xref: /linux/include/linux/sched.h (revision 00a6d7b6762c27d441e9ac8faff36384bc0fc180)
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3 
4 #include <uapi/linux/sched.h>
5 
6 #include <linux/sched/prio.h>
7 
8 
9 struct sched_param {
10 	int sched_priority;
11 };
12 
13 #include <asm/param.h>	/* for HZ */
14 
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt_mask.h>
29 
30 #include <asm/page.h>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
33 
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/signal.h>
37 #include <linux/compiler.h>
38 #include <linux/completion.h>
39 #include <linux/pid.h>
40 #include <linux/percpu.h>
41 #include <linux/topology.h>
42 #include <linux/proportions.h>
43 #include <linux/seccomp.h>
44 #include <linux/rcupdate.h>
45 #include <linux/rculist.h>
46 #include <linux/rtmutex.h>
47 
48 #include <linux/time.h>
49 #include <linux/param.h>
50 #include <linux/resource.h>
51 #include <linux/timer.h>
52 #include <linux/hrtimer.h>
53 #include <linux/task_io_accounting.h>
54 #include <linux/latencytop.h>
55 #include <linux/cred.h>
56 #include <linux/llist.h>
57 #include <linux/uidgid.h>
58 #include <linux/gfp.h>
59 
60 #include <asm/processor.h>
61 
62 #define SCHED_ATTR_SIZE_VER0	48	/* sizeof first published struct */
63 
64 /*
65  * Extended scheduling parameters data structure.
66  *
67  * This is needed because the original struct sched_param can not be
68  * altered without introducing ABI issues with legacy applications
69  * (e.g., in sched_getparam()).
70  *
71  * However, the possibility of specifying more than just a priority for
72  * the tasks may be useful for a wide variety of application fields, e.g.,
73  * multimedia, streaming, automation and control, and many others.
74  *
75  * This variant (sched_attr) is meant at describing a so-called
76  * sporadic time-constrained task. In such model a task is specified by:
77  *  - the activation period or minimum instance inter-arrival time;
78  *  - the maximum (or average, depending on the actual scheduling
79  *    discipline) computation time of all instances, a.k.a. runtime;
80  *  - the deadline (relative to the actual activation time) of each
81  *    instance.
82  * Very briefly, a periodic (sporadic) task asks for the execution of
83  * some specific computation --which is typically called an instance--
84  * (at most) every period. Moreover, each instance typically lasts no more
85  * than the runtime and must be completed by time instant t equal to
86  * the instance activation time + the deadline.
87  *
88  * This is reflected by the actual fields of the sched_attr structure:
89  *
90  *  @size		size of the structure, for fwd/bwd compat.
91  *
92  *  @sched_policy	task's scheduling policy
93  *  @sched_flags	for customizing the scheduler behaviour
94  *  @sched_nice		task's nice value      (SCHED_NORMAL/BATCH)
95  *  @sched_priority	task's static priority (SCHED_FIFO/RR)
96  *  @sched_deadline	representative of the task's deadline
97  *  @sched_runtime	representative of the task's runtime
98  *  @sched_period	representative of the task's period
99  *
100  * Given this task model, there are a multiplicity of scheduling algorithms
101  * and policies, that can be used to ensure all the tasks will make their
102  * timing constraints.
103  *
104  * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
105  * only user of this new interface. More information about the algorithm
106  * available in the scheduling class file or in Documentation/.
107  */
108 struct sched_attr {
109 	u32 size;
110 
111 	u32 sched_policy;
112 	u64 sched_flags;
113 
114 	/* SCHED_NORMAL, SCHED_BATCH */
115 	s32 sched_nice;
116 
117 	/* SCHED_FIFO, SCHED_RR */
118 	u32 sched_priority;
119 
120 	/* SCHED_DEADLINE */
121 	u64 sched_runtime;
122 	u64 sched_deadline;
123 	u64 sched_period;
124 };
125 
126 struct exec_domain;
127 struct futex_pi_state;
128 struct robust_list_head;
129 struct bio_list;
130 struct fs_struct;
131 struct perf_event_context;
132 struct blk_plug;
133 struct filename;
134 
135 #define VMACACHE_BITS 2
136 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
137 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
138 
139 /*
140  * List of flags we want to share for kernel threads,
141  * if only because they are not used by them anyway.
142  */
143 #define CLONE_KERNEL	(CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
144 
145 /*
146  * These are the constant used to fake the fixed-point load-average
147  * counting. Some notes:
148  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
149  *    a load-average precision of 10 bits integer + 11 bits fractional
150  *  - if you want to count load-averages more often, you need more
151  *    precision, or rounding will get you. With 2-second counting freq,
152  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
153  *    11 bit fractions.
154  */
155 extern unsigned long avenrun[];		/* Load averages */
156 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
157 
158 #define FSHIFT		11		/* nr of bits of precision */
159 #define FIXED_1		(1<<FSHIFT)	/* 1.0 as fixed-point */
160 #define LOAD_FREQ	(5*HZ+1)	/* 5 sec intervals */
161 #define EXP_1		1884		/* 1/exp(5sec/1min) as fixed-point */
162 #define EXP_5		2014		/* 1/exp(5sec/5min) */
163 #define EXP_15		2037		/* 1/exp(5sec/15min) */
164 
165 #define CALC_LOAD(load,exp,n) \
166 	load *= exp; \
167 	load += n*(FIXED_1-exp); \
168 	load >>= FSHIFT;
169 
170 extern unsigned long total_forks;
171 extern int nr_threads;
172 DECLARE_PER_CPU(unsigned long, process_counts);
173 extern int nr_processes(void);
174 extern unsigned long nr_running(void);
175 extern unsigned long nr_iowait(void);
176 extern unsigned long nr_iowait_cpu(int cpu);
177 extern unsigned long this_cpu_load(void);
178 
179 
180 extern void calc_global_load(unsigned long ticks);
181 extern void update_cpu_load_nohz(void);
182 
183 extern unsigned long get_parent_ip(unsigned long addr);
184 
185 extern void dump_cpu_task(int cpu);
186 
187 struct seq_file;
188 struct cfs_rq;
189 struct task_group;
190 #ifdef CONFIG_SCHED_DEBUG
191 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
192 extern void proc_sched_set_task(struct task_struct *p);
193 extern void
194 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
195 #endif
196 
197 /*
198  * Task state bitmask. NOTE! These bits are also
199  * encoded in fs/proc/array.c: get_task_state().
200  *
201  * We have two separate sets of flags: task->state
202  * is about runnability, while task->exit_state are
203  * about the task exiting. Confusing, but this way
204  * modifying one set can't modify the other one by
205  * mistake.
206  */
207 #define TASK_RUNNING		0
208 #define TASK_INTERRUPTIBLE	1
209 #define TASK_UNINTERRUPTIBLE	2
210 #define __TASK_STOPPED		4
211 #define __TASK_TRACED		8
212 /* in tsk->exit_state */
213 #define EXIT_DEAD		16
214 #define EXIT_ZOMBIE		32
215 #define EXIT_TRACE		(EXIT_ZOMBIE | EXIT_DEAD)
216 /* in tsk->state again */
217 #define TASK_DEAD		64
218 #define TASK_WAKEKILL		128
219 #define TASK_WAKING		256
220 #define TASK_PARKED		512
221 #define TASK_STATE_MAX		1024
222 
223 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
224 
225 extern char ___assert_task_state[1 - 2*!!(
226 		sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
227 
228 /* Convenience macros for the sake of set_task_state */
229 #define TASK_KILLABLE		(TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
230 #define TASK_STOPPED		(TASK_WAKEKILL | __TASK_STOPPED)
231 #define TASK_TRACED		(TASK_WAKEKILL | __TASK_TRACED)
232 
233 /* Convenience macros for the sake of wake_up */
234 #define TASK_NORMAL		(TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
235 #define TASK_ALL		(TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
236 
237 /* get_task_state() */
238 #define TASK_REPORT		(TASK_RUNNING | TASK_INTERRUPTIBLE | \
239 				 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
240 				 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
241 
242 #define task_is_traced(task)	((task->state & __TASK_TRACED) != 0)
243 #define task_is_stopped(task)	((task->state & __TASK_STOPPED) != 0)
244 #define task_is_stopped_or_traced(task)	\
245 			((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
246 #define task_contributes_to_load(task)	\
247 				((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
248 				 (task->flags & PF_FROZEN) == 0)
249 
250 #define __set_task_state(tsk, state_value)		\
251 	do { (tsk)->state = (state_value); } while (0)
252 #define set_task_state(tsk, state_value)		\
253 	set_mb((tsk)->state, (state_value))
254 
255 /*
256  * set_current_state() includes a barrier so that the write of current->state
257  * is correctly serialised wrt the caller's subsequent test of whether to
258  * actually sleep:
259  *
260  *	set_current_state(TASK_UNINTERRUPTIBLE);
261  *	if (do_i_need_to_sleep())
262  *		schedule();
263  *
264  * If the caller does not need such serialisation then use __set_current_state()
265  */
266 #define __set_current_state(state_value)			\
267 	do { current->state = (state_value); } while (0)
268 #define set_current_state(state_value)		\
269 	set_mb(current->state, (state_value))
270 
271 /* Task command name length */
272 #define TASK_COMM_LEN 16
273 
274 #include <linux/spinlock.h>
275 
276 /*
277  * This serializes "schedule()" and also protects
278  * the run-queue from deletions/modifications (but
279  * _adding_ to the beginning of the run-queue has
280  * a separate lock).
281  */
282 extern rwlock_t tasklist_lock;
283 extern spinlock_t mmlist_lock;
284 
285 struct task_struct;
286 
287 #ifdef CONFIG_PROVE_RCU
288 extern int lockdep_tasklist_lock_is_held(void);
289 #endif /* #ifdef CONFIG_PROVE_RCU */
290 
291 extern void sched_init(void);
292 extern void sched_init_smp(void);
293 extern asmlinkage void schedule_tail(struct task_struct *prev);
294 extern void init_idle(struct task_struct *idle, int cpu);
295 extern void init_idle_bootup_task(struct task_struct *idle);
296 
297 extern int runqueue_is_locked(int cpu);
298 
299 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
300 extern void nohz_balance_enter_idle(int cpu);
301 extern void set_cpu_sd_state_idle(void);
302 extern int get_nohz_timer_target(int pinned);
303 #else
304 static inline void nohz_balance_enter_idle(int cpu) { }
305 static inline void set_cpu_sd_state_idle(void) { }
306 static inline int get_nohz_timer_target(int pinned)
307 {
308 	return smp_processor_id();
309 }
310 #endif
311 
312 /*
313  * Only dump TASK_* tasks. (0 for all tasks)
314  */
315 extern void show_state_filter(unsigned long state_filter);
316 
317 static inline void show_state(void)
318 {
319 	show_state_filter(0);
320 }
321 
322 extern void show_regs(struct pt_regs *);
323 
324 /*
325  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
326  * task), SP is the stack pointer of the first frame that should be shown in the back
327  * trace (or NULL if the entire call-chain of the task should be shown).
328  */
329 extern void show_stack(struct task_struct *task, unsigned long *sp);
330 
331 void io_schedule(void);
332 long io_schedule_timeout(long timeout);
333 
334 extern void cpu_init (void);
335 extern void trap_init(void);
336 extern void update_process_times(int user);
337 extern void scheduler_tick(void);
338 
339 extern void sched_show_task(struct task_struct *p);
340 
341 #ifdef CONFIG_LOCKUP_DETECTOR
342 extern void touch_softlockup_watchdog(void);
343 extern void touch_softlockup_watchdog_sync(void);
344 extern void touch_all_softlockup_watchdogs(void);
345 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
346 				  void __user *buffer,
347 				  size_t *lenp, loff_t *ppos);
348 extern unsigned int  softlockup_panic;
349 void lockup_detector_init(void);
350 #else
351 static inline void touch_softlockup_watchdog(void)
352 {
353 }
354 static inline void touch_softlockup_watchdog_sync(void)
355 {
356 }
357 static inline void touch_all_softlockup_watchdogs(void)
358 {
359 }
360 static inline void lockup_detector_init(void)
361 {
362 }
363 #endif
364 
365 #ifdef CONFIG_DETECT_HUNG_TASK
366 void reset_hung_task_detector(void);
367 #else
368 static inline void reset_hung_task_detector(void)
369 {
370 }
371 #endif
372 
373 /* Attach to any functions which should be ignored in wchan output. */
374 #define __sched		__attribute__((__section__(".sched.text")))
375 
376 /* Linker adds these: start and end of __sched functions */
377 extern char __sched_text_start[], __sched_text_end[];
378 
379 /* Is this address in the __sched functions? */
380 extern int in_sched_functions(unsigned long addr);
381 
382 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
383 extern signed long schedule_timeout(signed long timeout);
384 extern signed long schedule_timeout_interruptible(signed long timeout);
385 extern signed long schedule_timeout_killable(signed long timeout);
386 extern signed long schedule_timeout_uninterruptible(signed long timeout);
387 asmlinkage void schedule(void);
388 extern void schedule_preempt_disabled(void);
389 
390 struct nsproxy;
391 struct user_namespace;
392 
393 #ifdef CONFIG_MMU
394 extern void arch_pick_mmap_layout(struct mm_struct *mm);
395 extern unsigned long
396 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
397 		       unsigned long, unsigned long);
398 extern unsigned long
399 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
400 			  unsigned long len, unsigned long pgoff,
401 			  unsigned long flags);
402 #else
403 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
404 #endif
405 
406 #define SUID_DUMP_DISABLE	0	/* No setuid dumping */
407 #define SUID_DUMP_USER		1	/* Dump as user of process */
408 #define SUID_DUMP_ROOT		2	/* Dump as root */
409 
410 /* mm flags */
411 
412 /* for SUID_DUMP_* above */
413 #define MMF_DUMPABLE_BITS 2
414 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
415 
416 extern void set_dumpable(struct mm_struct *mm, int value);
417 /*
418  * This returns the actual value of the suid_dumpable flag. For things
419  * that are using this for checking for privilege transitions, it must
420  * test against SUID_DUMP_USER rather than treating it as a boolean
421  * value.
