struct backing_dev_info *backing_dev_info;
959 struct io_context *io_context;
962 unsigned long ptrace_message;
963 kernel_siginfo_t *last_siginfo;
965 struct task_io_accounting ioac;
967 /* Pressure stall state */
968 unsigned int psi_flags;
970 #ifdef CONFIG_TASK_XACCT
971 /* Accumulated RSS usage: */
973 /* Accumulated virtual memory usage: */
975 /* stime + utime since last update: */
978 #ifdef CONFIG_CPUSETS
979 /* Protected by ->alloc_lock: */
980 nodemask_t mems_allowed;
981 /* Seqence number to catch updates: */
982 seqcount_t mems_allowed_seq;
983 int cpuset_mem_spread_rotor;
984 int cpuset_slab_spread_rotor;
986 #ifdef CONFIG_CGROUPS
987 /* Control Group info protected by css_set_lock: */
988 struct css_set __rcu *cgroups;
989 /* cg_list protected by css_set_lock and tsk->alloc_lock: */
990 struct list_head cg_list;
992 #ifdef CONFIG_X86_CPU_RESCTRL
997 struct robust_list_head __user *robust_list;
999 struct compat_robust_list_head __user *compat_robust_list;
1001 struct list_head pi_state_list;
1002 struct futex_pi_state *pi_state_cache;
1004 #ifdef CONFIG_PERF_EVENTS
1005 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1006 struct mutex perf_event_mutex;
1007 struct list_head perf_event_list;
1009 #ifdef CONFIG_DEBUG_PREEMPT
1010 unsigned long preempt_disable_ip;
1013 /* Protected by alloc_lock: */
1014 struct mempolicy *mempolicy;
1016 short pref_node_fork;
1018 #ifdef CONFIG_NUMA_BALANCING
1020 unsigned int numa_scan_period;
1021 unsigned int numa_scan_period_max;
1022 int numa_preferred_nid;
1023 unsigned long numa_migrate_retry;
1024 /* Migration stamp: */
1026 u64 last_task_numa_placement;
1027 u64 last_sum_exec_runtime;
1028 struct callback_head numa_work;
1030 struct numa_group *numa_group;
1033 * numa_faults is an array split into four regions:
1034 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1035 * in this precise order.
1037 * faults_memory: Exponential decaying average of faults on a per-node
1038 * basis. Scheduling placement decisions are made based on these
1039 * counts. The values remain static for the duration of a PTE scan.
1040 * faults_cpu: Track the nodes the process was running on when a NUMA
1041 * hinting fault was incurred.
1042 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1043 * during the current scan window. When the scan completes, the counts
1044 * in faults_memory and faults_cpu decay and these values are copied.
1046 unsigned long *numa_faults;
1047 unsigned long total_numa_faults;
1050 * numa_faults_locality tracks if faults recorded during the last
1051 * scan window were remote/local or failed to migrate. The task scan
1052 * period is adapted based on the locality of the faults with different
1053 * weights depending on whether they were shared or private faults
1055 unsigned long numa_faults_locality[3];
1057 unsigned long numa_pages_migrated;
1058 #endif /* CONFIG_NUMA_BALANCING */
1061 struct rseq __user *rseq;
1065 * RmW on rseq_event_mask must be performed atomically
1066 * with respect to preemption.
1068 unsigned long rseq_event_mask;
1071 struct tlbflush_unmap_batch tlb_ubc;
1073 struct rcu_head rcu;
1075 /* Cache last used pipe for splice(): */
1076 struct pipe_inode_info *splice_pipe;
1078 struct page_frag task_frag;
1080 #ifdef CONFIG_TASK_DELAY_ACCT
1081 struct task_delay_info *delays;
1084 #ifdef CONFIG_FAULT_INJECTION
1086 unsigned int fail_nth;
1089 * When (nr_dirtied >= nr_dirtied_pause), it's time to call
1090 * balance_dirty_pages() for a dirty throttling pause:
1093 int nr_dirtied_pause;
1094 /* Start of a write-and-pause period: */
1095 unsigned long dirty_paused_when;
1097 #ifdef CONFIG_LATENCYTOP
1098 int latency_record_count;
1099 struct latency_record latency_record[LT_SAVECOUNT];
1102 * Time slack values; these are used to round up poll() and
1103 * select() etc timeout values. These are in nanoseconds.
