xref: /linux/include/linux/sched.h (revision 2c97b5ae83dca56718774e7b4bf9640f05d11867)
1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _LINUX_SCHED_H
3 #define _LINUX_SCHED_H
4 
5 /*
6  * Define 'struct task_struct' and provide the main scheduler
7  * APIs (schedule(), wakeup variants, etc.)
8  */
9 
10 #include <uapi/linux/sched.h>
11 
12 #include <asm/current.h>
13 
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/sched/types.h>
29 #include <linux/signal_types.h>
30 #include <linux/mm_types_task.h>
31 #include <linux/task_io_accounting.h>
32 #include <linux/posix-timers.h>
33 #include <linux/rseq.h>
34 
35 /* task_struct member predeclarations (sorted alphabetically): */
36 struct audit_context;
37 struct backing_dev_info;
38 struct bio_list;
39 struct blk_plug;
40 struct capture_control;
41 struct cfs_rq;
42 struct fs_struct;
43 struct futex_pi_state;
44 struct io_context;
45 struct mempolicy;
46 struct nameidata;
47 struct nsproxy;
48 struct perf_event_context;
49 struct pid_namespace;
50 struct pipe_inode_info;
51 struct rcu_node;
52 struct reclaim_state;
53 struct robust_list_head;
54 struct root_domain;
55 struct rq;
56 struct sched_attr;
57 struct sched_param;
58 struct seq_file;
59 struct sighand_struct;
60 struct signal_struct;
61 struct task_delay_info;
62 struct task_group;
63 
64 /*
65  * Task state bitmask. NOTE! These bits are also
66  * encoded in fs/proc/array.c: get_task_state().
67  *
68  * We have two separate sets of flags: task->state
69  * is about runnability, while task->exit_state are
70  * about the task exiting. Confusing, but this way
71  * modifying one set can't modify the other one by
72  * mistake.
73  */
74 
75 /* Used in tsk->state: */
76 #define TASK_RUNNING			0x0000
77 #define TASK_INTERRUPTIBLE		0x0001
78 #define TASK_UNINTERRUPTIBLE		0x0002
79 #define __TASK_STOPPED			0x0004
80 #define __TASK_TRACED			0x0008
81 /* Used in tsk->exit_state: */
82 #define EXIT_DEAD			0x0010
83 #define EXIT_ZOMBIE			0x0020
84 #define EXIT_TRACE			(EXIT_ZOMBIE | EXIT_DEAD)
85 /* Used in tsk->state again: */
86 #define TASK_PARKED			0x0040
87 #define TASK_DEAD			0x0080
88 #define TASK_WAKEKILL			0x0100
89 #define TASK_WAKING			0x0200
90 #define TASK_NOLOAD			0x0400
91 #define TASK_NEW			0x0800
92 #define TASK_STATE_MAX			0x1000
93 
94 /* Convenience macros for the sake of set_current_state: */
95 #define TASK_KILLABLE			(TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
96 #define TASK_STOPPED			(TASK_WAKEKILL | __TASK_STOPPED)
97 #define TASK_TRACED			(TASK_WAKEKILL | __TASK_TRACED)
98 
99 #define TASK_IDLE			(TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
100 
101 /* Convenience macros for the sake of wake_up(): */
102 #define TASK_NORMAL			(TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
103 
104 /* get_task_state(): */
105 #define TASK_REPORT			(TASK_RUNNING | TASK_INTERRUPTIBLE | \
106 					 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
107 					 __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \
108 					 TASK_PARKED)
109 
110 #define task_is_traced(task)		((task->state & __TASK_TRACED) != 0)
111 
112 #define task_is_stopped(task)		((task->state & __TASK_STOPPED) != 0)
113 
114 #define task_is_stopped_or_traced(task)	((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
115 
116 #define task_contributes_to_load(task)	((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
117 					 (task->flags & PF_FROZEN) == 0 && \
118 					 (task->state & TASK_NOLOAD) == 0)
119 
120 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
121 
122 /*
123  * Special states are those that do not use the normal wait-loop pattern. See
124  * the comment with set_special_state().
125  */
126 #define is_special_task_state(state)				\
127 	((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_PARKED | TASK_DEAD))
128 
129 #define __set_current_state(state_value)			\
130 	do {							\
131 		WARN_ON_ONCE(is_special_task_state(state_value));\
132 		current->task_state_change = _THIS_IP_;		\
133 		current->state = (state_value);			\
134 	} while (0)
135 
136 #define set_current_state(state_value)				\
137 	do {							\
138 		WARN_ON_ONCE(is_special_task_state(state_value));\
139 		current->task_state_change = _THIS_IP_;		\
140 		smp_store_mb(current->state, (state_value));	\
141 	} while (0)
142 
143 #define set_special_state(state_value)					\
144 	do {								\
145 		unsigned long flags; /* may shadow */			\
146 		WARN_ON_ONCE(!is_special_task_state(state_value));	\
147 		raw_spin_lock_irqsave(&current->pi_lock, flags);	\
148 		current->task_state_change = _THIS_IP_;			\
149 		current->state = (state_value);				\
150 		raw_spin_unlock_irqrestore(&current->pi_lock, flags);	\
151 	} while (0)
152 #else
153 /*
154  * set_current_state() includes a barrier so that the write of current->state
155  * is correctly serialised wrt the caller's subsequent test of whether to
156  * actually sleep:
157  *
158  *   for (;;) {
159  *	set_current_state(TASK_UNINTERRUPTIBLE);
160  *	if (!need_sleep)
161  *		break;
162  *
163  *	schedule();
164  *   }
165  *   __set_current_state(TASK_RUNNING);
166  *
167  * If the caller does not need such serialisation (because, for instance, the
168  * condition test and condition change and wakeup are under the same lock) then
169  * use __set_current_state().
170  *
171  * The above is typically ordered against the wakeup, which does:
172  *
173  *   need_sleep = false;
174  *   wake_up_state(p, TASK_UNINTERRUPTIBLE);
175  *
176  * where wake_up_state() executes a full memory barrier before accessing the
177  * task state.
178  *
179  * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
180  * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
181  * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
182  *
183  * However, with slightly different timing the wakeup TASK_RUNNING store can
184  * also collide with the TASK_UNINTERRUPTIBLE store. Losing that store is not
185  * a problem either because that will result in one extra go around the loop
186  * and our @cond test will save the day.
187  *
188  * Also see the comments of try_to_wake_up().
189  */
190 #define __set_current_state(state_value)				\
191 	current->state = (state_value)
192 
193 #define set_current_state(state_value)					\
194 	smp_store_mb(current->state, (state_value))
195 
196 /*
197  * set_special_state() should be used for those states when the blocking task
198  * can not use the regular condition based wait-loop. In that case we must
199  * serialize against wakeups such that any possible in-flight TASK_RUNNING stores
200  * will not collide with our state change.
201  */
202 #define set_special_state(state_value)					\
203 	do {								\
204 		unsigned long flags; /* may shadow */			\
205 		raw_spin_lock_irqsave(&current->pi_lock, flags);	\
206 		current->state = (state_value);				\
207 		raw_spin_unlock_irqrestore(&current->pi_lock, flags);	\
208 	} while (0)
209 
210 #endif
211 
212 /* Task command name length: */
213 #define TASK_COMM_LEN			16
214 
215 extern void scheduler_tick(void);
216 
217 #define	MAX_SCHEDULE_TIMEOUT		LONG_MAX
218 
219 extern long schedule_timeout(long timeout);
220 extern long schedule_timeout_interruptible(long timeout);
221 extern long schedule_timeout_killable(long timeout);
222 extern long schedule_timeout_uninterruptible(long timeout);
223 extern long schedule_timeout_idle(long timeout);
224 asmlinkage void schedule(void);
225 extern void schedule_preempt_disabled(void);
226 asmlinkage void preempt_schedule_irq(void);
227 
228 extern int __must_check io_schedule_prepare(void);
229 extern void io_schedule_finish(int token);
230 extern long io_schedule_timeout(long timeout);
231 extern void io_schedule(void);
232 
233 /**
234  * struct prev_cputime - snapshot of system and user cputime
235  * @utime: time spent in user mode
236  * @stime: time spent in system mode
237  * @lock: protects the above two fields
238  *
239  * Stores previous user/system time values such that we can guarantee
240  * monotonicity.
241  */
242 struct prev_cputime {
243 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
244 	u64				utime;
245 	u64				stime;
246 	raw_spinlock_t			lock;
247 #endif
248 };
249 
250 enum vtime_state {
251 	/* Task is sleeping or running in a CPU with VTIME inactive: */
252 	VTIME_INACTIVE = 0,
253 	/* Task is idle */
254 	VTIME_IDLE,
255 	/* Task runs in kernelspace in a CPU with VTIME active: */
256 	VTIME_SYS,
257 	/* Task runs in userspace in a CPU with VTIME active: */
258 	VTIME_USER,
259 	/* Task runs as guests in a CPU with VTIME active: */
260 	VTIME_GUEST,
261 };
262 
263 struct vtime {
264 	seqcount_t		seqcount;
265 	unsigned long long	starttime;
266 	enum vtime_state	state;
267 	unsigned int		cpu;
268 	u64			utime;
269 	u64			stime;
270 	u64			gtime;
271 };
272 
273 /*
274  * Utilization clamp constraints.
