xref: /linux-6.15/include/linux/sched.h (revision 84ed8a99)
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3 
4 #include <uapi/linux/sched.h>
5 
6 
7 struct sched_param {
8 	int sched_priority;
9 };
10 
11 #include <asm/param.h>	/* for HZ */
12 
13 #include <linux/capability.h>
14 #include <linux/threads.h>
15 #include <linux/kernel.h>
16 #include <linux/types.h>
17 #include <linux/timex.h>
18 #include <linux/jiffies.h>
19 #include <linux/rbtree.h>
20 #include <linux/thread_info.h>
21 #include <linux/cpumask.h>
22 #include <linux/errno.h>
23 #include <linux/nodemask.h>
24 #include <linux/mm_types.h>
25 #include <linux/preempt_mask.h>
26 
27 #include <asm/page.h>
28 #include <asm/ptrace.h>
29 #include <asm/cputime.h>
30 
31 #include <linux/smp.h>
32 #include <linux/sem.h>
33 #include <linux/signal.h>
34 #include <linux/compiler.h>
35 #include <linux/completion.h>
36 #include <linux/pid.h>
37 #include <linux/percpu.h>
38 #include <linux/topology.h>
39 #include <linux/proportions.h>
40 #include <linux/seccomp.h>
41 #include <linux/rcupdate.h>
42 #include <linux/rculist.h>
43 #include <linux/rtmutex.h>
44 
45 #include <linux/time.h>
46 #include <linux/param.h>
47 #include <linux/resource.h>
48 #include <linux/timer.h>
49 #include <linux/hrtimer.h>
50 #include <linux/task_io_accounting.h>
51 #include <linux/latencytop.h>
52 #include <linux/cred.h>
53 #include <linux/llist.h>
54 #include <linux/uidgid.h>
55 #include <linux/gfp.h>
56 
57 #include <asm/processor.h>
58 
59 struct exec_domain;
60 struct futex_pi_state;
61 struct robust_list_head;
62 struct bio_list;
63 struct fs_struct;
64 struct perf_event_context;
65 struct blk_plug;
66 
67 /*
68  * List of flags we want to share for kernel threads,
69  * if only because they are not used by them anyway.
70  */
71 #define CLONE_KERNEL	(CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
72 
73 /*
74  * These are the constant used to fake the fixed-point load-average
75  * counting. Some notes:
76  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
77  *    a load-average precision of 10 bits integer + 11 bits fractional
78  *  - if you want to count load-averages more often, you need more
79  *    precision, or rounding will get you. With 2-second counting freq,
80  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
81  *    11 bit fractions.
82  */
83 extern unsigned long avenrun[];		/* Load averages */
84 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
85 
86 #define FSHIFT		11		/* nr of bits of precision */
87 #define FIXED_1		(1<<FSHIFT)	/* 1.0 as fixed-point */
88 #define LOAD_FREQ	(5*HZ+1)	/* 5 sec intervals */
89 #define EXP_1		1884		/* 1/exp(5sec/1min) as fixed-point */
90 #define EXP_5		2014		/* 1/exp(5sec/5min) */
91 #define EXP_15		2037		/* 1/exp(5sec/15min) */
92 
93 #define CALC_LOAD(load,exp,n) \
94 	load *= exp; \
95 	load += n*(FIXED_1-exp); \
96 	load >>= FSHIFT;
97 
98 extern unsigned long total_forks;
99 extern int nr_threads;
100 DECLARE_PER_CPU(unsigned long, process_counts);
101 extern int nr_processes(void);
102 extern unsigned long nr_running(void);
103 extern unsigned long nr_iowait(void);
104 extern unsigned long nr_iowait_cpu(int cpu);
105 extern unsigned long this_cpu_load(void);
106 
107 
108 extern void calc_global_load(unsigned long ticks);
109 extern void update_cpu_load_nohz(void);
110 
111 extern unsigned long get_parent_ip(unsigned long addr);
112 
113 extern void dump_cpu_task(int cpu);
114 
115 struct seq_file;
116 struct cfs_rq;
117 struct task_group;
118 #ifdef CONFIG_SCHED_DEBUG
119 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
120 extern void proc_sched_set_task(struct task_struct *p);
121 extern void
122 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
123 #endif
124 
125 /*
126  * Task state bitmask. NOTE! These bits are also
127  * encoded in fs/proc/array.c: get_task_state().
128  *
129  * We have two separate sets of flags: task->state
130  * is about runnability, while task->exit_state are
131  * about the task exiting. Confusing, but this way
132  * modifying one set can't modify the other one by
133  * mistake.
134  */
135 #define TASK_RUNNING		0
136 #define TASK_INTERRUPTIBLE	1
137 #define TASK_UNINTERRUPTIBLE	2
138 #define __TASK_STOPPED		4
139 #define __TASK_TRACED		8
140 /* in tsk->exit_state */
141 #define EXIT_ZOMBIE		16
142 #define EXIT_DEAD		32
143 /* in tsk->state again */
144 #define TASK_DEAD		64
145 #define TASK_WAKEKILL		128
146 #define TASK_WAKING		256
147 #define TASK_PARKED		512
148 #define TASK_STATE_MAX		1024
149 
150 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
151 
152 extern char ___assert_task_state[1 - 2*!!(
153 		sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
154 
155 /* Convenience macros for the sake of set_task_state */
156 #define TASK_KILLABLE		(TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
157 #define TASK_STOPPED		(TASK_WAKEKILL | __TASK_STOPPED)
158 #define TASK_TRACED		(TASK_WAKEKILL | __TASK_TRACED)
159 
160 /* Convenience macros for the sake of wake_up */
161 #define TASK_NORMAL		(TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
162 #define TASK_ALL		(TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
163 
164 /* get_task_state() */
165 #define TASK_REPORT		(TASK_RUNNING | TASK_INTERRUPTIBLE | \
166 				 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
167 				 __TASK_TRACED)
168 
169 #define task_is_traced(task)	((task->state & __TASK_TRACED) != 0)
170 #define task_is_stopped(task)	((task->state & __TASK_STOPPED) != 0)
171 #define task_is_dead(task)	((task)->exit_state != 0)
172 #define task_is_stopped_or_traced(task)	\
173 			((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
174 #define task_contributes_to_load(task)	\
175 				((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
176 				 (task->flags & PF_FROZEN) == 0)
177 
178 #define __set_task_state(tsk, state_value)		\
179 	do { (tsk)->state = (state_value); } while (0)
180 #define set_task_state(tsk, state_value)		\
181 	set_mb((tsk)->state, (state_value))
182 
183 /*
184  * set_current_state() includes a barrier so that the write of current->state
185  * is correctly serialised wrt the caller's subsequent test of whether to
186  * actually sleep:
187  *
188  *	set_current_state(TASK_UNINTERRUPTIBLE);
189  *	if (do_i_need_to_sleep())
190  *		schedule();
191  *
192  * If the caller does not need such serialisation then use __set_current_state()
193  */
194 #define __set_current_state(state_value)			\
195 	do { current->state = (state_value); } while (0)
196 #define set_current_state(state_value)		\
197 	set_mb(current->state, (state_value))
198 
199 /* Task command name length */
200 #define TASK_COMM_LEN 16
201 
202 #include <linux/spinlock.h>
203 
204 /*
205  * This serializes "schedule()" and also protects
206  * the run-queue from deletions/modifications (but
207  * _adding_ to the beginning of the run-queue has
208  * a separate lock).
209  */
210 extern rwlock_t tasklist_lock;
211 extern spinlock_t mmlist_lock;
212 
213 struct task_struct;
214 
215 #ifdef CONFIG_PROVE_RCU
216 extern int lockdep_tasklist_lock_is_held(void);
217 #endif /* #ifdef CONFIG_PROVE_RCU */
218 
219 extern void sched_init(void);
220 extern void sched_init_smp(void);
221 extern asmlinkage void schedule_tail(struct task_struct *prev);
222 extern void init_idle(struct task_struct *idle, int cpu);
223 extern void init_idle_bootup_task(struct task_struct *idle);
224 
225 extern int runqueue_is_locked(int cpu);
226 
227 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
228 extern void nohz_balance_enter_idle(int cpu);
229 extern void set_cpu_sd_state_idle(void);
230 extern int get_nohz_timer_target(void);
231 #else
232 static inline void nohz_balance_enter_idle(int cpu) { }
233 static inline void set_cpu_sd_state_idle(void) { }
234 #endif
235 
236 /*
237  * Only dump TASK_* tasks. (0 for all tasks)
238  */
239 extern void show_state_filter(unsigned long state_filter);
240 
241 static inline void show_state(void)
242 {
243 	show_state_filter(0);
244 }
245 
246 extern void show_regs(struct pt_regs *);
247 
248 /*
249  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
250  * task), SP is the stack pointer of the first frame that should be shown in the back
251  * trace (or NULL if the entire call-chain of the task should be shown).
252  */
253 extern void show_stack(struct task_struct *task, unsigned long *sp);
254 
255 void io_schedule(void);
256 long io_schedule_timeout(long timeout);
257 
258 extern void cpu_init (void);
259 extern void trap_init(void);
260 extern void update_process_times(int user);
261 extern void scheduler_tick(void);
262 
263 extern void sched_show_task(struct task_struct *p);
264 
265 #ifdef CONFIG_LOCKUP_DETECTOR
266 extern void touch_softlockup_watchdog(void);
267 extern void touch_softlockup_watchdog_sync(void);
268 extern void touch_all_softlockup_watchdogs(void);
269 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
270 				  void __user *buffer,
271 				  size_t *lenp, loff_t *ppos);
272 extern unsigned int  softlockup_panic;
273 void lockup_detector_init(void);
274 #else
275 static inline void touch_softlockup_watchdog(void)
276 {
277 }
278 static inline void touch_softlockup_watchdog_sync(void)
279 {
280 }
281 static inline void touch_all_softlockup_watchdogs(void)
282 {
283 }
284 static inline void lockup_detector_init(void)
285 {
286 }
287 #endif
288 
289 #ifdef CONFIG_DETECT_HUNG_TASK
290 void reset_hung_task_detector(void);
291 #else
292 static inline void reset_hung_task_detector(void)
293 {
294 }
295 #endif
296 
297 /* Attach to any functions which should be ignored in wchan output. */
298 #define __sched		__attribute__((__section__(".sched.text")))
299 
300 /* Linker adds these: start and end of __sched functions */
301 extern char __sched_text_start[], __sched_text_end[];
302 
303 /* Is this address in the __sched functions? */
304 extern int in_sched_functions(unsigned long addr);
305 
306 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
307 extern signed long schedule_timeout(signed long timeout);
308 extern signed long schedule_timeout_interruptible(signed long timeout);
309 extern signed long schedule_timeout_killable(signed long timeout);
310 extern signed long schedule_timeout_uninterruptible(signed long timeout);
311 asmlinkage void schedule(void);
312 extern void schedule_preempt_disabled(void);
313 
314 struct nsproxy;
315 struct user_namespace;
316 
317 #ifdef CONFIG_MMU
318 extern void arch_pick_mmap_layout(struct mm_struct *mm);
319 extern unsigned long
320 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
321 		       unsigned long, unsigned long);
322 extern unsigned long
323 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
324 			  unsigned long len, unsigned long pgoff,
325 			  unsigned long flags);
326 #else
327 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
328 #endif
329 
330 
331 extern void set_dumpable(struct mm_struct *mm, int value);
332 extern int get_dumpable(struct mm_struct *mm);
333 
334 #define SUID_DUMP_DISABLE	0	/* No setuid dumping */
335 #define SUID_DUMP_USER		1	/* Dump as user of process */
336 #define SUID_DUMP_ROOT		2	/* Dump as root */
337 
338 /* mm flags */
339 /* dumpable bits */
340 #define MMF_DUMPABLE      0  /* core dump is permitted */
341 #define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
342 
343 #define MMF_DUMPABLE_BITS 2
344 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
345 
346 /* coredump filter bits */
347 #define MMF_DUMP_ANON_PRIVATE	2
348 #define MMF_DUMP_ANON_SHARED	3
349 #define MMF_DUMP_MAPPED_PRIVATE	4
350 #define MMF_DUMP_MAPPED_SHARED	5
351 #define MMF_DUMP_ELF_HEADERS	6
352 #define MMF_DUMP_HUGETLB_PRIVATE 7
353 #define MMF_DUMP_HUGETLB_SHARED  8
354 
355 #define MMF_DUMP_FILTER_SHIFT	MMF_DUMPABLE_BITS
356 #define MMF_DUMP_FILTER_BITS	7
357 #define MMF_DUMP_FILTER_MASK \
358 	(((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
359 #define MMF_DUMP_FILTER_DEFAULT \
360 	((1 << MMF_DUMP_ANON_PRIVATE) |	(1 << MMF_DUMP_ANON_SHARED) |\
361 	 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
362 
363 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
364 # define MMF_DUMP_MASK_DEFAULT_ELF	(1 << MMF_DUMP_ELF_HEADERS)
365 #else
366 # define MMF_DUMP_MASK_DEFAULT_ELF	0
367 #endif
368 					/* leave room for more dump flags */
369 #define MMF_VM_MERGEABLE	16	/* KSM may merge identical pages */
370 #define MMF_VM_HUGEPAGE		17	/* set when VM_HUGEPAGE is set on vma */
371 #define MMF_EXE_FILE_CHANGED	18	/* see prctl_set_mm_exe_file() */
372 
373 #define MMF_HAS_UPROBES		19	/* has uprobes */
374 #define MMF_RECALC_UPROBES	20	/* MMF_HAS_UPROBES can be wrong */
375 
376 #define MMF_INIT_MASK		(MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
377 
378 struct sighand_struct {
379 	atomic_t		count;
380 	struct k_sigaction	action[_NSIG];
381 	spinlock_t		siglock;
382 	wait_queue_head_t	signalfd_wqh;
383 };
384 
385 struct pacct_struct {
386 	int			ac_flag;
387 	long			ac_exitcode;
388 	unsigned long		ac_mem;
389 	cputime_t		ac_utime, ac_stime;
390 	unsigned long		ac_minflt, ac_majflt;
391 };
392 
393 struct cpu_itimer {
394 	cputime_t expires;
395 	cputime_t incr;
396 	u32 error;
397 	u32 incr_error;
398 };
399 
400 /**
401  * struct cputime - snaphsot of system and user cputime
402  * @utime: time spent in user mode
403  * @stime: time spent in system mode
404  *
405  * Gathers a generic snapshot of user and system time.
