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/refcount.h>
25 #include <linux/resource.h>
26 #include <linux/latencytop.h>
27 #include <linux/sched/prio.h>
28 #include <linux/signal_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 capture_control;
51 struct robust_list_head;
55 struct sighand_struct;
57 struct task_delay_info;
61 * Task state bitmask. NOTE! These bits are also
62 * encoded in fs/proc/array.c: get_task_state().
64 * We have two separate sets of flags: task->state
65 * is about runnability, while task->exit_state are
66 * about the task exiting. Confusing, but this way
67 * modifying one set can't modify the other one by
71 /* Used in tsk->state: */
72 #define TASK_RUNNING 0x0000
73 #define TASK_INTERRUPTIBLE 0x0001
74 #define TASK_UNINTERRUPTIBLE 0x0002
75 #define __TASK_STOPPED 0x0004
76 #define __TASK_TRACED 0x0008
77 /* Used in tsk->exit_state: */
78 #define EXIT_DEAD 0x0010
79 #define EXIT_ZOMBIE 0x0020
80 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
81 /* Used in tsk->state again: */
82 #define TASK_PARKED 0x0040
83 #define TASK_DEAD 0x0080
84 #define TASK_WAKEKILL 0x0100
85 #define TASK_WAKING 0x0200
86 #define TASK_NOLOAD 0x0400
87 #define TASK_NEW 0x0800
88 #define TASK_STATE_MAX 0x1000
90 /* Convenience macros for the sake of set_current_state: */
91 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
92 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
93 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
95 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
97 /* Convenience macros for the sake of wake_up(): */
98 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
100 /* get_task_state(): */
101 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
102 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
103 __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \
106 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
108 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
110 #define task_is_stopped_or_traced(task) ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
112 #define task_contributes_to_load(task) ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
113 (task->flags & PF_FROZEN) == 0 && \
114 (task->state & TASK_NOLOAD) == 0)
116 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
119 * Special states are those that do not use the normal wait-loop pattern. See
120 * the comment with set_special_state().
122 #define is_special_task_state(state) \
123 ((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_PARKED | TASK_DEAD))
125 #define __set_current_state(state_value) \
127 WARN_ON_ONCE(is_special_task_state(state_value));\
128 current->task_state_change = _THIS_IP_; \
129 current->state = (state_value); \
132 #define set_current_state(state_value) \
134 WARN_ON_ONCE(is_special_task_state(state_value));\
135 current->task_state_change = _THIS_IP_; \
136 smp_store_mb(current->state, (state_value)); \
139 #define set_special_state(state_value) \
141 unsigned long flags; /* may shadow */ \
142 WARN_ON_ONCE(!is_special_task_state(state_value)); \
143 raw_spin_lock_irqsave(¤t->pi_lock, flags); \
144 current->task_state_change = _THIS_IP_; \
145 current->state = (state_value); \
146 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
150 * set_current_state() includes a barrier so that the write of current->state
151 * is correctly serialised wrt the caller's subsequent test of whether to
155 * set_current_state(TASK_UNINTERRUPTIBLE);
161 * __set_current_state(TASK_RUNNING);
163 * If the caller does not need such serialisation (because, for instance, the
164 * condition test and condition change and wakeup are under the same lock) then
165 * use __set_current_state().
167 * The above is typically ordered against the wakeup, which does:
169 * need_sleep = false;
170 * wake_up_state(p, TASK_UNINTERRUPTIBLE);
172 * where wake_up_state() executes a full memory barrier before accessing the
175 * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
176 * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
177 * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
179 * However, with slightly different timing the wakeup TASK_RUNNING store can
180 * also collide with the TASK_UNINTERRUPTIBLE store. Losing that store is not
181 * a problem either because that will result in one extra go around the loop
182 * and our @cond test will save the day.
184 * Also see the comments of try_to_wake_up().
186 #define __set_current_state(state_value) \
187 current->state = (state_value)
189 #define set_current_state(state_value) \
190 smp_store_mb(current->state, (state_value))
193 * set_special_state() should be used for those states when the blocking task
194 * can not use the regular condition based wait-loop. In that case we must
195 * serialize against wakeups such that any possible in-flight TASK_RUNNING stores
196 * will not collide with our state change.
198 #define set_special_state(state_value) \
200 unsigned long flags; /* may shadow */ \
201 raw_spin_lock_irqsave(¤t->pi_lock, flags); \
202 current->state = (state_value); \
203 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
208 /* Task command name length: */
209 #define TASK_COMM_LEN 16
211 extern void scheduler_tick(void);
213 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
215 extern long schedule_timeout(long timeout);
216 extern long schedule_timeout_interruptible(long timeout);
217 extern long schedule_timeout_killable(long timeout);
218 extern long schedule_timeout_uninterruptible(long timeout);
219 extern long schedule_timeout_idle(long timeout);
220 asmlinkage void schedule(void);
221 extern void schedule_preempt_disabled(void);
223 extern int __must_check io_schedule_prepare(void);
224 extern void io_schedule_finish(int token);
225 extern long io_schedule_timeout(long timeout);
226 extern void io_schedule(void);
229 * struct prev_cputime - snapshot of system and user cputime
230 * @utime: time spent in user mode
231 * @stime: time spent in system mode
232 * @lock: protects the above two fields
234 * Stores previous user/system time values such that we can guarantee
237 struct prev_cputime {
238 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
246 * struct task_cputime - collected CPU time counts
247 * @utime: time spent in user mode, in nanoseconds
248 * @stime: time spent in kernel mode, in nanoseconds
249 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
251 * This structure groups together three kinds of CPU time that are tracked for
252 * threads and thread groups. Most things considering CPU time want to group
253 * these counts together and treat all three of them in parallel.