275  * @UCLAMP_MIN:	Minimum utilization
276  * @UCLAMP_MAX:	Maximum utilization
277  * @UCLAMP_CNT:	Utilization clamp constraints count
278  */
279 enum uclamp_id {
280 	UCLAMP_MIN = 0,
281 	UCLAMP_MAX,
282 	UCLAMP_CNT
283 };
284 
285 #ifdef CONFIG_SMP
286 extern struct root_domain def_root_domain;
287 extern struct mutex sched_domains_mutex;
288 #endif
289 
290 struct sched_info {
291 #ifdef CONFIG_SCHED_INFO
292 	/* Cumulative counters: */
293 
294 	/* # of times we have run on this CPU: */
295 	unsigned long			pcount;
296 
297 	/* Time spent waiting on a runqueue: */
298 	unsigned long long		run_delay;
299 
300 	/* Timestamps: */
301 
302 	/* When did we last run on a CPU? */
303 	unsigned long long		last_arrival;
304 
305 	/* When were we last queued to run? */
306 	unsigned long long		last_queued;
307 
308 #endif /* CONFIG_SCHED_INFO */
309 };
310 
311 /*
312  * Integer metrics need fixed point arithmetic, e.g., sched/fair
313  * has a few: load, load_avg, util_avg, freq, and capacity.
314  *
315  * We define a basic fixed point arithmetic range, and then formalize
316  * all these metrics based on that basic range.
317  */
318 # define SCHED_FIXEDPOINT_SHIFT		10
319 # define SCHED_FIXEDPOINT_SCALE		(1L << SCHED_FIXEDPOINT_SHIFT)
320 
321 /* Increase resolution of cpu_capacity calculations */
322 # define SCHED_CAPACITY_SHIFT		SCHED_FIXEDPOINT_SHIFT
323 # define SCHED_CAPACITY_SCALE		(1L << SCHED_CAPACITY_SHIFT)
324 
325 struct load_weight {
326 	unsigned long			weight;
327 	u32				inv_weight;
328 };
329 
330 /**
331  * struct util_est - Estimation utilization of FAIR tasks
332  * @enqueued: instantaneous estimated utilization of a task/cpu
333  * @ewma:     the Exponential Weighted Moving Average (EWMA)
334  *            utilization of a task
335  *
336  * Support data structure to track an Exponential Weighted Moving Average
337  * (EWMA) of a FAIR task's utilization. New samples are added to the moving
338  * average each time a task completes an activation. Sample's weight is chosen
339  * so that the EWMA will be relatively insensitive to transient changes to the
340  * task's workload.
341  *
342  * The enqueued attribute has a slightly different meaning for tasks and cpus:
343  * - task:   the task's util_avg at last task dequeue time
344  * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU
345  * Thus, the util_est.enqueued of a task represents the contribution on the
346  * estimated utilization of the CPU where that task is currently enqueued.
347  *
348  * Only for tasks we track a moving average of the past instantaneous
349  * estimated utilization. This allows to absorb sporadic drops in utilization
350  * of an otherwise almost periodic task.
351  */
352 struct util_est {
353 	unsigned int			enqueued;
354 	unsigned int			ewma;
355 #define UTIL_EST_WEIGHT_SHIFT		2
356 } __attribute__((__aligned__(sizeof(u64))));
357 
358 /*
359  * The load_avg/util_avg accumulates an infinite geometric series
360  * (see __update_load_avg() in kernel/sched/fair.c).
361  *
362  * [load_avg definition]
363  *
364  *   load_avg = runnable% * scale_load_down(load)
365  *
366  * where runnable% is the time ratio that a sched_entity is runnable.
367  * For cfs_rq, it is the aggregated load_avg of all runnable and
368  * blocked sched_entities.
369  *
370  * [util_avg definition]
371  *
372  *   util_avg = running% * SCHED_CAPACITY_SCALE
373  *
374  * where running% is the time ratio that a sched_entity is running on
375  * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
376  * and blocked sched_entities.
377  *
378  * load_avg and util_avg don't direcly factor frequency scaling and CPU
379  * capacity scaling. The scaling is done through the rq_clock_pelt that
380  * is used for computing those signals (see update_rq_clock_pelt())
381  *
382  * N.B., the above ratios (runnable% and running%) themselves are in the
383  * range of [0, 1]. To do fixed point arithmetics, we therefore scale them
384  * to as large a range as necessary. This is for example reflected by
385  * util_avg's SCHED_CAPACITY_SCALE.
386  *
387  * [Overflow issue]
388  *
389  * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
390  * with the highest load (=88761), always runnable on a single cfs_rq,
391  * and should not overflow as the number already hits PID_MAX_LIMIT.
392  *
393  * For all other cases (including 32-bit kernels), struct load_weight's
394  * weight will overflow first before we do, because:
395  *
396  *    Max(load_avg) <= Max(load.weight)
397  *
398  * Then it is the load_weight's responsibility to consider overflow
399  * issues.
400  */
401 struct sched_avg {
402 	u64				last_update_time;
403 	u64				load_sum;
404 	u64				runnable_load_sum;
405 	u32				util_sum;
406 	u32				period_contrib;
407 	unsigned long			load_avg;
408 	unsigned long			runnable_load_avg;
409 	unsigned long			util_avg;
410 	struct util_est			util_est;
411 } ____cacheline_aligned;
412 
413 struct sched_statistics {
414 #ifdef CONFIG_SCHEDSTATS
415 	u64				wait_start;
416 	u64				wait_max;
417 	u64				wait_count;
418 	u64				wait_sum;
419 	u64				iowait_count;
420 	u64				iowait_sum;
421 
422 	u64				sleep_start;
423 	u64				sleep_max;
424 	s64				sum_sleep_runtime;
425 
426 	u64				block_start;
427 	u64				block_max;
428 	u64				exec_max;
429 	u64				slice_max;
430 
431 	u64				nr_migrations_cold;
432 	u64				nr_failed_migrations_affine;
433 	u64				nr_failed_migrations_running;
434 	u64				nr_failed_migrations_hot;
435 	u64				nr_forced_migrations;
436 
437 	u64				nr_wakeups;
438 	u64				nr_wakeups_sync;
439 	u64				nr_wakeups_migrate;
440 	u64				nr_wakeups_local;
441 	u64				nr_wakeups_remote;
442 	u64				nr_wakeups_affine;
443 	u64				nr_wakeups_affine_attempts;
444 	u64				nr_wakeups_passive;
445 	u64				nr_wakeups_idle;
446 #endif
447 };
448 
449 struct sched_entity {
450 	/* For load-balancing: */
451 	struct load_weight		load;
452 	unsigned long			runnable_weight;
453 	struct rb_node			run_node;
454 	struct list_head		group_node;
455 	unsigned int			on_rq;
456 
457 	u64				exec_start;
458 	u64				sum_exec_runtime;
459 	u64				vruntime;
460 	u64				prev_sum_exec_runtime;
461 
462 	u64				nr_migrations;
463 
464 	struct sched_statistics		statistics;
465 
466 #ifdef CONFIG_FAIR_GROUP_SCHED
467 	int				depth;
468 	struct sched_entity		*parent;
469 	/* rq on which this entity is (to be) queued: */
470 	struct cfs_rq			*cfs_rq;
471 	/* rq "owned" by this entity/group: */
472 	struct cfs_rq			*my_q;
473 #endif
474 
475 #ifdef CONFIG_SMP
476 	/*
477 	 * Per entity load average tracking.
478 	 *
479 	 * Put into separate cache line so it does not
480 	 * collide with read-mostly values above.
481 	 */
482 	struct sched_avg		avg;
483 #endif
484 };
485 
486 struct sched_rt_entity {
487 	struct list_head		run_list;
488 	unsigned long			timeout;
489 	unsigned long			watchdog_stamp;
490 	unsigned int			time_slice;
491 	unsigned short			on_rq;
492 	unsigned short			on_list;
493 
494 	struct sched_rt_entity		*back;
495 #ifdef CONFIG_RT_GROUP_SCHED
496 	struct sched_rt_entity		*parent;
497 	/* rq on which this entity is (to be) queued: */
498 	struct rt_rq			*rt_rq;
499 	/* rq "owned" by this entity/group: */
500 	struct rt_rq			*my_q;
501 #endif
502 } __randomize_layout;
503 
504 struct sched_dl_entity {
505 	struct rb_node			rb_node;
506 
507 	/*
508 	 * Original scheduling parameters. Copied here from sched_attr
509 	 * during sched_setattr(), they will remain the same until
510 	 * the next sched_setattr().
