1#ifndef _LINUX_SCHED_H
2#define _LINUX_SCHED_H
3
4/*
5 * Define 'struct task_struct' and provide the main scheduler
6 * APIs (schedule(), wakeup variants, etc.)
7 */
8
9#include <uapi/linux/sched.h>
10
11#include <asm/current.h>
12
13#include <linux/pid.h>
14#include <linux/sem.h>
15#include <linux/shm.h>
16#include <linux/kcov.h>
17#include <linux/mutex.h>
18#include <linux/plist.h>
19#include <linux/hrtimer.h>
20#include <linux/seccomp.h>
21#include <linux/nodemask.h>
22#include <linux/rcupdate.h>
23#include <linux/resource.h>
24#include <linux/latencytop.h>
25#include <linux/sched/prio.h>
26#include <linux/signal_types.h>
27#include <linux/mm_types_task.h>
28#include <linux/task_io_accounting.h>
29
30/* task_struct member predeclarations (sorted alphabetically): */
31struct audit_context;
32struct backing_dev_info;
33struct bio_list;
34struct blk_plug;
35struct cfs_rq;
36struct fs_struct;
37struct futex_pi_state;
38struct io_context;
39struct mempolicy;
40struct nameidata;
41struct nsproxy;
42struct perf_event_context;
43struct pid_namespace;
44struct pipe_inode_info;
45struct rcu_node;
46struct reclaim_state;
47struct robust_list_head;
48struct sched_attr;
49struct sched_param;
50struct seq_file;
51struct sighand_struct;
52struct signal_struct;
53struct task_delay_info;
54struct task_group;
55
56/*
57 * Task state bitmask. NOTE! These bits are also
58 * encoded in fs/proc/array.c: get_task_state().
59 *
60 * We have two separate sets of flags: task->state
61 * is about runnability, while task->exit_state are
62 * about the task exiting. Confusing, but this way
63 * modifying one set can't modify the other one by
64 * mistake.
65 */
66
67/* Used in tsk->state: */
68#define TASK_RUNNING 0
69#define TASK_INTERRUPTIBLE 1
70#define TASK_UNINTERRUPTIBLE 2
71#define __TASK_STOPPED 4
72#define __TASK_TRACED 8
73/* Used in tsk->exit_state: */
74#define EXIT_DEAD 16
75#define EXIT_ZOMBIE 32
76#define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
77/* Used in tsk->state again: */
78#define TASK_DEAD 64
79#define TASK_WAKEKILL 128
80#define TASK_WAKING 256
81#define TASK_PARKED 512
82#define TASK_NOLOAD 1024
83#define TASK_NEW 2048
84#define TASK_STATE_MAX 4096
85
86#define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"
87
88/* Convenience macros for the sake of set_current_state: */
89#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
90#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
91#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
92
93#define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
94
95/* Convenience macros for the sake of wake_up(): */
96#define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
97#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
98
99/* get_task_state(): */
100#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
101 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
102 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
103
104#define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
105
106#define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
107
108#define task_is_stopped_or_traced(task) ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
109
110#define task_contributes_to_load(task) ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
111 (task->flags & PF_FROZEN) == 0 && \
112 (task->state & TASK_NOLOAD) == 0)
113
114#ifdef CONFIG_DEBUG_ATOMIC_SLEEP
115
116#define __set_current_state(state_value) \
117 do { \
118 current->task_state_change = _THIS_IP_; \
119 current->state = (state_value); \
120 } while (0)
121#define set_current_state(state_value) \
122 do { \
123 current->task_state_change = _THIS_IP_; \
124 smp_store_mb(current->state, (state_value)); \
125 } while (0)
126
127#else
128/*
129 * set_current_state() includes a barrier so that the write of current->state
130 * is correctly serialised wrt the caller's subsequent test of whether to
131 * actually sleep:
132 *
133 * for (;;) {
134 * set_current_state(TASK_UNINTERRUPTIBLE);
135 * if (!need_sleep)
136 * break;
137 *
138 * schedule();
139 * }
140 * __set_current_state(TASK_RUNNING);
141 *
142 * If the caller does not need such serialisation (because, for instance, the
143 * condition test and condition change and wakeup are under the same lock) then
144 * use __set_current_state().
145 *
146 * The above is typically ordered against the wakeup, which does:
147 *
148 * need_sleep = false;
149 * wake_up_state(p, TASK_UNINTERRUPTIBLE);
150 *
151 * Where wake_up_state() (and all other wakeup primitives) imply enough
152 * barriers to order the store of the variable against wakeup.
153 *
154 * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
155 * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
156 * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
157 *
158 * This is obviously fine, since they both store the exact same value.
159 *
160 * Also see the comments of try_to_wake_up().
161 */
162#define __set_current_state(state_value) do { current->state = (state_value); } while (0)
163#define set_current_state(state_value) smp_store_mb(current->state, (state_value))
164#endif
165
166/* Task command name length: */
167#define TASK_COMM_LEN 16
168
169extern cpumask_var_t cpu_isolated_map;
170
171extern void scheduler_tick(void);
172
173#define MAX_SCHEDULE_TIMEOUT LONG_MAX
174
175extern long schedule_timeout(long timeout);
176extern long schedule_timeout_interruptible(long timeout);
177extern long schedule_timeout_killable(long timeout);
178extern long schedule_timeout_uninterruptible(long timeout);
179extern long schedule_timeout_idle(long timeout);
180asmlinkage void schedule(void);
181extern void schedule_preempt_disabled(void);
182
183extern int __must_check io_schedule_prepare(void);
184extern void io_schedule_finish(int token);
185extern long io_schedule_timeout(long timeout);
186extern void io_schedule(void);
187
188/**
189 * struct prev_cputime - snaphsot of system and user cputime
190 * @utime: time spent in user mode
191 * @stime: time spent in system mode
192 * @lock: protects the above two fields
193 *
194 * Stores previous user/system time values such that we can guarantee
195 * monotonicity.