255 struct task_cputime {
258 unsigned long long sum_exec_runtime;
261 /* Alternate field names when used on cache expirations: */
262 #define virt_exp utime
263 #define prof_exp stime
264 #define sched_exp sum_exec_runtime
267 /* Task is sleeping or running in a CPU with VTIME inactive: */
269 /* Task runs in userspace in a CPU with VTIME active: */
271 /* Task runs in kernelspace in a CPU with VTIME active: */
277 unsigned long long starttime;
278 enum vtime_state state;
285 #ifdef CONFIG_SCHED_INFO
286 /* Cumulative counters: */
288 /* # of times we have run on this CPU: */
289 unsigned long pcount;
291 /* Time spent waiting on a runqueue: */
292 unsigned long long run_delay;
296 /* When did we last run on a CPU? */
297 unsigned long long last_arrival;
299 /* When were we last queued to run? */
300 unsigned long long last_queued;
302 #endif /* CONFIG_SCHED_INFO */
306 * Integer metrics need fixed point arithmetic, e.g., sched/fair
307 * has a few: load, load_avg, util_avg, freq, and capacity.
309 * We define a basic fixed point arithmetic range, and then formalize
310 * all these metrics based on that basic range.
312 # define SCHED_FIXEDPOINT_SHIFT 10
313 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
316 unsigned long weight;
321 * struct util_est - Estimation utilization of FAIR tasks
322 * @enqueued: instantaneous estimated utilization of a task/cpu
323 * @ewma: the Exponential Weighted Moving Average (EWMA)
324 * utilization of a task
326 * Support data structure to track an Exponential Weighted Moving Average
327 * (EWMA) of a FAIR task's utilization. New samples are added to the moving
328 * average each time a task completes an activation. Sample's weight is chosen
329 * so that the EWMA will be relatively insensitive to transient changes to the
332 * The enqueued attribute has a slightly different meaning for tasks and cpus:
333 * - task: the task's util_avg at last task dequeue time
334 * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU
335 * Thus, the util_est.enqueued of a task represents the contribution on the
336 * estimated utilization of the CPU where that task is currently enqueued.
338 * Only for tasks we track a moving average of the past instantaneous
339 * estimated utilization. This allows to absorb sporadic drops in utilization
340 * of an otherwise almost periodic task.
343 unsigned int enqueued;
345 #define UTIL_EST_WEIGHT_SHIFT 2
346 } __attribute__((__aligned__(sizeof(u64))));
349 * The load_avg/util_avg accumulates an infinite geometric series
350 * (see __update_load_avg() in kernel/sched/fair.c).
352 * [load_avg definition]
354 * load_avg = runnable% * scale_load_down(load)
356 * where runnable% is the time ratio that a sched_entity is runnable.
357 * For cfs_rq, it is the aggregated load_avg of all runnable and
358 * blocked sched_entities.
360 * [util_avg definition]
362 * util_avg = running% * SCHED_CAPACITY_SCALE
364 * where running% is the time ratio that a sched_entity is running on
365 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
366 * and blocked sched_entities.
368 * load_avg and util_avg don't direcly factor frequency scaling and CPU
369 * capacity scaling. The scaling is done through the rq_clock_pelt that
370 * is used for computing those signals (see update_rq_clock_pelt())
372 * N.B., the above ratios (runnable% and running%) themselves are in the
373 * range of [0, 1]. To do fixed point arithmetics, we therefore scale them
374 * to as large a range as necessary. This is for example reflected by
375 * util_avg's SCHED_CAPACITY_SCALE.
379 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
380 * with the highest load (=88761), always runnable on a single cfs_rq,
381 * and should not overflow as the number already hits PID_MAX_LIMIT.
383 * For all other cases (including 32-bit kernels), struct load_weight's
384 * weight will overflow first before we do, because:
386 * Max(load_avg) <= Max(load.weight)
388 * Then it is the load_weight's responsibility to consider overflow
392 u64 last_update_time;
394 u64 runnable_load_sum;
397 unsigned long load_avg;
398 unsigned long runnable_load_avg;
399 unsigned long util_avg;
400 struct util_est util_est;
401 } ____cacheline_aligned;
403 struct sched_statistics {
404 #ifdef CONFIG_SCHEDSTATS
414 s64 sum_sleep_runtime;
421 u64 nr_migrations_cold;
422 u64 nr_failed_migrations_affine;
423 u64 nr_failed_migrations_running;
424 u64 nr_failed_migrations_hot;
425 u64 nr_forced_migrations;
429 u64 nr_wakeups_migrate;
430 u64 nr_wakeups_local;
431 u64 nr_wakeups_remote;
432 u64 nr_wakeups_affine;
433 u64 nr_wakeups_affine_attempts;
434 u64 nr_wakeups_passive;
439 struct sched_entity {
440 /* For load-balancing: */
441 struct load_weight load;
442 unsigned long runnable_weight;
443 struct rb_node run_node;
444 struct list_head group_node;
448 u64 sum_exec_runtime;
450 u64 prev_sum_exec_runtime;
454 struct sched_statistics statistics;
456 #ifdef CONFIG_FAIR_GROUP_SCHED
458 struct sched_entity *parent;
459 /* rq on which this entity is (to be) queued: */
460 struct cfs_rq *cfs_rq;
461 /* rq "owned" by this entity/group: */
467 * Per entity load average tracking.
469 * Put into separate cache line so it does not
470 * collide with read-mostly values above.
472 struct sched_avg avg;
476 struct sched_rt_entity {
477 struct list_head run_list;
478 unsigned long timeout;
479 unsigned long watchdog_stamp;
480 unsigned int time_slice;
481 unsigned short on_rq;
482 unsigned short on_list;
484 struct sched_rt_entity *back;
485 #ifdef CONFIG_RT_GROUP_SCHED
486 struct sched_rt_entity *parent;
487 /* rq on which this entity is (to be) queued: */
489 /* rq "owned" by this entity/group: */
492 } __randomize_layout;
494 struct sched_dl_entity {
495 struct rb_node rb_node;
498 * Original scheduling parameters. Copied here from sched_attr
499 * during sched_setattr(), they will remain the same until
500 * the next sched_setattr().