422  */
423 static inline int __get_dumpable(unsigned long mm_flags)
424 {
425 	return mm_flags & MMF_DUMPABLE_MASK;
426 }
427 
428 static inline int get_dumpable(struct mm_struct *mm)
429 {
430 	return __get_dumpable(mm->flags);
431 }
432 
433 /* coredump filter bits */
434 #define MMF_DUMP_ANON_PRIVATE	2
435 #define MMF_DUMP_ANON_SHARED	3
436 #define MMF_DUMP_MAPPED_PRIVATE	4
437 #define MMF_DUMP_MAPPED_SHARED	5
438 #define MMF_DUMP_ELF_HEADERS	6
439 #define MMF_DUMP_HUGETLB_PRIVATE 7
440 #define MMF_DUMP_HUGETLB_SHARED  8
441 
442 #define MMF_DUMP_FILTER_SHIFT	MMF_DUMPABLE_BITS
443 #define MMF_DUMP_FILTER_BITS	7
444 #define MMF_DUMP_FILTER_MASK \
445 	(((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
446 #define MMF_DUMP_FILTER_DEFAULT \
447 	((1 << MMF_DUMP_ANON_PRIVATE) |	(1 << MMF_DUMP_ANON_SHARED) |\
448 	 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
449 
450 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
451 # define MMF_DUMP_MASK_DEFAULT_ELF	(1 << MMF_DUMP_ELF_HEADERS)
452 #else
453 # define MMF_DUMP_MASK_DEFAULT_ELF	0
454 #endif
455 					/* leave room for more dump flags */
456 #define MMF_VM_MERGEABLE	16	/* KSM may merge identical pages */
457 #define MMF_VM_HUGEPAGE		17	/* set when VM_HUGEPAGE is set on vma */
458 #define MMF_EXE_FILE_CHANGED	18	/* see prctl_set_mm_exe_file() */
459 
460 #define MMF_HAS_UPROBES		19	/* has uprobes */
461 #define MMF_RECALC_UPROBES	20	/* MMF_HAS_UPROBES can be wrong */
462 
463 #define MMF_INIT_MASK		(MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
464 
465 struct sighand_struct {
466 	atomic_t		count;
467 	struct k_sigaction	action[_NSIG];
468 	spinlock_t		siglock;
469 	wait_queue_head_t	signalfd_wqh;
470 };
471 
472 struct pacct_struct {
473 	int			ac_flag;
474 	long			ac_exitcode;
475 	unsigned long		ac_mem;
476 	cputime_t		ac_utime, ac_stime;
477 	unsigned long		ac_minflt, ac_majflt;
478 };
479 
480 struct cpu_itimer {
481 	cputime_t expires;
482 	cputime_t incr;
483 	u32 error;
484 	u32 incr_error;
485 };
486 
487 /**
488  * struct cputime - snaphsot of system and user cputime
489  * @utime: time spent in user mode
490  * @stime: time spent in system mode
491  *
492  * Gathers a generic snapshot of user and system time.
493  */
494 struct cputime {
495 	cputime_t utime;
496 	cputime_t stime;
497 };
498 
499 /**
500  * struct task_cputime - collected CPU time counts
501  * @utime:		time spent in user mode, in &cputime_t units
502  * @stime:		time spent in kernel mode, in &cputime_t units
503  * @sum_exec_runtime:	total time spent on the CPU, in nanoseconds
504  *
505  * This is an extension of struct cputime that includes the total runtime
506  * spent by the task from the scheduler point of view.
507  *
508  * As a result, this structure groups together three kinds of CPU time
509  * that are tracked for threads and thread groups.  Most things considering
510  * CPU time want to group these counts together and treat all three
511  * of them in parallel.
512  */
513 struct task_cputime {
514 	cputime_t utime;
515 	cputime_t stime;
516 	unsigned long long sum_exec_runtime;
517 };
518 /* Alternate field names when used to cache expirations. */
519 #define prof_exp	stime
520 #define virt_exp	utime
521 #define sched_exp	sum_exec_runtime
522 
523 #define INIT_CPUTIME	\
524 	(struct task_cputime) {					\
525 		.utime = 0,					\
526 		.stime = 0,					\
527 		.sum_exec_runtime = 0,				\
528 	}
529 
530 #ifdef CONFIG_PREEMPT_COUNT
531 #define PREEMPT_DISABLED	(1 + PREEMPT_ENABLED)
532 #else
533 #define PREEMPT_DISABLED	PREEMPT_ENABLED
534 #endif
535 
536 /*
537  * Disable preemption until the scheduler is running.
538  * Reset by start_kernel()->sched_init()->init_idle().
539  *
540  * We include PREEMPT_ACTIVE to avoid cond_resched() from working
541  * before the scheduler is active -- see should_resched().
542  */
543 #define INIT_PREEMPT_COUNT	(PREEMPT_DISABLED + PREEMPT_ACTIVE)
544 
545 /**
546  * struct thread_group_cputimer - thread group interval timer counts
547  * @cputime:		thread group interval timers.
548  * @running:		non-zero when there are timers running and
549  * 			@cputime receives updates.
550  * @lock:		lock for fields in this struct.
551  *
552  * This structure contains the version of task_cputime, above, that is
553  * used for thread group CPU timer calculations.
554  */
555 struct thread_group_cputimer {
556 	struct task_cputime cputime;
557 	int running;
558 	raw_spinlock_t lock;
559 };
560 
561 #include <linux/rwsem.h>
562 struct autogroup;
563 
564 /*
565  * NOTE! "signal_struct" does not have its own
566  * locking, because a shared signal_struct always
567  * implies a shared sighand_struct, so locking
568  * sighand_struct is always a proper superset of
569  * the locking of signal_struct.
570  */
571 struct signal_struct {
572 	atomic_t		sigcnt;
573 	atomic_t		live;
574 	int			nr_threads;
575 	struct list_head	thread_head;
576 
577 	wait_queue_head_t	wait_chldexit;	/* for wait4() */
578 
579 	/* current thread group signal load-balancing target: */
580 	struct task_struct	*curr_target;
581 
582 	/* shared signal handling: */
583 	struct sigpending	shared_pending;
584 
585 	/* thread group exit support */
586 	int			group_exit_code;
587 	/* overloaded:
588 	 * - notify group_exit_task when ->count is equal to notify_count
589 	 * - everyone except group_exit_task is stopped during signal delivery
590 	 *   of fatal signals, group_exit_task processes the signal.
591 	 */
592 	int			notify_count;
593 	struct task_struct	*group_exit_task;
594 
595 	/* thread group stop support, overloads group_exit_code too */
596 	int			group_stop_count;
597 	unsigned int		flags; /* see SIGNAL_* flags below */
598 
599 	/*
600 	 * PR_SET_CHILD_SUBREAPER marks a process, like a service
601 	 * manager, to re-parent orphan (double-forking) child processes
602 	 * to this process instead of 'init'. The service manager is
603 	 * able to receive SIGCHLD signals and is able to investigate
604 	 * the process until it calls wait(). All children of this
605 	 * process will inherit a flag if they should look for a
606 	 * child_subreaper process at exit.
607 	 */
608 	unsigned int		is_child_subreaper:1;
609 	unsigned int		has_child_subreaper:1;
610 
611 	/* POSIX.1b Interval Timers */
612 	int			posix_timer_id;
613 	struct list_head	posix_timers;
614 
615 	/* ITIMER_REAL timer for the process */
616 	struct hrtimer real_timer;
617 	struct pid *leader_pid;
618 	ktime_t it_real_incr;
619 
620 	/*
621 	 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
622 	 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
623 	 * values are defined to 0 and 1 respectively
624 	 */
625 	struct cpu_itimer it[2];
626 
627 	/*
628 	 * Thread group totals for process CPU timers.
629 	 * See thread_group_cputimer(), et al, for details.
630 	 */
631 	struct thread_group_cputimer cputimer;
632 
633 	/* Earliest-expiration cache. */
634 	struct task_cputime cputime_expires;
635 
636 	struct list_head cpu_timers[3];
637 
638 	struct pid *tty_old_pgrp;
639 
640 	/* boolean value for session group leader */
641 	int leader;
642 
643 	struct tty_struct *tty; /* NULL if no tty */
644 
645 #ifdef CONFIG_SCHED_AUTOGROUP
646 	struct autogroup *autogroup;
647 #endif
648 	/*
649 	 * Cumulative resource counters for dead threads in the group,
650 	 * and for reaped dead child processes forked by this group.
651 	 * Live threads maintain their own counters and add to these
652 	 * in __exit_signal, except for the group leader.
653 	 */
654 	cputime_t utime, stime, cutime, cstime;
655 	cputime_t gtime;
656 	cputime_t cgtime;
657 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
658 	struct cputime prev_cputime;
659 #endif
660 	unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
661 	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
662 	unsigned long inblock, oublock, cinblock, coublock;
663 	unsigned long maxrss, cmaxrss;
664 	struct task_io_accounting ioac;
665 
666 	/*
667 	 * Cumulative ns of schedule CPU time fo dead threads in the
668 	 * group, not including a zombie group leader, (This only differs
669 	 * from jiffies_to_ns(utime + stime) if sched_clock uses something
670 	 * other than jiffies.)
671 	 */
672 	unsigned long long sum_sched_runtime;
673 
674 	/*
675 	 * We don't bother to synchronize most readers of this at all,
676 	 * because there is no reader checking a limit that actually needs
677 	 * to get both rlim_cur and rlim_max atomically, and either one
678 	 * alone is a single word that can safely be read normally.
679 	 * getrlimit/setrlimit use task_lock(current->group_leader) to
680 	 * protect this instead of the siglock, because they really
681 	 * have no need to disable irqs.
682 	 */
683 	struct rlimit rlim[RLIM_NLIMITS];
684 
685 #ifdef CONFIG_BSD_PROCESS_ACCT
686 	struct pacct_struct pacct;	/* per-process accounting information */
687 #endif
688 #ifdef CONFIG_TASKSTATS
689 	struct taskstats *stats;
690 #endif
691 #ifdef CONFIG_AUDIT
692 	unsigned audit_tty;
693 	unsigned audit_tty_log_passwd;
694 	struct tty_audit_buf *tty_audit_buf;
695 #endif
696 #ifdef CONFIG_CGROUPS
697 	/*
698 	 * group_rwsem prevents new tasks from entering the threadgroup and
699 	 * member tasks from exiting,a more specifically, setting of
700 	 * PF_EXITING.  fork and exit paths are protected with this rwsem
701 	 * using threadgroup_change_begin/end().  Users which require
702 	 * threadgroup to remain stable should use threadgroup_[un]lock()
703 	 * which also takes care of exec path.  Currently, cgroup is the
704 	 * only user.
705 	 */
706 	struct rw_semaphore group_rwsem;
707 #endif
708 
709 	oom_flags_t oom_flags;
710 	short oom_score_adj;		/* OOM kill score adjustment */
711 	short oom_score_adj_min;	/* OOM kill score adjustment min value.
712 					 * Only settable by CAP_SYS_RESOURCE. */
713 
714 	struct mutex cred_guard_mutex;	/* guard against foreign influences on
715 					 * credential calculations
716 					 * (notably. ptrace) */
717 };
718 
719 /*
720  * Bits in flags field of signal_struct.
721  */
722 #define SIGNAL_STOP_STOPPED	0x00000001 /* job control stop in effect */
723 #define SIGNAL_STOP_CONTINUED	0x00000002 /* SIGCONT since WCONTINUED reap */
724 #define SIGNAL_GROUP_EXIT	0x00000004 /* group exit in progress */
725 #define SIGNAL_GROUP_COREDUMP	0x00000008 /* coredump in progress */
726 /*
727  * Pending notifications to parent.