196 */
197struct prev_cputime {
198#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
199 u64 utime;
200 u64 stime;
201 raw_spinlock_t lock;
202#endif
203};
204
205/**
206 * struct task_cputime - collected CPU time counts
207 * @utime: time spent in user mode, in nanoseconds
208 * @stime: time spent in kernel mode, in nanoseconds
209 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
210 *
211 * This structure groups together three kinds of CPU time that are tracked for
212 * threads and thread groups. Most things considering CPU time want to group
213 * these counts together and treat all three of them in parallel.
214 */
215struct task_cputime {
216 u64 utime;
217 u64 stime;
218 unsigned long long sum_exec_runtime;
219};
220
221/* Alternate field names when used on cache expirations: */
222#define virt_exp utime
223#define prof_exp stime
224#define sched_exp sum_exec_runtime
225
226struct sched_info {
227#ifdef CONFIG_SCHED_INFO
228 /* Cumulative counters: */
229
230 /* # of times we have run on this CPU: */
231 unsigned long pcount;
232
233 /* Time spent waiting on a runqueue: */
234 unsigned long long run_delay;
235
236 /* Timestamps: */
237
238 /* When did we last run on a CPU? */
239 unsigned long long last_arrival;
240
241 /* When were we last queued to run? */
242 unsigned long long last_queued;
243
244#endif /* CONFIG_SCHED_INFO */
245};
246
247/*
248 * Integer metrics need fixed point arithmetic, e.g., sched/fair
249 * has a few: load, load_avg, util_avg, freq, and capacity.
250 *
251 * We define a basic fixed point arithmetic range, and then formalize
252 * all these metrics based on that basic range.
253 */
254# define SCHED_FIXEDPOINT_SHIFT 10
255# define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
256
257struct load_weight {
258 unsigned long weight;
259 u32 inv_weight;
260};
261
262/*
263 * The load_avg/util_avg accumulates an infinite geometric series
264 * (see __update_load_avg() in kernel/sched/fair.c).
265 *
266 * [load_avg definition]
267 *
268 * load_avg = runnable% * scale_load_down(load)
269 *
270 * where runnable% is the time ratio that a sched_entity is runnable.
271 * For cfs_rq, it is the aggregated load_avg of all runnable and
272 * blocked sched_entities.
273 *
274 * load_avg may also take frequency scaling into account:
275 *
276 * load_avg = runnable% * scale_load_down(load) * freq%
277 *
278 * where freq% is the CPU frequency normalized to the highest frequency.
279 *
280 * [util_avg definition]
281 *
282 * util_avg = running% * SCHED_CAPACITY_SCALE
283 *
284 * where running% is the time ratio that a sched_entity is running on
285 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
286 * and blocked sched_entities.
287 *
288 * util_avg may also factor frequency scaling and CPU capacity scaling:
289 *
290 * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
291 *
292 * where freq% is the same as above, and capacity% is the CPU capacity
293 * normalized to the greatest capacity (due to uarch differences, etc).
294 *
295 * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
296 * themselves are in the range of [0, 1]. To do fixed point arithmetics,
297 * we therefore scale them to as large a range as necessary. This is for
298 * example reflected by util_avg's SCHED_CAPACITY_SCALE.
299 *
300 * [Overflow issue]
301 *
302 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
303 * with the highest load (=88761), always runnable on a single cfs_rq,
304 * and should not overflow as the number already hits PID_MAX_LIMIT.
305 *
306 * For all other cases (including 32-bit kernels), struct load_weight's
307 * weight will overflow first before we do, because:
308 *
309 * Max(load_avg) <= Max(load.weight)
310 *
311 * Then it is the load_weight's responsibility to consider overflow
312 * issues.
313 */
314struct sched_avg {
315 u64 last_update_time;
316 u64 load_sum;
317 u32 util_sum;
318 u32 period_contrib;
319 unsigned long load_avg;
320 unsigned long util_avg;
321};
322
323struct sched_statistics {
324#ifdef CONFIG_SCHEDSTATS
325 u64 wait_start;
326 u64 wait_max;
327 u64 wait_count;
328 u64 wait_sum;
329 u64 iowait_count;
330 u64 iowait_sum;
331
332 u64 sleep_start;
333 u64 sleep_max;
334 s64 sum_sleep_runtime;
335
336 u64 block_start;
337 u64 block_max;
338 u64 exec_max;
339 u64 slice_max;
340
341 u64 nr_migrations_cold;
342 u64 nr_failed_migrations_affine;
343 u64 nr_failed_migrations_running;
344 u64 nr_failed_migrations_hot;
345 u64 nr_forced_migrations;
346
347 u64 nr_wakeups;
348 u64 nr_wakeups_sync;
349 u64 nr_wakeups_migrate;
350 u64 nr_wakeups_local;
351 u64 nr_wakeups_remote;
352 u64 nr_wakeups_affine;
353 u64 nr_wakeups_affine_attempts;
354 u64 nr_wakeups_passive;
355 u64 nr_wakeups_idle;
356#endif
357};
358
359struct sched_entity {
360 /* For load-balancing: */
361 struct load_weight load;
362 struct rb_node run_node;
363 struct list_head group_node;
364 unsigned int on_rq;
365
366 u64 exec_start;
367 u64 sum_exec_runtime;
368 u64 vruntime;
369 u64 prev_sum_exec_runtime;
370
371 u64 nr_migrations;
372
373 struct sched_statistics statistics;
374
375#ifdef CONFIG_FAIR_GROUP_SCHED
376 int depth;
377 struct sched_entity *parent;
378 /* rq on which this entity is (to be) queued: */
379 struct cfs_rq *cfs_rq;
380 /* rq "owned" by this entity/group: */
381 struct cfs_rq *my_q;
382#endif
383
384#ifdef CONFIG_SMP
385 /*
386 * Per entity load average tracking.
387 *
388 * Put into separate cache line so it does not
389 * collide with read-mostly values above.