406  */
407 struct cputime {
408 	cputime_t utime;
409 	cputime_t stime;
410 };
411 
412 /**
413  * struct task_cputime - collected CPU time counts
414  * @utime:		time spent in user mode, in &cputime_t units
415  * @stime:		time spent in kernel mode, in &cputime_t units
416  * @sum_exec_runtime:	total time spent on the CPU, in nanoseconds
417  *
418  * This is an extension of struct cputime that includes the total runtime
419  * spent by the task from the scheduler point of view.
420  *
421  * As a result, this structure groups together three kinds of CPU time
422  * that are tracked for threads and thread groups.  Most things considering
423  * CPU time want to group these counts together and treat all three
424  * of them in parallel.
425  */
426 struct task_cputime {
427 	cputime_t utime;
428 	cputime_t stime;
429 	unsigned long long sum_exec_runtime;
430 };
431 /* Alternate field names when used to cache expirations. */
432 #define prof_exp	stime
433 #define virt_exp	utime
434 #define sched_exp	sum_exec_runtime
435 
436 #define INIT_CPUTIME	\
437 	(struct task_cputime) {					\
438 		.utime = 0,					\
439 		.stime = 0,					\
440 		.sum_exec_runtime = 0,				\
441 	}
442 
443 #ifdef CONFIG_PREEMPT_COUNT
444 #define PREEMPT_DISABLED	(1 + PREEMPT_ENABLED)
445 #else
446 #define PREEMPT_DISABLED	PREEMPT_ENABLED
447 #endif
448 
449 /*
450  * Disable preemption until the scheduler is running.
451  * Reset by start_kernel()->sched_init()->init_idle().
452  *
453  * We include PREEMPT_ACTIVE to avoid cond_resched() from working
454  * before the scheduler is active -- see should_resched().
455  */
456 #define INIT_PREEMPT_COUNT	(PREEMPT_DISABLED + PREEMPT_ACTIVE)
457 
458 /**
459  * struct thread_group_cputimer - thread group interval timer counts
460  * @cputime:		thread group interval timers.
461  * @running:		non-zero when there are timers running and
462  * 			@cputime receives updates.
463  * @lock:		lock for fields in this struct.
464  *
465  * This structure contains the version of task_cputime, above, that is
466  * used for thread group CPU timer calculations.
467  */
468 struct thread_group_cputimer {
469 	struct task_cputime cputime;
470 	int running;
471 	raw_spinlock_t lock;
472 };
473 
474 #include <linux/rwsem.h>
475 struct autogroup;
476 
477 /*
478  * NOTE! "signal_struct" does not have its own
479  * locking, because a shared signal_struct always
480  * implies a shared sighand_struct, so locking
481  * sighand_struct is always a proper superset of
482  * the locking of signal_struct.
483  */
484 struct signal_struct {
485 	atomic_t		sigcnt;
486 	atomic_t		live;
487 	int			nr_threads;
488 
489 	wait_queue_head_t	wait_chldexit;	/* for wait4() */
490 
491 	/* current thread group signal load-balancing target: */
492 	struct task_struct	*curr_target;
493 
494 	/* shared signal handling: */
495 	struct sigpending	shared_pending;
496 
497 	/* thread group exit support */
498 	int			group_exit_code;
499 	/* overloaded:
500 	 * - notify group_exit_task when ->count is equal to notify_count
501 	 * - everyone except group_exit_task is stopped during signal delivery
502 	 *   of fatal signals, group_exit_task processes the signal.
503 	 */
504 	int			notify_count;
505 	struct task_struct	*group_exit_task;
506 
507 	/* thread group stop support, overloads group_exit_code too */
508 	int			group_stop_count;
509 	unsigned int		flags; /* see SIGNAL_* flags below */
510 
511 	/*
512 	 * PR_SET_CHILD_SUBREAPER marks a process, like a service
513 	 * manager, to re-parent orphan (double-forking) child processes
514 	 * to this process instead of 'init'. The service manager is
515 	 * able to receive SIGCHLD signals and is able to investigate
516 	 * the process until it calls wait(). All children of this
517 	 * process will inherit a flag if they should look for a
518 	 * child_subreaper process at exit.
519 	 */
520 	unsigned int		is_child_subreaper:1;
521 	unsigned int		has_child_subreaper:1;
522 
523 	/* POSIX.1b Interval Timers */
524 	int			posix_timer_id;
525 	struct list_head	posix_timers;
526 
527 	/* ITIMER_REAL timer for the process */
528 	struct hrtimer real_timer;
529 	struct pid *leader_pid;
530 	ktime_t it_real_incr;
531 
532 	/*
533 	 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
534 	 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
535 	 * values are defined to 0 and 1 respectively
536 	 */
537 	struct cpu_itimer it[2];
538 
539 	/*
540 	 * Thread group totals for process CPU timers.
541 	 * See thread_group_cputimer(), et al, for details.
542 	 */
543 	struct thread_group_cputimer cputimer;
544 
545 	/* Earliest-expiration cache. */
546 	struct task_cputime cputime_expires;
547 
548 	struct list_head cpu_timers[3];
549 
550 	struct pid *tty_old_pgrp;
551 
552 	/* boolean value for session group leader */
553 	int leader;
554 
555 	struct tty_struct *tty; /* NULL if no tty */
556 
557 #ifdef CONFIG_SCHED_AUTOGROUP
558 	struct autogroup *autogroup;
559 #endif
560 	/*
561 	 * Cumulative resource counters for dead threads in the group,
562 	 * and for reaped dead child processes forked by this group.
563 	 * Live threads maintain their own counters and add to these
564 	 * in __exit_signal, except for the group leader.
565 	 */
566 	cputime_t utime, stime, cutime, cstime;
567 	cputime_t gtime;
568 	cputime_t cgtime;
569 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
570 	struct cputime prev_cputime;
571 #endif
572 	unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
573 	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
574 	unsigned long inblock, oublock, cinblock, coublock;
575 	unsigned long maxrss, cmaxrss;
576 	struct task_io_accounting ioac;
577 
578 	/*
579 	 * Cumulative ns of schedule CPU time fo dead threads in the
580 	 * group, not including a zombie group leader, (This only differs
581 	 * from jiffies_to_ns(utime + stime) if sched_clock uses something
582 	 * other than jiffies.)
583 	 */
584 	unsigned long long sum_sched_runtime;
585 
586 	/*
587 	 * We don't bother to synchronize most readers of this at all,
588 	 * because there is no reader checking a limit that actually needs
589 	 * to get both rlim_cur and rlim_max atomically, and either one
590 	 * alone is a single word that can safely be read normally.
591 	 * getrlimit/setrlimit use task_lock(current->group_leader) to
592 	 * protect this instead of the siglock, because they really
593 	 * have no need to disable irqs.
594 	 */
595 	struct rlimit rlim[RLIM_NLIMITS];
596 
597 #ifdef CONFIG_BSD_PROCESS_ACCT
598 	struct pacct_struct pacct;	/* per-process accounting information */
599 #endif
600 #ifdef CONFIG_TASKSTATS
601 	struct taskstats *stats;
602 #endif
603 #ifdef CONFIG_AUDIT
604 	unsigned audit_tty;
605 	unsigned audit_tty_log_passwd;
606 	struct tty_audit_buf *tty_audit_buf;
607 #endif
608 #ifdef CONFIG_CGROUPS
609 	/*
610 	 * group_rwsem prevents new tasks from entering the threadgroup and
611 	 * member tasks from exiting,a more specifically, setting of
612 	 * PF_EXITING.  fork and exit paths are protected with this rwsem
613 	 * using threadgroup_change_begin/end().  Users which require
614 	 * threadgroup to remain stable should use threadgroup_[un]lock()
615 	 * which also takes care of exec path.  Currently, cgroup is the
616 	 * only user.
617 	 */
618 	struct rw_semaphore group_rwsem;
619 #endif
620 
621 	oom_flags_t oom_flags;
622 	short oom_score_adj;		/* OOM kill score adjustment */
623 	short oom_score_adj_min;	/* OOM kill score adjustment min value.
624 					 * Only settable by CAP_SYS_RESOURCE. */
625 
626 	struct mutex cred_guard_mutex;	/* guard against foreign influences on
627 					 * credential calculations
628 					 * (notably. ptrace) */
629 };
630 
631 /*
632  * Bits in flags field of signal_struct.
633  */
634 #define SIGNAL_STOP_STOPPED	0x00000001 /* job control stop in effect */
635 #define SIGNAL_STOP_CONTINUED	0x00000002 /* SIGCONT since WCONTINUED reap */
636 #define SIGNAL_GROUP_EXIT	0x00000004 /* group exit in progress */
637 #define SIGNAL_GROUP_COREDUMP	0x00000008 /* coredump in progress */
638 /*
639  * Pending notifications to parent.
640  */
641 #define SIGNAL_CLD_STOPPED	0x00000010
642 #define SIGNAL_CLD_CONTINUED	0x00000020
643 #define SIGNAL_CLD_MASK		(SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
644 
645 #define SIGNAL_UNKILLABLE	0x00000040 /* for init: ignore fatal signals */
646 
647 /* If true, all threads except ->group_exit_task have pending SIGKILL */
648 static inline int signal_group_exit(const struct signal_struct *sig)
649 {
650 	return	(sig->flags & SIGNAL_GROUP_EXIT) ||
651 		(sig->group_exit_task != NULL);
652 }
653 
654 /*
655  * Some day this will be a full-fledged user tracking system..
