xref: /linux-6.15/include/linux/sched.h (revision ee665ecc)
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3 
4 /*
5  * cloning flags:
6  */
7 #define CSIGNAL		0x000000ff	/* signal mask to be sent at exit */
8 #define CLONE_VM	0x00000100	/* set if VM shared between processes */
9 #define CLONE_FS	0x00000200	/* set if fs info shared between processes */
10 #define CLONE_FILES	0x00000400	/* set if open files shared between processes */
11 #define CLONE_SIGHAND	0x00000800	/* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE	0x00002000	/* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK	0x00004000	/* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT	0x00008000	/* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD	0x00010000	/* Same thread group? */
16 #define CLONE_NEWNS	0x00020000	/* New namespace group? */
17 #define CLONE_SYSVSEM	0x00040000	/* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS	0x00080000	/* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID	0x00100000	/* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID	0x00200000	/* clear the TID in the child */
21 #define CLONE_DETACHED		0x00400000	/* Unused, ignored */
22 #define CLONE_UNTRACED		0x00800000	/* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID	0x01000000	/* set the TID in the child */
24 #define CLONE_STOPPED		0x02000000	/* Start in stopped state */
25 #define CLONE_NEWUTS		0x04000000	/* New utsname group? */
26 #define CLONE_NEWIPC		0x08000000	/* New ipcs */
27 #define CLONE_NEWUSER		0x10000000	/* New user namespace */
28 #define CLONE_NEWPID		0x20000000	/* New pid namespace */
29 #define CLONE_NEWNET		0x40000000	/* New network namespace */
30 #define CLONE_IO		0x80000000	/* Clone io context */
31 
32 /*
33  * Scheduling policies
34  */
35 #define SCHED_NORMAL		0
36 #define SCHED_FIFO		1
37 #define SCHED_RR		2
38 #define SCHED_BATCH		3
39 /* SCHED_ISO: reserved but not implemented yet */
40 #define SCHED_IDLE		5
41 
42 #ifdef __KERNEL__
43 
44 struct sched_param {
45 	int sched_priority;
46 };
47 
48 #include <asm/param.h>	/* for HZ */
49 
50 #include <linux/capability.h>
51 #include <linux/threads.h>
52 #include <linux/kernel.h>
53 #include <linux/types.h>
54 #include <linux/timex.h>
55 #include <linux/jiffies.h>
56 #include <linux/rbtree.h>
57 #include <linux/thread_info.h>
58 #include <linux/cpumask.h>
59 #include <linux/errno.h>
60 #include <linux/nodemask.h>
61 #include <linux/mm_types.h>
62 
63 #include <asm/system.h>
64 #include <asm/page.h>
65 #include <asm/ptrace.h>
66 #include <asm/cputime.h>
67 
68 #include <linux/smp.h>
69 #include <linux/sem.h>
70 #include <linux/signal.h>
71 #include <linux/fs_struct.h>
72 #include <linux/compiler.h>
73 #include <linux/completion.h>
74 #include <linux/pid.h>
75 #include <linux/percpu.h>
76 #include <linux/topology.h>
77 #include <linux/proportions.h>
78 #include <linux/seccomp.h>
79 #include <linux/rcupdate.h>
80 #include <linux/rtmutex.h>
81 
82 #include <linux/time.h>
83 #include <linux/param.h>
84 #include <linux/resource.h>
85 #include <linux/timer.h>
86 #include <linux/hrtimer.h>
87 #include <linux/task_io_accounting.h>
88 #include <linux/kobject.h>
89 #include <linux/latencytop.h>
90 #include <linux/cred.h>
91 
92 #include <asm/processor.h>
93 
94 struct mem_cgroup;
95 struct exec_domain;
96 struct futex_pi_state;
97 struct robust_list_head;
98 struct bio;
99 struct bts_tracer;
100 
101 /*
102  * List of flags we want to share for kernel threads,
103  * if only because they are not used by them anyway.
104  */
105 #define CLONE_KERNEL	(CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
106 
107 /*
108  * These are the constant used to fake the fixed-point load-average
109  * counting. Some notes:
110  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
111  *    a load-average precision of 10 bits integer + 11 bits fractional
112  *  - if you want to count load-averages more often, you need more
113  *    precision, or rounding will get you. With 2-second counting freq,
114  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
115  *    11 bit fractions.
116  */
117 extern unsigned long avenrun[];		/* Load averages */
118 
119 #define FSHIFT		11		/* nr of bits of precision */
120 #define FIXED_1		(1<<FSHIFT)	/* 1.0 as fixed-point */
121 #define LOAD_FREQ	(5*HZ+1)	/* 5 sec intervals */
122 #define EXP_1		1884		/* 1/exp(5sec/1min) as fixed-point */
123 #define EXP_5		2014		/* 1/exp(5sec/5min) */
124 #define EXP_15		2037		/* 1/exp(5sec/15min) */
125 
126 #define CALC_LOAD(load,exp,n) \
127 	load *= exp; \
128 	load += n*(FIXED_1-exp); \
129 	load >>= FSHIFT;
130 
131 extern unsigned long total_forks;
132 extern int nr_threads;
133 DECLARE_PER_CPU(unsigned long, process_counts);
134 extern int nr_processes(void);
135 extern unsigned long nr_running(void);
136 extern unsigned long nr_uninterruptible(void);
137 extern unsigned long nr_active(void);
138 extern unsigned long nr_iowait(void);
139 
140 struct seq_file;
141 struct cfs_rq;
142 struct task_group;
143 #ifdef CONFIG_SCHED_DEBUG
144 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
145 extern void proc_sched_set_task(struct task_struct *p);
146 extern void
147 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
148 #else
149 static inline void
150 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
151 {
152 }
153 static inline void proc_sched_set_task(struct task_struct *p)
154 {
155 }
156 static inline void
157 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
158 {
159 }
160 #endif
161 
162 extern unsigned long long time_sync_thresh;
163 
164 /*
165  * Task state bitmask. NOTE! These bits are also
166  * encoded in fs/proc/array.c: get_task_state().
