xref: /linux/include/linux/sched.h (revision 74ce1896c6c65b2f8cccbf59162d542988835835)
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): */
31 struct audit_context;
32 struct backing_dev_info;
33 struct bio_list;
34 struct blk_plug;
35 struct cfs_rq;
36 struct fs_struct;
37 struct futex_pi_state;
38 struct io_context;
39 struct mempolicy;
40 struct nameidata;
41 struct nsproxy;
42 struct perf_event_context;
43 struct pid_namespace;
44 struct pipe_inode_info;
45 struct rcu_node;
46 struct reclaim_state;
47 struct robust_list_head;
48 struct sched_attr;
49 struct sched_param;
50 struct seq_file;
51 struct sighand_struct;
52 struct signal_struct;
53 struct task_delay_info;
54 struct 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 
169 extern cpumask_var_t			cpu_isolated_map;
170 
171 extern void scheduler_tick(void);
172 
173 #define	MAX_SCHEDULE_TIMEOUT		LONG_MAX
174 
175 extern long schedule_timeout(long timeout);
176 extern long schedule_timeout_interruptible(long timeout);
177 extern long schedule_timeout_killable(long timeout);
178 extern long schedule_timeout_uninterruptible(long timeout);
179 extern long schedule_timeout_idle(long timeout);
180 asmlinkage void schedule(void);
181 extern void schedule_preempt_disabled(void);
182 
183 extern int __must_check io_schedule_prepare(void);
184 extern void io_schedule_finish(int token);
185 extern long io_schedule_timeout(long timeout);
186 extern void io_schedule(void);
187 
188 /**
189  * struct prev_cputime - snapshot 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  */
197 struct 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  */
215 struct 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 
226 enum vtime_state {
227 	/* Task is sleeping or running in a CPU with VTIME inactive: */
228 	VTIME_INACTIVE = 0,
229 	/* Task runs in userspace in a CPU with VTIME active: */
230 	VTIME_USER,
231 	/* Task runs in kernelspace in a CPU with VTIME active: */
232 	VTIME_SYS,
233 };
234 
235 struct vtime {
236 	seqcount_t		seqcount;
237 	unsigned long long	starttime;
238 	enum vtime_state	state;
239 	u64			utime;
240 	u64			stime;
241 	u64			gtime;
242 };
243 
244 struct sched_info {
245 #ifdef CONFIG_SCHED_INFO
246 	/* Cumulative counters: */
247 
248 	/* # of times we have run on this CPU: */
249 	unsigned long			pcount;
250 
251 	/* Time spent waiting on a runqueue: */
252 	unsigned long long		run_delay;
253 
254 	/* Timestamps: */
255 
256 	/* When did we last run on a CPU? */
257 	unsigned long long		last_arrival;
258 
259 	/* When were we last queued to run? */
260 	unsigned long long		last_queued;
261 
262 #endif /* CONFIG_SCHED_INFO */
263 };
264 
265 /*
266  * Integer metrics need fixed point arithmetic, e.g., sched/fair
267  * has a few: load, load_avg, util_avg, freq, and capacity.
268  *
269  * We define a basic fixed point arithmetic range, and then formalize
270  * all these metrics based on that basic range.
271  */
272 # define SCHED_FIXEDPOINT_SHIFT		10
273 # define SCHED_FIXEDPOINT_SCALE		(1L << SCHED_FIXEDPOINT_SHIFT)
274 
275 struct load_weight {
276 	unsigned long			weight;
277 	u32				inv_weight;
278 };
279 
280 /*
281  * The load_avg/util_avg accumulates an infinite geometric series
282  * (see __update_load_avg() in kernel/sched/fair.c).
283  *
284  * [load_avg definition]
285  *
286  *   load_avg = runnable% * scale_load_down(load)
287  *
288  * where runnable% is the time ratio that a sched_entity is runnable.
289  * For cfs_rq, it is the aggregated load_avg of all runnable and
290  * blocked sched_entities.
291  *
292  * load_avg may also take frequency scaling into account:
293  *
294  *   load_avg = runnable% * scale_load_down(load) * freq%
295  *
296  * where freq% is the CPU frequency normalized to the highest frequency.
297  *
298  * [util_avg definition]
299  *
300  *   util_avg = running% * SCHED_CAPACITY_SCALE
301  *
302  * where running% is the time ratio that a sched_entity is running on
303  * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
304  * and blocked sched_entities.
305  *
306  * util_avg may also factor frequency scaling and CPU capacity scaling:
307  *
308  *   util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
309  *
310  * where freq% is the same as above, and capacity% is the CPU capacity
311  * normalized to the greatest capacity (due to uarch differences, etc).
312  *
313  * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
314  * themselves are in the range of [0, 1]. To do fixed point arithmetics,
315  * we therefore scale them to as large a range as necessary. This is for
316  * example reflected by util_avg's SCHED_CAPACITY_SCALE.
317  *
318  * [Overflow issue]
319  *
320  * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
321  * with the highest load (=88761), always runnable on a single cfs_rq,
322  * and should not overflow as the number already hits PID_MAX_LIMIT.
323  *
324  * For all other cases (including 32-bit kernels), struct load_weight's
325  * weight will overflow first before we do, because:
326  *
327  *    Max(load_avg) <= Max(load.weight)
328  *
329  * Then it is the load_weight's responsibility to consider overflow
330  * issues.