656  */
657 struct user_struct {
658 	atomic_t __count;	/* reference count */
659 	atomic_t processes;	/* How many processes does this user have? */
660 	atomic_t files;		/* How many open files does this user have? */
661 	atomic_t sigpending;	/* How many pending signals does this user have? */
662 #ifdef CONFIG_INOTIFY_USER
663 	atomic_t inotify_watches; /* How many inotify watches does this user have? */
664 	atomic_t inotify_devs;	/* How many inotify devs does this user have opened? */
665 #endif
666 #ifdef CONFIG_FANOTIFY
667 	atomic_t fanotify_listeners;
668 #endif
669 #ifdef CONFIG_EPOLL
670 	atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
671 #endif
672 #ifdef CONFIG_POSIX_MQUEUE
673 	/* protected by mq_lock	*/
674 	unsigned long mq_bytes;	/* How many bytes can be allocated to mqueue? */
675 #endif
676 	unsigned long locked_shm; /* How many pages of mlocked shm ? */
677 
678 #ifdef CONFIG_KEYS
679 	struct key *uid_keyring;	/* UID specific keyring */
680 	struct key *session_keyring;	/* UID's default session keyring */
681 #endif
682 
683 	/* Hash table maintenance information */
684 	struct hlist_node uidhash_node;
685 	kuid_t uid;
686 
687 #ifdef CONFIG_PERF_EVENTS
688 	atomic_long_t locked_vm;
689 #endif
690 };
691 
692 extern int uids_sysfs_init(void);
693 
694 extern struct user_struct *find_user(kuid_t);
695 
696 extern struct user_struct root_user;
697 #define INIT_USER (&root_user)
698 
699 
700 struct backing_dev_info;
701 struct reclaim_state;
702 
703 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
704 struct sched_info {
705 	/* cumulative counters */
706 	unsigned long pcount;	      /* # of times run on this cpu */
707 	unsigned long long run_delay; /* time spent waiting on a runqueue */
708 
709 	/* timestamps */
710 	unsigned long long last_arrival,/* when we last ran on a cpu */
711 			   last_queued;	/* when we were last queued to run */
712 };
713 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
714 
715 #ifdef CONFIG_TASK_DELAY_ACCT
716 struct task_delay_info {
717 	spinlock_t	lock;
718 	unsigned int	flags;	/* Private per-task flags */
719 
720 	/* For each stat XXX, add following, aligned appropriately
721 	 *
722 	 * struct timespec XXX_start, XXX_end;
723 	 * u64 XXX_delay;
724 	 * u32 XXX_count;
725 	 *
726 	 * Atomicity of updates to XXX_delay, XXX_count protected by
727 	 * single lock above (split into XXX_lock if contention is an issue).
728 	 */
729 
730 	/*
731 	 * XXX_count is incremented on every XXX operation, the delay
732 	 * associated with the operation is added to XXX_delay.
733 	 * XXX_delay contains the accumulated delay time in nanoseconds.
734 	 */
735 	struct timespec blkio_start, blkio_end;	/* Shared by blkio, swapin */
736 	u64 blkio_delay;	/* wait for sync block io completion */
737 	u64 swapin_delay;	/* wait for swapin block io completion */
738 	u32 blkio_count;	/* total count of the number of sync block */
739 				/* io operations performed */
740 	u32 swapin_count;	/* total count of the number of swapin block */
741 				/* io operations performed */
742 
743 	struct timespec freepages_start, freepages_end;
744 	u64 freepages_delay;	/* wait for memory reclaim */
745 	u32 freepages_count;	/* total count of memory reclaim */
746 };
747 #endif	/* CONFIG_TASK_DELAY_ACCT */
748 
749 static inline int sched_info_on(void)
750 {
751 #ifdef CONFIG_SCHEDSTATS
752 	return 1;
753 #elif defined(CONFIG_TASK_DELAY_ACCT)
754 	extern int delayacct_on;
755 	return delayacct_on;
756 #else
757 	return 0;
758 #endif
759 }
760 
761 enum cpu_idle_type {
762 	CPU_IDLE,
763 	CPU_NOT_IDLE,
764 	CPU_NEWLY_IDLE,
765 	CPU_MAX_IDLE_TYPES
766 };
767 
768 /*
769  * Increase resolution of cpu_power calculations
770  */
771 #define SCHED_POWER_SHIFT	10
772 #define SCHED_POWER_SCALE	(1L << SCHED_POWER_SHIFT)
773 
774 /*
775  * sched-domains (multiprocessor balancing) declarations:
776  */
777 #ifdef CONFIG_SMP
778 #define SD_LOAD_BALANCE		0x0001	/* Do load balancing on this domain. */
779 #define SD_BALANCE_NEWIDLE	0x0002	/* Balance when about to become idle */
780 #define SD_BALANCE_EXEC		0x0004	/* Balance on exec */
781 #define SD_BALANCE_FORK		0x0008	/* Balance on fork, clone */
782 #define SD_BALANCE_WAKE		0x0010  /* Balance on wakeup */
783 #define SD_WAKE_AFFINE		0x0020	/* Wake task to waking CPU */
784 #define SD_SHARE_CPUPOWER	0x0080	/* Domain members share cpu power */
785 #define SD_SHARE_PKG_RESOURCES	0x0200	/* Domain members share cpu pkg resources */
786 #define SD_SERIALIZE		0x0400	/* Only a single load balancing instance */
787 #define SD_ASYM_PACKING		0x0800  /* Place busy groups earlier in the domain */
788 #define SD_PREFER_SIBLING	0x1000	/* Prefer to place tasks in a sibling domain */
789 #define SD_OVERLAP		0x2000	/* sched_domains of this level overlap */
790 #define SD_NUMA			0x4000	/* cross-node balancing */
791 
792 extern int __weak arch_sd_sibiling_asym_packing(void);
793 
794 struct sched_domain_attr {
795 	int relax_domain_level;
796 };
797 
798 #define SD_ATTR_INIT	(struct sched_domain_attr) {	\
799 	.relax_domain_level = -1,			\
800 }
801 
802 extern int sched_domain_level_max;
803 
804 struct sched_group;
805 
806 struct sched_domain {
807 	/* These fields must be setup */
808 	struct sched_domain *parent;	/* top domain must be null terminated */
809 	struct sched_domain *child;	/* bottom domain must be null terminated */
810 	struct sched_group *groups;	/* the balancing groups of the domain */
811 	unsigned long min_interval;	/* Minimum balance interval ms */
812 	unsigned long max_interval;	/* Maximum balance interval ms */
813 	unsigned int busy_factor;	/* less balancing by factor if busy */
814 	unsigned int imbalance_pct;	/* No balance until over watermark */
815 	unsigned int cache_nice_tries;	/* Leave cache hot tasks for # tries */
816 	unsigned int busy_idx;
817 	unsigned int idle_idx;
818 	unsigned int newidle_idx;
819 	unsigned int wake_idx;
820 	unsigned int forkexec_idx;
821 	unsigned int smt_gain;
822 
823 	int nohz_idle;			/* NOHZ IDLE status */
824 	int flags;			/* See SD_* */
825 	int level;
826 
827 	/* Runtime fields. */
828 	unsigned long last_balance;	/* init to jiffies. units in jiffies */
829 	unsigned int balance_interval;	/* initialise to 1. units in ms. */
830 	unsigned int nr_balance_failed; /* initialise to 0 */
831 
832 	/* idle_balance() stats */
833 	u64 max_newidle_lb_cost;
834 	unsigned long next_decay_max_lb_cost;
835 
836 #ifdef CONFIG_SCHEDSTATS
837 	/* load_balance() stats */
838 	unsigned int lb_count[CPU_MAX_IDLE_TYPES];
839 	unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
840 	unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
841 	unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
842 	unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
843 	unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
844 	unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
845 	unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
846 
847 	/* Active load balancing */
848 	unsigned int alb_count;
849 	unsigned int alb_failed;
850 	unsigned int alb_pushed;
851 
852 	/* SD_BALANCE_EXEC stats */
853 	unsigned int sbe_count;
854 	unsigned int sbe_balanced;
855 	unsigned int sbe_pushed;
856 
857 	/* SD_BALANCE_FORK stats */
858 	unsigned int sbf_count;
859 	unsigned int sbf_balanced;
860 	unsigned int sbf_pushed;
861 
862 	/* try_to_wake_up() stats */
863 	unsigned int ttwu_wake_remote;
864 	unsigned int ttwu_move_affine;
865 	unsigned int ttwu_move_balance;
866 #endif
867 #ifdef CONFIG_SCHED_DEBUG
868 	char *name;
869 #endif
870 	union {
871 		void *private;		/* used during construction */
872 		struct rcu_head rcu;	/* used during destruction */
873 	};
874 
875 	unsigned int span_weight;
876 	/*
877 	 * Span of all CPUs in this domain.
878 	 *
879 	 * NOTE: this field is variable length. (Allocated dynamically
880 	 * by attaching extra space to the end of the structure,
881 	 * depending on how many CPUs the kernel has booted up with)
882 	 */
883 	unsigned long span[0];
884 };
885 
886 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
887 {
888 	return to_cpumask(sd->span);
889 }
890 
891 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
892 				    struct sched_domain_attr *dattr_new);
893 
894 /* Allocate an array of sched domains, for partition_sched_domains(). */
895 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
896 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
897 
898 bool cpus_share_cache(int this_cpu, int that_cpu);
899 
900 #else /* CONFIG_SMP */
901 
902 struct sched_domain_attr;
903 
904 static inline void
905 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
906 			struct sched_domain_attr *dattr_new)
907 {
908 }
909 
910 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
911 {
912 	return true;
913 }
914 
915 #endif	/* !CONFIG_SMP */
916 
917 
918 struct io_context;			/* See blkdev.h */
919 
920 
921 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
922 extern void prefetch_stack(struct task_struct *t);
923 #else
924 static inline void prefetch_stack(struct task_struct *t) { }
925 #endif
926 
927 struct audit_context;		/* See audit.c */
928 struct mempolicy;
929 struct pipe_inode_info;
930 struct uts_namespace;
931 
932 struct load_weight {
933 	unsigned long weight;
934 	u32 inv_weight;
935 };
936 
937 struct sched_avg {
938 	/*
939 	 * These sums represent an infinite geometric series and so are bound
940 	 * above by 1024/(1-y).  Thus we only need a u32 to store them for all
941 	 * choices of y < 1-2^(-32)*1024.