511 	 */
512 	u64				dl_runtime;	/* Maximum runtime for each instance	*/
513 	u64				dl_deadline;	/* Relative deadline of each instance	*/
514 	u64				dl_period;	/* Separation of two instances (period) */
515 	u64				dl_bw;		/* dl_runtime / dl_period		*/
516 	u64				dl_density;	/* dl_runtime / dl_deadline		*/
517 
518 	/*
519 	 * Actual scheduling parameters. Initialized with the values above,
520 	 * they are continuously updated during task execution. Note that
521 	 * the remaining runtime could be < 0 in case we are in overrun.
522 	 */
523 	s64				runtime;	/* Remaining runtime for this instance	*/
524 	u64				deadline;	/* Absolute deadline for this instance	*/
525 	unsigned int			flags;		/* Specifying the scheduler behaviour	*/
526 
527 	/*
528 	 * Some bool flags:
529 	 *
530 	 * @dl_throttled tells if we exhausted the runtime. If so, the
531 	 * task has to wait for a replenishment to be performed at the
532 	 * next firing of dl_timer.
533 	 *
534 	 * @dl_boosted tells if we are boosted due to DI. If so we are
535 	 * outside bandwidth enforcement mechanism (but only until we
536 	 * exit the critical section);
537 	 *
538 	 * @dl_yielded tells if task gave up the CPU before consuming
539 	 * all its available runtime during the last job.
540 	 *
541 	 * @dl_non_contending tells if the task is inactive while still
542 	 * contributing to the active utilization. In other words, it
543 	 * indicates if the inactive timer has been armed and its handler
544 	 * has not been executed yet. This flag is useful to avoid race
545 	 * conditions between the inactive timer handler and the wakeup
546 	 * code.
547 	 *
548 	 * @dl_overrun tells if the task asked to be informed about runtime
549 	 * overruns.
550 	 */
551 	unsigned int			dl_throttled      : 1;
552 	unsigned int			dl_boosted        : 1;
553 	unsigned int			dl_yielded        : 1;
554 	unsigned int			dl_non_contending : 1;
555 	unsigned int			dl_overrun	  : 1;
556 
557 	/*
558 	 * Bandwidth enforcement timer. Each -deadline task has its
559 	 * own bandwidth to be enforced, thus we need one timer per task.
560 	 */
561 	struct hrtimer			dl_timer;
562 
563 	/*
564 	 * Inactive timer, responsible for decreasing the active utilization
565 	 * at the "0-lag time". When a -deadline task blocks, it contributes
566 	 * to GRUB's active utilization until the "0-lag time", hence a
567 	 * timer is needed to decrease the active utilization at the correct
568 	 * time.
569 	 */
570 	struct hrtimer inactive_timer;
571 };
572 
573 #ifdef CONFIG_UCLAMP_TASK
574 /* Number of utilization clamp buckets (shorter alias) */
575 #define UCLAMP_BUCKETS CONFIG_UCLAMP_BUCKETS_COUNT
576 
577 /*
578  * Utilization clamp for a scheduling entity
579  * @value:		clamp value "assigned" to a se
580  * @bucket_id:		bucket index corresponding to the "assigned" value
581  * @active:		the se is currently refcounted in a rq's bucket
582  * @user_defined:	the requested clamp value comes from user-space
583  *
584  * The bucket_id is the index of the clamp bucket matching the clamp value
585  * which is pre-computed and stored to avoid expensive integer divisions from
586  * the fast path.
587  *
588  * The active bit is set whenever a task has got an "effective" value assigned,
589  * which can be different from the clamp value "requested" from user-space.
590  * This allows to know a task is refcounted in the rq's bucket corresponding
591  * to the "effective" bucket_id.
592  *
593  * The user_defined bit is set whenever a task has got a task-specific clamp
594  * value requested from userspace, i.e. the system defaults apply to this task
595  * just as a restriction. This allows to relax default clamps when a less
596  * restrictive task-specific value has been requested, thus allowing to
597  * implement a "nice" semantic. For example, a task running with a 20%
598  * default boost can still drop its own boosting to 0%.
599  */
600 struct uclamp_se {
601 	unsigned int value		: bits_per(SCHED_CAPACITY_SCALE);
602 	unsigned int bucket_id		: bits_per(UCLAMP_BUCKETS);
603 	unsigned int active		: 1;
604 	unsigned int user_defined	: 1;
605 };
606 #endif /* CONFIG_UCLAMP_TASK */
607 
608 union rcu_special {
609 	struct {
610 		u8			blocked;
611 		u8			need_qs;
612 		u8			exp_hint; /* Hint for performance. */
613 		u8			deferred_qs;
614 	} b; /* Bits. */
615 	u32 s; /* Set of bits. */
616 };
617 
618 enum perf_event_task_context {
619 	perf_invalid_context = -1,
620 	perf_hw_context = 0,
621 	perf_sw_context,
622 	perf_nr_task_contexts,
623 };
624 
625 struct wake_q_node {
626 	struct wake_q_node *next;
627 };
628 
629 struct task_struct {
630 #ifdef CONFIG_THREAD_INFO_IN_TASK
631 	/*
632 	 * For reasons of header soup (see current_thread_info()), this
633 	 * must be the first element of task_struct.
634 	 */
635 	struct thread_info		thread_info;
636 #endif
637 	/* -1 unrunnable, 0 runnable, >0 stopped: */
638 	volatile long			state;
639 
640 	/*
641 	 * This begins the randomizable portion of task_struct. Only
642 	 * scheduling-critical items should be added above here.
643 	 */
644 	randomized_struct_fields_start
645 
646 	void				*stack;
647 	refcount_t			usage;
648 	/* Per task flags (PF_*), defined further below: */
649 	unsigned int			flags;
650 	unsigned int			ptrace;
651 
652 #ifdef CONFIG_SMP
653 	struct llist_node		wake_entry;
654 	int				on_cpu;
655 #ifdef CONFIG_THREAD_INFO_IN_TASK
656 	/* Current CPU: */
657 	unsigned int			cpu;
658 #endif
659 	unsigned int			wakee_flips;
660 	unsigned long			wakee_flip_decay_ts;
661 	struct task_struct		*last_wakee;
662 
663 	/*
664 	 * recent_used_cpu is initially set as the last CPU used by a task
665 	 * that wakes affine another task. Waker/wakee relationships can
666 	 * push tasks around a CPU where each wakeup moves to the next one.
667 	 * Tracking a recently used CPU allows a quick search for a recently
668 	 * used CPU that may be idle.
669 	 */
670 	int				recent_used_cpu;
671 	int				wake_cpu;
672 #endif
673 	int				on_rq;
674 
675 	int				prio;
676 	int				static_prio;
677 	int				normal_prio;
678 	unsigned int			rt_priority;
679 
680 	const struct sched_class	*sched_class;
681 	struct sched_entity		se;
682 	struct sched_rt_entity		rt;
683 #ifdef CONFIG_CGROUP_SCHED
684 	struct task_group		*sched_task_group;
685 #endif
686 	struct sched_dl_entity		dl;
687 
688 #ifdef CONFIG_UCLAMP_TASK
689 	/* Clamp values requested for a scheduling entity */
690 	struct uclamp_se		uclamp_req[UCLAMP_CNT];
691 	/* Effective clamp values used for a scheduling entity */
692 	struct uclamp_se		uclamp[UCLAMP_CNT];
693 #endif
694 
695 #ifdef CONFIG_PREEMPT_NOTIFIERS
696 	/* List of struct preempt_notifier: */
697 	struct hlist_head		preempt_notifiers;
698 #endif
699 
700 #ifdef CONFIG_BLK_DEV_IO_TRACE
701 	unsigned int			btrace_seq;
702 #endif
703 
704 	unsigned int			policy;
705 	int				nr_cpus_allowed;
706 	const cpumask_t			*cpus_ptr;
707 	cpumask_t			cpus_mask;
708 
709 #ifdef CONFIG_PREEMPT_RCU
710 	int				rcu_read_lock_nesting;
711 	union rcu_special		rcu_read_unlock_special;
712 	struct list_head		rcu_node_entry;
713 	struct rcu_node			*rcu_blocked_node;
714 #endif /* #ifdef CONFIG_PREEMPT_RCU */
715 
716 #ifdef CONFIG_TASKS_RCU
717 	unsigned long			rcu_tasks_nvcsw;
718 	u8				rcu_tasks_holdout;
719 	u8				rcu_tasks_idx;
720 	int				rcu_tasks_idle_cpu;
721 	struct list_head		rcu_tasks_holdout_list;
722 #endif /* #ifdef CONFIG_TASKS_RCU */
723 
724 	struct sched_info		sched_info;
725 
726 	struct list_head		tasks;
727 #ifdef CONFIG_SMP
728 	struct plist_node		pushable_tasks;
729 	struct rb_node			pushable_dl_tasks;
730 #endif
731 
732 	struct mm_struct		*mm;
733 	struct mm_struct		*active_mm;
734 
735 	/* Per-thread vma caching: */
736 	struct vmacache			vmacache;
737 
738 #ifdef SPLIT_RSS_COUNTING
739 	struct task_rss_stat		rss_stat;
740 #endif
741 	int				exit_state;
742 	int				exit_code;
743 	int				exit_signal;
744 	/* The signal sent when the parent dies: */