728  */
729 #define SIGNAL_CLD_STOPPED	0x00000010
730 #define SIGNAL_CLD_CONTINUED	0x00000020
731 #define SIGNAL_CLD_MASK		(SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
732 
733 #define SIGNAL_UNKILLABLE	0x00000040 /* for init: ignore fatal signals */
734 
735 /* If true, all threads except ->group_exit_task have pending SIGKILL */
736 static inline int signal_group_exit(const struct signal_struct *sig)
737 {
738 	return	(sig->flags & SIGNAL_GROUP_EXIT) ||
739 		(sig->group_exit_task != NULL);
740 }
741 
742 /*
743  * Some day this will be a full-fledged user tracking system..
744  */
745 struct user_struct {
746 	atomic_t __count;	/* reference count */
747 	atomic_t processes;	/* How many processes does this user have? */
748 	atomic_t files;		/* How many open files does this user have? */
749 	atomic_t sigpending;	/* How many pending signals does this user have? */
750 #ifdef CONFIG_INOTIFY_USER
751 	atomic_t inotify_watches; /* How many inotify watches does this user have? */
752 	atomic_t inotify_devs;	/* How many inotify devs does this user have opened? */
753 #endif
754 #ifdef CONFIG_FANOTIFY
755 	atomic_t fanotify_listeners;
756 #endif
757 #ifdef CONFIG_EPOLL
758 	atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
759 #endif
760 #ifdef CONFIG_POSIX_MQUEUE
761 	/* protected by mq_lock	*/
762 	unsigned long mq_bytes;	/* How many bytes can be allocated to mqueue? */
763 #endif
764 	unsigned long locked_shm; /* How many pages of mlocked shm ? */
765 
766 #ifdef CONFIG_KEYS
767 	struct key *uid_keyring;	/* UID specific keyring */
768 	struct key *session_keyring;	/* UID's default session keyring */
769 #endif
770 
771 	/* Hash table maintenance information */
772 	struct hlist_node uidhash_node;
773 	kuid_t uid;
774 
775 #ifdef CONFIG_PERF_EVENTS
776 	atomic_long_t locked_vm;
777 #endif
778 };
779 
780 extern int uids_sysfs_init(void);
781 
782 extern struct user_struct *find_user(kuid_t);
783 
784 extern struct user_struct root_user;
785 #define INIT_USER (&root_user)
786 
787 
788 struct backing_dev_info;
789 struct reclaim_state;
790 
791 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
792 struct sched_info {
793 	/* cumulative counters */
794 	unsigned long pcount;	      /* # of times run on this cpu */
795 	unsigned long long run_delay; /* time spent waiting on a runqueue */
796 
797 	/* timestamps */
798 	unsigned long long last_arrival,/* when we last ran on a cpu */
799 			   last_queued;	/* when we were last queued to run */
800 };
801 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
802 
803 #ifdef CONFIG_TASK_DELAY_ACCT
804 struct task_delay_info {
805 	spinlock_t	lock;
806 	unsigned int	flags;	/* Private per-task flags */
807 
808 	/* For each stat XXX, add following, aligned appropriately
809 	 *
810 	 * struct timespec XXX_start, XXX_end;
811 	 * u64 XXX_delay;
812 	 * u32 XXX_count;
813 	 *
814 	 * Atomicity of updates to XXX_delay, XXX_count protected by
815 	 * single lock above (split into XXX_lock if contention is an issue).
816 	 */
817 
818 	/*
819 	 * XXX_count is incremented on every XXX operation, the delay
820 	 * associated with the operation is added to XXX_delay.
821 	 * XXX_delay contains the accumulated delay time in nanoseconds.
822 	 */
823 	struct timespec blkio_start, blkio_end;	/* Shared by blkio, swapin */
824 	u64 blkio_delay;	/* wait for sync block io completion */
825 	u64 swapin_delay;	/* wait for swapin block io completion */
826 	u32 blkio_count;	/* total count of the number of sync block */
827 				/* io operations performed */
828 	u32 swapin_count;	/* total count of the number of swapin block */
829 				/* io operations performed */
830 
831 	struct timespec freepages_start, freepages_end;
832 	u64 freepages_delay;	/* wait for memory reclaim */
833 	u32 freepages_count;	/* total count of memory reclaim */
834 };
835 #endif	/* CONFIG_TASK_DELAY_ACCT */
836 
837 static inline int sched_info_on(void)
838 {
839 #ifdef CONFIG_SCHEDSTATS
840 	return 1;
841 #elif defined(CONFIG_TASK_DELAY_ACCT)
842 	extern int delayacct_on;
843 	return delayacct_on;
844 #else
845 	return 0;
846 #endif
847 }
848 
849 enum cpu_idle_type {
850 	CPU_IDLE,
851 	CPU_NOT_IDLE,
852 	CPU_NEWLY_IDLE,
853 	CPU_MAX_IDLE_TYPES
854 };
855 
856 /*
857  * Increase resolution of cpu_power calculations
858  */
859 #define SCHED_POWER_SHIFT	10
860 #define SCHED_POWER_SCALE	(1L << SCHED_POWER_SHIFT)
861 
862 /*
863  * sched-domains (multiprocessor balancing) declarations:
864  */
865 #ifdef CONFIG_SMP
866 #define SD_LOAD_BALANCE		0x0001	/* Do load balancing on this domain. */
867 #define SD_BALANCE_NEWIDLE	0x0002	/* Balance when about to become idle */
868 #define SD_BALANCE_EXEC		0x0004	/* Balance on exec */
869 #define SD_BALANCE_FORK		0x0008	/* Balance on fork, clone */
870 #define SD_BALANCE_WAKE		0x0010  /* Balance on wakeup */
871 #define SD_WAKE_AFFINE		0x0020	/* Wake task to waking CPU */
872 #define SD_SHARE_CPUPOWER	0x0080	/* Domain members share cpu power */
873 #define SD_SHARE_PKG_RESOURCES	0x0200	/* Domain members share cpu pkg resources */
874 #define SD_SERIALIZE		0x0400	/* Only a single load balancing instance */
875 #define SD_ASYM_PACKING		0x0800  /* Place busy groups earlier in the domain */
876 #define SD_PREFER_SIBLING	0x1000	/* Prefer to place tasks in a sibling domain */
877 #define SD_OVERLAP		0x2000	/* sched_domains of this level overlap */
878 #define SD_NUMA			0x4000	/* cross-node balancing */
879 
880 extern int __weak arch_sd_sibiling_asym_packing(void);
881 
882 struct sched_domain_attr {
883 	int relax_domain_level;
884 };
885 
886 #define SD_ATTR_INIT	(struct sched_domain_attr) {	\
887 	.relax_domain_level = -1,			\
888 }
889 
890 extern int sched_domain_level_max;
891 
892 struct sched_group;
893 
894 struct sched_domain {
895 	/* These fields must be setup */
896 	struct sched_domain *parent;	/* top domain must be null terminated */
897 	struct sched_domain *child;	/* bottom domain must be null terminated */
898 	struct sched_group *groups;	/* the balancing groups of the domain */
899 	unsigned long min_interval;	/* Minimum balance interval ms */
900 	unsigned long max_interval;	/* Maximum balance interval ms */
901 	unsigned int busy_factor;	/* less balancing by factor if busy */
902 	unsigned int imbalance_pct;	/* No balance until over watermark */
903 	unsigned int cache_nice_tries;	/* Leave cache hot tasks for # tries */
904 	unsigned int busy_idx;
905 	unsigned int idle_idx;
906 	unsigned int newidle_idx;
907 	unsigned int wake_idx;
908 	unsigned int forkexec_idx;
909 	unsigned int smt_gain;
910 
911 	int nohz_idle;			/* NOHZ IDLE status */
912 	int flags;			/* See SD_* */
913 	int level;
914 
915 	/* Runtime fields. */
916 	unsigned long last_balance;	/* init to jiffies. units in jiffies */
917 	unsigned int balance_interval;	/* initialise to 1. units in ms. */
918 	unsigned int nr_balance_failed; /* initialise to 0 */
919 
920 	/* idle_balance() stats */
921 	u64 max_newidle_lb_cost;
922 	unsigned long next_decay_max_lb_cost;
923 
924 #ifdef CONFIG_SCHEDSTATS
925 	/* load_balance() stats */
926 	unsigned int lb_count[CPU_MAX_IDLE_TYPES];
927 	unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
928 	unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
929 	unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
930 	unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
931 	unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
932 	unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
933 	unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
934 
935 	/* Active load balancing */
936 	unsigned int alb_count;
937 	unsigned int alb_failed;
938 	unsigned int alb_pushed;
939 
940 	/* SD_BALANCE_EXEC stats */
941 	unsigned int sbe_count;
942 	unsigned int sbe_balanced;
943 	unsigned int sbe_pushed;
944 
945 	/* SD_BALANCE_FORK stats */
946 	unsigned int sbf_count;
947 	unsigned int sbf_balanced;
948 	unsigned int sbf_pushed;
949 
950 	/* try_to_wake_up() stats */
951 	unsigned int ttwu_wake_remote;
952 	unsigned int ttwu_move_affine;
953 	unsigned int ttwu_move_balance;
954 #endif
955 #ifdef CONFIG_SCHED_DEBUG
956 	char *name;
957 #endif
958 	union {
959 		void *private;		/* used during construction */
960 		struct rcu_head rcu;	/* used during destruction */
961 	};
962 
963 	unsigned int span_weight;
964 	/*
965 	 * Span of all CPUs in this domain.
966 	 *
967 	 * NOTE: this field is variable length. (Allocated dynamically
968 	 * by attaching extra space to the end of the structure,
969 	 * depending on how many CPUs the kernel has booted up with)
970 	 */
971 	unsigned long span[0];
972 };
973 
974 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
975 {
976 	return to_cpumask(sd->span);
977 }
978 
979 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
980 				    struct sched_domain_attr *dattr_new);
981 
982 /* Allocate an array of sched domains, for partition_sched_domains(). */
983 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
984 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
985 
986 bool cpus_share_cache(int this_cpu, int that_cpu);
987 
988 #else /* CONFIG_SMP */
989 
990 struct sched_domain_attr;
991 
992 static inline void
993 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
994 			struct sched_domain_attr *dattr_new)
995 {
996 }
997 
998 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
999 {
1000 	return true;
1001 }
1002 
1003 #endif	/* !CONFIG_SMP */
1004 
1005 
1006 struct io_context;			/* See blkdev.h */
1007 
1008 
1009 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1010 extern void prefetch_stack(struct task_struct *t);
1011 #else
1012 static inline void prefetch_stack(struct task_struct *t) { }
1013 #endif
1014 
1015 struct audit_context;		/* See audit.c */
1016 struct mempolicy;
1017 struct pipe_inode_info;
1018 struct uts_namespace;
1019 
1020 struct load_weight {
1021 	unsigned long weight;
1022 	u32 inv_weight;
1023 };
1024 
1025 struct sched_avg {
1026 	/*
1027 	 * These sums represent an infinite geometric series and so are bound
1028 	 * above by 1024/(1-y).  Thus we only need a u32 to store them for all
1029 	 * choices of y < 1-2^(-32)*1024.
1030 	 */
1031 	u32 runnable_avg_sum, runnable_avg_period;
1032 	u64 last_runnable_update;
1033 	s64 decay_count;
1034 	unsigned long load_avg_contrib;
1035 };
1036 
1037 #ifdef CONFIG_SCHEDSTATS
1038 struct sched_statistics {
1039 	u64			wait_start;
1040 	u64			wait_max;
1041 	u64			wait_count;
1042 	u64			wait_sum;
1043 	u64			iowait_count;
1044 	u64			iowait_sum;
1045 
1046 	u64			sleep_start;
1047 	u64			sleep_max;
1048 	s64			sum_sleep_runtime;
1049 
1050 	u64			block_start;
1051 	u64			block_max;
1052 	u64			exec_max;
1053 	u64			slice_max;
1054 
1055 	u64			nr_migrations_cold;
1056 	u64			nr_failed_migrations_affine;
1057 	u64			nr_failed_migrations_running;
1058 	u64			nr_failed_migrations_hot;
1059 	u64			nr_forced_migrations;
1060 
1061 	u64			nr_wakeups;
1062 	u64			nr_wakeups_sync;
1063 	u64			nr_wakeups_migrate;
1064 	u64			nr_wakeups_local;
1065 	u64			nr_wakeups_remote;
1066 	u64			nr_wakeups_affine;
1067 	u64			nr_wakeups_affine_attempts;
1068 	u64			nr_wakeups_passive;
1069 	u64			nr_wakeups_idle;
1070 };
1071 #endif
1072 
1073 struct sched_entity {
1074 	struct load_weight	load;		/* for load-balancing */
1075 	struct rb_node		run_node;
1076 	struct list_head	group_node;
1077 	unsigned int		on_rq;
1078 
1079 	u64			exec_start;
1080 	u64			sum_exec_runtime;
1081 	u64			vruntime;
1082 	u64			prev_sum_exec_runtime;
1083 
1084 	u64			nr_migrations;
1085 
1086 #ifdef CONFIG_SCHEDSTATS
1087 	struct sched_statistics statistics;
1088 #endif
1089 
1090 #ifdef CONFIG_FAIR_GROUP_SCHED
1091 	int			depth;
1092 	struct sched_entity	*parent;
1093 	/* rq on which this entity is (to be) queued: */
1094 	struct cfs_rq		*cfs_rq;
1095 	/* rq "owned" by this entity/group: */
1096 	struct cfs_rq		*my_q;
1097 #endif
1098 
1099 #ifdef CONFIG_SMP
1100 	/* Per-entity load-tracking */
1101 	struct sched_avg	avg;
1102 #endif
1103 };
1104 
1105 struct sched_rt_entity {
1106 	struct list_head run_list;
1107 	unsigned long timeout;
1108 	unsigned long watchdog_stamp;
1109 	unsigned int time_slice;
1110 
1111 	struct sched_rt_entity *back;
1112 #ifdef CONFIG_RT_GROUP_SCHED
1113 	struct sched_rt_entity	*parent;
1114 	/* rq on which this entity is (to be) queued: */
1115 	struct rt_rq		*rt_rq;
1116 	/* rq "owned" by this entity/group: */
1117 	struct rt_rq		*my_q;
1118 #endif
1119 };
1120 
1121 struct sched_dl_entity {
1122 	struct rb_node	rb_node;
1123 
1124 	/*
1125 	 * Original scheduling parameters. Copied here from sched_attr
1126 	 * during sched_setscheduler2(), they will remain the same until
1127 	 * the next sched_setscheduler2().