942 	 */
943 	u32 runnable_avg_sum, runnable_avg_period;
944 	u64 last_runnable_update;
945 	s64 decay_count;
946 	unsigned long load_avg_contrib;
947 };
948 
949 #ifdef CONFIG_SCHEDSTATS
950 struct sched_statistics {
951 	u64			wait_start;
952 	u64			wait_max;
953 	u64			wait_count;
954 	u64			wait_sum;
955 	u64			iowait_count;
956 	u64			iowait_sum;
957 
958 	u64			sleep_start;
959 	u64			sleep_max;
960 	s64			sum_sleep_runtime;
961 
962 	u64			block_start;
963 	u64			block_max;
964 	u64			exec_max;
965 	u64			slice_max;
966 
967 	u64			nr_migrations_cold;
968 	u64			nr_failed_migrations_affine;
969 	u64			nr_failed_migrations_running;
970 	u64			nr_failed_migrations_hot;
971 	u64			nr_forced_migrations;
972 
973 	u64			nr_wakeups;
974 	u64			nr_wakeups_sync;
975 	u64			nr_wakeups_migrate;
976 	u64			nr_wakeups_local;
977 	u64			nr_wakeups_remote;
978 	u64			nr_wakeups_affine;
979 	u64			nr_wakeups_affine_attempts;
980 	u64			nr_wakeups_passive;
981 	u64			nr_wakeups_idle;
982 };
983 #endif
984 
985 struct sched_entity {
986 	struct load_weight	load;		/* for load-balancing */
987 	struct rb_node		run_node;
988 	struct list_head	group_node;
989 	unsigned int		on_rq;
990 
991 	u64			exec_start;
992 	u64			sum_exec_runtime;
993 	u64			vruntime;
994 	u64			prev_sum_exec_runtime;
995 
996 	u64			nr_migrations;
997 
998 #ifdef CONFIG_SCHEDSTATS
999 	struct sched_statistics statistics;
1000 #endif
1001 
1002 #ifdef CONFIG_FAIR_GROUP_SCHED
1003 	struct sched_entity	*parent;
1004 	/* rq on which this entity is (to be) queued: */
1005 	struct cfs_rq		*cfs_rq;
1006 	/* rq "owned" by this entity/group: */
1007 	struct cfs_rq		*my_q;
1008 #endif
1009 
1010 #ifdef CONFIG_SMP
1011 	/* Per-entity load-tracking */
1012 	struct sched_avg	avg;
1013 #endif
1014 };
1015 
1016 struct sched_rt_entity {
1017 	struct list_head run_list;
1018 	unsigned long timeout;
1019 	unsigned long watchdog_stamp;
1020 	unsigned int time_slice;
1021 
1022 	struct sched_rt_entity *back;
1023 #ifdef CONFIG_RT_GROUP_SCHED
1024 	struct sched_rt_entity	*parent;
1025 	/* rq on which this entity is (to be) queued: */
1026 	struct rt_rq		*rt_rq;
1027 	/* rq "owned" by this entity/group: */
1028 	struct rt_rq		*my_q;
1029 #endif
1030 };
1031 
1032 
1033 struct rcu_node;
1034 
1035 enum perf_event_task_context {
1036 	perf_invalid_context = -1,
1037 	perf_hw_context = 0,
1038 	perf_sw_context,
1039 	perf_nr_task_contexts,
1040 };
1041 
1042 struct task_struct {
1043 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
1044 	void *stack;
1045 	atomic_t usage;
1046 	unsigned int flags;	/* per process flags, defined below */
1047 	unsigned int ptrace;
1048 
1049 #ifdef CONFIG_SMP
1050 	struct llist_node wake_entry;
1051 	int on_cpu;
1052 	struct task_struct *last_wakee;
1053 	unsigned long wakee_flips;
1054 	unsigned long wakee_flip_decay_ts;
1055 
1056 	int wake_cpu;
1057 #endif
1058 	int on_rq;
1059 
1060 	int prio, static_prio, normal_prio;
1061 	unsigned int rt_priority;
1062 	const struct sched_class *sched_class;
1063 	struct sched_entity se;
1064 	struct sched_rt_entity rt;
1065 #ifdef CONFIG_CGROUP_SCHED
1066 	struct task_group *sched_task_group;
1067 #endif
1068 
1069 #ifdef CONFIG_PREEMPT_NOTIFIERS
1070 	/* list of struct preempt_notifier: */
1071 	struct hlist_head preempt_notifiers;
1072 #endif
1073 
1074 #ifdef CONFIG_BLK_DEV_IO_TRACE
1075 	unsigned int btrace_seq;
1076 #endif
1077 
1078 	unsigned int policy;
1079 	int nr_cpus_allowed;
1080 	cpumask_t cpus_allowed;
1081 
1082 #ifdef CONFIG_PREEMPT_RCU
1083 	int rcu_read_lock_nesting;
1084 	char rcu_read_unlock_special;
1085 	struct list_head rcu_node_entry;
1086 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1087 #ifdef CONFIG_TREE_PREEMPT_RCU
1088 	struct rcu_node *rcu_blocked_node;
1089 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1090 #ifdef CONFIG_RCU_BOOST
1091 	struct rt_mutex *rcu_boost_mutex;
1092 #endif /* #ifdef CONFIG_RCU_BOOST */
1093 
1094 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1095 	struct sched_info sched_info;
1096 #endif
1097 
1098 	struct list_head tasks;
1099 #ifdef CONFIG_SMP
1100 	struct plist_node pushable_tasks;
1101 #endif
1102 
1103 	struct mm_struct *mm, *active_mm;
1104 #ifdef CONFIG_COMPAT_BRK
1105 	unsigned brk_randomized:1;
1106 #endif
1107 #if defined(SPLIT_RSS_COUNTING)
1108 	struct task_rss_stat	rss_stat;
1109 #endif
1110 /* task state */
1111 	int exit_state;
1112 	int exit_code, exit_signal;
1113 	int pdeath_signal;  /*  The signal sent when the parent dies  */
1114 	unsigned int jobctl;	/* JOBCTL_*, siglock protected */
1115 
1116 	/* Used for emulating ABI behavior of previous Linux versions */
1117 	unsigned int personality;
1118 
1119 	unsigned did_exec:1;
1120 	unsigned in_execve:1;	/* Tell the LSMs that the process is doing an
1121 				 * execve */
1122 	unsigned in_iowait:1;
1123 
1124 	/* task may not gain privileges */
1125 	unsigned no_new_privs:1;
1126 
1127 	/* Revert to default priority/policy when forking */
1128 	unsigned sched_reset_on_fork:1;
1129 	unsigned sched_contributes_to_load:1;
1130 
1131 	pid_t pid;
1132 	pid_t tgid;
1133 
1134 #ifdef CONFIG_CC_STACKPROTECTOR
1135 	/* Canary value for the -fstack-protector gcc feature */
1136 	unsigned long stack_canary;
1137 #endif
1138 	/*
1139 	 * pointers to (original) parent process, youngest child, younger sibling,
1140 	 * older sibling, respectively.  (p->father can be replaced with
1141 	 * p->real_parent->pid)
1142 	 */
1143 	struct task_struct __rcu *real_parent; /* real parent process */
1144 	struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1145 	/*
1146 	 * children/sibling forms the list of my natural children
1147 	 */
1148 	struct list_head children;	/* list of my children */
1149 	struct list_head sibling;	/* linkage in my parent's children list */
1150 	struct task_struct *group_leader;	/* threadgroup leader */
1151 
1152 	/*
1153 	 * ptraced is the list of tasks this task is using ptrace on.
1154 	 * This includes both natural children and PTRACE_ATTACH targets.
1155 	 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1156 	 */
1157 	struct list_head ptraced;
1158 	struct list_head ptrace_entry;
1159 
1160 	/* PID/PID hash table linkage. */
1161 	struct pid_link pids[PIDTYPE_MAX];
1162 	struct list_head thread_group;
1163 
1164 	struct completion *vfork_done;		/* for vfork() */
1165 	int __user *set_child_tid;		/* CLONE_CHILD_SETTID */
1166 	int __user *clear_child_tid;		/* CLONE_CHILD_CLEARTID */
1167 
1168 	cputime_t utime, stime, utimescaled, stimescaled;
1169 	cputime_t gtime;
1170 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1171 	struct cputime prev_cputime;
1172 #endif
1173 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1174 	seqlock_t vtime_seqlock;
1175 	unsigned long long vtime_snap;
1176 	enum {
1177 		VTIME_SLEEPING = 0,
1178 		VTIME_USER,
1179 		VTIME_SYS,
1180 	} vtime_snap_whence;
1181 #endif
1182 	unsigned long nvcsw, nivcsw; /* context switch counts */
1183 	struct timespec start_time; 		/* monotonic time */
1184 	struct timespec real_start_time;	/* boot based time */
1185 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1186 	unsigned long min_flt, maj_flt;
1187 
1188 	struct task_cputime cputime_expires;
1189 	struct list_head cpu_timers[3];
1190 
1191 /* process credentials */
1192 	const struct cred __rcu *real_cred; /* objective and real subjective task
1193 					 * credentials (COW) */
1194 	const struct cred __rcu *cred;	/* effective (overridable) subjective task
1195 					 * credentials (COW) */
1196 	char comm[TASK_COMM_LEN]; /* executable name excluding path
1197 				     - access with [gs]et_task_comm (which lock
1198 				       it with task_lock())
1199 				     - initialized normally by setup_new_exec */
1200 /* file system info */
1201 	int link_count, total_link_count;
1202 #ifdef CONFIG_SYSVIPC
1203 /* ipc stuff */
1204 	struct sysv_sem sysvsem;
1205 #endif
1206 #ifdef CONFIG_DETECT_HUNG_TASK
1207 /* hung task detection */
1208 	unsigned long last_switch_count;
1209 #endif
1210 /* CPU-specific state of this task */
1211 	struct thread_struct thread;
1212 /* filesystem information */
1213 	struct fs_struct *fs;
1214 /* open file information */
1215 	struct files_struct *files;
1216 /* namespaces */
1217 	struct nsproxy *nsproxy;
1218 /* signal handlers */
1219 	struct signal_struct *signal;
1220 	struct sighand_struct *sighand;
1221 
1222 	sigset_t blocked, real_blocked;
1223 	sigset_t saved_sigmask;	/* restored if set_restore_sigmask() was used */
1224 	struct sigpending pending;
1225 
1226 	unsigned long sas_ss_sp;
1227 	size_t sas_ss_size;
1228 	int (*notifier)(void *priv);
1229 	void *notifier_data;
1230 	sigset_t *notifier_mask;
1231 	struct callback_head *task_works;
1232 
1233 	struct audit_context *audit_context;
1234 #ifdef CONFIG_AUDITSYSCALL
1235 	kuid_t loginuid;
1236 	unsigned int sessionid;
1237 #endif
1238 	struct seccomp seccomp;
1239 
1240 /* Thread group tracking */
1241    	u32 parent_exec_id;
1242    	u32 self_exec_id;
1243 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1244  * mempolicy */
1245 	spinlock_t alloc_lock;
1246 
1247 	/* Protection of the PI data structures: */
1248 	raw_spinlock_t pi_lock;
1249 
1250 #ifdef CONFIG_RT_MUTEXES
1251 	/* PI waiters blocked on a rt_mutex held by this task */
1252 	struct plist_head pi_waiters;
1253 	/* Deadlock detection and priority inheritance handling */
1254 	struct rt_mutex_waiter *pi_blocked_on;
1255 #endif
1256 
1257 #ifdef CONFIG_DEBUG_MUTEXES
1258 	/* mutex deadlock detection */
1259 	struct mutex_waiter *blocked_on;
1260 #endif
1261 #ifdef CONFIG_TRACE_IRQFLAGS
1262 	unsigned int irq_events;
1263 	unsigned long hardirq_enable_ip;
1264 	unsigned long hardirq_disable_ip;
1265 	unsigned int hardirq_enable_event;
1266 	unsigned int hardirq_disable_event;
1267 	int hardirqs_enabled;
1268 	int hardirq_context;
1269 	unsigned long softirq_disable_ip;
1270 	unsigned long softirq_enable_ip;
1271 	unsigned int softirq_disable_event;
1272 	unsigned int softirq_enable_event;
1273 	int softirqs_enabled;
1274 	int softirq_context;
1275 #endif
1276 #ifdef CONFIG_LOCKDEP
1277 # define MAX_LOCK_DEPTH 48UL
1278 	u64 curr_chain_key;
1279 	int lockdep_depth;
1280 	unsigned int lockdep_recursion;
1281 	struct held_lock held_locks[MAX_LOCK_DEPTH];
1282 	gfp_t lockdep_reclaim_gfp;
1283 #endif
1284 
1285 /* journalling filesystem info */
1286 	void *journal_info;
1287 
1288 /* stacked block device info */
1289 	struct bio_list *bio_list;
1290 
1291 #ifdef CONFIG_BLOCK
1292 /* stack plugging */
1293 	struct blk_plug *plug;
1294 #endif
1295 
1296 /* VM state */
1297 	struct reclaim_state *reclaim_state;
1298 
1299 	struct backing_dev_info *backing_dev_info;
1300 
1301 	struct io_context *io_context;
1302 
1303 	unsigned long ptrace_message;
1304 	siginfo_t *last_siginfo; /* For ptrace use.  */
1305 	struct task_io_accounting ioac;
1306 #if defined(CONFIG_TASK_XACCT)
1307 	u64 acct_rss_mem1;	/* accumulated rss usage */
1308 	u64 acct_vm_mem1;	/* accumulated virtual memory usage */
1309 	cputime_t acct_timexpd;	/* stime + utime since last update */
1310 #endif
1311 #ifdef CONFIG_CPUSETS
1312 	nodemask_t mems_allowed;	/* Protected by alloc_lock */
1313 	seqcount_t mems_allowed_seq;	/* Seqence no to catch updates */
1314 	int cpuset_mem_spread_rotor;
1315 	int cpuset_slab_spread_rotor;
1316 #endif
1317 #ifdef CONFIG_CGROUPS
1318 	/* Control Group info protected by css_set_lock */
1319 	struct css_set __rcu *cgroups;
1320 	/* cg_list protected by css_set_lock and tsk->alloc_lock */
1321 	struct list_head cg_list;
1322 #endif
1323 #ifdef CONFIG_FUTEX
1324 	struct robust_list_head __user *robust_list;
1325 #ifdef CONFIG_COMPAT
1326 	struct compat_robust_list_head __user *compat_robust_list;
1327 #endif
1328 	struct list_head pi_state_list;
1329 	struct futex_pi_state *pi_state_cache;
1330 #endif
1331 #ifdef CONFIG_PERF_EVENTS
1332 	struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1333 	struct mutex perf_event_mutex;
1334 	struct list_head perf_event_list;
1335 #endif
1336 #ifdef CONFIG_NUMA
1337 	struct mempolicy *mempolicy;	/* Protected by alloc_lock */
1338 	short il_next;
1339 	short pref_node_fork;
1340 #endif
1341 #ifdef CONFIG_NUMA_BALANCING
1342 	int numa_scan_seq;
1343 	unsigned int numa_scan_period;
1344 	unsigned int numa_scan_period_max;
1345 	int numa_preferred_nid;
1346 	int numa_migrate_deferred;
1347 	unsigned long numa_migrate_retry;
1348 	u64 node_stamp;			/* migration stamp  */
1349 	struct callback_head numa_work;
1350 
1351 	struct list_head numa_entry;
1352 	struct numa_group *numa_group;
1353 
1354 	/*
1355 	 * Exponential decaying average of faults on a per-node basis.