1106 u64 default_timer_slack_ns;
1109 unsigned int kasan_depth;
1112 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1113 /* Index of current stored address in ret_stack: */
1117 /* Stack of return addresses for return function tracing: */
1118 struct ftrace_ret_stack *ret_stack;
1120 /* Timestamp for last schedule: */
1121 unsigned long long ftrace_timestamp;
1124 * Number of functions that haven't been traced
1125 * because of depth overrun:
1127 atomic_t trace_overrun;
1129 /* Pause tracing: */
1130 atomic_t tracing_graph_pause;
1133 #ifdef CONFIG_TRACING
1134 /* State flags for use by tracers: */
1135 unsigned long trace;
1137 /* Bitmask and counter of trace recursion: */
1138 unsigned long trace_recursion;
1139 #endif /* CONFIG_TRACING */
1142 /* Coverage collection mode enabled for this task (0 if disabled): */
1143 unsigned int kcov_mode;
1145 /* Size of the kcov_area: */
1146 unsigned int kcov_size;
1148 /* Buffer for coverage collection: */
1151 /* KCOV descriptor wired with this task or NULL: */
1156 struct mem_cgroup *memcg_in_oom;
1157 gfp_t memcg_oom_gfp_mask;
1158 int memcg_oom_order;
1160 /* Number of pages to reclaim on returning to userland: */
1161 unsigned int memcg_nr_pages_over_high;
1163 /* Used by memcontrol for targeted memcg charge: */
1164 struct mem_cgroup *active_memcg;
1167 #ifdef CONFIG_BLK_CGROUP
1168 struct request_queue *throttle_queue;
1171 #ifdef CONFIG_UPROBES
1172 struct uprobe_task *utask;
1174 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1175 unsigned int sequential_io;
1176 unsigned int sequential_io_avg;
1178 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1179 unsigned long task_state_change;
1181 int pagefault_disabled;
1183 struct task_struct *oom_reaper_list;
1185 #ifdef CONFIG_VMAP_STACK
1186 struct vm_struct *stack_vm_area;
1188 #ifdef CONFIG_THREAD_INFO_IN_TASK
1189 /* A live task holds one reference: */
1190 atomic_t stack_refcount;
1192 #ifdef CONFIG_LIVEPATCH
1195 #ifdef CONFIG_SECURITY
1196 /* Used by LSM modules for access restriction: */
1200 #ifdef CONFIG_GCC_PLUGIN_STACKLEAK
1201 unsigned long lowest_stack;
1202 unsigned long prev_lowest_stack;
1206 * New fields for task_struct should be added above here, so that
1207 * they are included in the randomized portion of task_struct.
1209 randomized_struct_fields_end
1211 /* CPU-specific state of this task: */
1212 struct thread_struct thread;
1215 * WARNING: on x86, 'thread_struct' contains a variable-sized
1216 * structure. It *MUST* be at the end of 'task_struct'.
1218 * Do not put anything below here!
1222 static inline struct pid *task_pid(struct task_struct *task)
1224 return task->thread_pid;
1228 * the helpers to get the task's different pids as they are seen
1229 * from various namespaces
1231 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1232 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1234 * task_xid_nr_ns() : id seen from the ns specified;
1236 * see also pid_nr() etc in include/linux/pid.h
1238 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns);
1240 static inline pid_t task_pid_nr(struct task_struct *tsk)
1245 static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1247 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1250 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1252 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1256 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1262 * pid_alive - check that a task structure is not stale
1263 * @p: Task structure to be checked.