390 */
391 struct sched_avg avg ____cacheline_aligned_in_smp;
392#endif
393};
394
395struct sched_rt_entity {
396 struct list_head run_list;
397 unsigned long timeout;
398 unsigned long watchdog_stamp;
399 unsigned int time_slice;
400 unsigned short on_rq;
401 unsigned short on_list;
402
403 struct sched_rt_entity *back;
404#ifdef CONFIG_RT_GROUP_SCHED
405 struct sched_rt_entity *parent;
406 /* rq on which this entity is (to be) queued: */
407 struct rt_rq *rt_rq;
408 /* rq "owned" by this entity/group: */
409 struct rt_rq *my_q;
410#endif
411};
412
413struct sched_dl_entity {
414 struct rb_node rb_node;
415
416 /*
417 * Original scheduling parameters. Copied here from sched_attr
418 * during sched_setattr(), they will remain the same until
419 * the next sched_setattr().
420 */
421 u64 dl_runtime; /* Maximum runtime for each instance */
422 u64 dl_deadline; /* Relative deadline of each instance */
423 u64 dl_period; /* Separation of two instances (period) */
424 u64 dl_bw; /* dl_runtime / dl_deadline */
425
426 /*
427 * Actual scheduling parameters. Initialized with the values above,
428 * they are continously updated during task execution. Note that
429 * the remaining runtime could be < 0 in case we are in overrun.
430 */
431 s64 runtime; /* Remaining runtime for this instance */
432 u64 deadline; /* Absolute deadline for this instance */
433 unsigned int flags; /* Specifying the scheduler behaviour */
434
435 /*
436 * Some bool flags:
437 *
438 * @dl_throttled tells if we exhausted the runtime. If so, the
439 * task has to wait for a replenishment to be performed at the
440 * next firing of dl_timer.
441 *
442 * @dl_boosted tells if we are boosted due to DI. If so we are
443 * outside bandwidth enforcement mechanism (but only until we
444 * exit the critical section);
445 *
446 * @dl_yielded tells if task gave up the CPU before consuming
447 * all its available runtime during the last job.
448 */
449 int dl_throttled;
450 int dl_boosted;
451 int dl_yielded;
452
453 /*
454 * Bandwidth enforcement timer. Each -deadline task has its
455 * own bandwidth to be enforced, thus we need one timer per task.
456 */
457 struct hrtimer dl_timer;
458};
459
460union rcu_special {
461 struct {
462 u8 blocked;
463 u8 need_qs;
464 u8 exp_need_qs;
465
466 /* Otherwise the compiler can store garbage here: */
467 u8 pad;
468 } b; /* Bits. */
469 u32 s; /* Set of bits. */
470};
471
472enum perf_event_task_context {
473 perf_invalid_context = -1,
474 perf_hw_context = 0,
475 perf_sw_context,
476 perf_nr_task_contexts,
477};
478
479struct wake_q_node {
480 struct wake_q_node *next;
481};
482
483struct task_struct {
484#ifdef CONFIG_THREAD_INFO_IN_TASK
485 /*
486 * For reasons of header soup (see current_thread_info()), this
487 * must be the first element of task_struct.
488 */
489 struct thread_info thread_info;
490#endif
491 /* -1 unrunnable, 0 runnable, >0 stopped: */
492 volatile long state;
493 void *stack;
494 atomic_t usage;
495 /* Per task flags (PF_*), defined further below: */
496 unsigned int flags;
497 unsigned int ptrace;
498
499#ifdef CONFIG_SMP
500 struct llist_node wake_entry;
501 int on_cpu;
502#ifdef CONFIG_THREAD_INFO_IN_TASK
503 /* Current CPU: */
504 unsigned int cpu;
505#endif
506 unsigned int wakee_flips;
507 unsigned long wakee_flip_decay_ts;
508 struct task_struct *last_wakee;
509
510 int wake_cpu;
511#endif
512 int on_rq;
513
514 int prio;
515 int static_prio;
516 int normal_prio;
517 unsigned int rt_priority;
518
519 const struct sched_class *sched_class;
520 struct sched_entity se;
521 struct sched_rt_entity rt;
522#ifdef CONFIG_CGROUP_SCHED
523 struct task_group *sched_task_group;
524#endif
525 struct sched_dl_entity dl;
526
527#ifdef CONFIG_PREEMPT_NOTIFIERS
528 /* List of struct preempt_notifier: */
529 struct hlist_head preempt_notifiers;
530#endif
531
532#ifdef CONFIG_BLK_DEV_IO_TRACE
533 unsigned int btrace_seq;
534#endif
535
536 unsigned int policy;
537 int nr_cpus_allowed;
538 cpumask_t cpus_allowed;
539
540#ifdef CONFIG_PREEMPT_RCU
541 int rcu_read_lock_nesting;
542 union rcu_special rcu_read_unlock_special;
543 struct list_head rcu_node_entry;
544 struct rcu_node *rcu_blocked_node;
545#endif /* #ifdef CONFIG_PREEMPT_RCU */
546
547#ifdef CONFIG_TASKS_RCU
548 unsigned long rcu_tasks_nvcsw;
549 bool rcu_tasks_holdout;
550 struct list_head rcu_tasks_holdout_list;
551 int rcu_tasks_idle_cpu;
552#endif /* #ifdef CONFIG_TASKS_RCU */
553