502 u64 dl_runtime; /* Maximum runtime for each instance */
503 u64 dl_deadline; /* Relative deadline of each instance */
504 u64 dl_period; /* Separation of two instances (period) */
505 u64 dl_bw; /* dl_runtime / dl_period */
506 u64 dl_density; /* dl_runtime / dl_deadline */
509 * Actual scheduling parameters. Initialized with the values above,
510 * they are continuously updated during task execution. Note that
511 * the remaining runtime could be < 0 in case we are in overrun.
513 s64 runtime; /* Remaining runtime for this instance */
514 u64 deadline; /* Absolute deadline for this instance */
515 unsigned int flags; /* Specifying the scheduler behaviour */
520 * @dl_throttled tells if we exhausted the runtime. If so, the
521 * task has to wait for a replenishment to be performed at the
522 * next firing of dl_timer.
524 * @dl_boosted tells if we are boosted due to DI. If so we are
525 * outside bandwidth enforcement mechanism (but only until we
526 * exit the critical section);
528 * @dl_yielded tells if task gave up the CPU before consuming
529 * all its available runtime during the last job.
531 * @dl_non_contending tells if the task is inactive while still
532 * contributing to the active utilization. In other words, it
533 * indicates if the inactive timer has been armed and its handler
534 * has not been executed yet. This flag is useful to avoid race
535 * conditions between the inactive timer handler and the wakeup
538 * @dl_overrun tells if the task asked to be informed about runtime
541 unsigned int dl_throttled : 1;
542 unsigned int dl_boosted : 1;
543 unsigned int dl_yielded : 1;
544 unsigned int dl_non_contending : 1;
545 unsigned int dl_overrun : 1;
548 * Bandwidth enforcement timer. Each -deadline task has its
549 * own bandwidth to be enforced, thus we need one timer per task.
551 struct hrtimer dl_timer;
554 * Inactive timer, responsible for decreasing the active utilization
555 * at the "0-lag time". When a -deadline task blocks, it contributes
556 * to GRUB's active utilization until the "0-lag time", hence a
557 * timer is needed to decrease the active utilization at the correct
560 struct hrtimer inactive_timer;
567 u8 exp_hint; /* Hint for performance. */
568 u8 pad; /* No garbage from compiler! */
570 u32 s; /* Set of bits. */
573 enum perf_event_task_context {
574 perf_invalid_context = -1,
577 perf_nr_task_contexts,
581 struct wake_q_node *next;
585 #ifdef CONFIG_THREAD_INFO_IN_TASK
587 * For reasons of header soup (see current_thread_info()), this
588 * must be the first element of task_struct.
590 struct thread_info thread_info;
592 /* -1 unrunnable, 0 runnable, >0 stopped: */
596 * This begins the randomizable portion of task_struct. Only
597 * scheduling-critical items should be added above here.
599 randomized_struct_fields_start
603 /* Per task flags (PF_*), defined further below: */
608 struct llist_node wake_entry;
610 #ifdef CONFIG_THREAD_INFO_IN_TASK
614 unsigned int wakee_flips;
615 unsigned long wakee_flip_decay_ts;
616 struct task_struct *last_wakee;
619 * recent_used_cpu is initially set as the last CPU used by a task
620 * that wakes affine another task. Waker/wakee relationships can
621 * push tasks around a CPU where each wakeup moves to the next one.
622 * Tracking a recently used CPU allows a quick search for a recently
623 * used CPU that may be idle.
633 unsigned int rt_priority;
635 const struct sched_class *sched_class;
636 struct sched_entity se;
637 struct sched_rt_entity rt;
638 #ifdef CONFIG_CGROUP_SCHED
639 struct task_group *sched_task_group;
641 struct sched_dl_entity dl;
643 #ifdef CONFIG_PREEMPT_NOTIFIERS
644 /* List of struct preempt_notifier: */
645 struct hlist_head preempt_notifiers;
648 #ifdef CONFIG_BLK_DEV_IO_TRACE
649 unsigned int btrace_seq;
654 cpumask_t cpus_allowed;
656 #ifdef CONFIG_PREEMPT_RCU
657 int rcu_read_lock_nesting;
658 union rcu_special rcu_read_unlock_special;
659 struct list_head rcu_node_entry;
660 struct rcu_node *rcu_blocked_node;
661 #endif /* #ifdef CONFIG_PREEMPT_RCU */
663 #ifdef CONFIG_TASKS_RCU
664 unsigned long rcu_tasks_nvcsw;
665 u8 rcu_tasks_holdout;
667 int rcu_tasks_idle_cpu;
668 struct list_head rcu_tasks_holdout_list;
669 #endif /* #ifdef CONFIG_TASKS_RCU */
671 struct sched_info sched_info;
673 struct list_head tasks;
675 struct plist_node pushable_tasks;
676 struct rb_node pushable_dl_tasks;
679 struct mm_struct *mm;
680 struct mm_struct *active_mm;
682 /* Per-thread vma caching: */
683 struct vmacache vmacache;
685 #ifdef SPLIT_RSS_COUNTING
686 struct task_rss_stat rss_stat;
691 /* The signal sent when the parent dies: */
693 /* JOBCTL_*, siglock protected: */
694 unsigned long jobctl;
696 /* Used for emulating ABI behavior of previous Linux versions: */
697 unsigned int personality;
699 /* Scheduler bits, serialized by scheduler locks: */
700 unsigned sched_reset_on_fork:1;
701 unsigned sched_contributes_to_load:1;
702 unsigned sched_migrated:1;
703 unsigned sched_remote_wakeup:1;
705 unsigned sched_psi_wake_requeue:1;
708 /* Force alignment to the next boundary: */
711 /* Unserialized, strictly 'current' */
713 /* Bit to tell LSMs we're in execve(): */
714 unsigned in_execve:1;
715 unsigned in_iowait:1;
716 #ifndef TIF_RESTORE_SIGMASK
717 unsigned restore_sigmask:1;
720 unsigned in_user_fault:1;
722 #ifdef CONFIG_COMPAT_BRK
723 unsigned brk_randomized:1;
725 #ifdef CONFIG_CGROUPS
726 /* disallow userland-initiated cgroup migration */
727 unsigned no_cgroup_migration:1;
728 /* task is frozen/stopped (used by the cgroup freezer) */
731 #ifdef CONFIG_BLK_CGROUP
732 /* to be used once the psi infrastructure lands upstream. */
733 unsigned use_memdelay:1;
736 unsigned long atomic_flags; /* Flags requiring atomic access. */
738 struct restart_block restart_block;
743 #ifdef CONFIG_STACKPROTECTOR
744 /* Canary value for the -fstack-protector GCC feature: */
745 unsigned long stack_canary;
748 * Pointers to the (original) parent process, youngest child, younger sibling,
749 * older sibling, respectively. (p->father can be replaced with
750 * p->real_parent->pid)
753 /* Real parent process: */
754 struct task_struct __rcu *real_parent;
756 /* Recipient of SIGCHLD, wait4() reports: */
757 struct task_struct __rcu *parent;
760 * Children/sibling form the list of natural children:
762 struct list_head children;
763 struct list_head sibling;
764 struct task_struct *group_leader;
767 * 'ptraced' is the list of tasks this task is using ptrace() on.