167  *
168  * We have two separate sets of flags: task->state
169  * is about runnability, while task->exit_state are
170  * about the task exiting. Confusing, but this way
171  * modifying one set can't modify the other one by
172  * mistake.
173  */
174 #define TASK_RUNNING		0
175 #define TASK_INTERRUPTIBLE	1
176 #define TASK_UNINTERRUPTIBLE	2
177 #define __TASK_STOPPED		4
178 #define __TASK_TRACED		8
179 /* in tsk->exit_state */
180 #define EXIT_ZOMBIE		16
181 #define EXIT_DEAD		32
182 /* in tsk->state again */
183 #define TASK_DEAD		64
184 #define TASK_WAKEKILL		128
185 
186 /* Convenience macros for the sake of set_task_state */
187 #define TASK_KILLABLE		(TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
188 #define TASK_STOPPED		(TASK_WAKEKILL | __TASK_STOPPED)
189 #define TASK_TRACED		(TASK_WAKEKILL | __TASK_TRACED)
190 
191 /* Convenience macros for the sake of wake_up */
192 #define TASK_NORMAL		(TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
193 #define TASK_ALL		(TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
194 
195 /* get_task_state() */
196 #define TASK_REPORT		(TASK_RUNNING | TASK_INTERRUPTIBLE | \
197 				 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
198 				 __TASK_TRACED)
199 
200 #define task_is_traced(task)	((task->state & __TASK_TRACED) != 0)
201 #define task_is_stopped(task)	((task->state & __TASK_STOPPED) != 0)
202 #define task_is_stopped_or_traced(task)	\
203 			((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
204 #define task_contributes_to_load(task)	\
205 				((task->state & TASK_UNINTERRUPTIBLE) != 0)
206 
207 #define __set_task_state(tsk, state_value)		\
208 	do { (tsk)->state = (state_value); } while (0)
209 #define set_task_state(tsk, state_value)		\
210 	set_mb((tsk)->state, (state_value))
211 
212 /*
213  * set_current_state() includes a barrier so that the write of current->state
214  * is correctly serialised wrt the caller's subsequent test of whether to
215  * actually sleep:
216  *
217  *	set_current_state(TASK_UNINTERRUPTIBLE);
218  *	if (do_i_need_to_sleep())
219  *		schedule();
220  *
221  * If the caller does not need such serialisation then use __set_current_state()
222  */
223 #define __set_current_state(state_value)			\
224 	do { current->state = (state_value); } while (0)
225 #define set_current_state(state_value)		\
226 	set_mb(current->state, (state_value))
227 
228 /* Task command name length */
229 #define TASK_COMM_LEN 16
230 
231 #include <linux/spinlock.h>
232 
233 /*
234  * This serializes "schedule()" and also protects
235  * the run-queue from deletions/modifications (but
236  * _adding_ to the beginning of the run-queue has
237  * a separate lock).
238  */
239 extern rwlock_t tasklist_lock;
240 extern spinlock_t mmlist_lock;
241 
242 struct task_struct;
243 
244 extern void sched_init(void);
245 extern void sched_init_smp(void);
246 extern asmlinkage void schedule_tail(struct task_struct *prev);
247 extern void init_idle(struct task_struct *idle, int cpu);
248 extern void init_idle_bootup_task(struct task_struct *idle);
249 
250 extern int runqueue_is_locked(void);
251 extern void task_rq_unlock_wait(struct task_struct *p);
252 
253 extern cpumask_var_t nohz_cpu_mask;
254 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
255 extern int select_nohz_load_balancer(int cpu);
256 #else
257 static inline int select_nohz_load_balancer(int cpu)
258 {
259 	return 0;
260 }
261 #endif
262 
263 /*
264  * Only dump TASK_* tasks. (0 for all tasks)
265  */
266 extern void show_state_filter(unsigned long state_filter);
267 
268 static inline void show_state(void)
269 {
270 	show_state_filter(0);
271 }
272 
273 extern void show_regs(struct pt_regs *);
274 
275 /*
276  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
277  * task), SP is the stack pointer of the first frame that should be shown in the back
278  * trace (or NULL if the entire call-chain of the task should be shown).