554 struct sched_info sched_info;
555
556 struct list_head tasks;
557#ifdef CONFIG_SMP
558 struct plist_node pushable_tasks;
559 struct rb_node pushable_dl_tasks;
560#endif
561
562 struct mm_struct *mm;
563 struct mm_struct *active_mm;
564
565 /* Per-thread vma caching: */
566 struct vmacache vmacache;
567
568#ifdef SPLIT_RSS_COUNTING
569 struct task_rss_stat rss_stat;
570#endif
571 int exit_state;
572 int exit_code;
573 int exit_signal;
574 /* The signal sent when the parent dies: */
575 int pdeath_signal;
576 /* JOBCTL_*, siglock protected: */
577 unsigned long jobctl;
578
579 /* Used for emulating ABI behavior of previous Linux versions: */
580 unsigned int personality;
581
582 /* Scheduler bits, serialized by scheduler locks: */
583 unsigned sched_reset_on_fork:1;
584 unsigned sched_contributes_to_load:1;
585 unsigned sched_migrated:1;
586 unsigned sched_remote_wakeup:1;
587 /* Force alignment to the next boundary: */
588 unsigned :0;
589
590 /* Unserialized, strictly 'current' */
591
592 /* Bit to tell LSMs we're in execve(): */
593 unsigned in_execve:1;
594 unsigned in_iowait:1;
595#ifndef TIF_RESTORE_SIGMASK
596 unsigned restore_sigmask:1;
597#endif
598#ifdef CONFIG_MEMCG
599 unsigned memcg_may_oom:1;
600#ifndef CONFIG_SLOB
601 unsigned memcg_kmem_skip_account:1;
602#endif
603#endif
604#ifdef CONFIG_COMPAT_BRK
605 unsigned brk_randomized:1;
606#endif
607#ifdef CONFIG_CGROUPS
608 /* disallow userland-initiated cgroup migration */
609 unsigned no_cgroup_migration:1;
610#endif
611
612 unsigned long atomic_flags; /* Flags requiring atomic access. */
613
614 struct restart_block restart_block;
615
616 pid_t pid;
617 pid_t tgid;
618
619#ifdef CONFIG_CC_STACKPROTECTOR
620 /* Canary value for the -fstack-protector GCC feature: */
621 unsigned long stack_canary;
622#endif
623 /*
624 * Pointers to the (original) parent process, youngest child, younger sibling,
625 * older sibling, respectively. (p->father can be replaced with
626 * p->real_parent->pid)
627 */
628
629 /* Real parent process: */
630 struct task_struct __rcu *real_parent;
631
632 /* Recipient of SIGCHLD, wait4() reports: */
633 struct task_struct __rcu *parent;
634
635 /*
636 * Children/sibling form the list of natural children:
637 */
638 struct list_head children;
639 struct list_head sibling;
640 struct task_struct *group_leader;
641
642 /*
643 * 'ptraced' is the list of tasks this task is using ptrace() on.
644 *
645 * This includes both natural children and PTRACE_ATTACH targets.
646 * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
647 */
648 struct list_head ptraced;
649 struct list_head ptrace_entry;
650
651 /* PID/PID hash table linkage. */
652 struct pid_link pids[PIDTYPE_MAX];
653 struct list_head thread_group;
654 struct list_head thread_node;
655
656 struct completion *vfork_done;
657
658 /* CLONE_CHILD_SETTID: */
659 int __user *set_child_tid;
660
661 /* CLONE_CHILD_CLEARTID: */
662 int __user *clear_child_tid;
663
664 u64 utime;
665 u64 stime;
666#ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
667 u64 utimescaled;
668 u64 stimescaled;
669#endif
670 u64 gtime;
671 struct prev_cputime prev_cputime;
672#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
673 seqcount_t vtime_seqcount;
674 unsigned long long vtime_snap;
675 enum {
676 /* Task is sleeping or running in a CPU with VTIME inactive: */
677 VTIME_INACTIVE = 0,
678 /* Task runs in userspace in a CPU with VTIME active: */
679 VTIME_USER,
680 /* Task runs in kernelspace in a CPU with VTIME active: */
681 VTIME_SYS,
682 } vtime_snap_whence;
683#endif
684
685#ifdef CONFIG_NO_HZ_FULL
686 atomic_t tick_dep_mask;
687#endif
688 /* Context switch counts: */
689 unsigned long nvcsw;
690 unsigned long nivcsw;
691
692 /* Monotonic time in nsecs: */
693 u64 start_time;
694
695 /* Boot based time in nsecs: */
696 u64 real_start_time;
697
698 /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
699 unsigned long min_flt;
700 unsigned long maj_flt;
701
702#ifdef CONFIG_POSIX_TIMERS
703 struct task_cputime cputime_expires;
704 struct list_head cpu_timers[3];
705#endif
706
707 /* Process credentials: */
708
709 /* Tracer's credentials at attach: */
710 const struct cred __rcu *ptracer_cred;
711
712 /* Objective and real subjective task credentials (COW): */
713 const struct cred __rcu *real_cred;
714
715 /* Effective (overridable) subjective task credentials (COW): */
716 const struct cred __rcu *cred;
717
718 /*
719 * executable name, excluding path.