769 * This includes both natural children and PTRACE_ATTACH targets.
770 * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
772 struct list_head ptraced;
773 struct list_head ptrace_entry;
775 /* PID/PID hash table linkage. */
776 struct pid *thread_pid;
777 struct hlist_node pid_links[PIDTYPE_MAX];
778 struct list_head thread_group;
779 struct list_head thread_node;
781 struct completion *vfork_done;
783 /* CLONE_CHILD_SETTID: */
784 int __user *set_child_tid;
786 /* CLONE_CHILD_CLEARTID: */
787 int __user *clear_child_tid;
791 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
796 struct prev_cputime prev_cputime;
797 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
801 #ifdef CONFIG_NO_HZ_FULL
802 atomic_t tick_dep_mask;
804 /* Context switch counts: */
806 unsigned long nivcsw;
808 /* Monotonic time in nsecs: */
811 /* Boot based time in nsecs: */
814 /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
815 unsigned long min_flt;
816 unsigned long maj_flt;
818 #ifdef CONFIG_POSIX_TIMERS
819 struct task_cputime cputime_expires;
820 struct list_head cpu_timers[3];
823 /* Process credentials: */
825 /* Tracer's credentials at attach: */
826 const struct cred __rcu *ptracer_cred;
828 /* Objective and real subjective task credentials (COW): */
829 const struct cred __rcu *real_cred;
831 /* Effective (overridable) subjective task credentials (COW): */
832 const struct cred __rcu *cred;
835 * executable name, excluding path.
837 * - normally initialized setup_new_exec()
838 * - access it with [gs]et_task_comm()
839 * - lock it with task_lock()
841 char comm[TASK_COMM_LEN];
843 struct nameidata *nameidata;
845 #ifdef CONFIG_SYSVIPC
846 struct sysv_sem sysvsem;
847 struct sysv_shm sysvshm;
849 #ifdef CONFIG_DETECT_HUNG_TASK
850 unsigned long last_switch_count;
851 unsigned long last_switch_time;
853 /* Filesystem information: */
854 struct fs_struct *fs;
856 /* Open file information: */
857 struct files_struct *files;
860 struct nsproxy *nsproxy;
862 /* Signal handlers: */
863 struct signal_struct *signal;
864 struct sighand_struct *sighand;
866 sigset_t real_blocked;
867 /* Restored if set_restore_sigmask() was used: */
868 sigset_t saved_sigmask;
869 struct sigpending pending;
870 unsigned long sas_ss_sp;
872 unsigned int sas_ss_flags;
874 struct callback_head *task_works;
877 #ifdef CONFIG_AUDITSYSCALL
878 struct audit_context *audit_context;
881 unsigned int sessionid;
883 struct seccomp seccomp;
885 /* Thread group tracking: */
889 /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
890 spinlock_t alloc_lock;
892 /* Protection of the PI data structures: */
893 raw_spinlock_t pi_lock;
895 struct wake_q_node wake_q;
897 #ifdef CONFIG_RT_MUTEXES
898 /* PI waiters blocked on a rt_mutex held by this task: */
899 struct rb_root_cached pi_waiters;
900 /* Updated under owner's pi_lock and rq lock */
901 struct task_struct *pi_top_task;
902 /* Deadlock detection and priority inheritance handling: */
903 struct rt_mutex_waiter *pi_blocked_on;
906 #ifdef CONFIG_DEBUG_MUTEXES
907 /* Mutex deadlock detection: */
908 struct mutex_waiter *blocked_on;
911 #ifdef CONFIG_TRACE_IRQFLAGS
912 unsigned int irq_events;
913 unsigned long hardirq_enable_ip;
914 unsigned long hardirq_disable_ip;
915 unsigned int hardirq_enable_event;
916 unsigned int hardirq_disable_event;
917 int hardirqs_enabled;
919 unsigned long softirq_disable_ip;
920 unsigned long softirq_enable_ip;
921 unsigned int softirq_disable_event;
922 unsigned int softirq_enable_event;
923 int softirqs_enabled;
927 #ifdef CONFIG_LOCKDEP
928 # define MAX_LOCK_DEPTH 48UL
931 unsigned int lockdep_recursion;
932 struct held_lock held_locks[MAX_LOCK_DEPTH];
936 unsigned int in_ubsan;
939 /* Journalling filesystem info: */
942 /* Stacked block device info: */
943 struct bio_list *bio_list;
946 /* Stack plugging: */
947 struct blk_plug *plug;
951 struct reclaim_state *reclaim_state;