1128 	 */
1129 	u64 dl_runtime;		/* maximum runtime for each instance	*/
1130 	u64 dl_deadline;	/* relative deadline of each instance	*/
1131 	u64 dl_period;		/* separation of two instances (period) */
1132 	u64 dl_bw;		/* dl_runtime / dl_deadline		*/
1133 
1134 	/*
1135 	 * Actual scheduling parameters. Initialized with the values above,
1136 	 * they are continously updated during task execution. Note that
1137 	 * the remaining runtime could be < 0 in case we are in overrun.
1138 	 */
1139 	s64 runtime;		/* remaining runtime for this instance	*/
1140 	u64 deadline;		/* absolute deadline for this instance	*/
1141 	unsigned int flags;	/* specifying the scheduler behaviour	*/
1142 
1143 	/*
1144 	 * Some bool flags:
1145 	 *
1146 	 * @dl_throttled tells if we exhausted the runtime. If so, the
1147 	 * task has to wait for a replenishment to be performed at the
1148 	 * next firing of dl_timer.
1149 	 *
1150 	 * @dl_new tells if a new instance arrived. If so we must
1151 	 * start executing it with full runtime and reset its absolute
1152 	 * deadline;
1153 	 *
1154 	 * @dl_boosted tells if we are boosted due to DI. If so we are
1155 	 * outside bandwidth enforcement mechanism (but only until we
1156 	 * exit the critical section).
1157 	 */
1158 	int dl_throttled, dl_new, dl_boosted;
1159 
1160 	/*
1161 	 * Bandwidth enforcement timer. Each -deadline task has its
1162 	 * own bandwidth to be enforced, thus we need one timer per task.
1163 	 */
1164 	struct hrtimer dl_timer;
1165 };
1166 
1167 struct rcu_node;
1168 
1169 enum perf_event_task_context {
1170 	perf_invalid_context = -1,
1171 	perf_hw_context = 0,
1172 	perf_sw_context,
1173 	perf_nr_task_contexts,
1174 };
1175 
1176 struct task_struct {
1177 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
1178 	void *stack;
1179 	atomic_t usage;
1180 	unsigned int flags;	/* per process flags, defined below */
1181 	unsigned int ptrace;
1182 
1183 #ifdef CONFIG_SMP
1184 	struct llist_node wake_entry;
1185 	int on_cpu;
1186 	struct task_struct *last_wakee;
1187 	unsigned long wakee_flips;
1188 	unsigned long wakee_flip_decay_ts;
1189 
1190 	int wake_cpu;
1191 #endif
1192 	int on_rq;
1193 
1194 	int prio, static_prio, normal_prio;
1195 	unsigned int rt_priority;
1196 	const struct sched_class *sched_class;
1197 	struct sched_entity se;
1198 	struct sched_rt_entity rt;
1199 #ifdef CONFIG_CGROUP_SCHED
1200 	struct task_group *sched_task_group;
1201 #endif
1202 	struct sched_dl_entity dl;
1203 
1204 #ifdef CONFIG_PREEMPT_NOTIFIERS
1205 	/* list of struct preempt_notifier: */
1206 	struct hlist_head preempt_notifiers;
1207 #endif
1208 
1209 #ifdef CONFIG_BLK_DEV_IO_TRACE
1210 	unsigned int btrace_seq;
1211 #endif
1212 
1213 	unsigned int policy;
1214 	int nr_cpus_allowed;
1215 	cpumask_t cpus_allowed;
1216 
1217 #ifdef CONFIG_PREEMPT_RCU
1218 	int rcu_read_lock_nesting;
1219 	char rcu_read_unlock_special;
1220 	struct list_head rcu_node_entry;
1221 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1222 #ifdef CONFIG_TREE_PREEMPT_RCU
1223 	struct rcu_node *rcu_blocked_node;
1224 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1225 #ifdef CONFIG_RCU_BOOST
1226 	struct rt_mutex *rcu_boost_mutex;
1227 #endif /* #ifdef CONFIG_RCU_BOOST */
1228 
1229 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1230 	struct sched_info sched_info;
1231 #endif
1232 
1233 	struct list_head tasks;
1234 #ifdef CONFIG_SMP
1235 	struct plist_node pushable_tasks;
1236 	struct rb_node pushable_dl_tasks;
1237 #endif
1238 
1239 	struct mm_struct *mm, *active_mm;
1240 #ifdef CONFIG_COMPAT_BRK
1241 	unsigned brk_randomized:1;
1242 #endif
1243 	/* per-thread vma caching */
1244 	u32 vmacache_seqnum;
1245 	struct vm_area_struct *vmacache[VMACACHE_SIZE];
1246 #if defined(SPLIT_RSS_COUNTING)
1247 	struct task_rss_stat	rss_stat;
1248 #endif
1249 /* task state */
1250 	int exit_state;
1251 	int exit_code, exit_signal;
1252 	int pdeath_signal;  /*  The signal sent when the parent dies  */
1253 	unsigned int jobctl;	/* JOBCTL_*, siglock protected */
1254 
1255 	/* Used for emulating ABI behavior of previous Linux versions */
1256 	unsigned int personality;
1257 
1258 	unsigned in_execve:1;	/* Tell the LSMs that the process is doing an
1259 				 * execve */
1260 	unsigned in_iowait:1;
1261 
1262 	/* task may not gain privileges */
1263 	unsigned no_new_privs:1;
1264 
1265 	/* Revert to default priority/policy when forking */
1266 	unsigned sched_reset_on_fork:1;
1267 	unsigned sched_contributes_to_load:1;
1268 
1269 	pid_t pid;
1270 	pid_t tgid;
1271 
1272 #ifdef CONFIG_CC_STACKPROTECTOR
1273 	/* Canary value for the -fstack-protector gcc feature */
1274 	unsigned long stack_canary;
1275 #endif
1276 	/*
1277 	 * pointers to (original) parent process, youngest child, younger sibling,
1278 	 * older sibling, respectively.  (p->father can be replaced with
1279 	 * p->real_parent->pid)
1280 	 */
1281 	struct task_struct __rcu *real_parent; /* real parent process */
1282 	struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1283 	/*
1284 	 * children/sibling forms the list of my natural children
1285 	 */
1286 	struct list_head children;	/* list of my children */
1287 	struct list_head sibling;	/* linkage in my parent's children list */
1288 	struct task_struct *group_leader;	/* threadgroup leader */
1289 
1290 	/*
1291 	 * ptraced is the list of tasks this task is using ptrace on.
1292 	 * This includes both natural children and PTRACE_ATTACH targets.
1293 	 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1294 	 */
1295 	struct list_head ptraced;
1296 	struct list_head ptrace_entry;
1297 
1298 	/* PID/PID hash table linkage. */
1299 	struct pid_link pids[PIDTYPE_MAX];
1300 	struct list_head thread_group;
1301 	struct list_head thread_node;
1302 
1303 	struct completion *vfork_done;		/* for vfork() */
1304 	int __user *set_child_tid;		/* CLONE_CHILD_SETTID */
1305 	int __user *clear_child_tid;		/* CLONE_CHILD_CLEARTID */
1306 
1307 	cputime_t utime, stime, utimescaled, stimescaled;
1308 	cputime_t gtime;
1309 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1310 	struct cputime prev_cputime;
1311 #endif
1312 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1313 	seqlock_t vtime_seqlock;
1314 	unsigned long long vtime_snap;
1315 	enum {
1316 		VTIME_SLEEPING = 0,
1317 		VTIME_USER,
1318 		VTIME_SYS,
1319 	} vtime_snap_whence;
1320 #endif
1321 	unsigned long nvcsw, nivcsw; /* context switch counts */
1322 	struct timespec start_time; 		/* monotonic time */
1323 	struct timespec real_start_time;	/* boot based time */
1324 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1325 	unsigned long min_flt, maj_flt;
1326 
1327 	struct task_cputime cputime_expires;
1328 	struct list_head cpu_timers[3];
1329 
1330 /* process credentials */
1331 	const struct cred __rcu *real_cred; /* objective and real subjective task
1332 					 * credentials (COW) */
1333 	const struct cred __rcu *cred;	/* effective (overridable) subjective task
1334 					 * credentials (COW) */
1335 	char comm[TASK_COMM_LEN]; /* executable name excluding path
1336 				     - access with [gs]et_task_comm (which lock
1337 				       it with task_lock())
1338 				     - initialized normally by setup_new_exec */
1339 /* file system info */
1340 	int link_count, total_link_count;
1341 #ifdef CONFIG_SYSVIPC
1342 /* ipc stuff */
1343 	struct sysv_sem sysvsem;
1344 #endif
1345 #ifdef CONFIG_DETECT_HUNG_TASK
1346 /* hung task detection */
1347 	unsigned long last_switch_count;
1348 #endif
1349 /* CPU-specific state of this task */
1350 	struct thread_struct thread;
1351 /* filesystem information */
1352 	struct fs_struct *fs;
1353 /* open file information */
1354 	struct files_struct *files;
1355 /* namespaces */
1356 	struct nsproxy *nsproxy;
1357 /* signal handlers */
1358 	struct signal_struct *signal;
1359 	struct sighand_struct *sighand;
1360 
1361 	sigset_t blocked, real_blocked;
1362 	sigset_t saved_sigmask;	/* restored if set_restore_sigmask() was used */
1363 	struct sigpending pending;
1364 
1365 	unsigned long sas_ss_sp;
1366 	size_t sas_ss_size;
1367 	int (*notifier)(void *priv);
1368 	void *notifier_data;
1369 	sigset_t *notifier_mask;
1370 	struct callback_head *task_works;
1371 
1372 	struct audit_context *audit_context;
1373 #ifdef CONFIG_AUDITSYSCALL
1374 	kuid_t loginuid;
1375 	unsigned int sessionid;
1376 #endif
1377 	struct seccomp seccomp;
1378 
1379 /* Thread group tracking */
1380    	u32 parent_exec_id;
1381    	u32 self_exec_id;
1382 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1383  * mempolicy */
1384 	spinlock_t alloc_lock;
1385 
1386 	/* Protection of the PI data structures: */
1387 	raw_spinlock_t pi_lock;
1388 
1389 #ifdef CONFIG_RT_MUTEXES
1390 	/* PI waiters blocked on a rt_mutex held by this task */
1391 	struct rb_root pi_waiters;
1392 	struct rb_node *pi_waiters_leftmost;
1393 	/* Deadlock detection and priority inheritance handling */
1394 	struct rt_mutex_waiter *pi_blocked_on;
1395 	/* Top pi_waiters task */
1396 	struct task_struct *pi_top_task;
1397 #endif
1398 
1399 #ifdef CONFIG_DEBUG_MUTEXES
1400 	/* mutex deadlock detection */
1401 	struct mutex_waiter *blocked_on;
1402 #endif
1403 #ifdef CONFIG_TRACE_IRQFLAGS
1404 	unsigned int irq_events;
1405 	unsigned long hardirq_enable_ip;
1406 	unsigned long hardirq_disable_ip;
1407 	unsigned int hardirq_enable_event;
1408 	unsigned int hardirq_disable_event;
1409 	int hardirqs_enabled;
1410 	int hardirq_context;
1411 	unsigned long softirq_disable_ip;
1412 	unsigned long softirq_enable_ip;
1413 	unsigned int softirq_disable_event;
1414 	unsigned int softirq_enable_event;
1415 	int softirqs_enabled;
1416 	int softirq_context;
1417 #endif
1418 #ifdef CONFIG_LOCKDEP
1419 # define MAX_LOCK_DEPTH 48UL
1420 	u64 curr_chain_key;
1421 	int lockdep_depth;
1422 	unsigned int lockdep_recursion;
1423 	struct held_lock held_locks[MAX_LOCK_DEPTH];
1424 	gfp_t lockdep_reclaim_gfp;
1425 #endif
1426 
1427 /* journalling filesystem info */
1428 	void *journal_info;
1429 
1430 /* stacked block device info */
1431 	struct bio_list *bio_list;
1432 
1433 #ifdef CONFIG_BLOCK
1434 /* stack plugging */
1435 	struct blk_plug *plug;
1436 #endif
1437 
1438 /* VM state */
1439 	struct reclaim_state *reclaim_state;
1440 
1441 	struct backing_dev_info *backing_dev_info;
1442 
1443 	struct io_context *io_context;
1444 
1445 	unsigned long ptrace_message;
1446 	siginfo_t *last_siginfo; /* For ptrace use.  */
1447 	struct task_io_accounting ioac;
1448 #if defined(CONFIG_TASK_XACCT)
1449 	u64 acct_rss_mem1;	/* accumulated rss usage */
1450 	u64 acct_vm_mem1;	/* accumulated virtual memory usage */
1451 	cputime_t acct_timexpd;	/* stime + utime since last update */
1452 #endif
1453 #ifdef CONFIG_CPUSETS
1454 	nodemask_t mems_allowed;	/* Protected by alloc_lock */
1455 	seqcount_t mems_allowed_seq;	/* Seqence no to catch updates */
1456 	int cpuset_mem_spread_rotor;
1457 	int cpuset_slab_spread_rotor;
1458 #endif
1459 #ifdef CONFIG_CGROUPS
1460 	/* Control Group info protected by css_set_lock */
1461 	struct css_set __rcu *cgroups;
1462 	/* cg_list protected by css_set_lock and tsk->alloc_lock */
1463 	struct list_head cg_list;
1464 #endif
1465 #ifdef CONFIG_FUTEX
1466 	struct robust_list_head __user *robust_list;
1467 #ifdef CONFIG_COMPAT
1468 	struct compat_robust_list_head __user *compat_robust_list;
1469 #endif
1470 	struct list_head pi_state_list;
1471 	struct futex_pi_state *pi_state_cache;
1472 #endif
1473 #ifdef CONFIG_PERF_EVENTS
1474 	struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1475 	struct mutex perf_event_mutex;
1476 	struct list_head perf_event_list;
1477 #endif
1478 #ifdef CONFIG_DEBUG_PREEMPT
1479 	unsigned long preempt_disable_ip;
1480 #endif
1481 #ifdef CONFIG_NUMA
1482 	struct mempolicy *mempolicy;	/* Protected by alloc_lock */
1483 	short il_next;
1484 	short pref_node_fork;
1485 #endif
1486 #ifdef CONFIG_NUMA_BALANCING
1487 	int numa_scan_seq;
1488 	unsigned int numa_scan_period;
1489 	unsigned int numa_scan_period_max;
1490 	int numa_preferred_nid;
1491 	unsigned long numa_migrate_retry;
1492 	u64 node_stamp;			/* migration stamp  */
1493 	u64 last_task_numa_placement;
1494 	u64 last_sum_exec_runtime;
1495 	struct callback_head numa_work;
1496 
1497 	struct list_head numa_entry;
1498 	struct numa_group *numa_group;
1499 
1500 	/*
1501 	 * Exponential decaying average of faults on a per-node basis.