745 	int				pdeath_signal;
746 	/* JOBCTL_*, siglock protected: */
747 	unsigned long			jobctl;
748 
749 	/* Used for emulating ABI behavior of previous Linux versions: */
750 	unsigned int			personality;
751 
752 	/* Scheduler bits, serialized by scheduler locks: */
753 	unsigned			sched_reset_on_fork:1;
754 	unsigned			sched_contributes_to_load:1;
755 	unsigned			sched_migrated:1;
756 	unsigned			sched_remote_wakeup:1;
757 #ifdef CONFIG_PSI
758 	unsigned			sched_psi_wake_requeue:1;
759 #endif
760 
761 	/* Force alignment to the next boundary: */
762 	unsigned			:0;
763 
764 	/* Unserialized, strictly 'current' */
765 
766 	/* Bit to tell LSMs we're in execve(): */
767 	unsigned			in_execve:1;
768 	unsigned			in_iowait:1;
769 #ifndef TIF_RESTORE_SIGMASK
770 	unsigned			restore_sigmask:1;
771 #endif
772 #ifdef CONFIG_MEMCG
773 	unsigned			in_user_fault:1;
774 #endif
775 #ifdef CONFIG_COMPAT_BRK
776 	unsigned			brk_randomized:1;
777 #endif
778 #ifdef CONFIG_CGROUPS
779 	/* disallow userland-initiated cgroup migration */
780 	unsigned			no_cgroup_migration:1;
781 	/* task is frozen/stopped (used by the cgroup freezer) */
782 	unsigned			frozen:1;
783 #endif
784 #ifdef CONFIG_BLK_CGROUP
785 	/* to be used once the psi infrastructure lands upstream. */
786 	unsigned			use_memdelay:1;
787 #endif
788 
789 	unsigned long			atomic_flags; /* Flags requiring atomic access. */
790 
791 	struct restart_block		restart_block;
792 
793 	pid_t				pid;
794 	pid_t				tgid;
795 
796 #ifdef CONFIG_STACKPROTECTOR
797 	/* Canary value for the -fstack-protector GCC feature: */
798 	unsigned long			stack_canary;
799 #endif
800 	/*
801 	 * Pointers to the (original) parent process, youngest child, younger sibling,
802 	 * older sibling, respectively.  (p->father can be replaced with
803 	 * p->real_parent->pid)
804 	 */
805 
806 	/* Real parent process: */
807 	struct task_struct __rcu	*real_parent;
808 
809 	/* Recipient of SIGCHLD, wait4() reports: */
810 	struct task_struct __rcu	*parent;
811 
812 	/*
813 	 * Children/sibling form the list of natural children:
814 	 */
815 	struct list_head		children;
816 	struct list_head		sibling;
817 	struct task_struct		*group_leader;
818 
819 	/*
820 	 * 'ptraced' is the list of tasks this task is using ptrace() on.
821 	 *
822 	 * This includes both natural children and PTRACE_ATTACH targets.
823 	 * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
824 	 */
825 	struct list_head		ptraced;
826 	struct list_head		ptrace_entry;
827 
828 	/* PID/PID hash table linkage. */
829 	struct pid			*thread_pid;
830 	struct hlist_node		pid_links[PIDTYPE_MAX];
831 	struct list_head		thread_group;
832 	struct list_head		thread_node;
833 
834 	struct completion		*vfork_done;
835 
836 	/* CLONE_CHILD_SETTID: */
837 	int __user			*set_child_tid;
838 
839 	/* CLONE_CHILD_CLEARTID: */
840 	int __user			*clear_child_tid;
841 
842 	u64				utime;
843 	u64				stime;
844 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
845 	u64				utimescaled;
846 	u64				stimescaled;
847 #endif
848 	u64				gtime;
849 	struct prev_cputime		prev_cputime;
850 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
851 	struct vtime			vtime;
852 #endif
853 
854 #ifdef CONFIG_NO_HZ_FULL
855 	atomic_t			tick_dep_mask;
856 #endif
857 	/* Context switch counts: */
858 	unsigned long			nvcsw;
859 	unsigned long			nivcsw;
860 
861 	/* Monotonic time in nsecs: */
862 	u64				start_time;
863 
864 	/* Boot based time in nsecs: */
865 	u64				real_start_time;
866 
867 	/* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
868 	unsigned long			min_flt;
869 	unsigned long			maj_flt;
870 
871 	/* Empty if CONFIG_POSIX_CPUTIMERS=n */
872 	struct posix_cputimers		posix_cputimers;
873 
874 	/* Process credentials: */
875 
876 	/* Tracer's credentials at attach: */
877 	const struct cred __rcu		*ptracer_cred;
878 
879 	/* Objective and real subjective task credentials (COW): */
880 	const struct cred __rcu		*real_cred;
881 
882 	/* Effective (overridable) subjective task credentials (COW): */
883 	const struct cred __rcu		*cred;
884 
885 #ifdef CONFIG_KEYS
886 	/* Cached requested key. */
887 	struct key			*cached_requested_key;
888 #endif
889 
890 	/*
891 	 * executable name, excluding path.
892 	 *
893 	 * - normally initialized setup_new_exec()
894 	 * - access it with [gs]et_task_comm()
895 	 * - lock it with task_lock()
896 	 */
897 	char				comm[TASK_COMM_LEN];
898 
899 	struct nameidata		*nameidata;
900 
901 #ifdef CONFIG_SYSVIPC
902 	struct sysv_sem			sysvsem;
903 	struct sysv_shm			sysvshm;
904 #endif
905 #ifdef CONFIG_DETECT_HUNG_TASK
906 	unsigned long			last_switch_count;
907 	unsigned long			last_switch_time;
908 #endif
909 	/* Filesystem information: */
910 	struct fs_struct		*fs;
911 
912 	/* Open file information: */
913 	struct files_struct		*files;
914 
915 	/* Namespaces: */
916 	struct nsproxy			*nsproxy;
917 
918 	/* Signal handlers: */
919 	struct signal_struct		*signal;
920 	struct sighand_struct		*sighand;
921 	sigset_t			blocked;
922 	sigset_t			real_blocked;
923 	/* Restored if set_restore_sigmask() was used: */
924 	sigset_t			saved_sigmask;
925 	struct sigpending		pending;
926 	unsigned long			sas_ss_sp;
927 	size_t				sas_ss_size;
928 	unsigned int			sas_ss_flags;
929 
930 	struct callback_head		*task_works;
931 
932 #ifdef CONFIG_AUDIT
933 #ifdef CONFIG_AUDITSYSCALL
934 	struct audit_context		*audit_context;
935 #endif
936 	kuid_t				loginuid;
937 	unsigned int			sessionid;
938 #endif
939 	struct seccomp			seccomp;
940 
941 	/* Thread group tracking: */
942 	u32				parent_exec_id;
943 	u32				self_exec_id;
944 
945 	/* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
946 	spinlock_t			alloc_lock;
947 
948 	/* Protection of the PI data structures: */
949 	raw_spinlock_t			pi_lock;
950 
951 	struct wake_q_node		wake_q;
952 
953 #ifdef CONFIG_RT_MUTEXES
954 	/* PI waiters blocked on a rt_mutex held by this task: */
955 	struct rb_root_cached		pi_waiters;
956 	/* Updated under owner's pi_lock and rq lock */
957 	struct task_struct		*pi_top_task;
958 	/* Deadlock detection and priority inheritance handling: */
959 	struct rt_mutex_waiter		*pi_blocked_on;
960 #endif
961 
962 #ifdef CONFIG_DEBUG_MUTEXES
963 	/* Mutex deadlock detection: */
964 	struct mutex_waiter		*blocked_on;
965 #endif
966 
967 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
968 	int				non_block_count;
969 #endif
970 
971 #ifdef CONFIG_TRACE_IRQFLAGS
972 	unsigned int			irq_events;
973 	unsigned long			hardirq_enable_ip;
974 	unsigned long			hardirq_disable_ip;
975 	unsigned int			hardirq_enable_event;
976 	unsigned int			hardirq_disable_event;
977 	int				hardirqs_enabled;
978 	int				hardirq_context;
979 	unsigned long			softirq_disable_ip;
980 	unsigned long			softirq_enable_ip;
981 	unsigned int			softirq_disable_event;
982 	unsigned int			softirq_enable_event;
983 	int				softirqs_enabled;
984 	int				softirq_context;
985 #endif
986 
987 #ifdef CONFIG_LOCKDEP
988 # define MAX_LOCK_DEPTH			48UL
989 	u64				curr_chain_key;
990 	int				lockdep_depth;
991 	unsigned int			lockdep_recursion;
992 	struct held_lock		held_locks[MAX_LOCK_DEPTH];
993 #endif
994 
995 #ifdef CONFIG_UBSAN
996 	unsigned int			in_ubsan;
997 #endif
998 
999 	/* Journalling filesystem info: */
1000 	void				*journal_info;
1001 
1002 	/* Stacked block device info: */
1003 	struct bio_list			*bio_list;
1004 
1005 #ifdef CONFIG_BLOCK
1006 	/* Stack plugging: */
1007 	struct blk_plug			*plug;
1008 #endif
1009 
1010 	/* VM state: */
1011 	struct reclaim_state		*reclaim_state;
1012 
1013 	struct backing_dev_info		*backing_dev_info;
1014 
1015 	struct io_context		*io_context;
1016 
1017 #ifdef CONFIG_COMPACTION
1018 	struct capture_control		*capture_control;
1019 #endif
1020 	/* Ptrace state: */
1021 	unsigned long			ptrace_message;
1022 	kernel_siginfo_t		*last_siginfo;
1023 
1024 	struct task_io_accounting	ioac;
1025 #ifdef CONFIG_PSI
1026 	/* Pressure stall state */
1027 	unsigned int			psi_flags;
1028 #endif