331  */
332 struct sched_avg {
333 	u64				last_update_time;
334 	u64				load_sum;
335 	u32				util_sum;
336 	u32				period_contrib;
337 	unsigned long			load_avg;
338 	unsigned long			util_avg;
339 };
340 
341 struct sched_statistics {
342 #ifdef CONFIG_SCHEDSTATS
343 	u64				wait_start;
344 	u64				wait_max;
345 	u64				wait_count;
346 	u64				wait_sum;
347 	u64				iowait_count;
348 	u64				iowait_sum;
349 
350 	u64				sleep_start;
351 	u64				sleep_max;
352 	s64				sum_sleep_runtime;
353 
354 	u64				block_start;
355 	u64				block_max;
356 	u64				exec_max;
357 	u64				slice_max;
358 
359 	u64				nr_migrations_cold;
360 	u64				nr_failed_migrations_affine;
361 	u64				nr_failed_migrations_running;
362 	u64				nr_failed_migrations_hot;
363 	u64				nr_forced_migrations;
364 
365 	u64				nr_wakeups;
366 	u64				nr_wakeups_sync;
367 	u64				nr_wakeups_migrate;
368 	u64				nr_wakeups_local;
369 	u64				nr_wakeups_remote;
370 	u64				nr_wakeups_affine;
371 	u64				nr_wakeups_affine_attempts;
372 	u64				nr_wakeups_passive;
373 	u64				nr_wakeups_idle;
374 #endif
375 };
376 
377 struct sched_entity {
378 	/* For load-balancing: */
379 	struct load_weight		load;
380 	struct rb_node			run_node;
381 	struct list_head		group_node;
382 	unsigned int			on_rq;
383 
384 	u64				exec_start;
385 	u64				sum_exec_runtime;
386 	u64				vruntime;
387 	u64				prev_sum_exec_runtime;
388 
389 	u64				nr_migrations;
390 
391 	struct sched_statistics		statistics;
392 
393 #ifdef CONFIG_FAIR_GROUP_SCHED
394 	int				depth;
395 	struct sched_entity		*parent;
396 	/* rq on which this entity is (to be) queued: */
397 	struct cfs_rq			*cfs_rq;
398 	/* rq "owned" by this entity/group: */
399 	struct cfs_rq			*my_q;
400 #endif
401 
402 #ifdef CONFIG_SMP
403 	/*
404 	 * Per entity load average tracking.
405 	 *
406 	 * Put into separate cache line so it does not
407 	 * collide with read-mostly values above.
408 	 */
409 	struct sched_avg		avg ____cacheline_aligned_in_smp;
410 #endif
411 };
412 
413 struct sched_rt_entity {
414 	struct list_head		run_list;
415 	unsigned long			timeout;
416 	unsigned long			watchdog_stamp;
417 	unsigned int			time_slice;
418 	unsigned short			on_rq;
419 	unsigned short			on_list;
420 
421 	struct sched_rt_entity		*back;
422 #ifdef CONFIG_RT_GROUP_SCHED
423 	struct sched_rt_entity		*parent;
424 	/* rq on which this entity is (to be) queued: */
425 	struct rt_rq			*rt_rq;
426 	/* rq "owned" by this entity/group: */
427 	struct rt_rq			*my_q;
428 #endif
429 } __randomize_layout;
430 
431 struct sched_dl_entity {
432 	struct rb_node			rb_node;
433 
434 	/*
435 	 * Original scheduling parameters. Copied here from sched_attr
436 	 * during sched_setattr(), they will remain the same until
437 	 * the next sched_setattr().
438 	 */
439 	u64				dl_runtime;	/* Maximum runtime for each instance	*/
440 	u64				dl_deadline;	/* Relative deadline of each instance	*/
441 	u64				dl_period;	/* Separation of two instances (period) */
442 	u64				dl_bw;		/* dl_runtime / dl_period		*/
443 	u64				dl_density;	/* dl_runtime / dl_deadline		*/
444 
445 	/*
446 	 * Actual scheduling parameters. Initialized with the values above,
447 	 * they are continously updated during task execution. Note that
448 	 * the remaining runtime could be < 0 in case we are in overrun.
449 	 */
450 	s64				runtime;	/* Remaining runtime for this instance	*/
451 	u64				deadline;	/* Absolute deadline for this instance	*/
452 	unsigned int			flags;		/* Specifying the scheduler behaviour	*/
453 
454 	/*
455 	 * Some bool flags:
456 	 *
457 	 * @dl_throttled tells if we exhausted the runtime. If so, the
458 	 * task has to wait for a replenishment to be performed at the
459 	 * next firing of dl_timer.
460 	 *
461 	 * @dl_boosted tells if we are boosted due to DI. If so we are
462 	 * outside bandwidth enforcement mechanism (but only until we
463 	 * exit the critical section);
464 	 *
465 	 * @dl_yielded tells if task gave up the CPU before consuming
466 	 * all its available runtime during the last job.
467 	 *
468 	 * @dl_non_contending tells if the task is inactive while still
469 	 * contributing to the active utilization. In other words, it
470 	 * indicates if the inactive timer has been armed and its handler
471 	 * has not been executed yet. This flag is useful to avoid race
472 	 * conditions between the inactive timer handler and the wakeup
473 	 * code.
474 	 */
475 	int				dl_throttled;
476 	int				dl_boosted;
477 	int				dl_yielded;
478 	int				dl_non_contending;
479 
480 	/*
481 	 * Bandwidth enforcement timer. Each -deadline task has its
482 	 * own bandwidth to be enforced, thus we need one timer per task.
483 	 */
484 	struct hrtimer			dl_timer;
485 
486 	/*
487 	 * Inactive timer, responsible for decreasing the active utilization
488 	 * at the "0-lag time". When a -deadline task blocks, it contributes
489 	 * to GRUB's active utilization until the "0-lag time", hence a
490 	 * timer is needed to decrease the active utilization at the correct
491 	 * time.
492 	 */
493 	struct hrtimer inactive_timer;
494 };
495 
496 union rcu_special {
497 	struct {
498 		u8			blocked;
499 		u8			need_qs;
500 		u8			exp_need_qs;
501 
502 		/* Otherwise the compiler can store garbage here: */
503 		u8			pad;
504 	} b; /* Bits. */
505 	u32 s; /* Set of bits. */
506 };
507 
508 enum perf_event_task_context {
509 	perf_invalid_context = -1,
510 	perf_hw_context = 0,
511 	perf_sw_context,
512 	perf_nr_task_contexts,
513 };
514 
515 struct wake_q_node {
516 	struct wake_q_node *next;
517 };
518 
519 struct task_struct {
520 #ifdef CONFIG_THREAD_INFO_IN_TASK
521 	/*
522 	 * For reasons of header soup (see current_thread_info()), this
523 	 * must be the first element of task_struct.
524 	 */
525 	struct thread_info		thread_info;
526 #endif
527 	/* -1 unrunnable, 0 runnable, >0 stopped: */
528 	volatile long			state;
529 
530 	/*
531 	 * This begins the randomizable portion of task_struct. Only
532 	 * scheduling-critical items should be added above here.