953 struct backing_dev_info *backing_dev_info;
955 struct io_context *io_context;
957 #ifdef CONFIG_COMPACTION
958 struct capture_control *capture_control;
961 unsigned long ptrace_message;
962 kernel_siginfo_t *last_siginfo;
964 struct task_io_accounting ioac;
966 /* Pressure stall state */
967 unsigned int psi_flags;
969 #ifdef CONFIG_TASK_XACCT
970 /* Accumulated RSS usage: */
972 /* Accumulated virtual memory usage: */
974 /* stime + utime since last update: */
977 #ifdef CONFIG_CPUSETS
978 /* Protected by ->alloc_lock: */
979 nodemask_t mems_allowed;
980 /* Seqence number to catch updates: */
981 seqcount_t mems_allowed_seq;
982 int cpuset_mem_spread_rotor;
983 int cpuset_slab_spread_rotor;
985 #ifdef CONFIG_CGROUPS
986 /* Control Group info protected by css_set_lock: */
987 struct css_set __rcu *cgroups;
988 /* cg_list protected by css_set_lock and tsk->alloc_lock: */
989 struct list_head cg_list;
991 #ifdef CONFIG_X86_CPU_RESCTRL
996 struct robust_list_head __user *robust_list;
998 struct compat_robust_list_head __user *compat_robust_list;
1000 struct list_head pi_state_list;
1001 struct futex_pi_state *pi_state_cache;
1003 #ifdef CONFIG_PERF_EVENTS
1004 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1005 struct mutex perf_event_mutex;
1006 struct list_head perf_event_list;
1008 #ifdef CONFIG_DEBUG_PREEMPT
1009 unsigned long preempt_disable_ip;
1012 /* Protected by alloc_lock: */
1013 struct mempolicy *mempolicy;
1015 short pref_node_fork;
1017 #ifdef CONFIG_NUMA_BALANCING
1019 unsigned int numa_scan_period;
1020 unsigned int numa_scan_period_max;
1021 int numa_preferred_nid;
1022 unsigned long numa_migrate_retry;
1023 /* Migration stamp: */
1025 u64 last_task_numa_placement;
1026 u64 last_sum_exec_runtime;
1027 struct callback_head numa_work;
1029 struct numa_group *numa_group;
1032 * numa_faults is an array split into four regions:
1033 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1034 * in this precise order.
1036 * faults_memory: Exponential decaying average of faults on a per-node
1037 * basis. Scheduling placement decisions are made based on these
1038 * counts. The values remain static for the duration of a PTE scan.
1039 * faults_cpu: Track the nodes the process was running on when a NUMA
1040 * hinting fault was incurred.
1041 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1042 * during the current scan window. When the scan completes, the counts
1043 * in faults_memory and faults_cpu decay and these values are copied.
1045 unsigned long *numa_faults;
1046 unsigned long total_numa_faults;
1049 * numa_faults_locality tracks if faults recorded during the last
1050 * scan window were remote/local or failed to migrate. The task scan
1051 * period is adapted based on the locality of the faults with different
1052 * weights depending on whether they were shared or private faults
1054 unsigned long numa_faults_locality[3];
1056 unsigned long numa_pages_migrated;
1057 #endif /* CONFIG_NUMA_BALANCING */
1060 struct rseq __user *rseq;
1063 * RmW on rseq_event_mask must be performed atomically
1064 * with respect to preemption.
1066 unsigned long rseq_event_mask;
1069 struct tlbflush_unmap_batch tlb_ubc;
1071 struct rcu_head rcu;
1073 /* Cache last used pipe for splice(): */
1074 struct pipe_inode_info *splice_pipe;
1076 struct page_frag task_frag;
1078 #ifdef CONFIG_TASK_DELAY_ACCT
1079 struct task_delay_info *delays;
1082 #ifdef CONFIG_FAULT_INJECTION
1084 unsigned int fail_nth;
1087 * When (nr_dirtied >= nr_dirtied_pause), it's time to call
1088 * balance_dirty_pages() for a dirty throttling pause:
1091 int nr_dirtied_pause;
1092 /* Start of a write-and-pause period: */
1093 unsigned long dirty_paused_when;
1095 #ifdef CONFIG_LATENCYTOP
1096 int latency_record_count;
1097 struct latency_record latency_record[LT_SAVECOUNT];
1100 * Time slack values; these are used to round up poll() and
1101 * select() etc timeout values. These are in nanoseconds.