1029 #ifdef CONFIG_TASK_XACCT
1030 	/* Accumulated RSS usage: */
1031 	u64				acct_rss_mem1;
1032 	/* Accumulated virtual memory usage: */
1033 	u64				acct_vm_mem1;
1034 	/* stime + utime since last update: */
1035 	u64				acct_timexpd;
1036 #endif
1037 #ifdef CONFIG_CPUSETS
1038 	/* Protected by ->alloc_lock: */
1039 	nodemask_t			mems_allowed;
1040 	/* Seqence number to catch updates: */
1041 	seqcount_t			mems_allowed_seq;
1042 	int				cpuset_mem_spread_rotor;
1043 	int				cpuset_slab_spread_rotor;
1044 #endif
1045 #ifdef CONFIG_CGROUPS
1046 	/* Control Group info protected by css_set_lock: */
1047 	struct css_set __rcu		*cgroups;
1048 	/* cg_list protected by css_set_lock and tsk->alloc_lock: */
1049 	struct list_head		cg_list;
1050 #endif
1051 #ifdef CONFIG_X86_CPU_RESCTRL
1052 	u32				closid;
1053 	u32				rmid;
1054 #endif
1055 #ifdef CONFIG_FUTEX
1056 	struct robust_list_head __user	*robust_list;
1057 #ifdef CONFIG_COMPAT
1058 	struct compat_robust_list_head __user *compat_robust_list;
1059 #endif
1060 	struct list_head		pi_state_list;
1061 	struct futex_pi_state		*pi_state_cache;
1062 	struct mutex			futex_exit_mutex;
1063 	unsigned int			futex_state;
1064 #endif
1065 #ifdef CONFIG_PERF_EVENTS
1066 	struct perf_event_context	*perf_event_ctxp[perf_nr_task_contexts];
1067 	struct mutex			perf_event_mutex;
1068 	struct list_head		perf_event_list;
1069 #endif
1070 #ifdef CONFIG_DEBUG_PREEMPT
1071 	unsigned long			preempt_disable_ip;
1072 #endif
1073 #ifdef CONFIG_NUMA
1074 	/* Protected by alloc_lock: */
1075 	struct mempolicy		*mempolicy;
1076 	short				il_prev;
1077 	short				pref_node_fork;
1078 #endif
1079 #ifdef CONFIG_NUMA_BALANCING
1080 	int				numa_scan_seq;
1081 	unsigned int			numa_scan_period;
1082 	unsigned int			numa_scan_period_max;
1083 	int				numa_preferred_nid;
1084 	unsigned long			numa_migrate_retry;
1085 	/* Migration stamp: */
1086 	u64				node_stamp;
1087 	u64				last_task_numa_placement;
1088 	u64				last_sum_exec_runtime;
1089 	struct callback_head		numa_work;
1090 
1091 	/*
1092 	 * This pointer is only modified for current in syscall and
1093 	 * pagefault context (and for tasks being destroyed), so it can be read
1094 	 * from any of the following contexts:
1095 	 *  - RCU read-side critical section
1096 	 *  - current->numa_group from everywhere
1097 	 *  - task's runqueue locked, task not running
1098 	 */
1099 	struct numa_group __rcu		*numa_group;
1100 
1101 	/*
1102 	 * numa_faults is an array split into four regions:
1103 	 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1104 	 * in this precise order.
1105 	 *
1106 	 * faults_memory: Exponential decaying average of faults on a per-node
1107 	 * basis. Scheduling placement decisions are made based on these
1108 	 * counts. The values remain static for the duration of a PTE scan.
1109 	 * faults_cpu: Track the nodes the process was running on when a NUMA
1110 	 * hinting fault was incurred.
1111 	 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1112 	 * during the current scan window. When the scan completes, the counts
1113 	 * in faults_memory and faults_cpu decay and these values are copied.
1114 	 */
1115 	unsigned long			*numa_faults;
1116 	unsigned long			total_numa_faults;
1117 
1118 	/*
1119 	 * numa_faults_locality tracks if faults recorded during the last
1120 	 * scan window were remote/local or failed to migrate. The task scan
1121 	 * period is adapted based on the locality of the faults with different
1122 	 * weights depending on whether they were shared or private faults
1123 	 */
1124 	unsigned long			numa_faults_locality[3];
1125 
1126 	unsigned long			numa_pages_migrated;
1127 #endif /* CONFIG_NUMA_BALANCING */
1128 
1129 #ifdef CONFIG_RSEQ
1130 	struct rseq __user *rseq;
1131 	u32 rseq_sig;
1132 	/*
1133 	 * RmW on rseq_event_mask must be performed atomically
1134 	 * with respect to preemption.
1135 	 */
1136 	unsigned long rseq_event_mask;
1137 #endif
1138 
1139 	struct tlbflush_unmap_batch	tlb_ubc;
1140 
1141 	union {
1142 		refcount_t		rcu_users;
1143 		struct rcu_head		rcu;
1144 	};
1145 
1146 	/* Cache last used pipe for splice(): */
1147 	struct pipe_inode_info		*splice_pipe;
1148 
1149 	struct page_frag		task_frag;
1150 
1151 #ifdef CONFIG_TASK_DELAY_ACCT
1152 	struct task_delay_info		*delays;
1153 #endif
1154 
1155 #ifdef CONFIG_FAULT_INJECTION
1156 	int				make_it_fail;
1157 	unsigned int			fail_nth;
1158 #endif
1159 	/*
1160 	 * When (nr_dirtied >= nr_dirtied_pause), it's time to call
1161 	 * balance_dirty_pages() for a dirty throttling pause:
1162 	 */
1163 	int				nr_dirtied;
1164 	int				nr_dirtied_pause;
1165 	/* Start of a write-and-pause period: */
1166 	unsigned long			dirty_paused_when;
1167 
1168 #ifdef CONFIG_LATENCYTOP
1169 	int				latency_record_count;
1170 	struct latency_record		latency_record[LT_SAVECOUNT];
1171 #endif
1172 	/*
1173 	 * Time slack values; these are used to round up poll() and
1174 	 * select() etc timeout values. These are in nanoseconds.
1175 	 */
1176 	u64				timer_slack_ns;
1177 	u64				default_timer_slack_ns;
1178 
1179 #ifdef CONFIG_KASAN
1180 	unsigned int			kasan_depth;
1181 #endif
1182 
1183 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1184 	/* Index of current stored address in ret_stack: */
1185 	int				curr_ret_stack;
1186 	int				curr_ret_depth;
1187 
1188 	/* Stack of return addresses for return function tracing: */
1189 	struct ftrace_ret_stack		*ret_stack;
1190 
1191 	/* Timestamp for last schedule: */
1192 	unsigned long long		ftrace_timestamp;
1193 
1194 	/*
1195 	 * Number of functions that haven't been traced
1196 	 * because of depth overrun:
1197 	 */
1198 	atomic_t			trace_overrun;
1199 
1200 	/* Pause tracing: */
1201 	atomic_t			tracing_graph_pause;
1202 #endif
1203 
1204 #ifdef CONFIG_TRACING
1205 	/* State flags for use by tracers: */
1206 	unsigned long			trace;
1207 
1208 	/* Bitmask and counter of trace recursion: */
1209 	unsigned long			trace_recursion;
1210 #endif /* CONFIG_TRACING */
1211 
1212 #ifdef CONFIG_KCOV
1213 	/* Coverage collection mode enabled for this task (0 if disabled): */
1214 	unsigned int			kcov_mode;
1215 
1216 	/* Size of the kcov_area: */
1217 	unsigned int			kcov_size;
1218 
1219 	/* Buffer for coverage collection: */
1220 	void				*kcov_area;
1221 
1222 	/* KCOV descriptor wired with this task or NULL: */
1223 	struct kcov			*kcov;
1224 #endif
1225 
1226 #ifdef CONFIG_MEMCG
1227 	struct mem_cgroup		*memcg_in_oom;
1228 	gfp_t				memcg_oom_gfp_mask;
1229 	int				memcg_oom_order;
1230 
1231 	/* Number of pages to reclaim on returning to userland: */
1232 	unsigned int			memcg_nr_pages_over_high;
1233 
1234 	/* Used by memcontrol for targeted memcg charge: */
1235 	struct mem_cgroup		*active_memcg;
1236 #endif
1237 
1238 #ifdef CONFIG_BLK_CGROUP
1239 	struct request_queue		*throttle_queue;
1240 #endif
1241 
1242 #ifdef CONFIG_UPROBES
1243 	struct uprobe_task		*utask;
1244 #endif
1245 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1246 	unsigned int			sequential_io;
1247 	unsigned int			sequential_io_avg;
1248 #endif
1249 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1250 	unsigned long			task_state_change;
1251 #endif
1252 	int				pagefault_disabled;
1253 #ifdef CONFIG_MMU
1254 	struct task_struct		*oom_reaper_list;
1255 #endif
1256 #ifdef CONFIG_VMAP_STACK
1257 	struct vm_struct		*stack_vm_area;
1258 #endif
1259 #ifdef CONFIG_THREAD_INFO_IN_TASK
1260 	/* A live task holds one reference: */
1261 	refcount_t			stack_refcount;
1262 #endif
1263 #ifdef CONFIG_LIVEPATCH
1264 	int patch_state;
1265 #endif
1266 #ifdef CONFIG_SECURITY
1267 	/* Used by LSM modules for access restriction: */
1268 	void				*security;
1269 #endif
1270 
1271 #ifdef CONFIG_GCC_PLUGIN_STACKLEAK
1272 	unsigned long			lowest_stack;
1273 	unsigned long			prev_lowest_stack;
1274 #endif
1275 
1276 	/*
1277 	 * New fields for task_struct should be added above here, so that
1278 	 * they are included in the randomized portion of task_struct.