533 	 */
534 	randomized_struct_fields_start
535 
536 	void				*stack;
537 	atomic_t			usage;
538 	/* Per task flags (PF_*), defined further below: */
539 	unsigned int			flags;
540 	unsigned int			ptrace;
541 
542 #ifdef CONFIG_SMP
543 	struct llist_node		wake_entry;
544 	int				on_cpu;
545 #ifdef CONFIG_THREAD_INFO_IN_TASK
546 	/* Current CPU: */
547 	unsigned int			cpu;
548 #endif
549 	unsigned int			wakee_flips;
550 	unsigned long			wakee_flip_decay_ts;
551 	struct task_struct		*last_wakee;
552 
553 	int				wake_cpu;
554 #endif
555 	int				on_rq;
556 
557 	int				prio;
558 	int				static_prio;
559 	int				normal_prio;
560 	unsigned int			rt_priority;
561 
562 	const struct sched_class	*sched_class;
563 	struct sched_entity		se;
564 	struct sched_rt_entity		rt;
565 #ifdef CONFIG_CGROUP_SCHED
566 	struct task_group		*sched_task_group;
567 #endif
568 	struct sched_dl_entity		dl;
569 
570 #ifdef CONFIG_PREEMPT_NOTIFIERS
571 	/* List of struct preempt_notifier: */
572 	struct hlist_head		preempt_notifiers;
573 #endif
574 
575 #ifdef CONFIG_BLK_DEV_IO_TRACE
576 	unsigned int			btrace_seq;
577 #endif
578 
579 	unsigned int			policy;
580 	int				nr_cpus_allowed;
581 	cpumask_t			cpus_allowed;
582 
583 #ifdef CONFIG_PREEMPT_RCU
584 	int				rcu_read_lock_nesting;
585 	union rcu_special		rcu_read_unlock_special;
586 	struct list_head		rcu_node_entry;
587 	struct rcu_node			*rcu_blocked_node;
588 #endif /* #ifdef CONFIG_PREEMPT_RCU */
589 
590 #ifdef CONFIG_TASKS_RCU
591 	unsigned long			rcu_tasks_nvcsw;
592 	u8				rcu_tasks_holdout;
593 	u8				rcu_tasks_idx;
594 	int				rcu_tasks_idle_cpu;
595 	struct list_head		rcu_tasks_holdout_list;
596 #endif /* #ifdef CONFIG_TASKS_RCU */
597 
598 	struct sched_info		sched_info;
599 
600 	struct list_head		tasks;
601 #ifdef CONFIG_SMP
602 	struct plist_node		pushable_tasks;
603 	struct rb_node			pushable_dl_tasks;
604 #endif
605 
606 	struct mm_struct		*mm;
607 	struct mm_struct		*active_mm;
608 
609 	/* Per-thread vma caching: */
610 	struct vmacache			vmacache;
611 
612 #ifdef SPLIT_RSS_COUNTING
613 	struct task_rss_stat		rss_stat;
614 #endif
615 	int				exit_state;
616 	int				exit_code;
617 	int				exit_signal;
618 	/* The signal sent when the parent dies: */
619 	int				pdeath_signal;
620 	/* JOBCTL_*, siglock protected: */
621 	unsigned long			jobctl;
622 
623 	/* Used for emulating ABI behavior of previous Linux versions: */
624 	unsigned int			personality;
625 
626 	/* Scheduler bits, serialized by scheduler locks: */
627 	unsigned			sched_reset_on_fork:1;
628 	unsigned			sched_contributes_to_load:1;
629 	unsigned			sched_migrated:1;
630 	unsigned			sched_remote_wakeup:1;
631 	/* Force alignment to the next boundary: */
632 	unsigned			:0;
633 
634 	/* Unserialized, strictly 'current' */
635 
636 	/* Bit to tell LSMs we're in execve(): */
637 	unsigned			in_execve:1;
638 	unsigned			in_iowait:1;
639 #ifndef TIF_RESTORE_SIGMASK
640 	unsigned			restore_sigmask:1;
641 #endif
642 #ifdef CONFIG_MEMCG
643 	unsigned			memcg_may_oom:1;
644 #ifndef CONFIG_SLOB
645 	unsigned			memcg_kmem_skip_account:1;
646 #endif
647 #endif
648 #ifdef CONFIG_COMPAT_BRK
649 	unsigned			brk_randomized:1;
650 #endif
651 #ifdef CONFIG_CGROUPS
652 	/* disallow userland-initiated cgroup migration */
653 	unsigned			no_cgroup_migration:1;
654 #endif
655 
656 	unsigned long			atomic_flags; /* Flags requiring atomic access. */
657 
658 	struct restart_block		restart_block;
659 
660 	pid_t				pid;
661 	pid_t				tgid;
662 
663 #ifdef CONFIG_CC_STACKPROTECTOR
664 	/* Canary value for the -fstack-protector GCC feature: */
665 	unsigned long			stack_canary;
666 #endif
667 	/*
668 	 * Pointers to the (original) parent process, youngest child, younger sibling,
669 	 * older sibling, respectively.  (p->father can be replaced with
670 	 * p->real_parent->pid)
671 	 */
672 
673 	/* Real parent process: */
674 	struct task_struct __rcu	*real_parent;
675 
676 	/* Recipient of SIGCHLD, wait4() reports: */
677 	struct task_struct __rcu	*parent;
678 
679 	/*
680 	 * Children/sibling form the list of natural children:
681 	 */
682 	struct list_head		children;
683 	struct list_head		sibling;
684 	struct task_struct		*group_leader;
685 
686 	/*
687 	 * 'ptraced' is the list of tasks this task is using ptrace() on.
688 	 *
689 	 * This includes both natural children and PTRACE_ATTACH targets.
690 	 * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
691 	 */
692 	struct list_head		ptraced;
693 	struct list_head		ptrace_entry;
694 
695 	/* PID/PID hash table linkage. */
696 	struct pid_link			pids[PIDTYPE_MAX];
697 	struct list_head		thread_group;
698 	struct list_head		thread_node;
699 
700 	struct completion		*vfork_done;
701 
702 	/* CLONE_CHILD_SETTID: */
703 	int __user			*set_child_tid;
704 
705 	/* CLONE_CHILD_CLEARTID: */
706 	int __user			*clear_child_tid;
707 
708 	u64				utime;
709 	u64				stime;
710 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
711 	u64				utimescaled;
712 	u64				stimescaled;
713 #endif
714 	u64				gtime;
715 	struct prev_cputime		prev_cputime;
716 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
717 	struct vtime			vtime;
718 #endif
719 
720 #ifdef CONFIG_NO_HZ_FULL
721 	atomic_t			tick_dep_mask;
722 #endif
723 	/* Context switch counts: */
724 	unsigned long			nvcsw;
725 	unsigned long			nivcsw;
726 
727 	/* Monotonic time in nsecs: */
728 	u64				start_time;
729 
730 	/* Boot based time in nsecs: */
731 	u64				real_start_time;
732 
733 	/* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
734 	unsigned long			min_flt;
735 	unsigned long			maj_flt;
736 
737 #ifdef CONFIG_POSIX_TIMERS
738 	struct task_cputime		cputime_expires;
739 	struct list_head		cpu_timers[3];
740 #endif
741 
742 	/* Process credentials: */
743 
744 	/* Tracer's credentials at attach: */
745 	const struct cred __rcu		*ptracer_cred;
746 
747 	/* Objective and real subjective task credentials (COW): */
748 	const struct cred __rcu		*real_cred;
749 
750 	/* Effective (overridable) subjective task credentials (COW): */
751 	const struct cred __rcu		*cred;
752 
753 	/*
754 	 * executable name, excluding path.