1356 	 * Scheduling placement decisions are made based on the these counts.
1357 	 * The values remain static for the duration of a PTE scan
1358 	 */
1359 	unsigned long *numa_faults;
1360 	unsigned long total_numa_faults;
1361 
1362 	/*
1363 	 * numa_faults_buffer records faults per node during the current
1364 	 * scan window. When the scan completes, the counts in numa_faults
1365 	 * decay and these values are copied.
1366 	 */
1367 	unsigned long *numa_faults_buffer;
1368 
1369 	/*
1370 	 * numa_faults_locality tracks if faults recorded during the last
1371 	 * scan window were remote/local. The task scan period is adapted
1372 	 * based on the locality of the faults with different weights
1373 	 * depending on whether they were shared or private faults
1374 	 */
1375 	unsigned long numa_faults_locality[2];
1376 
1377 	unsigned long numa_pages_migrated;
1378 #endif /* CONFIG_NUMA_BALANCING */
1379 
1380 	struct rcu_head rcu;
1381 
1382 	/*
1383 	 * cache last used pipe for splice
1384 	 */
1385 	struct pipe_inode_info *splice_pipe;
1386 
1387 	struct page_frag task_frag;
1388 
1389 #ifdef	CONFIG_TASK_DELAY_ACCT
1390 	struct task_delay_info *delays;
1391 #endif
1392 #ifdef CONFIG_FAULT_INJECTION
1393 	int make_it_fail;
1394 #endif
1395 	/*
1396 	 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1397 	 * balance_dirty_pages() for some dirty throttling pause
1398 	 */
1399 	int nr_dirtied;
1400 	int nr_dirtied_pause;
1401 	unsigned long dirty_paused_when; /* start of a write-and-pause period */
1402 
1403 #ifdef CONFIG_LATENCYTOP
1404 	int latency_record_count;
1405 	struct latency_record latency_record[LT_SAVECOUNT];
1406 #endif
1407 	/*
1408 	 * time slack values; these are used to round up poll() and
1409 	 * select() etc timeout values. These are in nanoseconds.
1410 	 */
1411 	unsigned long timer_slack_ns;
1412 	unsigned long default_timer_slack_ns;
1413 
1414 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1415 	/* Index of current stored address in ret_stack */
1416 	int curr_ret_stack;
1417 	/* Stack of return addresses for return function tracing */
1418 	struct ftrace_ret_stack	*ret_stack;
1419 	/* time stamp for last schedule */
1420 	unsigned long long ftrace_timestamp;
1421 	/*
1422 	 * Number of functions that haven't been traced
1423 	 * because of depth overrun.
1424 	 */
1425 	atomic_t trace_overrun;
1426 	/* Pause for the tracing */
1427 	atomic_t tracing_graph_pause;
1428 #endif
1429 #ifdef CONFIG_TRACING
1430 	/* state flags for use by tracers */
1431 	unsigned long trace;
1432 	/* bitmask and counter of trace recursion */
1433 	unsigned long trace_recursion;
1434 #endif /* CONFIG_TRACING */
1435 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1436 	struct memcg_batch_info {
1437 		int do_batch;	/* incremented when batch uncharge started */
1438 		struct mem_cgroup *memcg; /* target memcg of uncharge */
1439 		unsigned long nr_pages;	/* uncharged usage */
1440 		unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1441 	} memcg_batch;
1442 	unsigned int memcg_kmem_skip_account;
1443 	struct memcg_oom_info {
1444 		struct mem_cgroup *memcg;
1445 		gfp_t gfp_mask;
1446 		int order;
1447 		unsigned int may_oom:1;
1448 	} memcg_oom;
1449 #endif
1450 #ifdef CONFIG_UPROBES
1451 	struct uprobe_task *utask;
1452 #endif
1453 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1454 	unsigned int	sequential_io;
1455 	unsigned int	sequential_io_avg;
1456 #endif
1457 };
1458 
1459 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1460 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1461 
1462 #define TNF_MIGRATED	0x01
1463 #define TNF_NO_GROUP	0x02
1464 #define TNF_SHARED	0x04
1465 #define TNF_FAULT_LOCAL	0x08
1466 
1467 #ifdef CONFIG_NUMA_BALANCING
1468 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1469 extern pid_t task_numa_group_id(struct task_struct *p);
1470 extern void set_numabalancing_state(bool enabled);
1471 extern void task_numa_free(struct task_struct *p);
1472 
1473 extern unsigned int sysctl_numa_balancing_migrate_deferred;
1474 #else
1475 static inline void task_numa_fault(int last_node, int node, int pages,
1476 				   int flags)
1477 {
1478 }
1479 static inline pid_t task_numa_group_id(struct task_struct *p)
1480 {
1481 	return 0;
1482 }
1483 static inline void set_numabalancing_state(bool enabled)
1484 {
1485 }
1486 static inline void task_numa_free(struct task_struct *p)
1487 {
1488 }
1489 #endif
1490 
1491 static inline struct pid *task_pid(struct task_struct *task)
1492 {
1493 	return task->pids[PIDTYPE_PID].pid;
1494 }
1495 
1496 static inline struct pid *task_tgid(struct task_struct *task)
1497 {
1498 	return task->group_leader->pids[PIDTYPE_PID].pid;
1499 }
1500 
1501 /*
1502  * Without tasklist or rcu lock it is not safe to dereference
1503  * the result of task_pgrp/task_session even if task == current,
1504  * we can race with another thread doing sys_setsid/sys_setpgid.
1505  */
1506 static inline struct pid *task_pgrp(struct task_struct *task)
1507 {
1508 	return task->group_leader->pids[PIDTYPE_PGID].pid;
1509 }
1510 
1511 static inline struct pid *task_session(struct task_struct *task)
1512 {
1513 	return task->group_leader->pids[PIDTYPE_SID].pid;
1514 }
1515 
1516 struct pid_namespace;
1517 
1518 /*
1519  * the helpers to get the task's different pids as they are seen
1520  * from various namespaces
1521  *
1522  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1523  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1524  *                     current.
1525  * task_xid_nr_ns()  : id seen from the ns specified;
1526  *
1527  * set_task_vxid()   : assigns a virtual id to a task;
1528  *
1529  * see also pid_nr() etc in include/linux/pid.h
1530  */
1531 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1532 			struct pid_namespace *ns);
1533 
1534 static inline pid_t task_pid_nr(struct task_struct *tsk)
1535 {
1536 	return tsk->pid;
1537 }
1538 
1539 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1540 					struct pid_namespace *ns)
1541 {
1542 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1543 }
1544 
1545 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1546 {
1547 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1548 }
1549 
1550 
1551 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1552 {
1553 	return tsk->tgid;
1554 }
1555 
1556 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1557 
1558 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1559 {
1560 	return pid_vnr(task_tgid(tsk));
1561 }
1562 
1563 
1564 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1565 					struct pid_namespace *ns)
1566 {
1567 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1568 }
1569 
1570 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1571 {
1572 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1573 }
1574 
1575 
1576 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1577 					struct pid_namespace *ns)
1578 {
1579 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1580 }
1581 
1582 static inline pid_t task_session_vnr(struct task_struct *tsk)
1583 {
1584 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1585 }
1586 
1587 /* obsolete, do not use */
1588 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1589 {
1590 	return task_pgrp_nr_ns(tsk, &init_pid_ns);
1591 }
1592 
1593 /**
1594  * pid_alive - check that a task structure is not stale
1595  * @p: Task structure to be checked.
1596  *
1597  * Test if a process is not yet dead (at most zombie state)
1598  * If pid_alive fails, then pointers within the task structure
1599  * can be stale and must not be dereferenced.
1600  *
1601  * Return: 1 if the process is alive. 0 otherwise.
1602  */
1603 static inline int pid_alive(struct task_struct *p)
1604 {
1605 	return p->pids[PIDTYPE_PID].pid != NULL;
1606 }
1607 
1608 /**
1609  * is_global_init - check if a task structure is init
1610  * @tsk: Task structure to be checked.
1611  *
1612  * Check if a task structure is the first user space task the kernel created.
1613  *
1614  * Return: 1 if the task structure is init. 0 otherwise.
1615  */
1616 static inline int is_global_init(struct task_struct *tsk)
1617 {
1618 	return tsk->pid == 1;
1619 }
1620 
1621 extern struct pid *cad_pid;
1622 
1623 extern void free_task(struct task_struct *tsk);
1624 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1625 
1626 extern void __put_task_struct(struct task_struct *t);
1627 
1628 static inline void put_task_struct(struct task_struct *t)
1629 {
1630 	if (atomic_dec_and_test(&t->usage))
1631 		__put_task_struct(t);
1632 }
1633 
1634 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1635 extern void task_cputime(struct task_struct *t,
1636 			 cputime_t *utime, cputime_t *stime);
1637 extern void task_cputime_scaled(struct task_struct *t,
1638 				cputime_t *utimescaled, cputime_t *stimescaled);
1639 extern cputime_t task_gtime(struct task_struct *t);
1640 #else
1641 static inline void task_cputime(struct task_struct *t,
1642 				cputime_t *utime, cputime_t *stime)
1643 {
1644 	if (utime)
1645 		*utime = t->utime;
1646 	if (stime)
1647 		*stime = t->stime;
1648 }
1649 
1650 static inline void task_cputime_scaled(struct task_struct *t,
1651 				       cputime_t *utimescaled,
1652 				       cputime_t *stimescaled)
1653 {
1654 	if (utimescaled)
1655 		*utimescaled = t->utimescaled;
1656 	if (stimescaled)
1657 		*stimescaled = t->stimescaled;
1658 }
1659 
1660 static inline cputime_t task_gtime(struct task_struct *t)
1661 {
1662 	return t->gtime;
1663 }
1664 #endif
1665 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1666 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1667 
1668 /*
1669  * Per process flags
1670  */
1671 #define PF_EXITING	0x00000004	/* getting shut down */
1672 #define PF_EXITPIDONE	0x00000008	/* pi exit done on shut down */
1673 #define PF_VCPU		0x00000010	/* I'm a virtual CPU */
1674 #define PF_WQ_WORKER	0x00000020	/* I'm a workqueue worker */
1675 #define PF_FORKNOEXEC	0x00000040	/* forked but didn't exec */
1676 #define PF_MCE_PROCESS  0x00000080      /* process policy on mce errors */
1677 #define PF_SUPERPRIV	0x00000100	/* used super-user privileges */
1678 #define PF_DUMPCORE	0x00000200	/* dumped core */
1679 #define PF_SIGNALED	0x00000400	/* killed by a signal */
1680 #define PF_MEMALLOC	0x00000800	/* Allocating memory */
1681 #define PF_NPROC_EXCEEDED 0x00001000	/* set_user noticed that RLIMIT_NPROC was exceeded */
1682 #define PF_USED_MATH	0x00002000	/* if unset the fpu must be initialized before use */
1683 #define PF_USED_ASYNC	0x00004000	/* used async_schedule*(), used by module init */
1684 #define PF_NOFREEZE	0x00008000	/* this thread should not be frozen */
1685 #define PF_FROZEN	0x00010000	/* frozen for system suspend */
1686 #define PF_FSTRANS	0x00020000	/* inside a filesystem transaction */
1687 #define PF_KSWAPD	0x00040000	/* I am kswapd */
1688 #define PF_MEMALLOC_NOIO 0x00080000	/* Allocating memory without IO involved */
1689 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
1690 #define PF_KTHREAD	0x00200000	/* I am a kernel thread */
1691 #define PF_RANDOMIZE	0x00400000	/* randomize virtual address space */
1692 #define PF_SWAPWRITE	0x00800000	/* Allowed to write to swap */
1693 #define PF_SPREAD_PAGE	0x01000000	/* Spread page cache over cpuset */
1694 #define PF_SPREAD_SLAB	0x02000000	/* Spread some slab caches over cpuset */
1695 #define PF_NO_SETAFFINITY 0x04000000	/* Userland is not allowed to meddle with cpus_allowed */
1696 #define PF_MCE_EARLY    0x08000000      /* Early kill for mce process policy */
1697 #define PF_MEMPOLICY	0x10000000	/* Non-default NUMA mempolicy */
1698 #define PF_MUTEX_TESTER	0x20000000	/* Thread belongs to the rt mutex tester */
1699 #define PF_FREEZER_SKIP	0x40000000	/* Freezer should not count it as freezable */
1700 #define PF_SUSPEND_TASK 0x80000000      /* this thread called freeze_processes and should not be frozen */
1701 
1702 /*
1703  * Only the _current_ task can read/write to tsk->flags, but other
1704  * tasks can access tsk->flags in readonly mode for example
1705  * with tsk_used_math (like during threaded core dumping).