279  */
280 extern void show_stack(struct task_struct *task, unsigned long *sp);
281 
282 void io_schedule(void);
283 long io_schedule_timeout(long timeout);
284 
285 extern void cpu_init (void);
286 extern void trap_init(void);
287 extern void update_process_times(int user);
288 extern void scheduler_tick(void);
289 
290 extern void sched_show_task(struct task_struct *p);
291 
292 #ifdef CONFIG_DETECT_SOFTLOCKUP
293 extern void softlockup_tick(void);
294 extern void touch_softlockup_watchdog(void);
295 extern void touch_all_softlockup_watchdogs(void);
296 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
297 				    struct file *filp, void __user *buffer,
298 				    size_t *lenp, loff_t *ppos);
299 extern unsigned int  softlockup_panic;
300 extern unsigned long sysctl_hung_task_check_count;
301 extern unsigned long sysctl_hung_task_timeout_secs;
302 extern unsigned long sysctl_hung_task_warnings;
303 extern int softlockup_thresh;
304 #else
305 static inline void softlockup_tick(void)
306 {
307 }
308 static inline void spawn_softlockup_task(void)
309 {
310 }
311 static inline void touch_softlockup_watchdog(void)
312 {
313 }
314 static inline void touch_all_softlockup_watchdogs(void)
315 {
316 }
317 #endif
318 
319 
320 /* Attach to any functions which should be ignored in wchan output. */
321 #define __sched		__attribute__((__section__(".sched.text")))
322 
323 /* Linker adds these: start and end of __sched functions */
324 extern char __sched_text_start[], __sched_text_end[];
325 
326 /* Is this address in the __sched functions? */
327 extern int in_sched_functions(unsigned long addr);
328 
329 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
330 extern signed long schedule_timeout(signed long timeout);
331 extern signed long schedule_timeout_interruptible(signed long timeout);
332 extern signed long schedule_timeout_killable(signed long timeout);
333 extern signed long schedule_timeout_uninterruptible(signed long timeout);
334 asmlinkage void schedule(void);
335 
336 struct nsproxy;
337 struct user_namespace;
338 
339 /* Maximum number of active map areas.. This is a random (large) number */
340 #define DEFAULT_MAX_MAP_COUNT	65536
341 
342 extern int sysctl_max_map_count;
343 
344 #include <linux/aio.h>
345 
346 extern unsigned long
347 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
348 		       unsigned long, unsigned long);
349 extern unsigned long
350 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
351 			  unsigned long len, unsigned long pgoff,
352 			  unsigned long flags);
353 extern void arch_unmap_area(struct mm_struct *, unsigned long);
354 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
355 
356 #if USE_SPLIT_PTLOCKS
357 /*
358  * The mm counters are not protected by its page_table_lock,
359  * so must be incremented atomically.
360  */
361 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
362 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
363 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
364 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
365 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
366 
367 #else  /* !USE_SPLIT_PTLOCKS */
368 /*
369  * The mm counters are protected by its page_table_lock,
370  * so can be incremented directly.
371  */
372 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
373 #define get_mm_counter(mm, member) ((mm)->_##member)
374 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
375 #define inc_mm_counter(mm, member) (mm)->_##member++
376 #define dec_mm_counter(mm, member) (mm)->_##member--
377 
378 #endif /* !USE_SPLIT_PTLOCKS */
379 
380 #define get_mm_rss(mm)					\
381 	(get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
382 #define update_hiwater_rss(mm)	do {			\
383 	unsigned long _rss = get_mm_rss(mm);		\
384 	if ((mm)->hiwater_rss < _rss)			\
385 		(mm)->hiwater_rss = _rss;		\
386 } while (0)
387 #define update_hiwater_vm(mm)	do {			\
388 	if ((mm)->hiwater_vm < (mm)->total_vm)		\
389 		(mm)->hiwater_vm = (mm)->total_vm;	\
390 } while (0)
391 
392 #define get_mm_hiwater_rss(mm)	max((mm)->hiwater_rss, get_mm_rss(mm))
393 #define get_mm_hiwater_vm(mm)	max((mm)->hiwater_vm, (mm)->total_vm)
394 
395 extern void set_dumpable(struct mm_struct *mm, int value);
396 extern int get_dumpable(struct mm_struct *mm);
397 
398 /* mm flags */
399 /* dumpable bits */
400 #define MMF_DUMPABLE      0  /* core dump is permitted */
401 #define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
402 #define MMF_DUMPABLE_BITS 2
403 
404 /* coredump filter bits */
405 #define MMF_DUMP_ANON_PRIVATE	2
406 #define MMF_DUMP_ANON_SHARED	3
407 #define MMF_DUMP_MAPPED_PRIVATE	4
408 #define MMF_DUMP_MAPPED_SHARED	5
409 #define MMF_DUMP_ELF_HEADERS	6
410 #define MMF_DUMP_HUGETLB_PRIVATE 7
411 #define MMF_DUMP_HUGETLB_SHARED  8
412 #define MMF_DUMP_FILTER_SHIFT	MMF_DUMPABLE_BITS
413 #define MMF_DUMP_FILTER_BITS	7
414 #define MMF_DUMP_FILTER_MASK \
415 	(((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
416 #define MMF_DUMP_FILTER_DEFAULT \
417 	((1 << MMF_DUMP_ANON_PRIVATE) |	(1 << MMF_DUMP_ANON_SHARED) |\
418 	 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
419 
420 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
421 # define MMF_DUMP_MASK_DEFAULT_ELF	(1 << MMF_DUMP_ELF_HEADERS)
422 #else
423 # define MMF_DUMP_MASK_DEFAULT_ELF	0
424 #endif
425 
426 struct sighand_struct {
427 	atomic_t		count;
428 	struct k_sigaction	action[_NSIG];
429 	spinlock_t		siglock;
430 	wait_queue_head_t	signalfd_wqh;
431 };
432 
433 struct pacct_struct {
434 	int			ac_flag;
435 	long			ac_exitcode;
436 	unsigned long		ac_mem;
437 	cputime_t		ac_utime, ac_stime;
438 	unsigned long		ac_minflt, ac_majflt;
439 };
440 
441 /**
442  * struct task_cputime - collected CPU time counts
443  * @utime:		time spent in user mode, in &cputime_t units
444  * @stime:		time spent in kernel mode, in &cputime_t units
445  * @sum_exec_runtime:	total time spent on the CPU, in nanoseconds
446  *
447  * This structure groups together three kinds of CPU time that are
448  * tracked for threads and thread groups.  Most things considering
449  * CPU time want to group these counts together and treat all three
450  * of them in parallel.