720 *
721 * - normally initialized setup_new_exec()
722 * - access it with [gs]et_task_comm()
723 * - lock it with task_lock()
724 */
725 char comm[TASK_COMM_LEN];
726
727 struct nameidata *nameidata;
728
729#ifdef CONFIG_SYSVIPC
730 struct sysv_sem sysvsem;
731 struct sysv_shm sysvshm;
732#endif
733#ifdef CONFIG_DETECT_HUNG_TASK
734 unsigned long last_switch_count;
735#endif
736 /* Filesystem information: */
737 struct fs_struct *fs;
738
739 /* Open file information: */
740 struct files_struct *files;
741
742 /* Namespaces: */
743 struct nsproxy *nsproxy;
744
745 /* Signal handlers: */
746 struct signal_struct *signal;
747 struct sighand_struct *sighand;
748 sigset_t blocked;
749 sigset_t real_blocked;
750 /* Restored if set_restore_sigmask() was used: */
751 sigset_t saved_sigmask;
752 struct sigpending pending;
753 unsigned long sas_ss_sp;
754 size_t sas_ss_size;
755 unsigned int sas_ss_flags;
756
757 struct callback_head *task_works;
758
759 struct audit_context *audit_context;
760#ifdef CONFIG_AUDITSYSCALL
761 kuid_t loginuid;
762 unsigned int sessionid;
763#endif
764 struct seccomp seccomp;
765
766 /* Thread group tracking: */
767 u32 parent_exec_id;
768 u32 self_exec_id;
769
770 /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
771 spinlock_t alloc_lock;
772
773 /* Protection of the PI data structures: */
774 raw_spinlock_t pi_lock;
775
776 struct wake_q_node wake_q;
777
778#ifdef CONFIG_RT_MUTEXES
779 /* PI waiters blocked on a rt_mutex held by this task: */
780 struct rb_root pi_waiters;
781 struct rb_node *pi_waiters_leftmost;
782 /* Deadlock detection and priority inheritance handling: */
783 struct rt_mutex_waiter *pi_blocked_on;
784#endif
785
786#ifdef CONFIG_DEBUG_MUTEXES
787 /* Mutex deadlock detection: */
788 struct mutex_waiter *blocked_on;
789#endif
790
791#ifdef CONFIG_TRACE_IRQFLAGS
792 unsigned int irq_events;
793 unsigned long hardirq_enable_ip;
794 unsigned long hardirq_disable_ip;
795 unsigned int hardirq_enable_event;
796 unsigned int hardirq_disable_event;
797 int hardirqs_enabled;
798 int hardirq_context;
799 unsigned long softirq_disable_ip;
800 unsigned long softirq_enable_ip;
801 unsigned int softirq_disable_event;
802 unsigned int softirq_enable_event;
803 int softirqs_enabled;
804 int softirq_context;
805#endif
806
807#ifdef CONFIG_LOCKDEP
808# define MAX_LOCK_DEPTH 48UL
809 u64 curr_chain_key;
810 int lockdep_depth;
811 unsigned int lockdep_recursion;
812 struct held_lock held_locks[MAX_LOCK_DEPTH];
813 gfp_t lockdep_reclaim_gfp;
814#endif
815
816#ifdef CONFIG_UBSAN
817 unsigned int in_ubsan;
818#endif
819
820 /* Journalling filesystem info: */
821 void *journal_info;
822
823 /* Stacked block device info: */
824 struct bio_list *bio_list;
825
826#ifdef CONFIG_BLOCK
827 /* Stack plugging: */
828 struct blk_plug *plug;
829#endif
830
831 /* VM state: */
832 struct reclaim_state *reclaim_state;
833
834 struct backing_dev_info *backing_dev_info;
835
836 struct io_context *io_context;
837
838 /* Ptrace state: */
839 unsigned long ptrace_message;
840 siginfo_t *last_siginfo;
841
842 struct task_io_accounting ioac;
843#ifdef CONFIG_TASK_XACCT
844 /* Accumulated RSS usage: */
845 u64 acct_rss_mem1;
846 /* Accumulated virtual memory usage: */
847 u64 acct_vm_mem1;
848 /* stime + utime since last update: */
849 u64 acct_timexpd;
850#endif
851#ifdef CONFIG_CPUSETS
852 /* Protected by ->alloc_lock: */
853 nodemask_t mems_allowed;
854 /* Seqence number to catch updates: */
855 seqcount_t mems_allowed_seq;
856 int cpuset_mem_spread_rotor;
857 int cpuset_slab_spread_rotor;
858#endif
859#ifdef CONFIG_CGROUPS
860 /* Control Group info protected by css_set_lock: */
861 struct css_set __rcu *cgroups;
862 /* cg_list protected by css_set_lock and tsk->alloc_lock: */
863 struct list_head cg_list;
864#endif
865#ifdef CONFIG_INTEL_RDT_A
866 int closid;
867#endif
868#ifdef CONFIG_FUTEX
869 struct robust_list_head __user *robust_list;
870#ifdef CONFIG_COMPAT
871 struct compat_robust_list_head __user *compat_robust_list;
872#endif
873 struct list_head pi_state_list;
874 struct futex_pi_state *pi_state_cache;
875#endif
876#ifdef CONFIG_PERF_EVENTS
877 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
878 struct mutex perf_event_mutex;
879 struct list_head perf_event_list;
880#endif
881#ifdef CONFIG_DEBUG_PREEMPT
882 unsigned long preempt_disable_ip;
883#endif
884#ifdef CONFIG_NUMA
885 /* Protected by alloc_lock: */
886 struct mempolicy *mempolicy;
887 short il_next;
888 short pref_node_fork;
889#endif
890#ifdef CONFIG_NUMA_BALANCING
891 int numa_scan_seq;
892 unsigned int numa_scan_period;
893 unsigned int numa_scan_period_max;
894 int numa_preferred_nid;
895 unsigned long numa_migrate_retry;
896 /* Migration stamp: */
897 u64 node_stamp;
898 u64 last_task_numa_placement;
899 u64 last_sum_exec_runtime;
900 struct callback_head numa_work;
901
902 struct list_head numa_entry;
903 struct numa_group *numa_group;
904
905 /*
906 * numa_faults is an array split into four regions:
907 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
908 * in this precise order.
909 *
910 * faults_memory: Exponential decaying average of faults on a per-node
911 * basis. Scheduling placement decisions are made based on these
912 * counts. The values remain static for the duration of a PTE scan.
913 * faults_cpu: Track the nodes the process was running on when a NUMA
914 * hinting fault was incurred.
915 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
916 * during the current scan window. When the scan completes, the counts
917 * in faults_memory and faults_cpu decay and these values are copied.
918 */
919 unsigned long *numa_faults;
920 unsigned long total_numa_faults;
921
922 /*
923 * numa_faults_locality tracks if faults recorded during the last
924 * scan window were remote/local or failed to migrate. The task scan
925 * period is adapted based on the locality of the faults with different
926 * weights depending on whether they were shared or private faults
927 */
928 unsigned long numa_faults_locality[3];
929
930 unsigned long numa_pages_migrated;
931#endif /* CONFIG_NUMA_BALANCING */
932
933 struct tlbflush_unmap_batch tlb_ubc;
934
935 struct rcu_head rcu;
936
937 /* Cache last used pipe for splice(): */
938 struct pipe_inode_info *splice_pipe;
939
940 struct page_frag task_frag;
941
942#ifdef CONFIG_TASK_DELAY_ACCT
943 struct task_delay_info *delays;
944#endif
945
946#ifdef CONFIG_FAULT_INJECTION
947 int make_it_fail;
948#endif
949 /*
950 * When (nr_dirtied >= nr_dirtied_pause), it's time to call
951 * balance_dirty_pages() for a dirty throttling pause:
952 */
953 int nr_dirtied;
954 int nr_dirtied_pause;
955 /* Start of a write-and-pause period: */
956 unsigned long dirty_paused_when;
957
958#ifdef CONFIG_LATENCYTOP
959 int latency_record_count;
960 struct latency_record latency_record[LT_SAVECOUNT];
961#endif
962 /*
963 * Time slack values; these are used to round up poll() and
964 * select() etc timeout values. These are in nanoseconds.