755 	 *
756 	 * - normally initialized setup_new_exec()
757 	 * - access it with [gs]et_task_comm()
758 	 * - lock it with task_lock()
759 	 */
760 	char				comm[TASK_COMM_LEN];
761 
762 	struct nameidata		*nameidata;
763 
764 #ifdef CONFIG_SYSVIPC
765 	struct sysv_sem			sysvsem;
766 	struct sysv_shm			sysvshm;
767 #endif
768 #ifdef CONFIG_DETECT_HUNG_TASK
769 	unsigned long			last_switch_count;
770 #endif
771 	/* Filesystem information: */
772 	struct fs_struct		*fs;
773 
774 	/* Open file information: */
775 	struct files_struct		*files;
776 
777 	/* Namespaces: */
778 	struct nsproxy			*nsproxy;
779 
780 	/* Signal handlers: */
781 	struct signal_struct		*signal;
782 	struct sighand_struct		*sighand;
783 	sigset_t			blocked;
784 	sigset_t			real_blocked;
785 	/* Restored if set_restore_sigmask() was used: */
786 	sigset_t			saved_sigmask;
787 	struct sigpending		pending;
788 	unsigned long			sas_ss_sp;
789 	size_t				sas_ss_size;
790 	unsigned int			sas_ss_flags;
791 
792 	struct callback_head		*task_works;
793 
794 	struct audit_context		*audit_context;
795 #ifdef CONFIG_AUDITSYSCALL
796 	kuid_t				loginuid;
797 	unsigned int			sessionid;
798 #endif
799 	struct seccomp			seccomp;
800 
801 	/* Thread group tracking: */
802 	u32				parent_exec_id;
803 	u32				self_exec_id;
804 
805 	/* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
806 	spinlock_t			alloc_lock;
807 
808 	/* Protection of the PI data structures: */
809 	raw_spinlock_t			pi_lock;
810 
811 	struct wake_q_node		wake_q;
812 
813 #ifdef CONFIG_RT_MUTEXES
814 	/* PI waiters blocked on a rt_mutex held by this task: */
815 	struct rb_root_cached		pi_waiters;
816 	/* Updated under owner's pi_lock and rq lock */
817 	struct task_struct		*pi_top_task;
818 	/* Deadlock detection and priority inheritance handling: */
819 	struct rt_mutex_waiter		*pi_blocked_on;
820 #endif
821 
822 #ifdef CONFIG_DEBUG_MUTEXES
823 	/* Mutex deadlock detection: */
824 	struct mutex_waiter		*blocked_on;
825 #endif
826 
827 #ifdef CONFIG_TRACE_IRQFLAGS
828 	unsigned int			irq_events;
829 	unsigned long			hardirq_enable_ip;
830 	unsigned long			hardirq_disable_ip;
831 	unsigned int			hardirq_enable_event;
832 	unsigned int			hardirq_disable_event;
833 	int				hardirqs_enabled;
834 	int				hardirq_context;
835 	unsigned long			softirq_disable_ip;
836 	unsigned long			softirq_enable_ip;
837 	unsigned int			softirq_disable_event;
838 	unsigned int			softirq_enable_event;
839 	int				softirqs_enabled;
840 	int				softirq_context;
841 #endif
842 
843 #ifdef CONFIG_LOCKDEP
844 # define MAX_LOCK_DEPTH			48UL
845 	u64				curr_chain_key;
846 	int				lockdep_depth;
847 	unsigned int			lockdep_recursion;
848 	struct held_lock		held_locks[MAX_LOCK_DEPTH];
849 #endif
850 
851 #ifdef CONFIG_LOCKDEP_CROSSRELEASE
852 #define MAX_XHLOCKS_NR 64UL
853 	struct hist_lock *xhlocks; /* Crossrelease history locks */
854 	unsigned int xhlock_idx;
855 	/* For restoring at history boundaries */
856 	unsigned int xhlock_idx_hist[XHLOCK_CTX_NR];
857 	unsigned int hist_id;
858 	/* For overwrite check at each context exit */
859 	unsigned int hist_id_save[XHLOCK_CTX_NR];
860 #endif
861 
862 #ifdef CONFIG_UBSAN
863 	unsigned int			in_ubsan;
864 #endif
865 
866 	/* Journalling filesystem info: */
867 	void				*journal_info;
868 
869 	/* Stacked block device info: */
870 	struct bio_list			*bio_list;
871 
872 #ifdef CONFIG_BLOCK
873 	/* Stack plugging: */
874 	struct blk_plug			*plug;
875 #endif
876 
877 	/* VM state: */
878 	struct reclaim_state		*reclaim_state;
879 
880 	struct backing_dev_info		*backing_dev_info;
881 
882 	struct io_context		*io_context;
883 
884 	/* Ptrace state: */
885 	unsigned long			ptrace_message;
886 	siginfo_t			*last_siginfo;
887 
888 	struct task_io_accounting	ioac;
889 #ifdef CONFIG_TASK_XACCT
890 	/* Accumulated RSS usage: */
891 	u64				acct_rss_mem1;
892 	/* Accumulated virtual memory usage: */
893 	u64				acct_vm_mem1;
894 	/* stime + utime since last update: */
895 	u64				acct_timexpd;
896 #endif
897 #ifdef CONFIG_CPUSETS
898 	/* Protected by ->alloc_lock: */
899 	nodemask_t			mems_allowed;
900 	/* Seqence number to catch updates: */
901 	seqcount_t			mems_allowed_seq;
902 	int				cpuset_mem_spread_rotor;
903 	int				cpuset_slab_spread_rotor;
904 #endif
905 #ifdef CONFIG_CGROUPS
906 	/* Control Group info protected by css_set_lock: */
907 	struct css_set __rcu		*cgroups;
908 	/* cg_list protected by css_set_lock and tsk->alloc_lock: */
909 	struct list_head		cg_list;
910 #endif
911 #ifdef CONFIG_INTEL_RDT
912 	u32				closid;
913 	u32				rmid;
914 #endif
915 #ifdef CONFIG_FUTEX
916 	struct robust_list_head __user	*robust_list;
917 #ifdef CONFIG_COMPAT
918 	struct compat_robust_list_head __user *compat_robust_list;
919 #endif
920 	struct list_head		pi_state_list;
921 	struct futex_pi_state		*pi_state_cache;
922 #endif
923 #ifdef CONFIG_PERF_EVENTS
924 	struct perf_event_context	*perf_event_ctxp[perf_nr_task_contexts];
925 	struct mutex			perf_event_mutex;
926 	struct list_head		perf_event_list;
927 #endif
928 #ifdef CONFIG_DEBUG_PREEMPT
929 	unsigned long			preempt_disable_ip;
930 #endif
931 #ifdef CONFIG_NUMA
932 	/* Protected by alloc_lock: */
933 	struct mempolicy		*mempolicy;
934 	short				il_prev;
935 	short				pref_node_fork;
936 #endif
937 #ifdef CONFIG_NUMA_BALANCING
938 	int				numa_scan_seq;
939 	unsigned int			numa_scan_period;
940 	unsigned int			numa_scan_period_max;
941 	int				numa_preferred_nid;
942 	unsigned long			numa_migrate_retry;
943 	/* Migration stamp: */
944 	u64				node_stamp;
945 	u64				last_task_numa_placement;
946 	u64				last_sum_exec_runtime;
947 	struct callback_head		numa_work;
948 
949 	struct list_head		numa_entry;
950 	struct numa_group		*numa_group;
951 
952 	/*
953 	 * numa_faults is an array split into four regions:
954 	 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
955 	 * in this precise order.