451  */
452 struct task_cputime {
453 	cputime_t utime;
454 	cputime_t stime;
455 	unsigned long long sum_exec_runtime;
456 };
457 /* Alternate field names when used to cache expirations. */
458 #define prof_exp	stime
459 #define virt_exp	utime
460 #define sched_exp	sum_exec_runtime
461 
462 #define INIT_CPUTIME	\
463 	(struct task_cputime) {					\
464 		.utime = cputime_zero,				\
465 		.stime = cputime_zero,				\
466 		.sum_exec_runtime = 0,				\
467 	}
468 
469 /**
470  * struct thread_group_cputimer - thread group interval timer counts
471  * @cputime:		thread group interval timers.
472  * @running:		non-zero when there are timers running and
473  * 			@cputime receives updates.
474  * @lock:		lock for fields in this struct.
475  *
476  * This structure contains the version of task_cputime, above, that is
477  * used for thread group CPU timer calculations.
478  */
479 struct thread_group_cputimer {
480 	struct task_cputime cputime;
481 	int running;
482 	spinlock_t lock;
483 };
484 
485 /*
486  * NOTE! "signal_struct" does not have it's own
487  * locking, because a shared signal_struct always
488  * implies a shared sighand_struct, so locking
489  * sighand_struct is always a proper superset of
490  * the locking of signal_struct.
491  */
492 struct signal_struct {
493 	atomic_t		count;
494 	atomic_t		live;
495 
496 	wait_queue_head_t	wait_chldexit;	/* for wait4() */
497 
498 	/* current thread group signal load-balancing target: */
499 	struct task_struct	*curr_target;
500 
501 	/* shared signal handling: */
502 	struct sigpending	shared_pending;
503 
504 	/* thread group exit support */
505 	int			group_exit_code;
506 	/* overloaded:
507 	 * - notify group_exit_task when ->count is equal to notify_count
508 	 * - everyone except group_exit_task is stopped during signal delivery
509 	 *   of fatal signals, group_exit_task processes the signal.
510 	 */
511 	int			notify_count;
512 	struct task_struct	*group_exit_task;
513 
514 	/* thread group stop support, overloads group_exit_code too */
515 	int			group_stop_count;
516 	unsigned int		flags; /* see SIGNAL_* flags below */
517 
518 	/* POSIX.1b Interval Timers */
519 	struct list_head posix_timers;
520 
521 	/* ITIMER_REAL timer for the process */
522 	struct hrtimer real_timer;
523 	struct pid *leader_pid;
524 	ktime_t it_real_incr;
525 
526 	/* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
527 	cputime_t it_prof_expires, it_virt_expires;
528 	cputime_t it_prof_incr, it_virt_incr;
529 
530 	/*
531 	 * Thread group totals for process CPU timers.
532 	 * See thread_group_cputimer(), et al, for details.
533 	 */
534 	struct thread_group_cputimer cputimer;
535 
536 	/* Earliest-expiration cache. */
537 	struct task_cputime cputime_expires;
538 
539 	struct list_head cpu_timers[3];
540 
541 	/* job control IDs */
542 
543 	/*
544 	 * pgrp and session fields are deprecated.
545 	 * use the task_session_Xnr and task_pgrp_Xnr routines below
546 	 */
547 
548 	union {
549 		pid_t pgrp __deprecated;
550 		pid_t __pgrp;
551 	};
552 
553 	struct pid *tty_old_pgrp;
554 
555 	union {
556 		pid_t session __deprecated;
557 		pid_t __session;
558 	};
559 
560 	/* boolean value for session group leader */
561 	int leader;
562 
563 	struct tty_struct *tty; /* NULL if no tty */
564 
565 	/*
566 	 * Cumulative resource counters for dead threads in the group,
567 	 * and for reaped dead child processes forked by this group.
568 	 * Live threads maintain their own counters and add to these
569 	 * in __exit_signal, except for the group leader.
570 	 */
571 	cputime_t utime, stime, cutime, cstime;
572 	cputime_t gtime;
573 	cputime_t cgtime;
574 	unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
575 	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
576 	unsigned long inblock, oublock, cinblock, coublock;
577 	struct task_io_accounting ioac;
578 
579 	/*
580 	 * Cumulative ns of schedule CPU time fo dead threads in the
581 	 * group, not including a zombie group leader, (This only differs
582 	 * from jiffies_to_ns(utime + stime) if sched_clock uses something
583 	 * other than jiffies.)
584 	 */
585 	unsigned long long sum_sched_runtime;
586 
587 	/*
588 	 * We don't bother to synchronize most readers of this at all,
589 	 * because there is no reader checking a limit that actually needs
590 	 * to get both rlim_cur and rlim_max atomically, and either one
591 	 * alone is a single word that can safely be read normally.
592 	 * getrlimit/setrlimit use task_lock(current->group_leader) to
593 	 * protect this instead of the siglock, because they really
594 	 * have no need to disable irqs.
595 	 */
596 	struct rlimit rlim[RLIM_NLIMITS];
597 
598 #ifdef CONFIG_BSD_PROCESS_ACCT
599 	struct pacct_struct pacct;	/* per-process accounting information */
600 #endif
601 #ifdef CONFIG_TASKSTATS
602 	struct taskstats *stats;
603 #endif
604 #ifdef CONFIG_AUDIT
605 	unsigned audit_tty;
606 	struct tty_audit_buf *tty_audit_buf;
607 #endif
608 };
609 
610 /* Context switch must be unlocked if interrupts are to be enabled */
611 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
612 # define __ARCH_WANT_UNLOCKED_CTXSW
613 #endif
614 
615 /*
616  * Bits in flags field of signal_struct.