1265 * Test if a process is not yet dead (at most zombie state)
1266 * If pid_alive fails, then pointers within the task structure
1267 * can be stale and must not be dereferenced.
1269 * Return: 1 if the process is alive. 0 otherwise.
1271 static inline int pid_alive(const struct task_struct *p)
1273 return p->thread_pid != NULL;
1276 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1278 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1281 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1283 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1287 static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1289 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1292 static inline pid_t task_session_vnr(struct task_struct *tsk)
1294 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1297 static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1299 return __task_pid_nr_ns(tsk, PIDTYPE_TGID, ns);
1302 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1304 return __task_pid_nr_ns(tsk, PIDTYPE_TGID, NULL);
1307 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1313 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1319 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1321 return task_ppid_nr_ns(tsk, &init_pid_ns);
1324 /* Obsolete, do not use: */
1325 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1327 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1330 #define TASK_REPORT_IDLE (TASK_REPORT + 1)
1331 #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1)
1333 static inline unsigned int task_state_index(struct task_struct *tsk)
1335 unsigned int tsk_state = READ_ONCE(tsk->state);
1336 unsigned int state = (tsk_state | tsk->exit_state) & TASK_REPORT;
1338 BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX);
1340 if (tsk_state == TASK_IDLE)
1341 state = TASK_REPORT_IDLE;
1346 static inline char task_index_to_char(unsigned int state)
1348 static const char state_char[] = "RSDTtXZPI";
1350 BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX) != sizeof(state_char) - 1);
1352 return state_char[state];
1355 static inline char task_state_to_char(struct task_struct *tsk)
1357 return task_index_to_char(task_state_index(tsk));
1361 * is_global_init - check if a task structure is init. Since init
1362 * is free to have sub-threads we need to check tgid.
1363 * @tsk: Task structure to be checked.
1365 * Check if a task structure is the first user space task the kernel created.
1367 * Return: 1 if the task structure is init. 0 otherwise.
1369 static inline int is_global_init(struct task_struct *tsk)
1371 return task_tgid_nr(tsk) == 1;
1374 extern struct pid *cad_pid;
1379 #define PF_IDLE 0x00000002 /* I am an IDLE thread */
1380 #define PF_EXITING 0x00000004 /* Getting shut down */
1381 #define PF_EXITPIDONE 0x00000008 /* PI exit done on shut down */
1382 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1383 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1384 #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */
1385 #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */
1386 #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */
1387 #define PF_DUMPCORE 0x00000200 /* Dumped core */
1388 #define PF_SIGNALED 0x00000400 /* Killed by a signal */
1389 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1390 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */
1391 #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */
1392 #define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */
1393 #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */
1394 #define PF_FROZEN 0x00010000 /* Frozen for system suspend */
1395 #define PF_KSWAPD 0x00020000 /* I am kswapd */
1396 #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */
1397 #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */
1398 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1399 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1400 #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */
1401 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1402 #define PF_MEMSTALL 0x01000000 /* Stalled due to lack of memory */
1403 #define PF_UMH 0x02000000 /* I'm an Usermodehelper process */
1404 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1405 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1406 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1407 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1408 #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */
1411 * Only the _current_ task can read/write to tsk->flags, but other
1412 * tasks can access tsk->flags in readonly mode for example
1413 * with tsk_used_math (like during threaded core dumping).
1414 * There is however an exception to this rule during ptrace
1415 * or during fork: the ptracer task is allowed to write to the
1416 * child->flags of its traced child (same goes for fork, the parent
1417 * can write to the child->flags), because we're guaranteed the
1418 * child is not running and in turn not changing child->flags
1419 * at the same time the parent does it.