956 	 *
957 	 * faults_memory: Exponential decaying average of faults on a per-node
958 	 * basis. Scheduling placement decisions are made based on these
959 	 * counts. The values remain static for the duration of a PTE scan.
960 	 * faults_cpu: Track the nodes the process was running on when a NUMA
961 	 * hinting fault was incurred.
962 	 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
963 	 * during the current scan window. When the scan completes, the counts
964 	 * in faults_memory and faults_cpu decay and these values are copied.
965 	 */
966 	unsigned long			*numa_faults;
967 	unsigned long			total_numa_faults;
968 
969 	/*
970 	 * numa_faults_locality tracks if faults recorded during the last
971 	 * scan window were remote/local or failed to migrate. The task scan
972 	 * period is adapted based on the locality of the faults with different
973 	 * weights depending on whether they were shared or private faults
974 	 */
975 	unsigned long			numa_faults_locality[3];
976 
977 	unsigned long			numa_pages_migrated;
978 #endif /* CONFIG_NUMA_BALANCING */
979 
980 	struct tlbflush_unmap_batch	tlb_ubc;
981 
982 	struct rcu_head			rcu;
983 
984 	/* Cache last used pipe for splice(): */
985 	struct pipe_inode_info		*splice_pipe;
986 
987 	struct page_frag		task_frag;
988 
989 #ifdef CONFIG_TASK_DELAY_ACCT
990 	struct task_delay_info		*delays;
991 #endif
992 
993 #ifdef CONFIG_FAULT_INJECTION
994 	int				make_it_fail;
995 	unsigned int			fail_nth;
996 #endif
997 	/*
998 	 * When (nr_dirtied >= nr_dirtied_pause), it's time to call
999 	 * balance_dirty_pages() for a dirty throttling pause:
1000 	 */
1001 	int				nr_dirtied;
1002 	int				nr_dirtied_pause;
1003 	/* Start of a write-and-pause period: */
1004 	unsigned long			dirty_paused_when;
1005 
1006 #ifdef CONFIG_LATENCYTOP
1007 	int				latency_record_count;
1008 	struct latency_record		latency_record[LT_SAVECOUNT];
1009 #endif
1010 	/*
1011 	 * Time slack values; these are used to round up poll() and
1012 	 * select() etc timeout values. These are in nanoseconds.
1013 	 */
1014 	u64				timer_slack_ns;
1015 	u64				default_timer_slack_ns;
1016 
1017 #ifdef CONFIG_KASAN
1018 	unsigned int			kasan_depth;
1019 #endif
1020 
1021 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1022 	/* Index of current stored address in ret_stack: */
1023 	int				curr_ret_stack;
1024 
1025 	/* Stack of return addresses for return function tracing: */
1026 	struct ftrace_ret_stack		*ret_stack;
1027 
1028 	/* Timestamp for last schedule: */
1029 	unsigned long long		ftrace_timestamp;
1030 
1031 	/*
1032 	 * Number of functions that haven't been traced
1033 	 * because of depth overrun:
1034 	 */
1035 	atomic_t			trace_overrun;
1036 
1037 	/* Pause tracing: */
1038 	atomic_t			tracing_graph_pause;
1039 #endif
1040 
1041 #ifdef CONFIG_TRACING
1042 	/* State flags for use by tracers: */
1043 	unsigned long			trace;
1044 
1045 	/* Bitmask and counter of trace recursion: */
1046 	unsigned long			trace_recursion;
1047 #endif /* CONFIG_TRACING */
1048 
1049 #ifdef CONFIG_KCOV
1050 	/* Coverage collection mode enabled for this task (0 if disabled): */
1051 	enum kcov_mode			kcov_mode;
1052 
1053 	/* Size of the kcov_area: */
1054 	unsigned int			kcov_size;
1055 
1056 	/* Buffer for coverage collection: */
1057 	void				*kcov_area;
1058 
1059 	/* KCOV descriptor wired with this task or NULL: */
1060 	struct kcov			*kcov;
1061 #endif
1062 
1063 #ifdef CONFIG_MEMCG
1064 	struct mem_cgroup		*memcg_in_oom;
1065 	gfp_t				memcg_oom_gfp_mask;
1066 	int				memcg_oom_order;
1067 
1068 	/* Number of pages to reclaim on returning to userland: */
1069 	unsigned int			memcg_nr_pages_over_high;
1070 #endif
1071 
1072 #ifdef CONFIG_UPROBES
1073 	struct uprobe_task		*utask;
1074 #endif
1075 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1076 	unsigned int			sequential_io;
1077 	unsigned int			sequential_io_avg;
1078 #endif
1079 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1080 	unsigned long			task_state_change;
1081 #endif
1082 	int				pagefault_disabled;
1083 #ifdef CONFIG_MMU
1084 	struct task_struct		*oom_reaper_list;
1085 #endif
1086 #ifdef CONFIG_VMAP_STACK
1087 	struct vm_struct		*stack_vm_area;
1088 #endif
1089 #ifdef CONFIG_THREAD_INFO_IN_TASK
1090 	/* A live task holds one reference: */
1091 	atomic_t			stack_refcount;
1092 #endif
1093 #ifdef CONFIG_LIVEPATCH
1094 	int patch_state;
1095 #endif
1096 #ifdef CONFIG_SECURITY
1097 	/* Used by LSM modules for access restriction: */
1098 	void				*security;
1099 #endif
1100 
1101 	/*
1102 	 * New fields for task_struct should be added above here, so that
1103 	 * they are included in the randomized portion of task_struct.