617  */
618 #define SIGNAL_STOP_STOPPED	0x00000001 /* job control stop in effect */
619 #define SIGNAL_STOP_DEQUEUED	0x00000002 /* stop signal dequeued */
620 #define SIGNAL_STOP_CONTINUED	0x00000004 /* SIGCONT since WCONTINUED reap */
621 #define SIGNAL_GROUP_EXIT	0x00000008 /* group exit in progress */
622 /*
623  * Pending notifications to parent.
624  */
625 #define SIGNAL_CLD_STOPPED	0x00000010
626 #define SIGNAL_CLD_CONTINUED	0x00000020
627 #define SIGNAL_CLD_MASK		(SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
628 
629 #define SIGNAL_UNKILLABLE	0x00000040 /* for init: ignore fatal signals */
630 
631 /* If true, all threads except ->group_exit_task have pending SIGKILL */
632 static inline int signal_group_exit(const struct signal_struct *sig)
633 {
634 	return	(sig->flags & SIGNAL_GROUP_EXIT) ||
635 		(sig->group_exit_task != NULL);
636 }
637 
638 /*
639  * Some day this will be a full-fledged user tracking system..
640  */
641 struct user_struct {
642 	atomic_t __count;	/* reference count */
643 	atomic_t processes;	/* How many processes does this user have? */
644 	atomic_t files;		/* How many open files does this user have? */
645 	atomic_t sigpending;	/* How many pending signals does this user have? */
646 #ifdef CONFIG_INOTIFY_USER
647 	atomic_t inotify_watches; /* How many inotify watches does this user have? */
648 	atomic_t inotify_devs;	/* How many inotify devs does this user have opened? */
649 #endif
650 #ifdef CONFIG_EPOLL
651 	atomic_t epoll_watches;	/* The number of file descriptors currently watched */
652 #endif
653 #ifdef CONFIG_POSIX_MQUEUE
654 	/* protected by mq_lock	*/
655 	unsigned long mq_bytes;	/* How many bytes can be allocated to mqueue? */
656 #endif
657 	unsigned long locked_shm; /* How many pages of mlocked shm ? */
658 
659 #ifdef CONFIG_KEYS
660 	struct key *uid_keyring;	/* UID specific keyring */
661 	struct key *session_keyring;	/* UID's default session keyring */
662 #endif
663 
664 	/* Hash table maintenance information */
665 	struct hlist_node uidhash_node;
666 	uid_t uid;
667 	struct user_namespace *user_ns;
668 
669 #ifdef CONFIG_USER_SCHED
670 	struct task_group *tg;
671 #ifdef CONFIG_SYSFS
672 	struct kobject kobj;
673 	struct work_struct work;
674 #endif
675 #endif
676 };
677 
678 extern int uids_sysfs_init(void);
679 
680 extern struct user_struct *find_user(uid_t);
681 
682 extern struct user_struct root_user;
683 #define INIT_USER (&root_user)
684 
685 
686 struct backing_dev_info;
687 struct reclaim_state;
688 
689 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
690 struct sched_info {
691 	/* cumulative counters */
692 	unsigned long pcount;	      /* # of times run on this cpu */
693 	unsigned long long run_delay; /* time spent waiting on a runqueue */
694 
695 	/* timestamps */
696 	unsigned long long last_arrival,/* when we last ran on a cpu */
697 			   last_queued;	/* when we were last queued to run */
698 #ifdef CONFIG_SCHEDSTATS
699 	/* BKL stats */
700 	unsigned int bkl_count;
701 #endif
702 };
703 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
704 
705 #ifdef CONFIG_TASK_DELAY_ACCT
706 struct task_delay_info {
707 	spinlock_t	lock;
708 	unsigned int	flags;	/* Private per-task flags */
709 
710 	/* For each stat XXX, add following, aligned appropriately
711 	 *
712 	 * struct timespec XXX_start, XXX_end;
713 	 * u64 XXX_delay;
714 	 * u32 XXX_count;
715 	 *
716 	 * Atomicity of updates to XXX_delay, XXX_count protected by
717 	 * single lock above (split into XXX_lock if contention is an issue).
718 	 */
719 
720 	/*
721 	 * XXX_count is incremented on every XXX operation, the delay
722 	 * associated with the operation is added to XXX_delay.
723 	 * XXX_delay contains the accumulated delay time in nanoseconds.