1421 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1422 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1423 #define clear_used_math() clear_stopped_child_used_math(current)
1424 #define set_used_math() set_stopped_child_used_math(current)
1426 #define conditional_stopped_child_used_math(condition, child) \
1427 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1429 #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current)
1431 #define copy_to_stopped_child_used_math(child) \
1432 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1434 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1435 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1436 #define used_math() tsk_used_math(current)
1438 static inline bool is_percpu_thread(void)
1441 return (current->flags & PF_NO_SETAFFINITY) &&
1442 (current->nr_cpus_allowed == 1);
1448 /* Per-process atomic flags. */
1449 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
1450 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
1451 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
1452 #define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */
1453 #define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/
1454 #define PFA_SPEC_IB_DISABLE 5 /* Indirect branch speculation restricted */
1455 #define PFA_SPEC_IB_FORCE_DISABLE 6 /* Indirect branch speculation permanently restricted */
1457 #define TASK_PFA_TEST(name, func) \
1458 static inline bool task_##func(struct task_struct *p) \
1459 { return test_bit(PFA_##name, &p->atomic_flags); }
1461 #define TASK_PFA_SET(name, func) \
1462 static inline void task_set_##func(struct task_struct *p) \
1463 { set_bit(PFA_##name, &p->atomic_flags); }
1465 #define TASK_PFA_CLEAR(name, func) \
1466 static inline void task_clear_##func(struct task_struct *p) \
1467 { clear_bit(PFA_##name, &p->atomic_flags); }
1469 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
1470 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
1472 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
1473 TASK_PFA_SET(SPREAD_PAGE, spread_page)
1474 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
1476 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
1477 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
1478 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
1480 TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable)
1481 TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable)
1482 TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable)
1484 TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1485 TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1487 TASK_PFA_TEST(SPEC_IB_DISABLE, spec_ib_disable)
1488 TASK_PFA_SET(SPEC_IB_DISABLE, spec_ib_disable)
1489 TASK_PFA_CLEAR(SPEC_IB_DISABLE, spec_ib_disable)
1491 TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1492 TASK_PFA_SET(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1495 current_restore_flags(unsigned long orig_flags, unsigned long flags)
1497 current->flags &= ~flags;
1498 current->flags |= orig_flags & flags;
1501 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial);
1502 extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed);
1504 extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask);
1505 extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask);
1507 static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
1510 static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
1512 if (!cpumask_test_cpu(0, new_mask))
1518 #ifndef cpu_relax_yield
1519 #define cpu_relax_yield() cpu_relax()
1522 extern int yield_to(struct task_struct *p, bool preempt);
1523 extern void set_user_nice(struct task_struct *p, long nice);
1524 extern int task_prio(const struct task_struct *p);
1527 * task_nice - return the nice value of a given task.
1528 * @p: the task in question.
1530 * Return: The nice value [ -20 ... 0 ... 19 ].
1532 static inline int task_nice(const struct task_struct *p)
1534 return PRIO_TO_NICE((p)->static_prio);
1537 extern int can_nice(const struct task_struct *p, const int nice);
1538 extern int task_curr(const struct task_struct *p);
1539 extern int idle_cpu(int cpu);
1540 extern int available_idle_cpu(int cpu);
1541 extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *);
1542 extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *);
1543 extern int sched_setattr(struct task_struct *, const struct sched_attr *);
1544 extern int sched_setattr_nocheck(struct task_struct *, const struct sched_attr *);
1545 extern struct task_struct *idle_task(int cpu);
1548 * is_idle_task - is the specified task an idle task?
1549 * @p: the task in question.
1551 * Return: 1 if @p is an idle task. 0 otherwise.