1279 	 */
1280 	randomized_struct_fields_end
1281 
1282 	/* CPU-specific state of this task: */
1283 	struct thread_struct		thread;
1284 
1285 	/*
1286 	 * WARNING: on x86, 'thread_struct' contains a variable-sized
1287 	 * structure.  It *MUST* be at the end of 'task_struct'.
1288 	 *
1289 	 * Do not put anything below here!
1290 	 */
1291 };
1292 
1293 static inline struct pid *task_pid(struct task_struct *task)
1294 {
1295 	return task->thread_pid;
1296 }
1297 
1298 /*
1299  * the helpers to get the task's different pids as they are seen
1300  * from various namespaces
1301  *
1302  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1303  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1304  *                     current.
1305  * task_xid_nr_ns()  : id seen from the ns specified;
1306  *
1307  * see also pid_nr() etc in include/linux/pid.h
1308  */
1309 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns);
1310 
1311 static inline pid_t task_pid_nr(struct task_struct *tsk)
1312 {
1313 	return tsk->pid;
1314 }
1315 
1316 static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1317 {
1318 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1319 }
1320 
1321 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1322 {
1323 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1324 }
1325 
1326 
1327 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1328 {
1329 	return tsk->tgid;
1330 }
1331 
1332 /**
1333  * pid_alive - check that a task structure is not stale
1334  * @p: Task structure to be checked.
1335  *
1336  * Test if a process is not yet dead (at most zombie state)
1337  * If pid_alive fails, then pointers within the task structure
1338  * can be stale and must not be dereferenced.
1339  *
1340  * Return: 1 if the process is alive. 0 otherwise.
1341  */
1342 static inline int pid_alive(const struct task_struct *p)
1343 {
1344 	return p->thread_pid != NULL;
1345 }
1346 
1347 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1348 {
1349 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1350 }
1351 
1352 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1353 {
1354 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1355 }
1356 
1357 
1358 static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1359 {
1360 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1361 }
1362 
1363 static inline pid_t task_session_vnr(struct task_struct *tsk)
1364 {
1365 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1366 }
1367 
1368 static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1369 {
1370 	return __task_pid_nr_ns(tsk, PIDTYPE_TGID, ns);
1371 }
1372 
1373 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1374 {
1375 	return __task_pid_nr_ns(tsk, PIDTYPE_TGID, NULL);
1376 }
1377 
1378 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1379 {
1380 	pid_t pid = 0;
1381 
1382 	rcu_read_lock();
1383 	if (pid_alive(tsk))
1384 		pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1385 	rcu_read_unlock();
1386 
1387 	return pid;
1388 }
1389 
1390 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1391 {
1392 	return task_ppid_nr_ns(tsk, &init_pid_ns);
1393 }
1394 
1395 /* Obsolete, do not use: */
1396 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1397 {
1398 	return task_pgrp_nr_ns(tsk, &init_pid_ns);
1399 }
1400 
1401 #define TASK_REPORT_IDLE	(TASK_REPORT + 1)
1402 #define TASK_REPORT_MAX		(TASK_REPORT_IDLE << 1)
1403 
1404 static inline unsigned int task_state_index(struct task_struct *tsk)
1405 {
1406 	unsigned int tsk_state = READ_ONCE(tsk->state);
1407 	unsigned int state = (tsk_state | tsk->exit_state) & TASK_REPORT;
1408 
1409 	BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX);
1410 
1411 	if (tsk_state == TASK_IDLE)
1412 		state = TASK_REPORT_IDLE;
1413 
1414 	return fls(state);
1415 }
1416 
1417 static inline char task_index_to_char(unsigned int state)
1418 {
1419 	static const char state_char[] = "RSDTtXZPI";
1420 
1421 	BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX) != sizeof(state_char) - 1);
1422 
1423 	return state_char[state];
1424 }
1425 
1426 static inline char task_state_to_char(struct task_struct *tsk)
1427 {
1428 	return task_index_to_char(task_state_index(tsk));
1429 }
1430 
1431 /**
1432  * is_global_init - check if a task structure is init. Since init
1433  * is free to have sub-threads we need to check tgid.
1434  * @tsk: Task structure to be checked.
1435  *
1436  * Check if a task structure is the first user space task the kernel created.
1437  *
1438  * Return: 1 if the task structure is init. 0 otherwise.
1439  */
1440 static inline int is_global_init(struct task_struct *tsk)
1441 {
1442 	return task_tgid_nr(tsk) == 1;
1443 }
1444 
1445 extern struct pid *cad_pid;
1446 
1447 /*
1448  * Per process flags
1449  */
1450 #define PF_IDLE			0x00000002	/* I am an IDLE thread */
1451 #define PF_EXITING		0x00000004	/* Getting shut down */
1452 #define PF_VCPU			0x00000010	/* I'm a virtual CPU */
1453 #define PF_WQ_WORKER		0x00000020	/* I'm a workqueue worker */
1454 #define PF_FORKNOEXEC		0x00000040	/* Forked but didn't exec */
1455 #define PF_MCE_PROCESS		0x00000080      /* Process policy on mce errors */
1456 #define PF_SUPERPRIV		0x00000100	/* Used super-user privileges */
1457 #define PF_DUMPCORE		0x00000200	/* Dumped core */
1458 #define PF_SIGNALED		0x00000400	/* Killed by a signal */
1459 #define PF_MEMALLOC		0x00000800	/* Allocating memory */
1460 #define PF_NPROC_EXCEEDED	0x00001000	/* set_user() noticed that RLIMIT_NPROC was exceeded */
1461 #define PF_USED_MATH		0x00002000	/* If unset the fpu must be initialized before use */
1462 #define PF_USED_ASYNC		0x00004000	/* Used async_schedule*(), used by module init */
1463 #define PF_NOFREEZE		0x00008000	/* This thread should not be frozen */
1464 #define PF_FROZEN		0x00010000	/* Frozen for system suspend */
1465 #define PF_KSWAPD		0x00020000	/* I am kswapd */
1466 #define PF_MEMALLOC_NOFS	0x00040000	/* All allocation requests will inherit GFP_NOFS */
1467 #define PF_MEMALLOC_NOIO	0x00080000	/* All allocation requests will inherit GFP_NOIO */
1468 #define PF_LESS_THROTTLE	0x00100000	/* Throttle me less: I clean memory */
1469 #define PF_KTHREAD		0x00200000	/* I am a kernel thread */
1470 #define PF_RANDOMIZE		0x00400000	/* Randomize virtual address space */
1471 #define PF_SWAPWRITE		0x00800000	/* Allowed to write to swap */
1472 #define PF_MEMSTALL		0x01000000	/* Stalled due to lack of memory */
1473 #define PF_UMH			0x02000000	/* I'm an Usermodehelper process */
1474 #define PF_NO_SETAFFINITY	0x04000000	/* Userland is not allowed to meddle with cpus_mask */
1475 #define PF_MCE_EARLY		0x08000000      /* Early kill for mce process policy */
1476 #define PF_MEMALLOC_NOCMA	0x10000000	/* All allocation request will have _GFP_MOVABLE cleared */
1477 #define PF_IO_WORKER		0x20000000	/* Task is an IO worker */
1478 #define PF_FREEZER_SKIP		0x40000000	/* Freezer should not count it as freezable */
1479 #define PF_SUSPEND_TASK		0x80000000      /* This thread called freeze_processes() and should not be frozen */
1480 
1481 /*
1482  * Only the _current_ task can read/write to tsk->flags, but other
1483  * tasks can access tsk->flags in readonly mode for example
1484  * with tsk_used_math (like during threaded core dumping).