1706  * There is however an exception to this rule during ptrace
1707  * or during fork: the ptracer task is allowed to write to the
1708  * child->flags of its traced child (same goes for fork, the parent
1709  * can write to the child->flags), because we're guaranteed the
1710  * child is not running and in turn not changing child->flags
1711  * at the same time the parent does it.
1712  */
1713 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1714 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1715 #define clear_used_math() clear_stopped_child_used_math(current)
1716 #define set_used_math() set_stopped_child_used_math(current)
1717 #define conditional_stopped_child_used_math(condition, child) \
1718 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1719 #define conditional_used_math(condition) \
1720 	conditional_stopped_child_used_math(condition, current)
1721 #define copy_to_stopped_child_used_math(child) \
1722 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1723 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1724 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1725 #define used_math() tsk_used_math(current)
1726 
1727 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1728 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1729 {
1730 	if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1731 		flags &= ~__GFP_IO;
1732 	return flags;
1733 }
1734 
1735 static inline unsigned int memalloc_noio_save(void)
1736 {
1737 	unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1738 	current->flags |= PF_MEMALLOC_NOIO;
1739 	return flags;
1740 }
1741 
1742 static inline void memalloc_noio_restore(unsigned int flags)
1743 {
1744 	current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1745 }
1746 
1747 /*
1748  * task->jobctl flags
1749  */
1750 #define JOBCTL_STOP_SIGMASK	0xffff	/* signr of the last group stop */
1751 
1752 #define JOBCTL_STOP_DEQUEUED_BIT 16	/* stop signal dequeued */
1753 #define JOBCTL_STOP_PENDING_BIT	17	/* task should stop for group stop */
1754 #define JOBCTL_STOP_CONSUME_BIT	18	/* consume group stop count */
1755 #define JOBCTL_TRAP_STOP_BIT	19	/* trap for STOP */
1756 #define JOBCTL_TRAP_NOTIFY_BIT	20	/* trap for NOTIFY */
1757 #define JOBCTL_TRAPPING_BIT	21	/* switching to TRACED */
1758 #define JOBCTL_LISTENING_BIT	22	/* ptracer is listening for events */
1759 
1760 #define JOBCTL_STOP_DEQUEUED	(1 << JOBCTL_STOP_DEQUEUED_BIT)
1761 #define JOBCTL_STOP_PENDING	(1 << JOBCTL_STOP_PENDING_BIT)
1762 #define JOBCTL_STOP_CONSUME	(1 << JOBCTL_STOP_CONSUME_BIT)
1763 #define JOBCTL_TRAP_STOP	(1 << JOBCTL_TRAP_STOP_BIT)
1764 #define JOBCTL_TRAP_NOTIFY	(1 << JOBCTL_TRAP_NOTIFY_BIT)
1765 #define JOBCTL_TRAPPING		(1 << JOBCTL_TRAPPING_BIT)
1766 #define JOBCTL_LISTENING	(1 << JOBCTL_LISTENING_BIT)
1767 
1768 #define JOBCTL_TRAP_MASK	(JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1769 #define JOBCTL_PENDING_MASK	(JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1770 
1771 extern bool task_set_jobctl_pending(struct task_struct *task,
1772 				    unsigned int mask);
1773 extern void task_clear_jobctl_trapping(struct task_struct *task);
1774 extern void task_clear_jobctl_pending(struct task_struct *task,
1775 				      unsigned int mask);
1776 
1777 #ifdef CONFIG_PREEMPT_RCU
1778 
1779 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1780 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1781 
1782 static inline void rcu_copy_process(struct task_struct *p)
1783 {
1784 	p->rcu_read_lock_nesting = 0;
1785 	p->rcu_read_unlock_special = 0;
1786 #ifdef CONFIG_TREE_PREEMPT_RCU
1787 	p->rcu_blocked_node = NULL;
1788 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1789 #ifdef CONFIG_RCU_BOOST
1790 	p->rcu_boost_mutex = NULL;
1791 #endif /* #ifdef CONFIG_RCU_BOOST */
1792 	INIT_LIST_HEAD(&p->rcu_node_entry);
1793 }
1794 
1795 #else
1796 
1797 static inline void rcu_copy_process(struct task_struct *p)
1798 {
1799 }
1800 
1801 #endif
1802 
1803 static inline void tsk_restore_flags(struct task_struct *task,
1804 				unsigned long orig_flags, unsigned long flags)
1805 {
1806 	task->flags &= ~flags;
1807 	task->flags |= orig_flags & flags;
1808 }
1809 
1810 #ifdef CONFIG_SMP
1811 extern void do_set_cpus_allowed(struct task_struct *p,
1812 			       const struct cpumask *new_mask);
1813 
1814 extern int set_cpus_allowed_ptr(struct task_struct *p,
1815 				const struct cpumask *new_mask);
1816 #else
1817 static inline void do_set_cpus_allowed(struct task_struct *p,
1818 				      const struct cpumask *new_mask)
1819 {
1820 }
1821 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1822 				       const struct cpumask *new_mask)
1823 {
1824 	if (!cpumask_test_cpu(0, new_mask))
1825 		return -EINVAL;
1826 	return 0;
1827 }
1828 #endif
1829 
1830 #ifdef CONFIG_NO_HZ_COMMON
1831 void calc_load_enter_idle(void);
1832 void calc_load_exit_idle(void);
1833 #else
1834 static inline void calc_load_enter_idle(void) { }
1835 static inline void calc_load_exit_idle(void) { }
1836 #endif /* CONFIG_NO_HZ_COMMON */
1837 
1838 #ifndef CONFIG_CPUMASK_OFFSTACK
1839 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1840 {
1841 	return set_cpus_allowed_ptr(p, &new_mask);
1842 }
1843 #endif
1844 
1845 /*
1846  * Do not use outside of architecture code which knows its limitations.
1847  *
1848  * sched_clock() has no promise of monotonicity or bounded drift between
1849  * CPUs, use (which you should not) requires disabling IRQs.
1850  *
1851  * Please use one of the three interfaces below.
1852  */
1853 extern unsigned long long notrace sched_clock(void);
1854 /*
1855  * See the comment in kernel/sched/clock.c
1856  */
1857 extern u64 cpu_clock(int cpu);
1858 extern u64 local_clock(void);
1859 extern u64 sched_clock_cpu(int cpu);
1860 
1861 
1862 extern void sched_clock_init(void);
1863 
1864 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1865 static inline void sched_clock_tick(void)
1866 {
1867 }
1868 
1869 static inline void sched_clock_idle_sleep_event(void)
1870 {
1871 }
1872 
1873 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1874 {
1875 }
1876 #else
1877 /*
1878  * Architectures can set this to 1 if they have specified
1879  * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1880  * but then during bootup it turns out that sched_clock()
1881  * is reliable after all:
1882  */
1883 extern int sched_clock_stable;
1884 
1885 extern void sched_clock_tick(void);
1886 extern void sched_clock_idle_sleep_event(void);
1887 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1888 #endif
1889 
1890 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1891 /*
1892  * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1893  * The reason for this explicit opt-in is not to have perf penalty with
1894  * slow sched_clocks.
1895  */
1896 extern void enable_sched_clock_irqtime(void);
1897 extern void disable_sched_clock_irqtime(void);
1898 #else
1899 static inline void enable_sched_clock_irqtime(void) {}
1900 static inline void disable_sched_clock_irqtime(void) {}
1901 #endif
1902 
1903 extern unsigned long long
1904 task_sched_runtime(struct task_struct *task);
1905 
1906 /* sched_exec is called by processes performing an exec */
1907 #ifdef CONFIG_SMP
1908 extern void sched_exec(void);
1909 #else
1910 #define sched_exec()   {}
1911 #endif
1912 
1913 extern void sched_clock_idle_sleep_event(void);
1914 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1915 
1916 #ifdef CONFIG_HOTPLUG_CPU
1917 extern void idle_task_exit(void);
1918 #else
1919 static inline void idle_task_exit(void) {}
1920 #endif
1921 
1922 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1923 extern void wake_up_nohz_cpu(int cpu);
1924 #else
1925 static inline void wake_up_nohz_cpu(int cpu) { }
1926 #endif
1927 
1928 #ifdef CONFIG_NO_HZ_FULL
1929 extern bool sched_can_stop_tick(void);
1930 extern u64 scheduler_tick_max_deferment(void);
1931 #else
1932 static inline bool sched_can_stop_tick(void) { return false; }
1933 #endif
1934 
1935 #ifdef CONFIG_SCHED_AUTOGROUP
1936 extern void sched_autogroup_create_attach(struct task_struct *p);
1937 extern void sched_autogroup_detach(struct task_struct *p);
1938 extern void sched_autogroup_fork(struct signal_struct *sig);
1939 extern void sched_autogroup_exit(struct signal_struct *sig);
1940 #ifdef CONFIG_PROC_FS
1941 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1942 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
1943 #endif
1944 #else
1945 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1946 static inline void sched_autogroup_detach(struct task_struct *p) { }
1947 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1948 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1949 #endif
1950 
1951 extern bool yield_to(struct task_struct *p, bool preempt);
1952 extern void set_user_nice(struct task_struct *p, long nice);
1953 extern int task_prio(const struct task_struct *p);
1954 extern int task_nice(const struct task_struct *p);
1955 extern int can_nice(const struct task_struct *p, const int nice);
1956 extern int task_curr(const struct task_struct *p);
1957 extern int idle_cpu(int cpu);
1958 extern int sched_setscheduler(struct task_struct *, int,
1959 			      const struct sched_param *);
1960 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1961 				      const struct sched_param *);
1962 extern struct task_struct *idle_task(int cpu);
1963 /**
1964  * is_idle_task - is the specified task an idle task?
1965  * @p: the task in question.
1966  *
1967  * Return: 1 if @p is an idle task. 0 otherwise.
1968  */
1969 static inline bool is_idle_task(const struct task_struct *p)
1970 {
1971 	return p->pid == 0;
1972 }
1973 extern struct task_struct *curr_task(int cpu);
1974 extern void set_curr_task(int cpu, struct task_struct *p);
1975 
1976 void yield(void);
1977 
1978 /*
1979  * The default (Linux) execution domain.