724 	 */
725 	struct timespec blkio_start, blkio_end;	/* Shared by blkio, swapin */
726 	u64 blkio_delay;	/* wait for sync block io completion */
727 	u64 swapin_delay;	/* wait for swapin block io completion */
728 	u32 blkio_count;	/* total count of the number of sync block */
729 				/* io operations performed */
730 	u32 swapin_count;	/* total count of the number of swapin block */
731 				/* io operations performed */
732 
733 	struct timespec freepages_start, freepages_end;
734 	u64 freepages_delay;	/* wait for memory reclaim */
735 	u32 freepages_count;	/* total count of memory reclaim */
736 };
737 #endif	/* CONFIG_TASK_DELAY_ACCT */
738 
739 static inline int sched_info_on(void)
740 {
741 #ifdef CONFIG_SCHEDSTATS
742 	return 1;
743 #elif defined(CONFIG_TASK_DELAY_ACCT)
744 	extern int delayacct_on;
745 	return delayacct_on;
746 #else
747 	return 0;
748 #endif
749 }
750 
751 enum cpu_idle_type {
752 	CPU_IDLE,
753 	CPU_NOT_IDLE,
754 	CPU_NEWLY_IDLE,
755 	CPU_MAX_IDLE_TYPES
756 };
757 
758 /*
759  * sched-domains (multiprocessor balancing) declarations:
760  */
761 
762 /*
763  * Increase resolution of nice-level calculations:
764  */
765 #define SCHED_LOAD_SHIFT	10
766 #define SCHED_LOAD_SCALE	(1L << SCHED_LOAD_SHIFT)
767 
768 #define SCHED_LOAD_SCALE_FUZZ	SCHED_LOAD_SCALE
769 
770 #ifdef CONFIG_SMP
771 #define SD_LOAD_BALANCE		1	/* Do load balancing on this domain. */
772 #define SD_BALANCE_NEWIDLE	2	/* Balance when about to become idle */
773 #define SD_BALANCE_EXEC		4	/* Balance on exec */
774 #define SD_BALANCE_FORK		8	/* Balance on fork, clone */
775 #define SD_WAKE_IDLE		16	/* Wake to idle CPU on task wakeup */
776 #define SD_WAKE_AFFINE		32	/* Wake task to waking CPU */
777 #define SD_WAKE_BALANCE		64	/* Perform balancing at task wakeup */
778 #define SD_SHARE_CPUPOWER	128	/* Domain members share cpu power */
779 #define SD_POWERSAVINGS_BALANCE	256	/* Balance for power savings */
780 #define SD_SHARE_PKG_RESOURCES	512	/* Domain members share cpu pkg resources */
781 #define SD_SERIALIZE		1024	/* Only a single load balancing instance */
782 #define SD_WAKE_IDLE_FAR	2048	/* Gain latency sacrificing cache hit */
783 
784 enum powersavings_balance_level {
785 	POWERSAVINGS_BALANCE_NONE = 0,  /* No power saving load balance */
786 	POWERSAVINGS_BALANCE_BASIC,	/* Fill one thread/core/package
787 					 * first for long running threads
788 					 */
789 	POWERSAVINGS_BALANCE_WAKEUP,	/* Also bias task wakeups to semi-idle
790 					 * cpu package for power savings
791 					 */
792 	MAX_POWERSAVINGS_BALANCE_LEVELS
793 };
794 
795 extern int sched_mc_power_savings, sched_smt_power_savings;
796 
797 static inline int sd_balance_for_mc_power(void)
798 {
799 	if (sched_smt_power_savings)
800 		return SD_POWERSAVINGS_BALANCE;
801 
802 	return 0;
803 }
804 
805 static inline int sd_balance_for_package_power(void)
806 {
807 	if (sched_mc_power_savings | sched_smt_power_savings)
808 		return SD_POWERSAVINGS_BALANCE;
809 
810 	return 0;
811 }
812 
813 /*
814  * Optimise SD flags for power savings:
815  * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
816  * Keep default SD flags if sched_{smt,mc}_power_saving=0
817  */
818 
819 static inline int sd_power_saving_flags(void)
820 {
821 	if (sched_mc_power_savings | sched_smt_power_savings)
822 		return SD_BALANCE_NEWIDLE;
823 
824 	return 0;
825 }
826 
827 struct sched_group {
828 	struct sched_group *next;	/* Must be a circular list */
829 
830 	/*
831 	 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
832 	 * single CPU. This is read only (except for setup, hotplug CPU).
833 	 * Note : Never change cpu_power without recompute its reciprocal
834 	 */
835 	unsigned int __cpu_power;
836 	/*
837 	 * reciprocal value of cpu_power to avoid expensive divides
838 	 * (see include/linux/reciprocal_div.h)
839 	 */
840 	u32 reciprocal_cpu_power;
841 
842 	unsigned long cpumask[];
843 };
844 
845 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
846 {
847 	return to_cpumask(sg->cpumask);
848 }
849 
850 enum sched_domain_level {
851 	SD_LV_NONE = 0,
852 	SD_LV_SIBLING,
853 	SD_LV_MC,
854 	SD_LV_CPU,
855 	SD_LV_NODE,
856 	SD_LV_ALLNODES,
857 	SD_LV_MAX
858 };
859 
860 struct sched_domain_attr {
861 	int relax_domain_level;
862 };
863 
864 #define SD_ATTR_INIT	(struct sched_domain_attr) {	\
865 	.relax_domain_level = -1,			\
866 }
867 
868 struct sched_domain {
869 	/* These fields must be setup */
870 	struct sched_domain *parent;	/* top domain must be null terminated */
871 	struct sched_domain *child;	/* bottom domain must be null terminated */
872 	struct sched_group *groups;	/* the balancing groups of the domain */
873 	unsigned long min_interval;	/* Minimum balance interval ms */
874 	unsigned long max_interval;	/* Maximum balance interval ms */
875 	unsigned int busy_factor;	/* less balancing by factor if busy */
876 	unsigned int imbalance_pct;	/* No balance until over watermark */
877 	unsigned int cache_nice_tries;	/* Leave cache hot tasks for # tries */
878 	unsigned int busy_idx;
879 	unsigned int idle_idx;
880 	unsigned int newidle_idx;
881 	unsigned int wake_idx;
882 	unsigned int forkexec_idx;
883 	int flags;			/* See SD_* */
884 	enum sched_domain_level level;
885 
886 	/* Runtime fields. */
887 	unsigned long last_balance;	/* init to jiffies. units in jiffies */