1104 	 */
1105 	randomized_struct_fields_end
1106 
1107 	/* CPU-specific state of this task: */
1108 	struct thread_struct		thread;
1109 
1110 	/*
1111 	 * WARNING: on x86, 'thread_struct' contains a variable-sized
1112 	 * structure.  It *MUST* be at the end of 'task_struct'.
1113 	 *
1114 	 * Do not put anything below here!
1115 	 */
1116 };
1117 
1118 static inline struct pid *task_pid(struct task_struct *task)
1119 {
1120 	return task->pids[PIDTYPE_PID].pid;
1121 }
1122 
1123 static inline struct pid *task_tgid(struct task_struct *task)
1124 {
1125 	return task->group_leader->pids[PIDTYPE_PID].pid;
1126 }
1127 
1128 /*
1129  * Without tasklist or RCU lock it is not safe to dereference
1130  * the result of task_pgrp/task_session even if task == current,
1131  * we can race with another thread doing sys_setsid/sys_setpgid.
1132  */
1133 static inline struct pid *task_pgrp(struct task_struct *task)
1134 {
1135 	return task->group_leader->pids[PIDTYPE_PGID].pid;
1136 }
1137 
1138 static inline struct pid *task_session(struct task_struct *task)
1139 {
1140 	return task->group_leader->pids[PIDTYPE_SID].pid;
1141 }
1142 
1143 /*
1144  * the helpers to get the task's different pids as they are seen
1145  * from various namespaces
1146  *
1147  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1148  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1149  *                     current.
1150  * task_xid_nr_ns()  : id seen from the ns specified;
1151  *
1152  * see also pid_nr() etc in include/linux/pid.h
1153  */
1154 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns);
1155 
1156 static inline pid_t task_pid_nr(struct task_struct *tsk)
1157 {
1158 	return tsk->pid;
1159 }
1160 
1161 static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1162 {
1163 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1164 }
1165 
1166 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1167 {
1168 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1169 }
1170 
1171 
1172 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1173 {
1174 	return tsk->tgid;
1175 }
1176 
1177 /**
1178  * pid_alive - check that a task structure is not stale
1179  * @p: Task structure to be checked.
1180  *
1181  * Test if a process is not yet dead (at most zombie state)
1182  * If pid_alive fails, then pointers within the task structure
1183  * can be stale and must not be dereferenced.
1184  *
1185  * Return: 1 if the process is alive. 0 otherwise.
1186  */
1187 static inline int pid_alive(const struct task_struct *p)
1188 {
1189 	return p->pids[PIDTYPE_PID].pid != NULL;
1190 }
1191 
1192 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1193 {
1194 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1195 }
1196 
1197 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1198 {
1199 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1200 }
1201 
1202 
1203 static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1204 {
1205 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1206 }
1207 
1208 static inline pid_t task_session_vnr(struct task_struct *tsk)
1209 {
1210 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1211 }
1212 
1213 static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1214 {
1215 	return __task_pid_nr_ns(tsk, __PIDTYPE_TGID, ns);
1216 }
1217 
1218 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1219 {
1220 	return __task_pid_nr_ns(tsk, __PIDTYPE_TGID, NULL);
1221 }
1222 
1223 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1224 {
1225 	pid_t pid = 0;
1226 
1227 	rcu_read_lock();
1228 	if (pid_alive(tsk))
1229 		pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1230 	rcu_read_unlock();
1231 
1232 	return pid;
1233 }
1234 
1235 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1236 {
1237 	return task_ppid_nr_ns(tsk, &init_pid_ns);
1238 }
1239 
1240 /* Obsolete, do not use: */
1241 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1242 {
1243 	return task_pgrp_nr_ns(tsk, &init_pid_ns);
1244 }
1245 
1246 static inline char task_state_to_char(struct task_struct *task)
1247 {
1248 	const char stat_nam[] = TASK_STATE_TO_CHAR_STR;
1249 	unsigned long state = task->state;
1250 
1251 	state = state ? __ffs(state) + 1 : 0;
1252 
1253 	/* Make sure the string lines up properly with the number of task states: */
1254 	BUILD_BUG_ON(sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1);
1255 
1256 	return state < sizeof(stat_nam) - 1 ? stat_nam[state] : '?';
1257 }
1258 
1259 /**
1260  * is_global_init - check if a task structure is init. Since init
1261  * is free to have sub-threads we need to check tgid.
1262  * @tsk: Task structure to be checked.
1263  *
1264  * Check if a task structure is the first user space task the kernel created.
1265  *
1266  * Return: 1 if the task structure is init. 0 otherwise.