1485  * There is however an exception to this rule during ptrace
1486  * or during fork: the ptracer task is allowed to write to the
1487  * child->flags of its traced child (same goes for fork, the parent
1488  * can write to the child->flags), because we're guaranteed the
1489  * child is not running and in turn not changing child->flags
1490  * at the same time the parent does it.
1491  */
1492 #define clear_stopped_child_used_math(child)	do { (child)->flags &= ~PF_USED_MATH; } while (0)
1493 #define set_stopped_child_used_math(child)	do { (child)->flags |= PF_USED_MATH; } while (0)
1494 #define clear_used_math()			clear_stopped_child_used_math(current)
1495 #define set_used_math()				set_stopped_child_used_math(current)
1496 
1497 #define conditional_stopped_child_used_math(condition, child) \
1498 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1499 
1500 #define conditional_used_math(condition)	conditional_stopped_child_used_math(condition, current)
1501 
1502 #define copy_to_stopped_child_used_math(child) \
1503 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1504 
1505 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1506 #define tsk_used_math(p)			((p)->flags & PF_USED_MATH)
1507 #define used_math()				tsk_used_math(current)
1508 
1509 static inline bool is_percpu_thread(void)
1510 {
1511 #ifdef CONFIG_SMP
1512 	return (current->flags & PF_NO_SETAFFINITY) &&
1513 		(current->nr_cpus_allowed  == 1);
1514 #else
1515 	return true;
1516 #endif
1517 }
1518 
1519 /* Per-process atomic flags. */
1520 #define PFA_NO_NEW_PRIVS		0	/* May not gain new privileges. */
1521 #define PFA_SPREAD_PAGE			1	/* Spread page cache over cpuset */
1522 #define PFA_SPREAD_SLAB			2	/* Spread some slab caches over cpuset */
1523 #define PFA_SPEC_SSB_DISABLE		3	/* Speculative Store Bypass disabled */
1524 #define PFA_SPEC_SSB_FORCE_DISABLE	4	/* Speculative Store Bypass force disabled*/
1525 #define PFA_SPEC_IB_DISABLE		5	/* Indirect branch speculation restricted */
1526 #define PFA_SPEC_IB_FORCE_DISABLE	6	/* Indirect branch speculation permanently restricted */
1527 #define PFA_SPEC_SSB_NOEXEC		7	/* Speculative Store Bypass clear on execve() */
1528 
1529 #define TASK_PFA_TEST(name, func)					\
1530 	static inline bool task_##func(struct task_struct *p)		\
1531 	{ return test_bit(PFA_##name, &p->atomic_flags); }
1532 
1533 #define TASK_PFA_SET(name, func)					\
1534 	static inline void task_set_##func(struct task_struct *p)	\
1535 	{ set_bit(PFA_##name, &p->atomic_flags); }
1536 
1537 #define TASK_PFA_CLEAR(name, func)					\
1538 	static inline void task_clear_##func(struct task_struct *p)	\
1539 	{ clear_bit(PFA_##name, &p->atomic_flags); }
1540 
1541 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
1542 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
1543 
1544 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
1545 TASK_PFA_SET(SPREAD_PAGE, spread_page)
1546 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
1547 
1548 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
1549 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
1550 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
1551 
1552 TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable)
1553 TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable)
1554 TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable)
1555 
1556 TASK_PFA_TEST(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1557 TASK_PFA_SET(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1558 TASK_PFA_CLEAR(SPEC_SSB_NOEXEC, spec_ssb_noexec)
1559 
1560 TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1561 TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1562 
1563 TASK_PFA_TEST(SPEC_IB_DISABLE, spec_ib_disable)
1564 TASK_PFA_SET(SPEC_IB_DISABLE, spec_ib_disable)
1565 TASK_PFA_CLEAR(SPEC_IB_DISABLE, spec_ib_disable)
1566 
1567 TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1568 TASK_PFA_SET(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable)
1569 
1570 static inline void
1571 current_restore_flags(unsigned long orig_flags, unsigned long flags)
1572 {
1573 	current->flags &= ~flags;
1574 	current->flags |= orig_flags & flags;
1575 }
1576 
1577 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial);
1578 extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed);
1579 #ifdef CONFIG_SMP
1580 extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask);
1581 extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask);
1582 #else
1583 static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
1584 {
1585 }
1586 static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
1587 {
1588 	if (!cpumask_test_cpu(0, new_mask))
1589 		return -EINVAL;
1590 	return 0;
1591 }
1592 #endif
1593 
1594 extern int yield_to(struct task_struct *p, bool preempt);
1595 extern void set_user_nice(struct task_struct *p, long nice);
1596 extern int task_prio(const struct task_struct *p);
1597 
1598 /**
1599  * task_nice - return the nice value of a given task.
1600  * @p: the task in question.
1601  *
1602  * Return: The nice value [ -20 ... 0 ... 19 ].
1603  */
1604 static inline int task_nice(const struct task_struct *p)
1605 {
1606 	return PRIO_TO_NICE((p)->static_prio);
1607 }
1608 
1609 extern int can_nice(const struct task_struct *p, const int nice);
1610 extern int task_curr(const struct task_struct *p);
1611 extern int idle_cpu(int cpu);
1612 extern int available_idle_cpu(int cpu);
1613 extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *);
1614 extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *);
1615 extern int sched_setattr(struct task_struct *, const struct sched_attr *);
1616 extern int sched_setattr_nocheck(struct task_struct *, const struct sched_attr *);
1617 extern struct task_struct *idle_task(int cpu);
1618 
1619 /**
1620  * is_idle_task - is the specified task an idle task?
1621  * @p: the task in question.
1622  *
1623  * Return: 1 if @p is an idle task. 0 otherwise.
1624  */
1625 static inline bool is_idle_task(const struct task_struct *p)
1626 {
1627 	return !!(p->flags & PF_IDLE);
1628 }
1629 
1630 extern struct task_struct *curr_task(int cpu);
1631 extern void ia64_set_curr_task(int cpu, struct task_struct *p);
1632 
1633 void yield(void);
1634 
1635 union thread_union {
1636 #ifndef CONFIG_ARCH_TASK_STRUCT_ON_STACK
1637 	struct task_struct task;
1638 #endif
1639 #ifndef CONFIG_THREAD_INFO_IN_TASK
1640 	struct thread_info thread_info;
1641 #endif
1642 	unsigned long stack[THREAD_SIZE/sizeof(long)];
1643 };
1644 
1645 #ifndef CONFIG_THREAD_INFO_IN_TASK
1646 extern struct thread_info init_thread_info;
1647 #endif
1648 
1649 extern unsigned long init_stack[THREAD_SIZE / sizeof(unsigned long)];
1650 
1651 #ifdef CONFIG_THREAD_INFO_IN_TASK
1652 static inline struct thread_info *task_thread_info(struct task_struct *task)
1653 {
1654 	return &task->thread_info;
1655 }
1656 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1657 # define task_thread_info(task)	((struct thread_info *)(task)->stack)
1658 #endif
1659 
1660 /*
1661  * find a task by one of its numerical ids
1662  *
1663  * find_task_by_pid_ns():
1664  *      finds a task by its pid in the specified namespace
1665  * find_task_by_vpid():
1666  *      finds a task by its virtual pid
1667  *
1668  * see also find_vpid() etc in include/linux/pid.h
1669  */
1670 
1671 extern struct task_struct *find_task_by_vpid(pid_t nr);
1672 extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns);
1673 
1674 /*
1675  * find a task by its virtual pid and get the task struct
1676  */
1677 extern struct task_struct *find_get_task_by_vpid(pid_t nr);
1678 
1679 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1680 extern int wake_up_process(struct task_struct *tsk);
1681 extern void wake_up_new_task(struct task_struct *tsk);
1682 
1683 #ifdef CONFIG_SMP
1684 extern void kick_process(struct task_struct *tsk);
1685 #else
1686 static inline void kick_process(struct task_struct *tsk) { }
1687 #endif
1688 
1689 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
1690 
1691 static inline void set_task_comm(struct task_struct *tsk, const char *from)
1692 {
1693 	__set_task_comm(tsk, from, false);
1694 }
1695 
1696 extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk);
1697 #define get_task_comm(buf, tsk) ({			\
1698 	BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN);	\
1699 	__get_task_comm(buf, sizeof(buf), tsk);		\
1700 })
1701 
1702 #ifdef CONFIG_SMP
1703 void scheduler_ipi(void);
1704 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1705 #else
1706 static inline void scheduler_ipi(void) { }
1707 static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state)
1708 {
1709 	return 1;
1710 }
1711 #endif
1712 
1713 /*
1714  * Set thread flags in other task's structures.