965 */
966 u64 timer_slack_ns;
967 u64 default_timer_slack_ns;
968
969#ifdef CONFIG_KASAN
970 unsigned int kasan_depth;
971#endif
972
973#ifdef CONFIG_FUNCTION_GRAPH_TRACER
974 /* Index of current stored address in ret_stack: */
975 int curr_ret_stack;
976
977 /* Stack of return addresses for return function tracing: */
978 struct ftrace_ret_stack *ret_stack;
979
980 /* Timestamp for last schedule: */
981 unsigned long long ftrace_timestamp;
982
983 /*
984 * Number of functions that haven't been traced
985 * because of depth overrun:
986 */
987 atomic_t trace_overrun;
988
989 /* Pause tracing: */
990 atomic_t tracing_graph_pause;
991#endif
992
993#ifdef CONFIG_TRACING
994 /* State flags for use by tracers: */
995 unsigned long trace;
996
997 /* Bitmask and counter of trace recursion: */
998 unsigned long trace_recursion;
999#endif /* CONFIG_TRACING */
1000
1001#ifdef CONFIG_KCOV
1002 /* Coverage collection mode enabled for this task (0 if disabled): */
1003 enum kcov_mode kcov_mode;
1004
1005 /* Size of the kcov_area: */
1006 unsigned int kcov_size;
1007
1008 /* Buffer for coverage collection: */
1009 void *kcov_area;
1010
1011 /* KCOV descriptor wired with this task or NULL: */
1012 struct kcov *kcov;
1013#endif
1014
1015#ifdef CONFIG_MEMCG
1016 struct mem_cgroup *memcg_in_oom;
1017 gfp_t memcg_oom_gfp_mask;
1018 int memcg_oom_order;
1019
1020 /* Number of pages to reclaim on returning to userland: */
1021 unsigned int memcg_nr_pages_over_high;
1022#endif
1023
1024#ifdef CONFIG_UPROBES
1025 struct uprobe_task *utask;
1026#endif
1027#if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1028 unsigned int sequential_io;
1029 unsigned int sequential_io_avg;
1030#endif
1031#ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1032 unsigned long task_state_change;
1033#endif
1034 int pagefault_disabled;
1035#ifdef CONFIG_MMU
1036 struct task_struct *oom_reaper_list;
1037#endif
1038#ifdef CONFIG_VMAP_STACK
1039 struct vm_struct *stack_vm_area;
1040#endif
1041#ifdef CONFIG_THREAD_INFO_IN_TASK
1042 /* A live task holds one reference: */
1043 atomic_t stack_refcount;
1044#endif
1045 /* CPU-specific state of this task: */
1046 struct thread_struct thread;
1047
1048 /*
1049 * WARNING: on x86, 'thread_struct' contains a variable-sized
1050 * structure. It *MUST* be at the end of 'task_struct'.
1051 *
1052 * Do not put anything below here!
1053 */
1054};
1055
1056static inline struct pid *task_pid(struct task_struct *task)
1057{
1058 return task->pids[PIDTYPE_PID].pid;
1059}
1060
1061static inline struct pid *task_tgid(struct task_struct *task)
1062{
1063 return task->group_leader->pids[PIDTYPE_PID].pid;
1064}
1065
1066/*
1067 * Without tasklist or RCU lock it is not safe to dereference
1068 * the result of task_pgrp/task_session even if task == current,
1069 * we can race with another thread doing sys_setsid/sys_setpgid.
1070 */
1071static inline struct pid *task_pgrp(struct task_struct *task)
1072{
1073 return task->group_leader->pids[PIDTYPE_PGID].pid;
1074}
1075
1076static inline struct pid *task_session(struct task_struct *task)
1077{
1078 return task->group_leader->pids[PIDTYPE_SID].pid;
1079}
1080
1081/*
1082 * the helpers to get the task's different pids as they are seen
1083 * from various namespaces
1084 *
1085 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1086 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1087 * current.
1088 * task_xid_nr_ns() : id seen from the ns specified;
1089 *
1090 * set_task_vxid() : assigns a virtual id to a task;
1091 *
1092 * see also pid_nr() etc in include/linux/pid.h
1093 */
1094pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns);
1095
1096static inline pid_t task_pid_nr(struct task_struct *tsk)
1097{
1098 return tsk->pid;
1099}
1100
1101static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1102{
1103 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1104}
1105
1106static inline pid_t task_pid_vnr(struct task_struct *tsk)
1107{
1108 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1109}
1110
1111
1112static inline pid_t task_tgid_nr(struct task_struct *tsk)
1113{
1114 return tsk->tgid;
1115}
1116
1117extern pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1118
1119static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1120{
1121 return pid_vnr(task_tgid(tsk));
1122}
1123
1124/**
1125 * pid_alive - check that a task structure is not stale
1126 * @p: Task structure to be checked.
1127 *
1128 * Test if a process is not yet dead (at most zombie state)
1129 * If pid_alive fails, then pointers within the task structure
1130 * can be stale and must not be dereferenced.
1131 *
1132 * Return: 1 if the process is alive. 0 otherwise.