1980  */
1981 extern struct exec_domain	default_exec_domain;
1982 
1983 union thread_union {
1984 	struct thread_info thread_info;
1985 	unsigned long stack[THREAD_SIZE/sizeof(long)];
1986 };
1987 
1988 #ifndef __HAVE_ARCH_KSTACK_END
1989 static inline int kstack_end(void *addr)
1990 {
1991 	/* Reliable end of stack detection:
1992 	 * Some APM bios versions misalign the stack
1993 	 */
1994 	return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1995 }
1996 #endif
1997 
1998 extern union thread_union init_thread_union;
1999 extern struct task_struct init_task;
2000 
2001 extern struct   mm_struct init_mm;
2002 
2003 extern struct pid_namespace init_pid_ns;
2004 
2005 /*
2006  * find a task by one of its numerical ids
2007  *
2008  * find_task_by_pid_ns():
2009  *      finds a task by its pid in the specified namespace
2010  * find_task_by_vpid():
2011  *      finds a task by its virtual pid
2012  *
2013  * see also find_vpid() etc in include/linux/pid.h
2014  */
2015 
2016 extern struct task_struct *find_task_by_vpid(pid_t nr);
2017 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2018 		struct pid_namespace *ns);
2019 
2020 /* per-UID process charging. */
2021 extern struct user_struct * alloc_uid(kuid_t);
2022 static inline struct user_struct *get_uid(struct user_struct *u)
2023 {
2024 	atomic_inc(&u->__count);
2025 	return u;
2026 }
2027 extern void free_uid(struct user_struct *);
2028 
2029 #include <asm/current.h>
2030 
2031 extern void xtime_update(unsigned long ticks);
2032 
2033 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2034 extern int wake_up_process(struct task_struct *tsk);
2035 extern void wake_up_new_task(struct task_struct *tsk);
2036 #ifdef CONFIG_SMP
2037  extern void kick_process(struct task_struct *tsk);
2038 #else
2039  static inline void kick_process(struct task_struct *tsk) { }
2040 #endif
2041 extern void sched_fork(unsigned long clone_flags, struct task_struct *p);
2042 extern void sched_dead(struct task_struct *p);
2043 
2044 extern void proc_caches_init(void);
2045 extern void flush_signals(struct task_struct *);
2046 extern void __flush_signals(struct task_struct *);
2047 extern void ignore_signals(struct task_struct *);
2048 extern void flush_signal_handlers(struct task_struct *, int force_default);
2049 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2050 
2051 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2052 {
2053 	unsigned long flags;
2054 	int ret;
2055 
2056 	spin_lock_irqsave(&tsk->sighand->siglock, flags);
2057 	ret = dequeue_signal(tsk, mask, info);
2058 	spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2059 
2060 	return ret;
2061 }
2062 
2063 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2064 			      sigset_t *mask);
2065 extern void unblock_all_signals(void);
2066 extern void release_task(struct task_struct * p);
2067 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2068 extern int force_sigsegv(int, struct task_struct *);
2069 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2070 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2071 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2072 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2073 				const struct cred *, u32);
2074 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2075 extern int kill_pid(struct pid *pid, int sig, int priv);
2076 extern int kill_proc_info(int, struct siginfo *, pid_t);
2077 extern __must_check bool do_notify_parent(struct task_struct *, int);
2078 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2079 extern void force_sig(int, struct task_struct *);
2080 extern int send_sig(int, struct task_struct *, int);
2081 extern int zap_other_threads(struct task_struct *p);
2082 extern struct sigqueue *sigqueue_alloc(void);
2083 extern void sigqueue_free(struct sigqueue *);
2084 extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
2085 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2086 
2087 static inline void restore_saved_sigmask(void)
2088 {
2089 	if (test_and_clear_restore_sigmask())
2090 		__set_current_blocked(&current->saved_sigmask);
2091 }
2092 
2093 static inline sigset_t *sigmask_to_save(void)
2094 {
2095 	sigset_t *res = &current->blocked;
2096 	if (unlikely(test_restore_sigmask()))
2097 		res = &current->saved_sigmask;
2098 	return res;
2099 }
2100 
2101 static inline int kill_cad_pid(int sig, int priv)
2102 {
2103 	return kill_pid(cad_pid, sig, priv);
2104 }
2105 
2106 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
2107 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2108 #define SEND_SIG_PRIV	((struct siginfo *) 1)
2109 #define SEND_SIG_FORCED	((struct siginfo *) 2)
2110 
2111 /*
2112  * True if we are on the alternate signal stack.
2113  */
2114 static inline int on_sig_stack(unsigned long sp)
2115 {
2116 #ifdef CONFIG_STACK_GROWSUP
2117 	return sp >= current->sas_ss_sp &&
2118 		sp - current->sas_ss_sp < current->sas_ss_size;
2119 #else
2120 	return sp > current->sas_ss_sp &&
2121 		sp - current->sas_ss_sp <= current->sas_ss_size;
2122 #endif
2123 }
2124 
2125 static inline int sas_ss_flags(unsigned long sp)
2126 {
2127 	return (current->sas_ss_size == 0 ? SS_DISABLE
2128 		: on_sig_stack(sp) ? SS_ONSTACK : 0);
2129 }
2130 
2131 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2132 {
2133 	if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2134 #ifdef CONFIG_STACK_GROWSUP
2135 		return current->sas_ss_sp;
2136 #else
2137 		return current->sas_ss_sp + current->sas_ss_size;
2138 #endif
2139 	return sp;
2140 }
2141 
2142 /*
2143  * Routines for handling mm_structs
2144  */
2145 extern struct mm_struct * mm_alloc(void);
2146 
2147 /* mmdrop drops the mm and the page tables */
2148 extern void __mmdrop(struct mm_struct *);
2149 static inline void mmdrop(struct mm_struct * mm)
2150 {
2151 	if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2152 		__mmdrop(mm);
2153 }
2154 
2155 /* mmput gets rid of the mappings and all user-space */
2156 extern void mmput(struct mm_struct *);
2157 /* Grab a reference to a task's mm, if it is not already going away */
2158 extern struct mm_struct *get_task_mm(struct task_struct *task);
2159 /*
2160  * Grab a reference to a task's mm, if it is not already going away
2161  * and ptrace_may_access with the mode parameter passed to it
2162  * succeeds.
2163  */
2164 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2165 /* Remove the current tasks stale references to the old mm_struct */
2166 extern void mm_release(struct task_struct *, struct mm_struct *);
2167 /* Allocate a new mm structure and copy contents from tsk->mm */
2168 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2169 
2170 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2171 			struct task_struct *);
2172 extern void flush_thread(void);
2173 extern void exit_thread(void);
2174 
2175 extern void exit_files(struct task_struct *);
2176 extern void __cleanup_sighand(struct sighand_struct *);
2177 
2178 extern void exit_itimers(struct signal_struct *);
2179 extern void flush_itimer_signals(void);
2180 
2181 extern void do_group_exit(int);
2182 
2183 extern int allow_signal(int);
2184 extern int disallow_signal(int);
2185 
2186 extern int do_execve(const char *,
2187 		     const char __user * const __user *,
2188 		     const char __user * const __user *);
2189 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2190 struct task_struct *fork_idle(int);
2191 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2192 
2193 extern void set_task_comm(struct task_struct *tsk, char *from);
2194 extern char *get_task_comm(char *to, struct task_struct *tsk);
2195 
2196 #ifdef CONFIG_SMP
2197 void scheduler_ipi(void);
2198 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2199 #else
2200 static inline void scheduler_ipi(void) { }
2201 static inline unsigned long wait_task_inactive(struct task_struct *p,
2202 					       long match_state)
2203 {
2204 	return 1;
2205 }
2206 #endif
2207 
2208 #define next_task(p) \
2209 	list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2210 
2211 #define for_each_process(p) \
2212 	for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2213 
2214 extern bool current_is_single_threaded(void);
2215 
2216 /*
2217  * Careful: do_each_thread/while_each_thread is a double loop so
2218  *          'break' will not work as expected - use goto instead.
2219  */
2220 #define do_each_thread(g, t) \
2221 	for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2222 
2223 #define while_each_thread(g, t) \
2224 	while ((t = next_thread(t)) != g)
2225 
2226 static inline int get_nr_threads(struct task_struct *tsk)
2227 {
2228 	return tsk->signal->nr_threads;
2229 }
2230 
2231 static inline bool thread_group_leader(struct task_struct *p)
2232 {
2233 	return p->exit_signal >= 0;
2234 }
2235 
2236 /* Do to the insanities of de_thread it is possible for a process
2237  * to have the pid of the thread group leader without actually being
2238  * the thread group leader.  For iteration through the pids in proc
2239  * all we care about is that we have a task with the appropriate
2240  * pid, we don't actually care if we have the right task.
2241  */
2242 static inline bool has_group_leader_pid(struct task_struct *p)
2243 {
2244 	return task_pid(p) == p->signal->leader_pid;
2245 }
2246 
2247 static inline
2248 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2249 {
2250 	return p1->signal == p2->signal;
2251 }
2252 
2253 static inline struct task_struct *next_thread(const struct task_struct *p)
2254 {
2255 	return list_entry_rcu(p->thread_group.next,
2256 			      struct task_struct, thread_group);
2257 }
2258 
2259 static inline int thread_group_empty(struct task_struct *p)
2260 {
2261 	return list_empty(&p->thread_group);
2262 }
2263 
2264 #define delay_group_leader(p) \
2265 		(thread_group_leader(p) && !thread_group_empty(p))
2266 
2267 /*
2268  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2269  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2270  * pins the final release of task.io_context.  Also protects ->cpuset and
2271  * ->cgroup.subsys[]. And ->vfork_done.
2272  *
2273  * Nests both inside and outside of read_lock(&tasklist_lock).
2274  * It must not be nested with write_lock_irq(&tasklist_lock),
2275  * neither inside nor outside.
2276  */
2277 static inline void task_lock(struct task_struct *p)
2278 {
2279 	spin_lock(&p->alloc_lock);
2280 }
2281 
2282 static inline void task_unlock(struct task_struct *p)
2283 {
2284 	spin_unlock(&p->alloc_lock);
2285 }
2286 
2287 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2288 							unsigned long *flags);
2289 
2290 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2291 						       unsigned long *flags)
2292 {
2293 	struct sighand_struct *ret;
2294 
2295 	ret = __lock_task_sighand(tsk, flags);
2296 	(void)__cond_lock(&tsk->sighand->siglock, ret);
2297 	return ret;
2298 }
2299 
2300 static inline void unlock_task_sighand(struct task_struct *tsk,
2301 						unsigned long *flags)
2302 {
2303 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2304 }
2305 
2306 #ifdef CONFIG_CGROUPS
2307 static inline void threadgroup_change_begin(struct task_struct *tsk)
2308 {
2309 	down_read(&tsk->signal->group_rwsem);
2310 }
2311 static inline void threadgroup_change_end(struct task_struct *tsk)
2312 {
2313 	up_read(&tsk->signal->group_rwsem);
2314 }
2315 
2316 /**
2317  * threadgroup_lock - lock threadgroup
2318  * @tsk: member task of the threadgroup to lock
2319  *
2320  * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
2321  * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2322  * change ->group_leader/pid.  This is useful for cases where the threadgroup
2323  * needs to stay stable across blockable operations.
2324  *
2325  * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2326  * synchronization.  While held, no new task will be added to threadgroup
2327  * and no existing live task will have its PF_EXITING set.
2328  *
2329  * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2330  * sub-thread becomes a new leader.
2331  */
2332 static inline void threadgroup_lock(struct task_struct *tsk)
2333 {
2334 	down_write(&tsk->signal->group_rwsem);
2335 }
2336 
2337 /**
2338  * threadgroup_unlock - unlock threadgroup
2339  * @tsk: member task of the threadgroup to unlock
2340  *
2341  * Reverse threadgroup_lock().