1715  * See asm/thread_info.h for TIF_xxxx flags available:
1716  */
1717 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1718 {
1719 	set_ti_thread_flag(task_thread_info(tsk), flag);
1720 }
1721 
1722 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1723 {
1724 	clear_ti_thread_flag(task_thread_info(tsk), flag);
1725 }
1726 
1727 static inline void update_tsk_thread_flag(struct task_struct *tsk, int flag,
1728 					  bool value)
1729 {
1730 	update_ti_thread_flag(task_thread_info(tsk), flag, value);
1731 }
1732 
1733 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1734 {
1735 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1736 }
1737 
1738 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1739 {
1740 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1741 }
1742 
1743 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1744 {
1745 	return test_ti_thread_flag(task_thread_info(tsk), flag);
1746 }
1747 
1748 static inline void set_tsk_need_resched(struct task_struct *tsk)
1749 {
1750 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1751 }
1752 
1753 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1754 {
1755 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1756 }
1757 
1758 static inline int test_tsk_need_resched(struct task_struct *tsk)
1759 {
1760 	return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
1761 }
1762 
1763 /*
1764  * cond_resched() and cond_resched_lock(): latency reduction via
1765  * explicit rescheduling in places that are safe. The return
1766  * value indicates whether a reschedule was done in fact.
1767  * cond_resched_lock() will drop the spinlock before scheduling,
1768  */
1769 #ifndef CONFIG_PREEMPTION
1770 extern int _cond_resched(void);
1771 #else
1772 static inline int _cond_resched(void) { return 0; }
1773 #endif
1774 
1775 #define cond_resched() ({			\
1776 	___might_sleep(__FILE__, __LINE__, 0);	\
1777 	_cond_resched();			\
1778 })
1779 
1780 extern int __cond_resched_lock(spinlock_t *lock);
1781 
1782 #define cond_resched_lock(lock) ({				\
1783 	___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
1784 	__cond_resched_lock(lock);				\
1785 })
1786 
1787 static inline void cond_resched_rcu(void)
1788 {
1789 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
1790 	rcu_read_unlock();
1791 	cond_resched();
1792 	rcu_read_lock();
1793 #endif
1794 }
1795 
1796 /*
1797  * Does a critical section need to be broken due to another
1798  * task waiting?: (technically does not depend on CONFIG_PREEMPTION,
1799  * but a general need for low latency)
1800  */
1801 static inline int spin_needbreak(spinlock_t *lock)
1802 {
1803 #ifdef CONFIG_PREEMPTION
1804 	return spin_is_contended(lock);
1805 #else
1806 	return 0;
1807 #endif
1808 }
1809 
1810 static __always_inline bool need_resched(void)
1811 {
1812 	return unlikely(tif_need_resched());
1813 }
1814 
1815 /*
1816  * Wrappers for p->thread_info->cpu access. No-op on UP.
1817  */
1818 #ifdef CONFIG_SMP
1819 
1820 static inline unsigned int task_cpu(const struct task_struct *p)
1821 {
1822 #ifdef CONFIG_THREAD_INFO_IN_TASK
1823 	return READ_ONCE(p->cpu);
1824 #else
1825 	return READ_ONCE(task_thread_info(p)->cpu);
1826 #endif
1827 }
1828 
1829 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1830 
1831 #else
1832 
1833 static inline unsigned int task_cpu(const struct task_struct *p)
1834 {
1835 	return 0;
1836 }
1837 
1838 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1839 {
1840 }
1841 
1842 #endif /* CONFIG_SMP */
1843 
1844 /*
1845  * In order to reduce various lock holder preemption latencies provide an
1846  * interface to see if a vCPU is currently running or not.
1847  *
1848  * This allows us to terminate optimistic spin loops and block, analogous to
1849  * the native optimistic spin heuristic of testing if the lock owner task is
1850  * running or not.
1851  */
1852 #ifndef vcpu_is_preempted
1853 static inline bool vcpu_is_preempted(int cpu)
1854 {
1855 	return false;
1856 }
1857 #endif
1858 
1859 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
1860 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
1861 
1862 #ifndef TASK_SIZE_OF
1863 #define TASK_SIZE_OF(tsk)	TASK_SIZE
1864 #endif
1865 
1866 #ifdef CONFIG_RSEQ
1867 
1868 /*
1869  * Map the event mask on the user-space ABI enum rseq_cs_flags
1870  * for direct mask checks.
1871  */
1872 enum rseq_event_mask_bits {
1873 	RSEQ_EVENT_PREEMPT_BIT	= RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT_BIT,
1874 	RSEQ_EVENT_SIGNAL_BIT	= RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL_BIT,
1875 	RSEQ_EVENT_MIGRATE_BIT	= RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE_BIT,
1876 };
1877 
1878 enum rseq_event_mask {
1879 	RSEQ_EVENT_PREEMPT	= (1U << RSEQ_EVENT_PREEMPT_BIT),
1880 	RSEQ_EVENT_SIGNAL	= (1U << RSEQ_EVENT_SIGNAL_BIT),
1881 	RSEQ_EVENT_MIGRATE	= (1U << RSEQ_EVENT_MIGRATE_BIT),
1882 };
1883 
1884 static inline void rseq_set_notify_resume(struct task_struct *t)
1885 {
1886 	if (t->rseq)
1887 		set_tsk_thread_flag(t, TIF_NOTIFY_RESUME);
1888 }
1889 
1890 void __rseq_handle_notify_resume(struct ksignal *sig, struct pt_regs *regs);
1891 
1892 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
1893 					     struct pt_regs *regs)
1894 {
1895 	if (current->rseq)
1896 		__rseq_handle_notify_resume(ksig, regs);
1897 }
1898 
1899 static inline void rseq_signal_deliver(struct ksignal *ksig,
1900 				       struct pt_regs *regs)
1901 {
1902 	preempt_disable();
1903 	__set_bit(RSEQ_EVENT_SIGNAL_BIT, &current->rseq_event_mask);
1904 	preempt_enable();
1905 	rseq_handle_notify_resume(ksig, regs);
1906 }
1907 
1908 /* rseq_preempt() requires preemption to be disabled. */
1909 static inline void rseq_preempt(struct task_struct *t)
1910 {
1911 	__set_bit(RSEQ_EVENT_PREEMPT_BIT, &t->rseq_event_mask);
1912 	rseq_set_notify_resume(t);
1913 }
1914 
1915 /* rseq_migrate() requires preemption to be disabled. */
1916 static inline void rseq_migrate(struct task_struct *t)
1917 {
1918 	__set_bit(RSEQ_EVENT_MIGRATE_BIT, &t->rseq_event_mask);
1919 	rseq_set_notify_resume(t);
1920 }
1921 
1922 /*
1923  * If parent process has a registered restartable sequences area, the
1924  * child inherits. Only applies when forking a process, not a thread.
1925  */
1926 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
1927 {
1928 	if (clone_flags & CLONE_THREAD) {
1929 		t->rseq = NULL;
1930 		t->rseq_sig = 0;
1931 		t->rseq_event_mask = 0;
1932 	} else {
1933 		t->rseq = current->rseq;
1934 		t->rseq_sig = current->rseq_sig;
1935 		t->rseq_event_mask = current->rseq_event_mask;
1936 	}
1937 }
1938 
1939 static inline void rseq_execve(struct task_struct *t)
1940 {
1941 	t->rseq = NULL;
1942 	t->rseq_sig = 0;
1943 	t->rseq_event_mask = 0;
1944 }
1945 
1946 #else
1947 
1948 static inline void rseq_set_notify_resume(struct task_struct *t)
1949 {
1950 }
1951 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
1952 					     struct pt_regs *regs)
1953 {
1954 }
1955 static inline void rseq_signal_deliver(struct ksignal *ksig,
1956 				       struct pt_regs *regs)
1957 {
1958 }
1959 static inline void rseq_preempt(struct task_struct *t)
1960 {
1961 }
1962 static inline void rseq_migrate(struct task_struct *t)
1963 {
1964 }
1965 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
1966 {
1967 }
1968 static inline void rseq_execve(struct task_struct *t)
1969 {
1970 }
1971 
1972 #endif
1973 
1974 void __exit_umh(struct task_struct *tsk);
1975 
1976 static inline void exit_umh(struct task_struct *tsk)
1977 {
1978 	if (unlikely(tsk->flags & PF_UMH))
1979 		__exit_umh(tsk);
1980 }
1981 
1982 #ifdef CONFIG_DEBUG_RSEQ
1983 
1984 void rseq_syscall(struct pt_regs *regs);
1985 
1986 #else
1987 
1988 static inline void rseq_syscall(struct pt_regs *regs)
1989 {
1990 }
1991 
1992 #endif
1993 
1994 const struct sched_avg *sched_trace_cfs_rq_avg(struct cfs_rq *cfs_rq);
1995 char *sched_trace_cfs_rq_path(struct cfs_rq *cfs_rq, char *str, int len);
1996 int sched_trace_cfs_rq_cpu(struct cfs_rq *cfs_rq);
1997 
1998 const struct sched_avg *sched_trace_rq_avg_rt(struct rq *rq);
1999 const struct sched_avg *sched_trace_rq_avg_dl(struct rq *rq);
2000 const struct sched_avg *sched_trace_rq_avg_irq(struct rq *rq);
2001 
2002 int sched_trace_rq_cpu(struct rq *rq);
2003 
2004 const struct cpumask *sched_trace_rd_span(struct root_domain *rd);
2005 
2006 #endif
2007