1104 u64 default_timer_slack_ns;
1107 unsigned int kasan_depth;
1110 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1111 /* Index of current stored address in ret_stack: */
1115 /* Stack of return addresses for return function tracing: */
1116 struct ftrace_ret_stack *ret_stack;
1118 /* Timestamp for last schedule: */
1119 unsigned long long ftrace_timestamp;
1122 * Number of functions that haven't been traced
1123 * because of depth overrun:
1125 atomic_t trace_overrun;
1127 /* Pause tracing: */
1128 atomic_t tracing_graph_pause;
1131 #ifdef CONFIG_TRACING
1132 /* State flags for use by tracers: */
1133 unsigned long trace;
1135 /* Bitmask and counter of trace recursion: */
1136 unsigned long trace_recursion;
1137 #endif /* CONFIG_TRACING */
1140 /* Coverage collection mode enabled for this task (0 if disabled): */
1141 unsigned int kcov_mode;
1143 /* Size of the kcov_area: */
1144 unsigned int kcov_size;
1146 /* Buffer for coverage collection: */
1149 /* KCOV descriptor wired with this task or NULL: */
1154 struct mem_cgroup *memcg_in_oom;
1155 gfp_t memcg_oom_gfp_mask;
1156 int memcg_oom_order;
1158 /* Number of pages to reclaim on returning to userland: */
1159 unsigned int memcg_nr_pages_over_high;
1161 /* Used by memcontrol for targeted memcg charge: */
1162 struct mem_cgroup *active_memcg;
1165 #ifdef CONFIG_BLK_CGROUP
1166 struct request_queue *throttle_queue;
1169 #ifdef CONFIG_UPROBES
1170 struct uprobe_task *utask;
1172 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1173 unsigned int sequential_io;
1174 unsigned int sequential_io_avg;
1176 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1177 unsigned long task_state_change;
1179 int pagefault_disabled;
1181 struct task_struct *oom_reaper_list;
1183 #ifdef CONFIG_VMAP_STACK
1184 struct vm_struct *stack_vm_area;
1186 #ifdef CONFIG_THREAD_INFO_IN_TASK
1187 /* A live task holds one reference: */
1188 refcount_t stack_refcount;
1190 #ifdef CONFIG_LIVEPATCH
1193 #ifdef CONFIG_SECURITY
1194 /* Used by LSM modules for access restriction: */
1198 #ifdef CONFIG_GCC_PLUGIN_STACKLEAK
1199 unsigned long lowest_stack;
1200 unsigned long prev_lowest_stack;
1204 * New fields for task_struct should be added above here, so that
1205 * they are included in the randomized portion of task_struct.
1207 randomized_struct_fields_end
1209 /* CPU-specific state of this task: */
1210 struct thread_struct thread;
1213 * WARNING: on x86, 'thread_struct' contains a variable-sized
1214 * structure. It *MUST* be at the end of 'task_struct'.
1216 * Do not put anything below here!
1220 static inline struct pid *task_pid(struct task_struct *task)
1222 return task->thread_pid;
1226 * the helpers to get the task's different pids as they are seen
1227 * from various namespaces
1229 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1230 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1232 * task_xid_nr_ns() : id seen from the ns specified;
1234 * see also pid_nr() etc in include/linux/pid.h
1236 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns);
1238 static inline pid_t task_pid_nr(struct task_struct *tsk)
1243 static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1245 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1248 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1250 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1254 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1260 * pid_alive - check that a task structure is not stale
1261 * @p: Task structure to be checked.
1263 * Test if a process is not yet dead (at most zombie state)
1264 * If pid_alive fails, then pointers within the task structure
1265 * can be stale and must not be dereferenced.
1267 * Return: 1 if the process is alive. 0 otherwise.
1269 static inline int pid_alive(const struct task_struct *p)
1271 return p->thread_pid != NULL;
1274 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1276 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1279 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1281 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1285 static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1287 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1290 static inline pid_t task_session_vnr(struct task_struct *tsk)
1292 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1295 static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1297 return __task_pid_nr_ns(tsk, PIDTYPE_TGID, ns);
1300 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1302 return __task_pid_nr_ns(tsk, PIDTYPE_TGID, NULL);
1305 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1311 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1317 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1319 return task_ppid_nr_ns(tsk, &init_pid_ns);
1322 /* Obsolete, do not use: */
1323 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1325 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1328 #define TASK_REPORT_IDLE (TASK_REPORT + 1)
1329 #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1)
1331 static inline unsigned int task_state_index(struct task_struct *tsk)
1333 unsigned int tsk_state = READ_ONCE(tsk->state);
1334 unsigned int state = (tsk_state | tsk->exit_state) & TASK_REPORT;
1336 BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX);
1338 if (tsk_state == TASK_IDLE)
1339 state = TASK_REPORT_IDLE;
1344 static inline char task_index_to_char(unsigned int state)
1346 static const char state_char[] = "RSDTtXZPI";
1348 BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX) != sizeof(state_char) - 1);
1350 return state_char[state];
1353 static inline char task_state_to_char(struct task_struct *tsk)
1355 return task_index_to_char(task_state_index(tsk));
1359 * is_global_init - check if a task structure is init. Since init
1360 * is free to have sub-threads we need to check tgid.
1361 * @tsk: Task structure to be checked.
1363 * Check if a task structure is the first user space task the kernel created.
1365 * Return: 1 if the task structure is init. 0 otherwise.
1367 static inline int is_global_init(struct task_struct *tsk)
1369 return task_tgid_nr(tsk) == 1;
1372 extern struct pid *cad_pid;
1377 #define PF_IDLE 0x00000002 /* I am an IDLE thread */
1378 #define PF_EXITING 0x00000004 /* Getting shut down */
1379 #define PF_EXITPIDONE 0x00000008 /* PI exit done on shut down */
1380 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1381 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1382 #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */
1383 #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */
1384 #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */
1385 #define PF_DUMPCORE 0x00000200 /* Dumped core */
1386 #define PF_SIGNALED 0x00000400 /* Killed by a signal */
1387 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1388 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */
1389 #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */
1390 #define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */
1391 #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */
1392 #define PF_FROZEN 0x00010000 /* Frozen for system suspend */
1393 #define PF_KSWAPD 0x00020000 /* I am kswapd */
1394 #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */
1395 #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */
1396 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1397 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1398 #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */
1399 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1400 #define PF_MEMSTALL 0x01000000 /* Stalled due to lack of memory */
1401 #define PF_UMH 0x02000000 /* I'm an Usermodehelper process */
1402 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1403 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1404 #define PF_MEMALLOC_NOCMA 0x10000000 /* All allocation request will have _GFP_MOVABLE cleared */
1405 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1406 #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */
1409 * Only the _current_ task can read/write to tsk->flags, but other
1410 * tasks can access tsk->flags in readonly mode for example
1411 * with tsk_used_math (like during threaded core dumping).