1502 	 * Scheduling placement decisions are made based on the these counts.
1503 	 * The values remain static for the duration of a PTE scan
1504 	 */
1505 	unsigned long *numa_faults_memory;
1506 	unsigned long total_numa_faults;
1507 
1508 	/*
1509 	 * numa_faults_buffer records faults per node during the current
1510 	 * scan window. When the scan completes, the counts in
1511 	 * numa_faults_memory decay and these values are copied.
1512 	 */
1513 	unsigned long *numa_faults_buffer_memory;
1514 
1515 	/*
1516 	 * Track the nodes the process was running on when a NUMA hinting
1517 	 * fault was incurred.
1518 	 */
1519 	unsigned long *numa_faults_cpu;
1520 	unsigned long *numa_faults_buffer_cpu;
1521 
1522 	/*
1523 	 * numa_faults_locality tracks if faults recorded during the last
1524 	 * scan window were remote/local. The task scan period is adapted
1525 	 * based on the locality of the faults with different weights
1526 	 * depending on whether they were shared or private faults
1527 	 */
1528 	unsigned long numa_faults_locality[2];
1529 
1530 	unsigned long numa_pages_migrated;
1531 #endif /* CONFIG_NUMA_BALANCING */
1532 
1533 	struct rcu_head rcu;
1534 
1535 	/*
1536 	 * cache last used pipe for splice
1537 	 */
1538 	struct pipe_inode_info *splice_pipe;
1539 
1540 	struct page_frag task_frag;
1541 
1542 #ifdef	CONFIG_TASK_DELAY_ACCT
1543 	struct task_delay_info *delays;
1544 #endif
1545 #ifdef CONFIG_FAULT_INJECTION
1546 	int make_it_fail;
1547 #endif
1548 	/*
1549 	 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1550 	 * balance_dirty_pages() for some dirty throttling pause
1551 	 */
1552 	int nr_dirtied;
1553 	int nr_dirtied_pause;
1554 	unsigned long dirty_paused_when; /* start of a write-and-pause period */
1555 
1556 #ifdef CONFIG_LATENCYTOP
1557 	int latency_record_count;
1558 	struct latency_record latency_record[LT_SAVECOUNT];
1559 #endif
1560 	/*
1561 	 * time slack values; these are used to round up poll() and
1562 	 * select() etc timeout values. These are in nanoseconds.
1563 	 */
1564 	unsigned long timer_slack_ns;
1565 	unsigned long default_timer_slack_ns;
1566 
1567 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1568 	/* Index of current stored address in ret_stack */
1569 	int curr_ret_stack;
1570 	/* Stack of return addresses for return function tracing */
1571 	struct ftrace_ret_stack	*ret_stack;
1572 	/* time stamp for last schedule */
1573 	unsigned long long ftrace_timestamp;
1574 	/*
1575 	 * Number of functions that haven't been traced
1576 	 * because of depth overrun.
1577 	 */
1578 	atomic_t trace_overrun;
1579 	/* Pause for the tracing */
1580 	atomic_t tracing_graph_pause;
1581 #endif
1582 #ifdef CONFIG_TRACING
1583 	/* state flags for use by tracers */
1584 	unsigned long trace;
1585 	/* bitmask and counter of trace recursion */
1586 	unsigned long trace_recursion;
1587 #endif /* CONFIG_TRACING */
1588 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1589 	struct memcg_batch_info {
1590 		int do_batch;	/* incremented when batch uncharge started */
1591 		struct mem_cgroup *memcg; /* target memcg of uncharge */
1592 		unsigned long nr_pages;	/* uncharged usage */
1593 		unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1594 	} memcg_batch;
1595 	unsigned int memcg_kmem_skip_account;
1596 	struct memcg_oom_info {
1597 		struct mem_cgroup *memcg;
1598 		gfp_t gfp_mask;
1599 		int order;
1600 		unsigned int may_oom:1;
1601 	} memcg_oom;
1602 #endif
1603 #ifdef CONFIG_UPROBES
1604 	struct uprobe_task *utask;
1605 #endif
1606 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1607 	unsigned int	sequential_io;
1608 	unsigned int	sequential_io_avg;
1609 #endif
1610 };
1611 
1612 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1613 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1614 
1615 #define TNF_MIGRATED	0x01
1616 #define TNF_NO_GROUP	0x02
1617 #define TNF_SHARED	0x04
1618 #define TNF_FAULT_LOCAL	0x08
1619 
1620 #ifdef CONFIG_NUMA_BALANCING
1621 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1622 extern pid_t task_numa_group_id(struct task_struct *p);
1623 extern void set_numabalancing_state(bool enabled);
1624 extern void task_numa_free(struct task_struct *p);
1625 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1626 					int src_nid, int dst_cpu);
1627 #else
1628 static inline void task_numa_fault(int last_node, int node, int pages,
1629 				   int flags)
1630 {
1631 }
1632 static inline pid_t task_numa_group_id(struct task_struct *p)
1633 {
1634 	return 0;
1635 }
1636 static inline void set_numabalancing_state(bool enabled)
1637 {
1638 }
1639 static inline void task_numa_free(struct task_struct *p)
1640 {
1641 }
1642 static inline bool should_numa_migrate_memory(struct task_struct *p,
1643 				struct page *page, int src_nid, int dst_cpu)
1644 {
1645 	return true;
1646 }
1647 #endif
1648 
1649 static inline struct pid *task_pid(struct task_struct *task)
1650 {
1651 	return task->pids[PIDTYPE_PID].pid;
1652 }
1653 
1654 static inline struct pid *task_tgid(struct task_struct *task)
1655 {
1656 	return task->group_leader->pids[PIDTYPE_PID].pid;
1657 }
1658 
1659 /*
1660  * Without tasklist or rcu lock it is not safe to dereference
1661  * the result of task_pgrp/task_session even if task == current,
1662  * we can race with another thread doing sys_setsid/sys_setpgid.
1663  */
1664 static inline struct pid *task_pgrp(struct task_struct *task)
1665 {
1666 	return task->group_leader->pids[PIDTYPE_PGID].pid;
1667 }
1668 
1669 static inline struct pid *task_session(struct task_struct *task)
1670 {
1671 	return task->group_leader->pids[PIDTYPE_SID].pid;
1672 }
1673 
1674 struct pid_namespace;
1675 
1676 /*
1677  * the helpers to get the task's different pids as they are seen
1678  * from various namespaces
1679  *
1680  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1681  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1682  *                     current.
1683  * task_xid_nr_ns()  : id seen from the ns specified;
1684  *
1685  * set_task_vxid()   : assigns a virtual id to a task;
1686  *
1687  * see also pid_nr() etc in include/linux/pid.h
1688  */
1689 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1690 			struct pid_namespace *ns);
1691 
1692 static inline pid_t task_pid_nr(struct task_struct *tsk)
1693 {
1694 	return tsk->pid;
1695 }
1696 
1697 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1698 					struct pid_namespace *ns)
1699 {
1700 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1701 }
1702 
1703 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1704 {
1705 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1706 }
1707 
1708 
1709 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1710 {
1711 	return tsk->tgid;
1712 }
1713 
1714 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1715 
1716 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1717 {
1718 	return pid_vnr(task_tgid(tsk));
1719 }
1720 
1721 
1722 static inline int pid_alive(const struct task_struct *p);
1723 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1724 {
1725 	pid_t pid = 0;
1726 
1727 	rcu_read_lock();
1728 	if (pid_alive(tsk))
1729 		pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1730 	rcu_read_unlock();
1731 
1732 	return pid;
1733 }
1734 
1735 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1736 {
1737 	return task_ppid_nr_ns(tsk, &init_pid_ns);
1738 }
1739 
1740 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1741 					struct pid_namespace *ns)
1742 {
1743 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1744 }
1745 
1746 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1747 {
1748 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1749 }
1750 
1751 
1752 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1753 					struct pid_namespace *ns)
1754 {
1755 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1756 }
1757 
1758 static inline pid_t task_session_vnr(struct task_struct *tsk)
1759 {
1760 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1761 }
1762 
1763 /* obsolete, do not use */
1764 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1765 {
1766 	return task_pgrp_nr_ns(tsk, &init_pid_ns);
1767 }
1768 
1769 /**
1770  * pid_alive - check that a task structure is not stale
1771  * @p: Task structure to be checked.
1772  *
1773  * Test if a process is not yet dead (at most zombie state)
1774  * If pid_alive fails, then pointers within the task structure
1775  * can be stale and must not be dereferenced.
1776  *
1777  * Return: 1 if the process is alive. 0 otherwise.
1778  */
1779 static inline int pid_alive(const struct task_struct *p)
1780 {
1781 	return p->pids[PIDTYPE_PID].pid != NULL;
1782 }
1783 
1784 /**
1785  * is_global_init - check if a task structure is init
1786  * @tsk: Task structure to be checked.
1787  *
1788  * Check if a task structure is the first user space task the kernel created.
1789  *
1790  * Return: 1 if the task structure is init. 0 otherwise.