1267  */
1268 static inline int is_global_init(struct task_struct *tsk)
1269 {
1270 	return task_tgid_nr(tsk) == 1;
1271 }
1272 
1273 extern struct pid *cad_pid;
1274 
1275 /*
1276  * Per process flags
1277  */
1278 #define PF_IDLE			0x00000002	/* I am an IDLE thread */
1279 #define PF_EXITING		0x00000004	/* Getting shut down */
1280 #define PF_EXITPIDONE		0x00000008	/* PI exit done on shut down */
1281 #define PF_VCPU			0x00000010	/* I'm a virtual CPU */
1282 #define PF_WQ_WORKER		0x00000020	/* I'm a workqueue worker */
1283 #define PF_FORKNOEXEC		0x00000040	/* Forked but didn't exec */
1284 #define PF_MCE_PROCESS		0x00000080      /* Process policy on mce errors */
1285 #define PF_SUPERPRIV		0x00000100	/* Used super-user privileges */
1286 #define PF_DUMPCORE		0x00000200	/* Dumped core */
1287 #define PF_SIGNALED		0x00000400	/* Killed by a signal */
1288 #define PF_MEMALLOC		0x00000800	/* Allocating memory */
1289 #define PF_NPROC_EXCEEDED	0x00001000	/* set_user() noticed that RLIMIT_NPROC was exceeded */
1290 #define PF_USED_MATH		0x00002000	/* If unset the fpu must be initialized before use */
1291 #define PF_USED_ASYNC		0x00004000	/* Used async_schedule*(), used by module init */
1292 #define PF_NOFREEZE		0x00008000	/* This thread should not be frozen */
1293 #define PF_FROZEN		0x00010000	/* Frozen for system suspend */
1294 #define PF_KSWAPD		0x00020000	/* I am kswapd */
1295 #define PF_MEMALLOC_NOFS	0x00040000	/* All allocation requests will inherit GFP_NOFS */
1296 #define PF_MEMALLOC_NOIO	0x00080000	/* All allocation requests will inherit GFP_NOIO */
1297 #define PF_LESS_THROTTLE	0x00100000	/* Throttle me less: I clean memory */
1298 #define PF_KTHREAD		0x00200000	/* I am a kernel thread */
1299 #define PF_RANDOMIZE		0x00400000	/* Randomize virtual address space */
1300 #define PF_SWAPWRITE		0x00800000	/* Allowed to write to swap */
1301 #define PF_NO_SETAFFINITY	0x04000000	/* Userland is not allowed to meddle with cpus_allowed */
1302 #define PF_MCE_EARLY		0x08000000      /* Early kill for mce process policy */
1303 #define PF_MUTEX_TESTER		0x20000000	/* Thread belongs to the rt mutex tester */
1304 #define PF_FREEZER_SKIP		0x40000000	/* Freezer should not count it as freezable */
1305 #define PF_SUSPEND_TASK		0x80000000      /* This thread called freeze_processes() and should not be frozen */
1306 
1307 /*
1308  * Only the _current_ task can read/write to tsk->flags, but other
1309  * tasks can access tsk->flags in readonly mode for example
1310  * with tsk_used_math (like during threaded core dumping).
1311  * There is however an exception to this rule during ptrace
1312  * or during fork: the ptracer task is allowed to write to the
1313  * child->flags of its traced child (same goes for fork, the parent
1314  * can write to the child->flags), because we're guaranteed the
1315  * child is not running and in turn not changing child->flags
1316  * at the same time the parent does it.
1317  */
1318 #define clear_stopped_child_used_math(child)	do { (child)->flags &= ~PF_USED_MATH; } while (0)
1319 #define set_stopped_child_used_math(child)	do { (child)->flags |= PF_USED_MATH; } while (0)
1320 #define clear_used_math()			clear_stopped_child_used_math(current)
1321 #define set_used_math()				set_stopped_child_used_math(current)
1322 
1323 #define conditional_stopped_child_used_math(condition, child) \
1324 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1325 
1326 #define conditional_used_math(condition)	conditional_stopped_child_used_math(condition, current)
1327 
1328 #define copy_to_stopped_child_used_math(child) \
1329 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1330 
1331 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1332 #define tsk_used_math(p)			((p)->flags & PF_USED_MATH)
1333 #define used_math()				tsk_used_math(current)
1334 
1335 static inline bool is_percpu_thread(void)
1336 {
1337 #ifdef CONFIG_SMP
1338 	return (current->flags & PF_NO_SETAFFINITY) &&
1339 		(current->nr_cpus_allowed  == 1);
1340 #else
1341 	return true;
1342 #endif
1343 }
1344 
1345 /* Per-process atomic flags. */
1346 #define PFA_NO_NEW_PRIVS		0	/* May not gain new privileges. */
1347 #define PFA_SPREAD_PAGE			1	/* Spread page cache over cpuset */
1348 #define PFA_SPREAD_SLAB			2	/* Spread some slab caches over cpuset */
1349 
1350 
1351 #define TASK_PFA_TEST(name, func)					\
1352 	static inline bool task_##func(struct task_struct *p)		\
1353 	{ return test_bit(PFA_##name, &p->atomic_flags); }
1354 
1355 #define TASK_PFA_SET(name, func)					\
1356 	static inline void task_set_##func(struct task_struct *p)	\
1357 	{ set_bit(PFA_##name, &p->atomic_flags); }
1358 
1359 #define TASK_PFA_CLEAR(name, func)					\
1360 	static inline void task_clear_##func(struct task_struct *p)	\
1361 	{ clear_bit(PFA_##name, &p->atomic_flags); }
1362 
1363 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
1364 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
1365 
1366 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
1367 TASK_PFA_SET(SPREAD_PAGE, spread_page)
1368 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
1369 
1370 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
1371 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
1372 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
1373 
1374 static inline void
1375 current_restore_flags(unsigned long orig_flags, unsigned long flags)
1376 {
1377 	current->flags &= ~flags;
1378 	current->flags |= orig_flags & flags;
1379 }
1380 
1381 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial);
1382 extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed);
1383 #ifdef CONFIG_SMP
1384 extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask);
1385 extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask);
1386 #else
1387 static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
1388 {
1389 }
1390 static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
1391 {
1392 	if (!cpumask_test_cpu(0, new_mask))
1393 		return -EINVAL;
1394 	return 0;
1395 }
1396 #endif
1397 
1398 #ifndef cpu_relax_yield
1399 #define cpu_relax_yield() cpu_relax()
1400 #endif
1401 
1402 extern int yield_to(struct task_struct *p, bool preempt);
1403 extern void set_user_nice(struct task_struct *p, long nice);
1404 extern int task_prio(const struct task_struct *p);
1405 
1406 /**
1407  * task_nice - return the nice value of a given task.
1408  * @p: the task in question.
1409  *
1410  * Return: The nice value [ -20 ... 0 ... 19 ].
1411  */
1412 static inline int task_nice(const struct task_struct *p)
1413 {
1414 	return PRIO_TO_NICE((p)->static_prio);
1415 }
1416 
1417 extern int can_nice(const struct task_struct *p, const int nice);
1418 extern int task_curr(const struct task_struct *p);
1419 extern int idle_cpu(int cpu);
1420 extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *);
1421 extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *);
1422 extern int sched_setattr(struct task_struct *, const struct sched_attr *);
1423 extern struct task_struct *idle_task(int cpu);
1424 
1425 /**
1426  * is_idle_task - is the specified task an idle task?
1427  * @p: the task in question.