1412 * There is however an exception to this rule during ptrace
1413 * or during fork: the ptracer task is allowed to write to the
1414 * child->flags of its traced child (same goes for fork, the parent
1415 * can write to the child->flags), because we're guaranteed the
1416 * child is not running and in turn not changing child->flags
1417 * at the same time the parent does it.
1419 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1420 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1421 #define clear_used_math() clear_stopped_child_used_math(current)
1422 #define set_used_math() set_stopped_child_used_math(current)
1424 #define conditional_stopped_child_used_math(condition, child) \
1425 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1427 #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current)
1429 #define copy_to_stopped_child_used_math(child) \
1430 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1432 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1433 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1434 #define used_math() tsk_used_math(current)
1436 static inline bool is_percpu_thread(void)
1439 return (current->flags & PF_NO_SETAFFINITY) &&
1440 (current->nr_cpus_allowed == 1);
1446 /* Per-process atomic flags. */
1447 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
1448 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
1449 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
1450 #define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */
1451 #define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/
1452 #define PFA_SPEC_IB_DISABLE 5 /* Indirect branch speculation restricted */
1453 #define PFA_SPEC_IB_FORCE_DISABLE 6 /* Indirect branch speculation permanently restricted */
1454 #define PFA_SPEC_SSB_NOEXEC 7 /* Speculative Store Bypass clear on execve() */
1456 #define TASK_PFA_TEST(name, func) \
1457 static inline bool task_##func(struct task_struct *p) \
1458 { return test_bit(PFA_##name, &p->atomic_flags); }
1460 #define TASK_PFA_SET(name, func) \
1461 static inline void task_set_##func(struct task_struct *p) \
1462 { set_bit(PFA_##name, &p->atomic_flags); }
1464 #define TASK_PFA_CLEAR(name, func) \
1465 static inline void task_clear_##func(struct task_struct *p) \
1466 { clear_bit(PFA_##name, &p->atomic_flags); }
1468 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
1469 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
1471 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
1472 TASK_PFA_SET(SPREAD_PAGE, spread_page)
1473 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
1475 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
1476 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
1477 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
1479 TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable)
1480 TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable)
1481 TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable)
1483 TASK_PFA_TEST(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1484 TASK_PFA_SET(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1485 TASK_PFA_CLEAR(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1487 TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1488 TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1490 TASK_PFA_TEST(SPEC_IB_DISABLE, spec_ib_disable)
1491 TASK_PFA_SET(SPEC_IB_DISABLE, spec_ib_disable)
1492 TASK_PFA_CLEAR(SPEC_IB_DISABLE, spec_ib_disable)
1494 TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1495 TASK_PFA_SET(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1498 current_restore_flags(unsigned long orig_flags, unsigned long flags)
1500 current->flags &= ~flags;
1501 current->flags |= orig_flags & flags;
1504 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial);
1505 extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed);
1507 extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask);
1508 extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask);
1510 static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
1513 static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
1515 if (!cpumask_test_cpu(0, new_mask))
1521 #ifndef cpu_relax_yield
1522 #define cpu_relax_yield() cpu_relax()
1525 extern int yield_to(struct task_struct *p, bool preempt);
1526 extern void set_user_nice(struct task_struct *p, long nice);
1527 extern int task_prio(const struct task_struct *p);
1530 * task_nice - return the nice value of a given task.
1531 * @p: the task in question.
1533 * Return: The nice value [ -20 ... 0 ... 19 ].
1535 static inline int task_nice(const struct task_struct *p)
1537 return PRIO_TO_NICE((p)->static_prio);
1540 extern int can_nice(const struct task_struct *p, const int nice);
1541 extern int task_curr(const struct task_struct *p);
1542 extern int idle_cpu(int cpu);
1543 extern int available_idle_cpu(int cpu);
1544 extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *);
1545 extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *);
1546 extern int sched_setattr(struct task_struct *, const struct sched_attr *);
1547 extern int sched_setattr_nocheck(struct task_struct *, const struct sched_attr *);
1548 extern struct task_struct *idle_task(int cpu);
1551 * is_idle_task - is the specified task an idle task?
1552 * @p: the task in question.
1554 * Return: 1 if @p is an idle task. 0 otherwise.
1556 static inline bool is_idle_task(const struct task_struct *p)
1558 return !!(p->flags & PF_IDLE);
1561 extern struct task_struct *curr_task(int cpu);
1562 extern void ia64_set_curr_task(int cpu, struct task_struct *p);
1566 union thread_union {
1567 #ifndef CONFIG_ARCH_TASK_STRUCT_ON_STACK
1568 struct task_struct task;
1570 #ifndef CONFIG_THREAD_INFO_IN_TASK
1571 struct thread_info thread_info;
1573 unsigned long stack[THREAD_SIZE/sizeof(long)];
1576 #ifndef CONFIG_THREAD_INFO_IN_TASK
1577 extern struct thread_info init_thread_info;
1580 extern unsigned long init_stack[THREAD_SIZE / sizeof(unsigned long)];
1582 #ifdef CONFIG_THREAD_INFO_IN_TASK
1583 static inline struct thread_info *task_thread_info(struct task_struct *task)
1585 return &task->thread_info;
1587 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1588 # define task_thread_info(task) ((struct thread_info *)(task)->stack)
1592 * find a task by one of its numerical ids
1594 * find_task_by_pid_ns():
1595 * finds a task by its pid in the specified namespace
1596 * find_task_by_vpid():
1597 * finds a task by its virtual pid
1599 * see also find_vpid() etc in include/linux/pid.h
1602 extern struct task_struct *find_task_by_vpid(pid_t nr);
1603 extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns);
1606 * find a task by its virtual pid and get the task struct
1608 extern struct task_struct *find_get_task_by_vpid(pid_t nr);
1610 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1611 extern int wake_up_process(struct task_struct *tsk);
1612 extern void wake_up_new_task(struct task_struct *tsk);
1615 extern void kick_process(struct task_struct *tsk);
1617 static inline void kick_process(struct task_struct *tsk) { }
1620 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
1622 static inline void set_task_comm(struct task_struct *tsk, const char *from)
1624 __set_task_comm(tsk, from, false);
1627 extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk);
1628 #define get_task_comm(buf, tsk) ({ \
1629 BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \
1630 __get_task_comm(buf, sizeof(buf), tsk); \
1634 void scheduler_ipi(void);
1635 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1637 static inline void scheduler_ipi(void) { }
1638 static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state)
1645 * Set thread flags in other task's structures.