1553 static inline bool is_idle_task(const struct task_struct *p)
1555 return !!(p->flags & PF_IDLE);
1558 extern struct task_struct *curr_task(int cpu);
1559 extern void ia64_set_curr_task(int cpu, struct task_struct *p);
1563 union thread_union {
1564 #ifndef CONFIG_ARCH_TASK_STRUCT_ON_STACK
1565 struct task_struct task;
1567 #ifndef CONFIG_THREAD_INFO_IN_TASK
1568 struct thread_info thread_info;
1570 unsigned long stack[THREAD_SIZE/sizeof(long)];
1573 #ifndef CONFIG_THREAD_INFO_IN_TASK
1574 extern struct thread_info init_thread_info;
1577 extern unsigned long init_stack[THREAD_SIZE / sizeof(unsigned long)];
1579 #ifdef CONFIG_THREAD_INFO_IN_TASK
1580 static inline struct thread_info *task_thread_info(struct task_struct *task)
1582 return &task->thread_info;
1584 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1585 # define task_thread_info(task) ((struct thread_info *)(task)->stack)
1589 * find a task by one of its numerical ids
1591 * find_task_by_pid_ns():
1592 * finds a task by its pid in the specified namespace
1593 * find_task_by_vpid():
1594 * finds a task by its virtual pid
1596 * see also find_vpid() etc in include/linux/pid.h
1599 extern struct task_struct *find_task_by_vpid(pid_t nr);
1600 extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns);
1603 * find a task by its virtual pid and get the task struct
1605 extern struct task_struct *find_get_task_by_vpid(pid_t nr);
1607 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1608 extern int wake_up_process(struct task_struct *tsk);
1609 extern void wake_up_new_task(struct task_struct *tsk);
1612 extern void kick_process(struct task_struct *tsk);
1614 static inline void kick_process(struct task_struct *tsk) { }
1617 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
1619 static inline void set_task_comm(struct task_struct *tsk, const char *from)
1621 __set_task_comm(tsk, from, false);
1624 extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk);
1625 #define get_task_comm(buf, tsk) ({ \
1626 BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \
1627 __get_task_comm(buf, sizeof(buf), tsk); \
1631 void scheduler_ipi(void);
1632 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1634 static inline void scheduler_ipi(void) { }
1635 static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state)
1642 * Set thread flags in other task's structures.
1643 * See asm/thread_info.h for TIF_xxxx flags available:
1645 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1647 set_ti_thread_flag(task_thread_info(tsk), flag);
1650 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1652 clear_ti_thread_flag(task_thread_info(tsk), flag);
1655 static inline void update_tsk_thread_flag(struct task_struct *tsk, int flag,
1658 update_ti_thread_flag(task_thread_info(tsk), flag, value);
1661 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1663 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1666 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1668 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1671 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1673 return test_ti_thread_flag(task_thread_info(tsk), flag);
1676 static inline void set_tsk_need_resched(struct task_struct *tsk)
1678 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1681 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1683 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1686 static inline int test_tsk_need_resched(struct task_struct *tsk)
1688 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
1692 * cond_resched() and cond_resched_lock(): latency reduction via
1693 * explicit rescheduling in places that are safe. The return
1694 * value indicates whether a reschedule was done in fact.
1695 * cond_resched_lock() will drop the spinlock before scheduling,
1697 #ifndef CONFIG_PREEMPT
1698 extern int _cond_resched(void);
1700 static inline int _cond_resched(void) { return 0; }
1703 #define cond_resched() ({ \
1704 ___might_sleep(__FILE__, __LINE__, 0); \
1708 extern int __cond_resched_lock(spinlock_t *lock);
1710 #define cond_resched_lock(lock) ({ \
1711 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
1712 __cond_resched_lock(lock); \
1715 static inline void cond_resched_rcu(void)
1717 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
1725 * Does a critical section need to be broken due to another
1726 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
1727 * but a general need for low latency)
1729 static inline int spin_needbreak(spinlock_t *lock)
1731 #ifdef CONFIG_PREEMPT
1732 return spin_is_contended(lock);
1738 static __always_inline bool need_resched(void)
1740 return unlikely(tif_need_resched());
1744 * Wrappers for p->thread_info->cpu access. No-op on UP.
1748 static inline unsigned int task_cpu(const struct task_struct *p)
1750 #ifdef CONFIG_THREAD_INFO_IN_TASK
1753 return task_thread_info(p)->cpu;
1757 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1761 static inline unsigned int task_cpu(const struct task_struct *p)
1766 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1770 #endif /* CONFIG_SMP */
1773 * In order to reduce various lock holder preemption latencies provide an
1774 * interface to see if a vCPU is currently running or not.