888 	unsigned int balance_interval;	/* initialise to 1. units in ms. */
889 	unsigned int nr_balance_failed; /* initialise to 0 */
890 
891 	u64 last_update;
892 
893 #ifdef CONFIG_SCHEDSTATS
894 	/* load_balance() stats */
895 	unsigned int lb_count[CPU_MAX_IDLE_TYPES];
896 	unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
897 	unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
898 	unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
899 	unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
900 	unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
901 	unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
902 	unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
903 
904 	/* Active load balancing */
905 	unsigned int alb_count;
906 	unsigned int alb_failed;
907 	unsigned int alb_pushed;
908 
909 	/* SD_BALANCE_EXEC stats */
910 	unsigned int sbe_count;
911 	unsigned int sbe_balanced;
912 	unsigned int sbe_pushed;
913 
914 	/* SD_BALANCE_FORK stats */
915 	unsigned int sbf_count;
916 	unsigned int sbf_balanced;
917 	unsigned int sbf_pushed;
918 
919 	/* try_to_wake_up() stats */
920 	unsigned int ttwu_wake_remote;
921 	unsigned int ttwu_move_affine;
922 	unsigned int ttwu_move_balance;
923 #endif
924 #ifdef CONFIG_SCHED_DEBUG
925 	char *name;
926 #endif
927 
928 	/* span of all CPUs in this domain */
929 	unsigned long span[];
930 };
931 
932 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
933 {
934 	return to_cpumask(sd->span);
935 }
936 
937 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
938 				    struct sched_domain_attr *dattr_new);
939 
940 /* Test a flag in parent sched domain */
941 static inline int test_sd_parent(struct sched_domain *sd, int flag)
942 {
943 	if (sd->parent && (sd->parent->flags & flag))
944 		return 1;
945 
946 	return 0;
947 }
948 
949 #else /* CONFIG_SMP */
950 
951 struct sched_domain_attr;
952 
953 static inline void
954 partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
955 			struct sched_domain_attr *dattr_new)
956 {
957 }
958 #endif	/* !CONFIG_SMP */
959 
960 struct io_context;			/* See blkdev.h */
961 
962 
963 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
964 extern void prefetch_stack(struct task_struct *t);
965 #else
966 static inline void prefetch_stack(struct task_struct *t) { }
967 #endif
968 
969 struct audit_context;		/* See audit.c */
970 struct mempolicy;
971 struct pipe_inode_info;
972 struct uts_namespace;
973 
974 struct rq;
975 struct sched_domain;
976 
977 struct sched_class {
978 	const struct sched_class *next;
979 
980 	void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
981 	void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
982 	void (*yield_task) (struct rq *rq);
983 
984 	void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync);
985 
986 	struct task_struct * (*pick_next_task) (struct rq *rq);
987 	void (*put_prev_task) (struct rq *rq, struct task_struct *p);
988 
989 #ifdef CONFIG_SMP
990 	int  (*select_task_rq)(struct task_struct *p, int sync);
991 
992 	unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
993 			struct rq *busiest, unsigned long max_load_move,
994 			struct sched_domain *sd, enum cpu_idle_type idle,
995 			int *all_pinned, int *this_best_prio);
996 
997 	int (*move_one_task) (struct rq *this_rq, int this_cpu,
998 			      struct rq *busiest, struct sched_domain *sd,
999 			      enum cpu_idle_type idle);
1000 	void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1001 	int (*needs_post_schedule) (struct rq *this_rq);
1002 	void (*post_schedule) (struct rq *this_rq);
1003 	void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
1004 
1005 	void (*set_cpus_allowed)(struct task_struct *p,
1006 				 const struct cpumask *newmask);
1007 
1008 	void (*rq_online)(struct rq *rq);
1009 	void (*rq_offline)(struct rq *rq);
1010 #endif
1011 
1012 	void (*set_curr_task) (struct rq *rq);
1013 	void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1014 	void (*task_new) (struct rq *rq, struct task_struct *p);
1015 
1016 	void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1017 			       int running);
1018 	void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1019 			     int running);
1020 	void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1021 			     int oldprio, int running);
1022 
1023 #ifdef CONFIG_FAIR_GROUP_SCHED
1024 	void (*moved_group) (struct task_struct *p);
1025 #endif
1026 };
1027 
1028 struct load_weight {
1029 	unsigned long weight, inv_weight;
1030 };
1031 
1032 /*
1033  * CFS stats for a schedulable entity (task, task-group etc)
1034  *
1035  * Current field usage histogram:
1036  *
1037  *     4 se->block_start
1038  *     4 se->run_node
1039  *     4 se->sleep_start
1040  *     6 se->load.weight
1041  */
1042 struct sched_entity {
1043 	struct load_weight	load;		/* for load-balancing */
1044 	struct rb_node		run_node;
1045 	struct list_head	group_node;
1046 	unsigned int		on_rq;
1047 
1048 	u64			exec_start;
1049 	u64			sum_exec_runtime;
1050 	u64			vruntime;
1051 	u64			prev_sum_exec_runtime;
1052 
1053 	u64			last_wakeup;
1054 	u64			avg_overlap;
1055 
1056 	u64			start_runtime;
1057 	u64			avg_wakeup;
1058 	u64			nr_migrations;
1059 
1060 #ifdef CONFIG_SCHEDSTATS
1061 	u64			wait_start;
1062 	u64			wait_max;
1063 	u64			wait_count;
1064 	u64			wait_sum;
1065 
1066 	u64			sleep_start;