1646 * See asm/thread_info.h for TIF_xxxx flags available:
1648 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1650 set_ti_thread_flag(task_thread_info(tsk), flag);
1653 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1655 clear_ti_thread_flag(task_thread_info(tsk), flag);
1658 static inline void update_tsk_thread_flag(struct task_struct *tsk, int flag,
1661 update_ti_thread_flag(task_thread_info(tsk), flag, value);
1664 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1666 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1669 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1671 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1674 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1676 return test_ti_thread_flag(task_thread_info(tsk), flag);
1679 static inline void set_tsk_need_resched(struct task_struct *tsk)
1681 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1684 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1686 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1689 static inline int test_tsk_need_resched(struct task_struct *tsk)
1691 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
1695 * cond_resched() and cond_resched_lock(): latency reduction via
1696 * explicit rescheduling in places that are safe. The return
1697 * value indicates whether a reschedule was done in fact.
1698 * cond_resched_lock() will drop the spinlock before scheduling,
1700 #ifndef CONFIG_PREEMPT
1701 extern int _cond_resched(void);
1703 static inline int _cond_resched(void) { return 0; }
1706 #define cond_resched() ({ \
1707 ___might_sleep(__FILE__, __LINE__, 0); \
1711 extern int __cond_resched_lock(spinlock_t *lock);
1713 #define cond_resched_lock(lock) ({ \
1714 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
1715 __cond_resched_lock(lock); \
1718 static inline void cond_resched_rcu(void)
1720 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
1728 * Does a critical section need to be broken due to another
1729 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
1730 * but a general need for low latency)
1732 static inline int spin_needbreak(spinlock_t *lock)
1734 #ifdef CONFIG_PREEMPT
1735 return spin_is_contended(lock);
1741 static __always_inline bool need_resched(void)
1743 return unlikely(tif_need_resched());
1747 * Wrappers for p->thread_info->cpu access. No-op on UP.
1751 static inline unsigned int task_cpu(const struct task_struct *p)
1753 #ifdef CONFIG_THREAD_INFO_IN_TASK
1754 return READ_ONCE(p->cpu);
1756 return READ_ONCE(task_thread_info(p)->cpu);
1760 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1764 static inline unsigned int task_cpu(const struct task_struct *p)
1769 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1773 #endif /* CONFIG_SMP */
1776 * In order to reduce various lock holder preemption latencies provide an
1777 * interface to see if a vCPU is currently running or not.
1779 * This allows us to terminate optimistic spin loops and block, analogous to
1780 * the native optimistic spin heuristic of testing if the lock owner task is
1783 #ifndef vcpu_is_preempted
1784 # define vcpu_is_preempted(cpu) false
1787 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
1788 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
1790 #ifndef TASK_SIZE_OF
1791 #define TASK_SIZE_OF(tsk) TASK_SIZE
1797 * Map the event mask on the user-space ABI enum rseq_cs_flags
1798 * for direct mask checks.
1800 enum rseq_event_mask_bits {
1801 RSEQ_EVENT_PREEMPT_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT_BIT,
1802 RSEQ_EVENT_SIGNAL_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL_BIT,
1803 RSEQ_EVENT_MIGRATE_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE_BIT,
1806 enum rseq_event_mask {
1807 RSEQ_EVENT_PREEMPT = (1U << RSEQ_EVENT_PREEMPT_BIT),
1808 RSEQ_EVENT_SIGNAL = (1U << RSEQ_EVENT_SIGNAL_BIT),
1809 RSEQ_EVENT_MIGRATE = (1U << RSEQ_EVENT_MIGRATE_BIT),
1812 static inline void rseq_set_notify_resume(struct task_struct *t)
1815 set_tsk_thread_flag(t, TIF_NOTIFY_RESUME);
1818 void __rseq_handle_notify_resume(struct ksignal *sig, struct pt_regs *regs);
1820 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
1821 struct pt_regs *regs)
1824 __rseq_handle_notify_resume(ksig, regs);
1827 static inline void rseq_signal_deliver(struct ksignal *ksig,
1828 struct pt_regs *regs)
1831 __set_bit(RSEQ_EVENT_SIGNAL_BIT, ¤t->rseq_event_mask);
1833 rseq_handle_notify_resume(ksig, regs);
1836 /* rseq_preempt() requires preemption to be disabled. */
1837 static inline void rseq_preempt(struct task_struct *t)
1839 __set_bit(RSEQ_EVENT_PREEMPT_BIT, &t->rseq_event_mask);
1840 rseq_set_notify_resume(t);
1843 /* rseq_migrate() requires preemption to be disabled. */
1844 static inline void rseq_migrate(struct task_struct *t)
1846 __set_bit(RSEQ_EVENT_MIGRATE_BIT, &t->rseq_event_mask);
1847 rseq_set_notify_resume(t);
1851 * If parent process has a registered restartable sequences area, the
1852 * child inherits. Only applies when forking a process, not a thread.
1854 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
1856 if (clone_flags & CLONE_THREAD) {
1859 t->rseq_event_mask = 0;
1861 t->rseq = current->rseq;
1862 t->rseq_sig = current->rseq_sig;
1863 t->rseq_event_mask = current->rseq_event_mask;
1867 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)