1791  */
1792 static inline int is_global_init(struct task_struct *tsk)
1793 {
1794 	return tsk->pid == 1;
1795 }
1796 
1797 extern struct pid *cad_pid;
1798 
1799 extern void free_task(struct task_struct *tsk);
1800 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1801 
1802 extern void __put_task_struct(struct task_struct *t);
1803 
1804 static inline void put_task_struct(struct task_struct *t)
1805 {
1806 	if (atomic_dec_and_test(&t->usage))
1807 		__put_task_struct(t);
1808 }
1809 
1810 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1811 extern void task_cputime(struct task_struct *t,
1812 			 cputime_t *utime, cputime_t *stime);
1813 extern void task_cputime_scaled(struct task_struct *t,
1814 				cputime_t *utimescaled, cputime_t *stimescaled);
1815 extern cputime_t task_gtime(struct task_struct *t);
1816 #else
1817 static inline void task_cputime(struct task_struct *t,
1818 				cputime_t *utime, cputime_t *stime)
1819 {
1820 	if (utime)
1821 		*utime = t->utime;
1822 	if (stime)
1823 		*stime = t->stime;
1824 }
1825 
1826 static inline void task_cputime_scaled(struct task_struct *t,
1827 				       cputime_t *utimescaled,
1828 				       cputime_t *stimescaled)
1829 {
1830 	if (utimescaled)
1831 		*utimescaled = t->utimescaled;
1832 	if (stimescaled)
1833 		*stimescaled = t->stimescaled;
1834 }
1835 
1836 static inline cputime_t task_gtime(struct task_struct *t)
1837 {
1838 	return t->gtime;
1839 }
1840 #endif
1841 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1842 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1843 
1844 /*
1845  * Per process flags
1846  */
1847 #define PF_EXITING	0x00000004	/* getting shut down */
1848 #define PF_EXITPIDONE	0x00000008	/* pi exit done on shut down */
1849 #define PF_VCPU		0x00000010	/* I'm a virtual CPU */
1850 #define PF_WQ_WORKER	0x00000020	/* I'm a workqueue worker */
1851 #define PF_FORKNOEXEC	0x00000040	/* forked but didn't exec */
1852 #define PF_MCE_PROCESS  0x00000080      /* process policy on mce errors */
1853 #define PF_SUPERPRIV	0x00000100	/* used super-user privileges */
1854 #define PF_DUMPCORE	0x00000200	/* dumped core */
1855 #define PF_SIGNALED	0x00000400	/* killed by a signal */
1856 #define PF_MEMALLOC	0x00000800	/* Allocating memory */
1857 #define PF_NPROC_EXCEEDED 0x00001000	/* set_user noticed that RLIMIT_NPROC was exceeded */
1858 #define PF_USED_MATH	0x00002000	/* if unset the fpu must be initialized before use */
1859 #define PF_USED_ASYNC	0x00004000	/* used async_schedule*(), used by module init */
1860 #define PF_NOFREEZE	0x00008000	/* this thread should not be frozen */
1861 #define PF_FROZEN	0x00010000	/* frozen for system suspend */
1862 #define PF_FSTRANS	0x00020000	/* inside a filesystem transaction */
1863 #define PF_KSWAPD	0x00040000	/* I am kswapd */
1864 #define PF_MEMALLOC_NOIO 0x00080000	/* Allocating memory without IO involved */
1865 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
1866 #define PF_KTHREAD	0x00200000	/* I am a kernel thread */
1867 #define PF_RANDOMIZE	0x00400000	/* randomize virtual address space */
1868 #define PF_SWAPWRITE	0x00800000	/* Allowed to write to swap */
1869 #define PF_SPREAD_PAGE	0x01000000	/* Spread page cache over cpuset */
1870 #define PF_SPREAD_SLAB	0x02000000	/* Spread some slab caches over cpuset */
1871 #define PF_NO_SETAFFINITY 0x04000000	/* Userland is not allowed to meddle with cpus_allowed */
1872 #define PF_MCE_EARLY    0x08000000      /* Early kill for mce process policy */
1873 #define PF_MUTEX_TESTER	0x20000000	/* Thread belongs to the rt mutex tester */
1874 #define PF_FREEZER_SKIP	0x40000000	/* Freezer should not count it as freezable */
1875 #define PF_SUSPEND_TASK 0x80000000      /* this thread called freeze_processes and should not be frozen */
1876 
1877 /*
1878  * Only the _current_ task can read/write to tsk->flags, but other
1879  * tasks can access tsk->flags in readonly mode for example
1880  * with tsk_used_math (like during threaded core dumping).
1881  * There is however an exception to this rule during ptrace
1882  * or during fork: the ptracer task is allowed to write to the
1883  * child->flags of its traced child (same goes for fork, the parent
1884  * can write to the child->flags), because we're guaranteed the
1885  * child is not running and in turn not changing child->flags
1886  * at the same time the parent does it.
1887  */
1888 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1889 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1890 #define clear_used_math() clear_stopped_child_used_math(current)
1891 #define set_used_math() set_stopped_child_used_math(current)
1892 #define conditional_stopped_child_used_math(condition, child) \
1893 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1894 #define conditional_used_math(condition) \
1895 	conditional_stopped_child_used_math(condition, current)
1896 #define copy_to_stopped_child_used_math(child) \
1897 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1898 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1899 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1900 #define used_math() tsk_used_math(current)
1901 
1902 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1903 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1904 {
1905 	if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1906 		flags &= ~__GFP_IO;
1907 	return flags;
1908 }
1909 
1910 static inline unsigned int memalloc_noio_save(void)
1911 {
1912 	unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1913 	current->flags |= PF_MEMALLOC_NOIO;
1914 	return flags;
1915 }
1916 
1917 static inline void memalloc_noio_restore(unsigned int flags)
1918 {
1919 	current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1920 }
1921 
1922 /*
1923  * task->jobctl flags
1924  */
1925 #define JOBCTL_STOP_SIGMASK	0xffff	/* signr of the last group stop */
1926 
1927 #define JOBCTL_STOP_DEQUEUED_BIT 16	/* stop signal dequeued */
1928 #define JOBCTL_STOP_PENDING_BIT	17	/* task should stop for group stop */
1929 #define JOBCTL_STOP_CONSUME_BIT	18	/* consume group stop count */
1930 #define JOBCTL_TRAP_STOP_BIT	19	/* trap for STOP */
1931 #define JOBCTL_TRAP_NOTIFY_BIT	20	/* trap for NOTIFY */
1932 #define JOBCTL_TRAPPING_BIT	21	/* switching to TRACED */
1933 #define JOBCTL_LISTENING_BIT	22	/* ptracer is listening for events */
1934 
1935 #define JOBCTL_STOP_DEQUEUED	(1 << JOBCTL_STOP_DEQUEUED_BIT)
1936 #define JOBCTL_STOP_PENDING	(1 << JOBCTL_STOP_PENDING_BIT)
1937 #define JOBCTL_STOP_CONSUME	(1 << JOBCTL_STOP_CONSUME_BIT)
1938 #define JOBCTL_TRAP_STOP	(1 << JOBCTL_TRAP_STOP_BIT)
1939 #define JOBCTL_TRAP_NOTIFY	(1 << JOBCTL_TRAP_NOTIFY_BIT)
1940 #define JOBCTL_TRAPPING		(1 << JOBCTL_TRAPPING_BIT)
1941 #define JOBCTL_LISTENING	(1 << JOBCTL_LISTENING_BIT)
1942 
1943 #define JOBCTL_TRAP_MASK	(JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1944 #define JOBCTL_PENDING_MASK	(JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1945 
1946 extern bool task_set_jobctl_pending(struct task_struct *task,
1947 				    unsigned int mask);
1948 extern void task_clear_jobctl_trapping(struct task_struct *task);
1949 extern void task_clear_jobctl_pending(struct task_struct *task,
1950 				      unsigned int mask);
1951 
1952 #ifdef CONFIG_PREEMPT_RCU
1953 
1954 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1955 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1956 
1957 static inline void rcu_copy_process(struct task_struct *p)
1958 {
1959 	p->rcu_read_lock_nesting = 0;
1960 	p->rcu_read_unlock_special = 0;
1961 #ifdef CONFIG_TREE_PREEMPT_RCU
1962 	p->rcu_blocked_node = NULL;
1963 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1964 #ifdef CONFIG_RCU_BOOST
1965 	p->rcu_boost_mutex = NULL;
1966 #endif /* #ifdef CONFIG_RCU_BOOST */
1967 	INIT_LIST_HEAD(&p->rcu_node_entry);
1968 }
1969 
1970 #else
1971 
1972 static inline void rcu_copy_process(struct task_struct *p)
1973 {
1974 }
1975 
1976 #endif
1977 
1978 static inline void tsk_restore_flags(struct task_struct *task,
1979 				unsigned long orig_flags, unsigned long flags)
1980 {
1981 	task->flags &= ~flags;
1982 	task->flags |= orig_flags & flags;
1983 }
1984 
1985 #ifdef CONFIG_SMP
1986 extern void do_set_cpus_allowed(struct task_struct *p,
1987 			       const struct cpumask *new_mask);
1988 
1989 extern int set_cpus_allowed_ptr(struct task_struct *p,
1990 				const struct cpumask *new_mask);
1991 #else
1992 static inline void do_set_cpus_allowed(struct task_struct *p,
1993 				      const struct cpumask *new_mask)
1994 {
1995 }
1996 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1997 				       const struct cpumask *new_mask)
1998 {
1999 	if (!cpumask_test_cpu(0, new_mask))
2000 		return -EINVAL;
2001 	return 0;
2002 }
2003 #endif
2004 
2005 #ifdef CONFIG_NO_HZ_COMMON
2006 void calc_load_enter_idle(void);
2007 void calc_load_exit_idle(void);
2008 #else
2009 static inline void calc_load_enter_idle(void) { }
2010 static inline void calc_load_exit_idle(void) { }
2011 #endif /* CONFIG_NO_HZ_COMMON */
2012 
2013 #ifndef CONFIG_CPUMASK_OFFSTACK
2014 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2015 {
2016 	return set_cpus_allowed_ptr(p, &new_mask);
2017 }
2018 #endif
2019 
2020 /*
2021  * Do not use outside of architecture code which knows its limitations.
2022  *
2023  * sched_clock() has no promise of monotonicity or bounded drift between
2024  * CPUs, use (which you should not) requires disabling IRQs.
2025  *
2026  * Please use one of the three interfaces below.
2027  */
2028 extern unsigned long long notrace sched_clock(void);
2029 /*
2030  * See the comment in kernel/sched/clock.c
2031  */
2032 extern u64 cpu_clock(int cpu);
2033 extern u64 local_clock(void);
2034 extern u64 sched_clock_cpu(int cpu);
2035 
2036 
2037 extern void sched_clock_init(void);
2038 
2039 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2040 static inline void sched_clock_tick(void)
2041 {
2042 }
2043 
2044 static inline void sched_clock_idle_sleep_event(void)
2045 {
2046 }
2047 
2048 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2049 {
2050 }
2051 #else
2052 /*
2053  * Architectures can set this to 1 if they have specified
2054  * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2055  * but then during bootup it turns out that sched_clock()
2056  * is reliable after all:
2057  */
2058 extern int sched_clock_stable(void);
2059 extern void set_sched_clock_stable(void);
2060 extern void clear_sched_clock_stable(void);
2061 
2062 extern void sched_clock_tick(void);
2063 extern void sched_clock_idle_sleep_event(void);
2064 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2065 #endif
2066 
2067 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2068 /*
2069  * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2070  * The reason for this explicit opt-in is not to have perf penalty with
2071  * slow sched_clocks.
2072  */
2073 extern void enable_sched_clock_irqtime(void);
2074 extern void disable_sched_clock_irqtime(void);
2075 #else
2076 static inline void enable_sched_clock_irqtime(void) {}
2077 static inline void disable_sched_clock_irqtime(void) {}
2078 #endif
2079 
2080 extern unsigned long long
2081 task_sched_runtime(struct task_struct *task);
2082 
2083 /* sched_exec is called by processes performing an exec */
2084 #ifdef CONFIG_SMP
2085 extern void sched_exec(void);
2086 #else
2087 #define sched_exec()   {}
2088 #endif
2089 
2090 extern void sched_clock_idle_sleep_event(void);
2091 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2092 
2093 #ifdef CONFIG_HOTPLUG_CPU
2094 extern void idle_task_exit(void);
2095 #else
2096 static inline void idle_task_exit(void) {}
2097 #endif
2098 
2099 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2100 extern void wake_up_nohz_cpu(int cpu);
2101 #else
2102 static inline void wake_up_nohz_cpu(int cpu) { }
2103 #endif
2104 
2105 #ifdef CONFIG_NO_HZ_FULL
2106 extern bool sched_can_stop_tick(void);
2107 extern u64 scheduler_tick_max_deferment(void);
2108 #else
2109 static inline bool sched_can_stop_tick(void) { return false; }
2110 #endif
2111 
2112 #ifdef CONFIG_SCHED_AUTOGROUP
2113 extern void sched_autogroup_create_attach(struct task_struct *p);
2114 extern void sched_autogroup_detach(struct task_struct *p);
2115 extern void sched_autogroup_fork(struct signal_struct *sig);
2116 extern void sched_autogroup_exit(struct signal_struct *sig);
2117 #ifdef CONFIG_PROC_FS
2118 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2119 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2120 #endif
2121 #else
2122 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2123 static inline void sched_autogroup_detach(struct task_struct *p) { }
2124 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2125 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2126 #endif
2127 
2128 extern bool yield_to(struct task_struct *p, bool preempt);
2129 extern void set_user_nice(struct task_struct *p, long nice);
2130 extern int task_prio(const struct task_struct *p);
2131 /**
2132  * task_nice - return the nice value of a given task.