1133 */
1134static inline int pid_alive(const struct task_struct *p)
1135{
1136 return p->pids[PIDTYPE_PID].pid != NULL;
1137}
1138
1139static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1140{
1141 pid_t pid = 0;
1142
1143 rcu_read_lock();
1144 if (pid_alive(tsk))
1145 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1146 rcu_read_unlock();
1147
1148 return pid;
1149}
1150
1151static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1152{
1153 return task_ppid_nr_ns(tsk, &init_pid_ns);
1154}
1155
1156static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1157{
1158 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1159}
1160
1161static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1162{
1163 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1164}
1165
1166
1167static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1168{
1169 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1170}
1171
1172static inline pid_t task_session_vnr(struct task_struct *tsk)
1173{
1174 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1175}
1176
1177/* Obsolete, do not use: */
1178static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1179{
1180 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1181}
1182
1183/**
1184 * is_global_init - check if a task structure is init. Since init
1185 * is free to have sub-threads we need to check tgid.
1186 * @tsk: Task structure to be checked.
1187 *
1188 * Check if a task structure is the first user space task the kernel created.
1189 *
1190 * Return: 1 if the task structure is init. 0 otherwise.
1191 */
1192static inline int is_global_init(struct task_struct *tsk)
1193{
1194 return task_tgid_nr(tsk) == 1;
1195}
1196
1197extern struct pid *cad_pid;
1198
1199/*
1200 * Per process flags
1201 */
1202#define PF_IDLE 0x00000002 /* I am an IDLE thread */
1203#define PF_EXITING 0x00000004 /* Getting shut down */
1204#define PF_EXITPIDONE 0x00000008 /* PI exit done on shut down */
1205#define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1206#define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1207#define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */
1208#define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */
1209#define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */
1210#define PF_DUMPCORE 0x00000200 /* Dumped core */
1211#define PF_SIGNALED 0x00000400 /* Killed by a signal */
1212#define PF_MEMALLOC 0x00000800 /* Allocating memory */
1213#define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */
1214#define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */
1215#define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */
1216#define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */
1217#define PF_FROZEN 0x00010000 /* Frozen for system suspend */
1218#define PF_FSTRANS 0x00020000 /* Inside a filesystem transaction */
1219#define PF_KSWAPD 0x00040000 /* I am kswapd */
1220#define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1221#define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1222#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1223#define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */
1224#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1225#define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1226#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1227#define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1228#define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1229#define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */
1230
1231/*
1232 * Only the _current_ task can read/write to tsk->flags, but other
1233 * tasks can access tsk->flags in readonly mode for example
1234 * with tsk_used_math (like during threaded core dumping).
1235 * There is however an exception to this rule during ptrace
1236 * or during fork: the ptracer task is allowed to write to the
1237 * child->flags of its traced child (same goes for fork, the parent
1238 * can write to the child->flags), because we're guaranteed the
1239 * child is not running and in turn not changing child->flags
1240 * at the same time the parent does it.
1241 */
1242#define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1243#define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1244#define clear_used_math() clear_stopped_child_used_math(current)
1245#define set_used_math() set_stopped_child_used_math(current)
1246
1247#define conditional_stopped_child_used_math(condition, child) \
1248 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1249
1250#define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current)
1251
1252#define copy_to_stopped_child_used_math(child) \
1253 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1254
1255/* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1256#define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1257#define used_math() tsk_used_math(current)
1258
1259/* Per-process atomic flags. */
1260#define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
1261#define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
1262#define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
1263#define PFA_LMK_WAITING 3 /* Lowmemorykiller is waiting */
1264
1265
1266#define TASK_PFA_TEST(name, func) \
1267 static inline bool task_##func(struct task_struct *p) \
1268 { return test_bit(PFA_##name, &p->atomic_flags); }
1269
1270#define TASK_PFA_SET(name, func) \
1271 static inline void task_set_##func(struct task_struct *p) \
1272 { set_bit(PFA_##name, &p->atomic_flags); }
1273
1274#define TASK_PFA_CLEAR(name, func) \
1275 static inline void task_clear_##func(struct task_struct *p) \
1276 { clear_bit(PFA_##name, &p->atomic_flags); }
1277
1278TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
1279TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
1280
1281TASK_PFA_TEST(SPREAD_PAGE, spread_page)
1282TASK_PFA_SET(SPREAD_PAGE, spread_page)
1283TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
1284
1285TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
1286TASK_PFA_SET(SPREAD_SLAB, spread_slab)
1287TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
1288
1289TASK_PFA_TEST(LMK_WAITING, lmk_waiting)
1290TASK_PFA_SET(LMK_WAITING, lmk_waiting)
1291
1292static inline void
1293tsk_restore_flags(struct task_struct *task, unsigned long orig_flags, unsigned long flags)
1294{
1295 task->flags &= ~flags;
1296 task->flags |= orig_flags & flags;
1297}
1298
1299extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial);
1300extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed);
1301#ifdef CONFIG_SMP
1302extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask);
1303extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask);
1304#else
1305static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
1306{
1307}
1308static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
1309{
1310 if (!cpumask_test_cpu(0, new_mask))
1311 return -EINVAL;
1312 return 0;
1313}
1314#endif
1315
1316#ifndef cpu_relax_yield
1317#define cpu_relax_yield() cpu_relax()
1318#endif
1319
1320extern int yield_to(struct task_struct *p, bool preempt);
1321extern void set_user_nice(struct task_struct *p, long nice);
1322extern int task_prio(const struct task_struct *p);
1323
1324/**
1325 * task_nice - return the nice value of a given task.
1326 * @p: the task in question.
1327 *
1328 * Return: The nice value [ -20 ... 0 ... 19 ].
1329 */
1330static inline int task_nice(const struct task_struct *p)
1331{
1332 return PRIO_TO_NICE((p)->static_prio);
1333}
1334
1335extern int can_nice(const struct task_struct *p, const int nice);
1336extern int task_curr(const struct task_struct *p);
1337extern int idle_cpu(int cpu);
1338extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *);
1339extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *);
1340extern int sched_setattr(struct task_struct *, const struct sched_attr *);
1341extern struct task_struct *idle_task(int cpu);
1342
1343/**
1344 * is_idle_task - is the specified task an idle task?
1345 * @p: the task in question.
1346 *
1347 * Return: 1 if @p is an idle task. 0 otherwise.