1776 * This allows us to terminate optimistic spin loops and block, analogous to
1777 * the native optimistic spin heuristic of testing if the lock owner task is
1780 #ifndef vcpu_is_preempted
1781 # define vcpu_is_preempted(cpu) false
1784 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
1785 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
1787 #ifndef TASK_SIZE_OF
1788 #define TASK_SIZE_OF(tsk) TASK_SIZE
1794 * Map the event mask on the user-space ABI enum rseq_cs_flags
1795 * for direct mask checks.
1797 enum rseq_event_mask_bits {
1798 RSEQ_EVENT_PREEMPT_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT_BIT,
1799 RSEQ_EVENT_SIGNAL_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL_BIT,
1800 RSEQ_EVENT_MIGRATE_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE_BIT,
1803 enum rseq_event_mask {
1804 RSEQ_EVENT_PREEMPT = (1U << RSEQ_EVENT_PREEMPT_BIT),
1805 RSEQ_EVENT_SIGNAL = (1U << RSEQ_EVENT_SIGNAL_BIT),
1806 RSEQ_EVENT_MIGRATE = (1U << RSEQ_EVENT_MIGRATE_BIT),
1809 static inline void rseq_set_notify_resume(struct task_struct *t)
1812 set_tsk_thread_flag(t, TIF_NOTIFY_RESUME);
1815 void __rseq_handle_notify_resume(struct ksignal *sig, struct pt_regs *regs);
1817 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
1818 struct pt_regs *regs)
1821 __rseq_handle_notify_resume(ksig, regs);
1824 static inline void rseq_signal_deliver(struct ksignal *ksig,
1825 struct pt_regs *regs)
1828 __set_bit(RSEQ_EVENT_SIGNAL_BIT, ¤t->rseq_event_mask);
1830 rseq_handle_notify_resume(ksig, regs);
1833 /* rseq_preempt() requires preemption to be disabled. */
1834 static inline void rseq_preempt(struct task_struct *t)
1836 __set_bit(RSEQ_EVENT_PREEMPT_BIT, &t->rseq_event_mask);
1837 rseq_set_notify_resume(t);
1840 /* rseq_migrate() requires preemption to be disabled. */
1841 static inline void rseq_migrate(struct task_struct *t)
1843 __set_bit(RSEQ_EVENT_MIGRATE_BIT, &t->rseq_event_mask);
1844 rseq_set_notify_resume(t);
1848 * If parent process has a registered restartable sequences area, the
1849 * child inherits. Only applies when forking a process, not a thread.
1851 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
1853 if (clone_flags & CLONE_THREAD) {
1857 t->rseq_event_mask = 0;
1859 t->rseq = current->rseq;
1860 t->rseq_len = current->rseq_len;
1861 t->rseq_sig = current->rseq_sig;
1862 t->rseq_event_mask = current->rseq_event_mask;
1866 static inline void rseq_execve(struct task_struct *t)
1871 t->rseq_event_mask = 0;
1876 static inline void rseq_set_notify_resume(struct task_struct *t)
1879 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
1880 struct pt_regs *regs)
1883 static inline void rseq_signal_deliver(struct ksignal *ksig,
1884 struct pt_regs *regs)
1887 static inline void rseq_preempt(struct task_struct *t)
1890 static inline void rseq_migrate(struct task_struct *t)
1893 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
1896 static inline void rseq_execve(struct task_struct *t)
1902 void __exit_umh(struct task_struct *tsk);
1904 static inline void exit_umh(struct task_struct *tsk)
1906 if (unlikely(tsk->flags & PF_UMH))
1910 #ifdef CONFIG_DEBUG_RSEQ
1912 void rseq_syscall(struct pt_regs *regs);
1916 static inline void rseq_syscall(struct pt_regs *regs)