2133  * @p: the task in question.
2134  *
2135  * Return: The nice value [ -20 ... 0 ... 19 ].
2136  */
2137 static inline int task_nice(const struct task_struct *p)
2138 {
2139 	return PRIO_TO_NICE((p)->static_prio);
2140 }
2141 extern int can_nice(const struct task_struct *p, const int nice);
2142 extern int task_curr(const struct task_struct *p);
2143 extern int idle_cpu(int cpu);
2144 extern int sched_setscheduler(struct task_struct *, int,
2145 			      const struct sched_param *);
2146 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2147 				      const struct sched_param *);
2148 extern int sched_setattr(struct task_struct *,
2149 			 const struct sched_attr *);
2150 extern struct task_struct *idle_task(int cpu);
2151 /**
2152  * is_idle_task - is the specified task an idle task?
2153  * @p: the task in question.
2154  *
2155  * Return: 1 if @p is an idle task. 0 otherwise.
2156  */
2157 static inline bool is_idle_task(const struct task_struct *p)
2158 {
2159 	return p->pid == 0;
2160 }
2161 extern struct task_struct *curr_task(int cpu);
2162 extern void set_curr_task(int cpu, struct task_struct *p);
2163 
2164 void yield(void);
2165 
2166 /*
2167  * The default (Linux) execution domain.
2168  */
2169 extern struct exec_domain	default_exec_domain;
2170 
2171 union thread_union {
2172 	struct thread_info thread_info;
2173 	unsigned long stack[THREAD_SIZE/sizeof(long)];
2174 };
2175 
2176 #ifndef __HAVE_ARCH_KSTACK_END
2177 static inline int kstack_end(void *addr)
2178 {
2179 	/* Reliable end of stack detection:
2180 	 * Some APM bios versions misalign the stack
2181 	 */
2182 	return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2183 }
2184 #endif
2185 
2186 extern union thread_union init_thread_union;
2187 extern struct task_struct init_task;
2188 
2189 extern struct   mm_struct init_mm;
2190 
2191 extern struct pid_namespace init_pid_ns;
2192 
2193 /*
2194  * find a task by one of its numerical ids
2195  *
2196  * find_task_by_pid_ns():
2197  *      finds a task by its pid in the specified namespace
2198  * find_task_by_vpid():
2199  *      finds a task by its virtual pid
2200  *
2201  * see also find_vpid() etc in include/linux/pid.h
2202  */
2203 
2204 extern struct task_struct *find_task_by_vpid(pid_t nr);
2205 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2206 		struct pid_namespace *ns);
2207 
2208 /* per-UID process charging. */
2209 extern struct user_struct * alloc_uid(kuid_t);
2210 static inline struct user_struct *get_uid(struct user_struct *u)
2211 {
2212 	atomic_inc(&u->__count);
2213 	return u;
2214 }
2215 extern void free_uid(struct user_struct *);
2216 
2217 #include <asm/current.h>
2218 
2219 extern void xtime_update(unsigned long ticks);
2220 
2221 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2222 extern int wake_up_process(struct task_struct *tsk);
2223 extern void wake_up_new_task(struct task_struct *tsk);
2224 #ifdef CONFIG_SMP
2225  extern void kick_process(struct task_struct *tsk);
2226 #else
2227  static inline void kick_process(struct task_struct *tsk) { }
2228 #endif
2229 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2230 extern void sched_dead(struct task_struct *p);
2231 
2232 extern void proc_caches_init(void);
2233 extern void flush_signals(struct task_struct *);
2234 extern void __flush_signals(struct task_struct *);
2235 extern void ignore_signals(struct task_struct *);
2236 extern void flush_signal_handlers(struct task_struct *, int force_default);
2237 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2238 
2239 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2240 {
2241 	unsigned long flags;
2242 	int ret;
2243 
2244 	spin_lock_irqsave(&tsk->sighand->siglock, flags);
2245 	ret = dequeue_signal(tsk, mask, info);
2246 	spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2247 
2248 	return ret;
2249 }
2250 
2251 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2252 			      sigset_t *mask);
2253 extern void unblock_all_signals(void);
2254 extern void release_task(struct task_struct * p);
2255 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2256 extern int force_sigsegv(int, struct task_struct *);
2257 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2258 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2259 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2260 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2261 				const struct cred *, u32);
2262 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2263 extern int kill_pid(struct pid *pid, int sig, int priv);
2264 extern int kill_proc_info(int, struct siginfo *, pid_t);
2265 extern __must_check bool do_notify_parent(struct task_struct *, int);
2266 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2267 extern void force_sig(int, struct task_struct *);
2268 extern int send_sig(int, struct task_struct *, int);
2269 extern int zap_other_threads(struct task_struct *p);
2270 extern struct sigqueue *sigqueue_alloc(void);
2271 extern void sigqueue_free(struct sigqueue *);
2272 extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
2273 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2274 
2275 static inline void restore_saved_sigmask(void)
2276 {
2277 	if (test_and_clear_restore_sigmask())
2278 		__set_current_blocked(&current->saved_sigmask);
2279 }
2280 
2281 static inline sigset_t *sigmask_to_save(void)
2282 {
2283 	sigset_t *res = &current->blocked;
2284 	if (unlikely(test_restore_sigmask()))
2285 		res = &current->saved_sigmask;
2286 	return res;
2287 }
2288 
2289 static inline int kill_cad_pid(int sig, int priv)
2290 {
2291 	return kill_pid(cad_pid, sig, priv);
2292 }
2293 
2294 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
2295 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2296 #define SEND_SIG_PRIV	((struct siginfo *) 1)
2297 #define SEND_SIG_FORCED	((struct siginfo *) 2)
2298 
2299 /*
2300  * True if we are on the alternate signal stack.
2301  */
2302 static inline int on_sig_stack(unsigned long sp)
2303 {
2304 #ifdef CONFIG_STACK_GROWSUP
2305 	return sp >= current->sas_ss_sp &&
2306 		sp - current->sas_ss_sp < current->sas_ss_size;
2307 #else
2308 	return sp > current->sas_ss_sp &&
2309 		sp - current->sas_ss_sp <= current->sas_ss_size;
2310 #endif
2311 }
2312 
2313 static inline int sas_ss_flags(unsigned long sp)
2314 {
2315 	return (current->sas_ss_size == 0 ? SS_DISABLE
2316 		: on_sig_stack(sp) ? SS_ONSTACK : 0);
2317 }
2318 
2319 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2320 {
2321 	if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2322 #ifdef CONFIG_STACK_GROWSUP
2323 		return current->sas_ss_sp;
2324 #else
2325 		return current->sas_ss_sp + current->sas_ss_size;
2326 #endif
2327 	return sp;
2328 }
2329 
2330 /*
2331  * Routines for handling mm_structs
2332  */
2333 extern struct mm_struct * mm_alloc(void);
2334 
2335 /* mmdrop drops the mm and the page tables */
2336 extern void __mmdrop(struct mm_struct *);
2337 static inline void mmdrop(struct mm_struct * mm)
2338 {
2339 	if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2340 		__mmdrop(mm);
2341 }
2342 
2343 /* mmput gets rid of the mappings and all user-space */
2344 extern void mmput(struct mm_struct *);
2345 /* Grab a reference to a task's mm, if it is not already going away */
2346 extern struct mm_struct *get_task_mm(struct task_struct *task);
2347 /*
2348  * Grab a reference to a task's mm, if it is not already going away
2349  * and ptrace_may_access with the mode parameter passed to it
2350  * succeeds.
2351  */
2352 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2353 /* Remove the current tasks stale references to the old mm_struct */
2354 extern void mm_release(struct task_struct *, struct mm_struct *);
2355 
2356 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2357 			struct task_struct *);
2358 extern void flush_thread(void);
2359 extern void exit_thread(void);
2360 
2361 extern void exit_files(struct task_struct *);
2362 extern void __cleanup_sighand(struct sighand_struct *);
2363 
2364 extern void exit_itimers(struct signal_struct *);
2365 extern void flush_itimer_signals(void);
2366 
2367 extern void do_group_exit(int);
2368 
2369 extern int allow_signal(int);
2370 extern int disallow_signal(int);
2371 
2372 extern int do_execve(struct filename *,
2373 		     const char __user * const __user *,
2374 		     const char __user * const __user *);
2375 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2376 struct task_struct *fork_idle(int);
2377 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2378 
2379 extern void set_task_comm(struct task_struct *tsk, const char *from);
2380 extern char *get_task_comm(char *to, struct task_struct *tsk);
2381 
2382 #ifdef CONFIG_SMP
2383 void scheduler_ipi(void);
2384 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2385 #else
2386 static inline void scheduler_ipi(void) { }
2387 static inline unsigned long wait_task_inactive(struct task_struct *p,
2388 					       long match_state)
2389 {
2390 	return 1;
2391 }
2392 #endif
2393 
2394 #define next_task(p) \
2395 	list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2396 
2397 #define for_each_process(p) \
2398 	for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2399 
2400 extern bool current_is_single_threaded(void);
2401 
2402 /*
2403  * Careful: do_each_thread/while_each_thread is a double loop so
2404  *          'break' will not work as expected - use goto instead.
2405  */
2406 #define do_each_thread(g, t) \
2407 	for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2408 
2409 #define while_each_thread(g, t) \
2410 	while ((t = next_thread(t)) != g)
2411 
2412 #define __for_each_thread(signal, t)	\
2413 	list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2414 
2415 #define for_each_thread(p, t)		\
2416 	__for_each_thread((p)->signal, t)
2417 
2418 /* Careful: this is a double loop, 'break' won't work as expected. */
2419 #define for_each_process_thread(p, t)	\
2420 	for_each_process(p) for_each_thread(p, t)
2421 
2422 static inline int get_nr_threads(struct task_struct *tsk)
2423 {
2424 	return tsk->signal->nr_threads;
2425 }
2426 
2427 static inline bool thread_group_leader(struct task_struct *p)
2428 {
2429 	return p->exit_signal >= 0;
2430 }
2431 
2432 /* Do to the insanities of de_thread it is possible for a process
2433  * to have the pid of the thread group leader without actually being
2434  * the thread group leader.  For iteration through the pids in proc
2435  * all we care about is that we have a task with the appropriate
2436  * pid, we don't actually care if we have the right task.
2437  */
2438 static inline bool has_group_leader_pid(struct task_struct *p)
2439 {
2440 	return task_pid(p) == p->signal->leader_pid;
2441 }
2442 
2443 static inline
2444 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2445 {
2446 	return p1->signal == p2->signal;
2447 }
2448 
2449 static inline struct task_struct *next_thread(const struct task_struct *p)
2450 {
2451 	return list_entry_rcu(p->thread_group.next,
2452 			      struct task_struct, thread_group);
2453 }
2454 
2455 static inline int thread_group_empty(struct task_struct *p)
2456 {
2457 	return list_empty(&p->thread_group);
2458 }
2459 
2460 #define delay_group_leader(p) \
2461 		(thread_group_leader(p) && !thread_group_empty(p))
2462 
2463 /*
2464  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2465  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2466  * pins the final release of task.io_context.  Also protects ->cpuset and
2467  * ->cgroup.subsys[]. And ->vfork_done.
2468  *
2469  * Nests both inside and outside of read_lock(&tasklist_lock).
2470  * It must not be nested with write_lock_irq(&tasklist_lock),
2471  * neither inside nor outside.
2472  */
2473 static inline void task_lock(struct task_struct *p)
2474 {
2475 	spin_lock(&p->alloc_lock);
2476 }
2477 
2478 static inline void task_unlock(struct task_struct *p)
2479 {
2480 	spin_unlock(&p->alloc_lock);
2481 }
2482 
2483 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2484 							unsigned long *flags);
2485 
2486 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2487 						       unsigned long *flags)
2488 {
2489 	struct sighand_struct *ret;
2490 
2491 	ret = __lock_task_sighand(tsk, flags);
2492 	(void)__cond_lock(&tsk->sighand->siglock, ret);
2493 	return ret;
2494 }
2495 
2496 static inline void unlock_task_sighand(struct task_struct *tsk,
2497 						unsigned long *flags)
2498 {
2499 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2500 }
2501 
2502 #ifdef CONFIG_CGROUPS
2503 static inline void threadgroup_change_begin(struct task_struct *tsk)
2504 {
2505 	down_read(&tsk->signal->group_rwsem);
2506 }
2507 static inline void threadgroup_change_end(struct task_struct *tsk)
2508 {
2509 	up_read(&tsk->signal->group_rwsem);
2510 }
2511 
2512 /**
2513  * threadgroup_lock - lock threadgroup
2514  * @tsk: member task of the threadgroup to lock
2515  *
2516  * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
2517  * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2518  * change ->group_leader/pid.  This is useful for cases where the threadgroup
2519  * needs to stay stable across blockable operations.
2520  *
2521  * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2522  * synchronization.  While held, no new task will be added to threadgroup
2523  * and no existing live task will have its PF_EXITING set.