2342  */
2343 static inline void threadgroup_unlock(struct task_struct *tsk)
2344 {
2345 	up_write(&tsk->signal->group_rwsem);
2346 }
2347 #else
2348 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2349 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2350 static inline void threadgroup_lock(struct task_struct *tsk) {}
2351 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2352 #endif
2353 
2354 #ifndef __HAVE_THREAD_FUNCTIONS
2355 
2356 #define task_thread_info(task)	((struct thread_info *)(task)->stack)
2357 #define task_stack_page(task)	((task)->stack)
2358 
2359 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2360 {
2361 	*task_thread_info(p) = *task_thread_info(org);
2362 	task_thread_info(p)->task = p;
2363 }
2364 
2365 static inline unsigned long *end_of_stack(struct task_struct *p)
2366 {
2367 	return (unsigned long *)(task_thread_info(p) + 1);
2368 }
2369 
2370 #endif
2371 
2372 static inline int object_is_on_stack(void *obj)
2373 {
2374 	void *stack = task_stack_page(current);
2375 
2376 	return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2377 }
2378 
2379 extern void thread_info_cache_init(void);
2380 
2381 #ifdef CONFIG_DEBUG_STACK_USAGE
2382 static inline unsigned long stack_not_used(struct task_struct *p)
2383 {
2384 	unsigned long *n = end_of_stack(p);
2385 
2386 	do { 	/* Skip over canary */
2387 		n++;
2388 	} while (!*n);
2389 
2390 	return (unsigned long)n - (unsigned long)end_of_stack(p);
2391 }
2392 #endif
2393 
2394 /* set thread flags in other task's structures
2395  * - see asm/thread_info.h for TIF_xxxx flags available
2396  */
2397 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2398 {
2399 	set_ti_thread_flag(task_thread_info(tsk), flag);
2400 }
2401 
2402 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2403 {
2404 	clear_ti_thread_flag(task_thread_info(tsk), flag);
2405 }
2406 
2407 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2408 {
2409 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2410 }
2411 
2412 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2413 {
2414 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2415 }
2416 
2417 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2418 {
2419 	return test_ti_thread_flag(task_thread_info(tsk), flag);
2420 }
2421 
2422 static inline void set_tsk_need_resched(struct task_struct *tsk)
2423 {
2424 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2425 }
2426 
2427 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2428 {
2429 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2430 }
2431 
2432 static inline int test_tsk_need_resched(struct task_struct *tsk)
2433 {
2434 	return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2435 }
2436 
2437 static inline int restart_syscall(void)
2438 {
2439 	set_tsk_thread_flag(current, TIF_SIGPENDING);
2440 	return -ERESTARTNOINTR;
2441 }
2442 
2443 static inline int signal_pending(struct task_struct *p)
2444 {
2445 	return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2446 }
2447 
2448 static inline int __fatal_signal_pending(struct task_struct *p)
2449 {
2450 	return unlikely(sigismember(&p->pending.signal, SIGKILL));
2451 }
2452 
2453 static inline int fatal_signal_pending(struct task_struct *p)
2454 {
2455 	return signal_pending(p) && __fatal_signal_pending(p);
2456 }
2457 
2458 static inline int signal_pending_state(long state, struct task_struct *p)
2459 {
2460 	if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2461 		return 0;
2462 	if (!signal_pending(p))
2463 		return 0;
2464 
2465 	return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2466 }
2467 
2468 /*
2469  * cond_resched() and cond_resched_lock(): latency reduction via
2470  * explicit rescheduling in places that are safe. The return
2471  * value indicates whether a reschedule was done in fact.
2472  * cond_resched_lock() will drop the spinlock before scheduling,
2473  * cond_resched_softirq() will enable bhs before scheduling.
2474  */
2475 extern int _cond_resched(void);
2476 
2477 #define cond_resched() ({			\
2478 	__might_sleep(__FILE__, __LINE__, 0);	\
2479 	_cond_resched();			\
2480 })
2481 
2482 extern int __cond_resched_lock(spinlock_t *lock);
2483 
2484 #ifdef CONFIG_PREEMPT_COUNT
2485 #define PREEMPT_LOCK_OFFSET	PREEMPT_OFFSET
2486 #else
2487 #define PREEMPT_LOCK_OFFSET	0
2488 #endif
2489 
2490 #define cond_resched_lock(lock) ({				\
2491 	__might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);	\
2492 	__cond_resched_lock(lock);				\
2493 })
2494 
2495 extern int __cond_resched_softirq(void);
2496 
2497 #define cond_resched_softirq() ({					\
2498 	__might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET);	\
2499 	__cond_resched_softirq();					\
2500 })
2501 
2502 static inline void cond_resched_rcu(void)
2503 {
2504 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2505 	rcu_read_unlock();
2506 	cond_resched();
2507 	rcu_read_lock();
2508 #endif
2509 }
2510 
2511 /*
2512  * Does a critical section need to be broken due to another
2513  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2514  * but a general need for low latency)
2515  */
2516 static inline int spin_needbreak(spinlock_t *lock)
2517 {
2518 #ifdef CONFIG_PREEMPT
2519 	return spin_is_contended(lock);
2520 #else
2521 	return 0;
2522 #endif
2523 }
2524 
2525 /*
2526  * Idle thread specific functions to determine the need_resched
2527  * polling state. We have two versions, one based on TS_POLLING in
2528  * thread_info.status and one based on TIF_POLLING_NRFLAG in
2529  * thread_info.flags
2530  */
2531 #ifdef TS_POLLING
2532 static inline int tsk_is_polling(struct task_struct *p)
2533 {
2534 	return task_thread_info(p)->status & TS_POLLING;
2535 }
2536 static inline void __current_set_polling(void)
2537 {
2538 	current_thread_info()->status |= TS_POLLING;
2539 }
2540 
2541 static inline bool __must_check current_set_polling_and_test(void)
2542 {
2543 	__current_set_polling();
2544 
2545 	/*
2546 	 * Polling state must be visible before we test NEED_RESCHED,
2547 	 * paired by resched_task()
2548 	 */
2549 	smp_mb();
2550 
2551 	return unlikely(tif_need_resched());
2552 }
2553 
2554 static inline void __current_clr_polling(void)
2555 {
2556 	current_thread_info()->status &= ~TS_POLLING;
2557 }
2558 
2559 static inline bool __must_check current_clr_polling_and_test(void)
2560 {
2561 	__current_clr_polling();
2562 
2563 	/*
2564 	 * Polling state must be visible before we test NEED_RESCHED,
2565 	 * paired by resched_task()
2566 	 */
2567 	smp_mb();
2568 
2569 	return unlikely(tif_need_resched());
2570 }
2571 #elif defined(TIF_POLLING_NRFLAG)
2572 static inline int tsk_is_polling(struct task_struct *p)
2573 {
2574 	return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2575 }
2576 
2577 static inline void __current_set_polling(void)
2578 {
2579 	set_thread_flag(TIF_POLLING_NRFLAG);
2580 }
2581 
2582 static inline bool __must_check current_set_polling_and_test(void)
2583 {
2584 	__current_set_polling();
2585 
2586 	/*
2587 	 * Polling state must be visible before we test NEED_RESCHED,
2588 	 * paired by resched_task()
2589 	 *
2590 	 * XXX: assumes set/clear bit are identical barrier wise.
2591 	 */
2592 	smp_mb__after_clear_bit();
2593 
2594 	return unlikely(tif_need_resched());
2595 }
2596 
2597 static inline void __current_clr_polling(void)
2598 {
2599 	clear_thread_flag(TIF_POLLING_NRFLAG);
2600 }
2601 
2602 static inline bool __must_check current_clr_polling_and_test(void)
2603 {
2604 	__current_clr_polling();
2605 
2606 	/*
2607 	 * Polling state must be visible before we test NEED_RESCHED,
2608 	 * paired by resched_task()
2609 	 */
2610 	smp_mb__after_clear_bit();
2611 
2612 	return unlikely(tif_need_resched());
2613 }
2614 
2615 #else
2616 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2617 static inline void __current_set_polling(void) { }
2618 static inline void __current_clr_polling(void) { }
2619 
2620 static inline bool __must_check current_set_polling_and_test(void)
2621 {
2622 	return unlikely(tif_need_resched());
2623 }
2624 static inline bool __must_check current_clr_polling_and_test(void)
2625 {
2626 	return unlikely(tif_need_resched());
2627 }
2628 #endif
2629 
2630 static __always_inline bool need_resched(void)
2631 {
2632 	return unlikely(tif_need_resched());
2633 }
2634 
2635 /*
2636  * Thread group CPU time accounting.
2637  */
2638 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2639 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2640 
2641 static inline void thread_group_cputime_init(struct signal_struct *sig)
2642 {
2643 	raw_spin_lock_init(&sig->cputimer.lock);
2644 }
2645 
2646 /*
2647  * Reevaluate whether the task has signals pending delivery.
2648  * Wake the task if so.
2649  * This is required every time the blocked sigset_t changes.
2650  * callers must hold sighand->siglock.
2651  */
2652 extern void recalc_sigpending_and_wake(struct task_struct *t);
2653 extern void recalc_sigpending(void);
2654 
2655 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2656 
2657 static inline void signal_wake_up(struct task_struct *t, bool resume)
2658 {
2659 	signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2660 }
2661 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2662 {
2663 	signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2664 }
2665 
2666 /*
2667  * Wrappers for p->thread_info->cpu access. No-op on UP.
2668  */
2669 #ifdef CONFIG_SMP
2670 
2671 static inline unsigned int task_cpu(const struct task_struct *p)
2672 {
2673 	return task_thread_info(p)->cpu;
2674 }
2675 
2676 static inline int task_node(const struct task_struct *p)
2677 {
2678 	return cpu_to_node(task_cpu(p));
2679 }
2680 
2681 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2682 
2683 #else
2684 
2685 static inline unsigned int task_cpu(const struct task_struct *p)
2686 {
2687 	return 0;
2688 }
2689 
2690 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2691 {
2692 }
2693 
2694 #endif /* CONFIG_SMP */
2695 
2696 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2697 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2698 
2699 #ifdef CONFIG_CGROUP_SCHED
2700 extern struct task_group root_task_group;
2701 #endif /* CONFIG_CGROUP_SCHED */
2702 
2703 extern int task_can_switch_user(struct user_struct *up,
2704 					struct task_struct *tsk);
2705 
2706 #ifdef CONFIG_TASK_XACCT
2707 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2708 {
2709 	tsk->ioac.rchar += amt;
2710 }
2711 
2712 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2713 {
2714 	tsk->ioac.wchar += amt;
2715 }
2716 
2717 static inline void inc_syscr(struct task_struct *tsk)
2718 {
2719 	tsk->ioac.syscr++;
2720 }
2721 
2722 static inline void inc_syscw(struct task_struct *tsk)
2723 {
2724 	tsk->ioac.syscw++;
2725 }
2726 #else
2727 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2728 {
2729 }
2730 
2731 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2732 {
2733 }
2734 
2735 static inline void inc_syscr(struct task_struct *tsk)
2736 {
2737 }
2738 
2739 static inline void inc_syscw(struct task_struct *tsk)
2740 {
2741 }
2742 #endif
2743 
2744 #ifndef TASK_SIZE_OF
2745 #define TASK_SIZE_OF(tsk)	TASK_SIZE
2746 #endif
2747 
2748 #ifdef CONFIG_MM_OWNER
2749 extern void mm_update_next_owner(struct mm_struct *mm);
2750 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2751 #else
2752 static inline void mm_update_next_owner(struct mm_struct *mm)
2753 {
2754 }
2755 
2756 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2757 {
2758 }
2759 #endif /* CONFIG_MM_OWNER */
2760 
2761 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2762 		unsigned int limit)
2763 {
2764 	return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2765 }
2766 
2767 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2768 		unsigned int limit)
2769 {
2770 	return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2771 }
2772 
2773 static inline unsigned long rlimit(unsigned int limit)
2774 {
2775 	return task_rlimit(current, limit);
2776 }
2777 
2778 static inline unsigned long rlimit_max(unsigned int limit)
2779 {
2780 	return task_rlimit_max(current, limit);
2781 }
2782 
2783 #endif
2784