1428  *
1429  * Return: 1 if @p is an idle task. 0 otherwise.
1430  */
1431 static inline bool is_idle_task(const struct task_struct *p)
1432 {
1433 	return !!(p->flags & PF_IDLE);
1434 }
1435 
1436 extern struct task_struct *curr_task(int cpu);
1437 extern void ia64_set_curr_task(int cpu, struct task_struct *p);
1438 
1439 void yield(void);
1440 
1441 union thread_union {
1442 #ifndef CONFIG_THREAD_INFO_IN_TASK
1443 	struct thread_info thread_info;
1444 #endif
1445 	unsigned long stack[THREAD_SIZE/sizeof(long)];
1446 };
1447 
1448 #ifdef CONFIG_THREAD_INFO_IN_TASK
1449 static inline struct thread_info *task_thread_info(struct task_struct *task)
1450 {
1451 	return &task->thread_info;
1452 }
1453 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1454 # define task_thread_info(task)	((struct thread_info *)(task)->stack)
1455 #endif
1456 
1457 /*
1458  * find a task by one of its numerical ids
1459  *
1460  * find_task_by_pid_ns():
1461  *      finds a task by its pid in the specified namespace
1462  * find_task_by_vpid():
1463  *      finds a task by its virtual pid
1464  *
1465  * see also find_vpid() etc in include/linux/pid.h
1466  */
1467 
1468 extern struct task_struct *find_task_by_vpid(pid_t nr);
1469 extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns);
1470 
1471 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1472 extern int wake_up_process(struct task_struct *tsk);
1473 extern void wake_up_new_task(struct task_struct *tsk);
1474 
1475 #ifdef CONFIG_SMP
1476 extern void kick_process(struct task_struct *tsk);
1477 #else
1478 static inline void kick_process(struct task_struct *tsk) { }
1479 #endif
1480 
1481 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
1482 
1483 static inline void set_task_comm(struct task_struct *tsk, const char *from)
1484 {
1485 	__set_task_comm(tsk, from, false);
1486 }
1487 
1488 extern char *get_task_comm(char *to, struct task_struct *tsk);
1489 
1490 #ifdef CONFIG_SMP
1491 void scheduler_ipi(void);
1492 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1493 #else
1494 static inline void scheduler_ipi(void) { }
1495 static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state)
1496 {
1497 	return 1;
1498 }
1499 #endif
1500 
1501 /*
1502  * Set thread flags in other task's structures.
1503  * See asm/thread_info.h for TIF_xxxx flags available:
1504  */
1505 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1506 {
1507 	set_ti_thread_flag(task_thread_info(tsk), flag);
1508 }
1509 
1510 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1511 {
1512 	clear_ti_thread_flag(task_thread_info(tsk), flag);
1513 }
1514 
1515 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1516 {
1517 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1518 }
1519 
1520 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1521 {
1522 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1523 }
1524 
1525 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1526 {
1527 	return test_ti_thread_flag(task_thread_info(tsk), flag);
1528 }
1529 
1530 static inline void set_tsk_need_resched(struct task_struct *tsk)
1531 {
1532 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1533 }
1534 
1535 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1536 {
1537 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1538 }
1539 
1540 static inline int test_tsk_need_resched(struct task_struct *tsk)
1541 {
1542 	return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
1543 }
1544 
1545 /*
1546  * cond_resched() and cond_resched_lock(): latency reduction via
1547  * explicit rescheduling in places that are safe. The return
1548  * value indicates whether a reschedule was done in fact.
1549  * cond_resched_lock() will drop the spinlock before scheduling,
1550  * cond_resched_softirq() will enable bhs before scheduling.
1551  */
1552 #ifndef CONFIG_PREEMPT
1553 extern int _cond_resched(void);
1554 #else
1555 static inline int _cond_resched(void) { return 0; }
1556 #endif
1557 
1558 #define cond_resched() ({			\
1559 	___might_sleep(__FILE__, __LINE__, 0);	\
1560 	_cond_resched();			\
1561 })
1562 
1563 extern int __cond_resched_lock(spinlock_t *lock);
1564 
1565 #define cond_resched_lock(lock) ({				\
1566 	___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
1567 	__cond_resched_lock(lock);				\
1568 })
1569 
1570 extern int __cond_resched_softirq(void);
1571 
1572 #define cond_resched_softirq() ({					\
1573 	___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET);	\
1574 	__cond_resched_softirq();					\
1575 })
1576 
1577 static inline void cond_resched_rcu(void)
1578 {
1579 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
1580 	rcu_read_unlock();
1581 	cond_resched();
1582 	rcu_read_lock();
1583 #endif
1584 }
1585 
1586 /*
1587  * Does a critical section need to be broken due to another
1588  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
1589  * but a general need for low latency)
1590  */
1591 static inline int spin_needbreak(spinlock_t *lock)
1592 {
1593 #ifdef CONFIG_PREEMPT
1594 	return spin_is_contended(lock);
1595 #else
1596 	return 0;
1597 #endif
1598 }
1599 
1600 static __always_inline bool need_resched(void)
1601 {
1602 	return unlikely(tif_need_resched());
1603 }
1604 
1605 /*
1606  * Wrappers for p->thread_info->cpu access. No-op on UP.
1607  */
1608 #ifdef CONFIG_SMP
1609 
1610 static inline unsigned int task_cpu(const struct task_struct *p)
1611 {
1612 #ifdef CONFIG_THREAD_INFO_IN_TASK
1613 	return p->cpu;
1614 #else
1615 	return task_thread_info(p)->cpu;
1616 #endif
1617 }
1618 
1619 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1620 
1621 #else
1622 
1623 static inline unsigned int task_cpu(const struct task_struct *p)
1624 {
1625 	return 0;
1626 }
1627 
1628 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1629 {
1630 }
1631 
1632 #endif /* CONFIG_SMP */
1633 
1634 /*
1635  * In order to reduce various lock holder preemption latencies provide an
1636  * interface to see if a vCPU is currently running or not.
1637  *
1638  * This allows us to terminate optimistic spin loops and block, analogous to
1639  * the native optimistic spin heuristic of testing if the lock owner task is
1640  * running or not.
1641  */
1642 #ifndef vcpu_is_preempted
1643 # define vcpu_is_preempted(cpu)	false
1644 #endif
1645 
1646 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
1647 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
1648 
1649 #ifndef TASK_SIZE_OF
1650 #define TASK_SIZE_OF(tsk)	TASK_SIZE
1651 #endif
1652 
1653 #endif
1654