2524  *
2525  * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2526  * sub-thread becomes a new leader.
2527  */
2528 static inline void threadgroup_lock(struct task_struct *tsk)
2529 {
2530 	down_write(&tsk->signal->group_rwsem);
2531 }
2532 
2533 /**
2534  * threadgroup_unlock - unlock threadgroup
2535  * @tsk: member task of the threadgroup to unlock
2536  *
2537  * Reverse threadgroup_lock().
2538  */
2539 static inline void threadgroup_unlock(struct task_struct *tsk)
2540 {
2541 	up_write(&tsk->signal->group_rwsem);
2542 }
2543 #else
2544 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2545 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2546 static inline void threadgroup_lock(struct task_struct *tsk) {}
2547 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2548 #endif
2549 
2550 #ifndef __HAVE_THREAD_FUNCTIONS
2551 
2552 #define task_thread_info(task)	((struct thread_info *)(task)->stack)
2553 #define task_stack_page(task)	((task)->stack)
2554 
2555 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2556 {
2557 	*task_thread_info(p) = *task_thread_info(org);
2558 	task_thread_info(p)->task = p;
2559 }
2560 
2561 static inline unsigned long *end_of_stack(struct task_struct *p)
2562 {
2563 	return (unsigned long *)(task_thread_info(p) + 1);
2564 }
2565 
2566 #endif
2567 
2568 static inline int object_is_on_stack(void *obj)
2569 {
2570 	void *stack = task_stack_page(current);
2571 
2572 	return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2573 }
2574 
2575 extern void thread_info_cache_init(void);
2576 
2577 #ifdef CONFIG_DEBUG_STACK_USAGE
2578 static inline unsigned long stack_not_used(struct task_struct *p)
2579 {
2580 	unsigned long *n = end_of_stack(p);
2581 
2582 	do { 	/* Skip over canary */
2583 		n++;
2584 	} while (!*n);
2585 
2586 	return (unsigned long)n - (unsigned long)end_of_stack(p);
2587 }
2588 #endif
2589 
2590 /* set thread flags in other task's structures
2591  * - see asm/thread_info.h for TIF_xxxx flags available
2592  */
2593 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2594 {
2595 	set_ti_thread_flag(task_thread_info(tsk), flag);
2596 }
2597 
2598 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2599 {
2600 	clear_ti_thread_flag(task_thread_info(tsk), flag);
2601 }
2602 
2603 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2604 {
2605 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2606 }
2607 
2608 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2609 {
2610 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2611 }
2612 
2613 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2614 {
2615 	return test_ti_thread_flag(task_thread_info(tsk), flag);
2616 }
2617 
2618 static inline void set_tsk_need_resched(struct task_struct *tsk)
2619 {
2620 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2621 }
2622 
2623 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2624 {
2625 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2626 }
2627 
2628 static inline int test_tsk_need_resched(struct task_struct *tsk)
2629 {
2630 	return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2631 }
2632 
2633 static inline int restart_syscall(void)
2634 {
2635 	set_tsk_thread_flag(current, TIF_SIGPENDING);
2636 	return -ERESTARTNOINTR;
2637 }
2638 
2639 static inline int signal_pending(struct task_struct *p)
2640 {
2641 	return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2642 }
2643 
2644 static inline int __fatal_signal_pending(struct task_struct *p)
2645 {
2646 	return unlikely(sigismember(&p->pending.signal, SIGKILL));
2647 }
2648 
2649 static inline int fatal_signal_pending(struct task_struct *p)
2650 {
2651 	return signal_pending(p) && __fatal_signal_pending(p);
2652 }
2653 
2654 static inline int signal_pending_state(long state, struct task_struct *p)
2655 {
2656 	if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2657 		return 0;
2658 	if (!signal_pending(p))
2659 		return 0;
2660 
2661 	return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2662 }
2663 
2664 /*
2665  * cond_resched() and cond_resched_lock(): latency reduction via
2666  * explicit rescheduling in places that are safe. The return
2667  * value indicates whether a reschedule was done in fact.
2668  * cond_resched_lock() will drop the spinlock before scheduling,
2669  * cond_resched_softirq() will enable bhs before scheduling.
2670  */
2671 extern int _cond_resched(void);
2672 
2673 #define cond_resched() ({			\
2674 	__might_sleep(__FILE__, __LINE__, 0);	\
2675 	_cond_resched();			\
2676 })
2677 
2678 extern int __cond_resched_lock(spinlock_t *lock);
2679 
2680 #ifdef CONFIG_PREEMPT_COUNT
2681 #define PREEMPT_LOCK_OFFSET	PREEMPT_OFFSET
2682 #else
2683 #define PREEMPT_LOCK_OFFSET	0
2684 #endif
2685 
2686 #define cond_resched_lock(lock) ({				\
2687 	__might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);	\
2688 	__cond_resched_lock(lock);				\
2689 })
2690 
2691 extern int __cond_resched_softirq(void);
2692 
2693 #define cond_resched_softirq() ({					\
2694 	__might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET);	\
2695 	__cond_resched_softirq();					\
2696 })
2697 
2698 static inline void cond_resched_rcu(void)
2699 {
2700 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2701 	rcu_read_unlock();
2702 	cond_resched();
2703 	rcu_read_lock();
2704 #endif
2705 }
2706 
2707 /*
2708  * Does a critical section need to be broken due to another
2709  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2710  * but a general need for low latency)
2711  */
2712 static inline int spin_needbreak(spinlock_t *lock)
2713 {
2714 #ifdef CONFIG_PREEMPT
2715 	return spin_is_contended(lock);
2716 #else
2717 	return 0;
2718 #endif
2719 }
2720 
2721 /*
2722  * Idle thread specific functions to determine the need_resched
2723  * polling state. We have two versions, one based on TS_POLLING in
2724  * thread_info.status and one based on TIF_POLLING_NRFLAG in
2725  * thread_info.flags
2726  */
2727 #ifdef TS_POLLING
2728 static inline int tsk_is_polling(struct task_struct *p)
2729 {
2730 	return task_thread_info(p)->status & TS_POLLING;
2731 }
2732 static inline void __current_set_polling(void)
2733 {
2734 	current_thread_info()->status |= TS_POLLING;
2735 }
2736 
2737 static inline bool __must_check current_set_polling_and_test(void)
2738 {
2739 	__current_set_polling();
2740 
2741 	/*
2742 	 * Polling state must be visible before we test NEED_RESCHED,
2743 	 * paired by resched_task()
2744 	 */
2745 	smp_mb();
2746 
2747 	return unlikely(tif_need_resched());
2748 }
2749 
2750 static inline void __current_clr_polling(void)
2751 {
2752 	current_thread_info()->status &= ~TS_POLLING;
2753 }
2754 
2755 static inline bool __must_check current_clr_polling_and_test(void)
2756 {
2757 	__current_clr_polling();
2758 
2759 	/*
2760 	 * Polling state must be visible before we test NEED_RESCHED,
2761 	 * paired by resched_task()
2762 	 */
2763 	smp_mb();
2764 
2765 	return unlikely(tif_need_resched());
2766 }
2767 #elif defined(TIF_POLLING_NRFLAG)
2768 static inline int tsk_is_polling(struct task_struct *p)
2769 {
2770 	return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2771 }
2772 
2773 static inline void __current_set_polling(void)
2774 {
2775 	set_thread_flag(TIF_POLLING_NRFLAG);
2776 }
2777 
2778 static inline bool __must_check current_set_polling_and_test(void)
2779 {
2780 	__current_set_polling();
2781 
2782 	/*
2783 	 * Polling state must be visible before we test NEED_RESCHED,
2784 	 * paired by resched_task()
2785 	 *
2786 	 * XXX: assumes set/clear bit are identical barrier wise.
2787 	 */
2788 	smp_mb__after_clear_bit();
2789 
2790 	return unlikely(tif_need_resched());
2791 }
2792 
2793 static inline void __current_clr_polling(void)
2794 {
2795 	clear_thread_flag(TIF_POLLING_NRFLAG);
2796 }
2797 
2798 static inline bool __must_check current_clr_polling_and_test(void)
2799 {
2800 	__current_clr_polling();
2801 
2802 	/*
2803 	 * Polling state must be visible before we test NEED_RESCHED,
2804 	 * paired by resched_task()
2805 	 */
2806 	smp_mb__after_clear_bit();
2807 
2808 	return unlikely(tif_need_resched());
2809 }
2810 
2811 #else
2812 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2813 static inline void __current_set_polling(void) { }
2814 static inline void __current_clr_polling(void) { }
2815 
2816 static inline bool __must_check current_set_polling_and_test(void)
2817 {
2818 	return unlikely(tif_need_resched());
2819 }
2820 static inline bool __must_check current_clr_polling_and_test(void)
2821 {
2822 	return unlikely(tif_need_resched());
2823 }
2824 #endif
2825 
2826 static inline void current_clr_polling(void)
2827 {
2828 	__current_clr_polling();
2829 
2830 	/*
2831 	 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2832 	 * Once the bit is cleared, we'll get IPIs with every new
2833 	 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2834 	 * fold.
2835 	 */
2836 	smp_mb(); /* paired with resched_task() */
2837 
2838 	preempt_fold_need_resched();
2839 }
2840 
2841 static __always_inline bool need_resched(void)
2842 {
2843 	return unlikely(tif_need_resched());
2844 }
2845 
2846 /*
2847  * Thread group CPU time accounting.
2848  */
2849 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2850 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2851 
2852 static inline void thread_group_cputime_init(struct signal_struct *sig)
2853 {
2854 	raw_spin_lock_init(&sig->cputimer.lock);
2855 }
2856 
2857 /*
2858  * Reevaluate whether the task has signals pending delivery.
2859  * Wake the task if so.
2860  * This is required every time the blocked sigset_t changes.
2861  * callers must hold sighand->siglock.
2862  */
2863 extern void recalc_sigpending_and_wake(struct task_struct *t);
2864 extern void recalc_sigpending(void);
2865 
2866 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2867 
2868 static inline void signal_wake_up(struct task_struct *t, bool resume)
2869 {
2870 	signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2871 }
2872 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2873 {
2874 	signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2875 }
2876 
2877 /*
2878  * Wrappers for p->thread_info->cpu access. No-op on UP.
2879  */
2880 #ifdef CONFIG_SMP
2881 
2882 static inline unsigned int task_cpu(const struct task_struct *p)
2883 {
2884 	return task_thread_info(p)->cpu;
2885 }
2886 
2887 static inline int task_node(const struct task_struct *p)
2888 {
2889 	return cpu_to_node(task_cpu(p));
2890 }
2891 
2892 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2893 
2894 #else
2895 
2896 static inline unsigned int task_cpu(const struct task_struct *p)
2897 {
2898 	return 0;
2899 }
2900 
2901 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2902 {
2903 }
2904 
2905 #endif /* CONFIG_SMP */
2906 
2907 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2908 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2909 
2910 #ifdef CONFIG_CGROUP_SCHED
2911 extern struct task_group root_task_group;
2912 #endif /* CONFIG_CGROUP_SCHED */
2913 
2914 extern int task_can_switch_user(struct user_struct *up,
2915 					struct task_struct *tsk);
2916 
2917 #ifdef CONFIG_TASK_XACCT
2918 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2919 {
2920 	tsk->ioac.rchar += amt;
2921 }
2922 
2923 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2924 {
2925 	tsk->ioac.wchar += amt;
2926 }
2927 
2928 static inline void inc_syscr(struct task_struct *tsk)
2929 {
2930 	tsk->ioac.syscr++;
2931 }
2932 
2933 static inline void inc_syscw(struct task_struct *tsk)
2934 {
2935 	tsk->ioac.syscw++;
2936 }
2937 #else
2938 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2939 {
2940 }
2941 
2942 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2943 {
2944 }
2945 
2946 static inline void inc_syscr(struct task_struct *tsk)
2947 {
2948 }
2949 
2950 static inline void inc_syscw(struct task_struct *tsk)
2951 {
2952 }
2953 #endif
2954 
2955 #ifndef TASK_SIZE_OF
2956 #define TASK_SIZE_OF(tsk)	TASK_SIZE
2957 #endif
2958 
2959 #ifdef CONFIG_MM_OWNER
2960 extern void mm_update_next_owner(struct mm_struct *mm);
2961 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2962 #else
2963 static inline void mm_update_next_owner(struct mm_struct *mm)
2964 {
2965 }
2966 
2967 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2968 {
2969 }
2970 #endif /* CONFIG_MM_OWNER */
2971 
2972 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2973 		unsigned int limit)
2974 {
2975 	return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2976 }
2977 
2978 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2979 		unsigned int limit)
2980 {
2981 	return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2982 }
2983 
2984 static inline unsigned long rlimit(unsigned int limit)
2985 {
2986 	return task_rlimit(current, limit);
2987 }
2988 
2989 static inline unsigned long rlimit_max(unsigned int limit)
2990 {
2991 	return task_rlimit_max(current, limit);
2992 }
2993 
2994 #endif
2995