1348 */
1349static inline bool is_idle_task(const struct task_struct *p)
1350{
1351 return !!(p->flags & PF_IDLE);
1352}
1353
1354extern struct task_struct *curr_task(int cpu);
1355extern void ia64_set_curr_task(int cpu, struct task_struct *p);
1356
1357void yield(void);
1358
1359union thread_union {
1360#ifndef CONFIG_THREAD_INFO_IN_TASK
1361 struct thread_info thread_info;
1362#endif
1363 unsigned long stack[THREAD_SIZE/sizeof(long)];
1364};
1365
1366#ifdef CONFIG_THREAD_INFO_IN_TASK
1367static inline struct thread_info *task_thread_info(struct task_struct *task)
1368{
1369 return &task->thread_info;
1370}
1371#elif !defined(__HAVE_THREAD_FUNCTIONS)
1372# define task_thread_info(task) ((struct thread_info *)(task)->stack)
1373#endif
1374
1375/*
1376 * find a task by one of its numerical ids
1377 *
1378 * find_task_by_pid_ns():
1379 * finds a task by its pid in the specified namespace
1380 * find_task_by_vpid():
1381 * finds a task by its virtual pid
1382 *
1383 * see also find_vpid() etc in include/linux/pid.h
1384 */
1385
1386extern struct task_struct *find_task_by_vpid(pid_t nr);
1387extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns);
1388
1389extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1390extern int wake_up_process(struct task_struct *tsk);
1391extern void wake_up_new_task(struct task_struct *tsk);
1392
1393#ifdef CONFIG_SMP
1394extern void kick_process(struct task_struct *tsk);
1395#else
1396static inline void kick_process(struct task_struct *tsk) { }
1397#endif
1398
1399extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
1400
1401static inline void set_task_comm(struct task_struct *tsk, const char *from)
1402{
1403 __set_task_comm(tsk, from, false);
1404}
1405
1406extern char *get_task_comm(char *to, struct task_struct *tsk);
1407
1408#ifdef CONFIG_SMP
1409void scheduler_ipi(void);
1410extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1411#else
1412static inline void scheduler_ipi(void) { }
1413static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state)
1414{
1415 return 1;
1416}
1417#endif
1418
1419/*
1420 * Set thread flags in other task's structures.
1421 * See asm/thread_info.h for TIF_xxxx flags available:
1422 */
1423static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1424{
1425 set_ti_thread_flag(task_thread_info(tsk), flag);
1426}
1427
1428static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1429{
1430 clear_ti_thread_flag(task_thread_info(tsk), flag);
1431}
1432
1433static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1434{
1435 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1436}
1437
1438static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1439{
1440 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1441}
1442
1443static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1444{
1445 return test_ti_thread_flag(task_thread_info(tsk), flag);
1446}
1447
1448static inline void set_tsk_need_resched(struct task_struct *tsk)
1449{
1450 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1451}
1452
1453static inline void clear_tsk_need_resched(struct task_struct *tsk)
1454{
1455 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1456}
1457
1458static inline int test_tsk_need_resched(struct task_struct *tsk)
1459{
1460 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
1461}
1462
1463/*
1464 * cond_resched() and cond_resched_lock(): latency reduction via
1465 * explicit rescheduling in places that are safe. The return
1466 * value indicates whether a reschedule was done in fact.
1467 * cond_resched_lock() will drop the spinlock before scheduling,
1468 * cond_resched_softirq() will enable bhs before scheduling.
1469 */
1470#ifndef CONFIG_PREEMPT
1471extern int _cond_resched(void);
1472#else
1473static inline int _cond_resched(void) { return 0; }
1474#endif
1475
1476#define cond_resched() ({ \
1477 ___might_sleep(__FILE__, __LINE__, 0); \
1478 _cond_resched(); \
1479})
1480
1481extern int __cond_resched_lock(spinlock_t *lock);
1482
1483#define cond_resched_lock(lock) ({ \
1484 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
1485 __cond_resched_lock(lock); \
1486})
1487
1488extern int __cond_resched_softirq(void);
1489
1490#define cond_resched_softirq() ({ \
1491 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
1492 __cond_resched_softirq(); \
1493})
1494
1495static inline void cond_resched_rcu(void)
1496{
1497#if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
1498 rcu_read_unlock();
1499 cond_resched();
1500 rcu_read_lock();
1501#endif
1502}
1503
1504/*
1505 * Does a critical section need to be broken due to another
1506 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
1507 * but a general need for low latency)
1508 */
1509static inline int spin_needbreak(spinlock_t *lock)
1510{
1511#ifdef CONFIG_PREEMPT
1512 return spin_is_contended(lock);
1513#else
1514 return 0;
1515#endif
1516}
1517
1518static __always_inline bool need_resched(void)
1519{
1520 return unlikely(tif_need_resched());
1521}
1522
1523/*
1524 * Wrappers for p->thread_info->cpu access. No-op on UP.
1525 */
1526#ifdef CONFIG_SMP
1527
1528static inline unsigned int task_cpu(const struct task_struct *p)
1529{
1530#ifdef CONFIG_THREAD_INFO_IN_TASK
1531 return p->cpu;
1532#else
1533 return task_thread_info(p)->cpu;
1534#endif
1535}
1536
1537extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1538
1539#else
1540
1541static inline unsigned int task_cpu(const struct task_struct *p)
1542{
1543 return 0;
1544}
1545
1546static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1547{
1548}
1549
1550#endif /* CONFIG_SMP */
1551
1552/*
1553 * In order to reduce various lock holder preemption latencies provide an
1554 * interface to see if a vCPU is currently running or not.
1555 *
1556 * This allows us to terminate optimistic spin loops and block, analogous to
1557 * the native optimistic spin heuristic of testing if the lock owner task is
1558 * running or not.
1559 */
1560#ifndef vcpu_is_preempted
1561# define vcpu_is_preempted(cpu) false
1562#endif
1563
1564extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
1565extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
1566
1567#ifndef TASK_SIZE_OF
1568#define TASK_SIZE_OF(tsk) TASK_SIZE
1569#endif
1570
1571#endif
1572