1067 	u64			sleep_max;
1068 	s64			sum_sleep_runtime;
1069 
1070 	u64			block_start;
1071 	u64			block_max;
1072 	u64			exec_max;
1073 	u64			slice_max;
1074 
1075 	u64			nr_migrations_cold;
1076 	u64			nr_failed_migrations_affine;
1077 	u64			nr_failed_migrations_running;
1078 	u64			nr_failed_migrations_hot;
1079 	u64			nr_forced_migrations;
1080 	u64			nr_forced2_migrations;
1081 
1082 	u64			nr_wakeups;
1083 	u64			nr_wakeups_sync;
1084 	u64			nr_wakeups_migrate;
1085 	u64			nr_wakeups_local;
1086 	u64			nr_wakeups_remote;
1087 	u64			nr_wakeups_affine;
1088 	u64			nr_wakeups_affine_attempts;
1089 	u64			nr_wakeups_passive;
1090 	u64			nr_wakeups_idle;
1091 #endif
1092 
1093 #ifdef CONFIG_FAIR_GROUP_SCHED
1094 	struct sched_entity	*parent;
1095 	/* rq on which this entity is (to be) queued: */
1096 	struct cfs_rq		*cfs_rq;
1097 	/* rq "owned" by this entity/group: */
1098 	struct cfs_rq		*my_q;
1099 #endif
1100 };
1101 
1102 struct sched_rt_entity {
1103 	struct list_head run_list;
1104 	unsigned long timeout;
1105 	unsigned int time_slice;
1106 	int nr_cpus_allowed;
1107 
1108 	struct sched_rt_entity *back;
1109 #ifdef CONFIG_RT_GROUP_SCHED
1110 	struct sched_rt_entity	*parent;
1111 	/* rq on which this entity is (to be) queued: */
1112 	struct rt_rq		*rt_rq;
1113 	/* rq "owned" by this entity/group: */
1114 	struct rt_rq		*my_q;
1115 #endif
1116 };
1117 
1118 struct task_struct {
1119 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
1120 	void *stack;
1121 	atomic_t usage;
1122 	unsigned int flags;	/* per process flags, defined below */
1123 	unsigned int ptrace;
1124 
1125 	int lock_depth;		/* BKL lock depth */
1126 
1127 #ifdef CONFIG_SMP
1128 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1129 	int oncpu;
1130 #endif
1131 #endif
1132 
1133 	int prio, static_prio, normal_prio;
1134 	unsigned int rt_priority;
1135 	const struct sched_class *sched_class;
1136 	struct sched_entity se;
1137 	struct sched_rt_entity rt;
1138 
1139 #ifdef CONFIG_PREEMPT_NOTIFIERS
1140 	/* list of struct preempt_notifier: */
1141 	struct hlist_head preempt_notifiers;
1142 #endif
1143 
1144 	/*
1145 	 * fpu_counter contains the number of consecutive context switches
1146 	 * that the FPU is used. If this is over a threshold, the lazy fpu
1147 	 * saving becomes unlazy to save the trap. This is an unsigned char
1148 	 * so that after 256 times the counter wraps and the behavior turns
1149 	 * lazy again; this to deal with bursty apps that only use FPU for
1150 	 * a short time
1151 	 */
1152 	unsigned char fpu_counter;
1153 	s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1154 #ifdef CONFIG_BLK_DEV_IO_TRACE
1155 	unsigned int btrace_seq;
1156 #endif
1157 
1158 	unsigned int policy;
1159 	cpumask_t cpus_allowed;
1160 
1161 #ifdef CONFIG_PREEMPT_RCU
1162 	int rcu_read_lock_nesting;
1163 	int rcu_flipctr_idx;
1164 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1165 
1166 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1167 	struct sched_info sched_info;
1168 #endif
1169 
1170 	struct list_head tasks;
1171 	struct plist_node pushable_tasks;
1172 
1173 	struct mm_struct *mm, *active_mm;
1174 
1175 /* task state */
1176 	struct linux_binfmt *binfmt;
1177 	int exit_state;
1178 	int exit_code, exit_signal;
1179 	int pdeath_signal;  /*  The signal sent when the parent dies  */
1180 	/* ??? */
1181 	unsigned int personality;
1182 	unsigned did_exec:1;
1183 	unsigned in_execve:1;	/* Tell the LSMs that the process is doing an
1184 				 * execve */
1185 	pid_t pid;
1186 	pid_t tgid;
1187 
1188 	/* Canary value for the -fstack-protector gcc feature */
1189 	unsigned long stack_canary;
1190 
1191 	/*
1192 	 * pointers to (original) parent process, youngest child, younger sibling,
1193 	 * older sibling, respectively.  (p->father can be replaced with
1194 	 * p->real_parent->pid)
1195 	 */
1196 	struct task_struct *real_parent; /* real parent process */
1197 	struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1198 	/*
1199 	 * children/sibling forms the list of my natural children
1200 	 */
1201 	struct list_head children;	/* list of my children */
1202 	struct list_head sibling;	/* linkage in my parent's children list */
1203 	struct task_struct *group_leader;	/* threadgroup leader */
1204 
1205 	/*
1206 	 * ptraced is the list of tasks this task is using ptrace on.
1207 	 * This includes both natural children and PTRACE_ATTACH targets.
1208 	 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1209 	 */
1210 	struct list_head ptraced;
1211 	struct list_head ptrace_entry;
1212 
1213 #ifdef CONFIG_X86_PTRACE_BTS
1214 	/*
1215 	 * This is the tracer handle for the ptrace BTS extension.
1216 	 * This field actually belongs to the ptracer task.
1217 	 */
1218 	struct bts_tracer *bts;
1219 	/*
1220 	 * The buffer to hold the BTS data.
1221 	 */
1222 	void *bts_buffer;
1223 	size_t bts_size;
1224 #endif /* CONFIG_X86_PTRACE_BTS */
1225 
1226 	/* PID/PID hash table linkage. */
1227 	struct pid_link pids[PIDTYPE_MAX];
1228 	struct list_head thread_group;
1229 
1230 	struct completion *vfork_done;		/* for vfork() */
1231 	int __user *set_child_tid;		/* CLONE_CHILD_SETTID */
1232 	int __user *clear_child_tid;		/* CLONE_CHILD_CLEARTID */
1233 
1234 	cputime_t utime, stime, utimescaled, stimescaled;
1235 	cputime_t gtime;
1236 	cputime_t prev_utime, prev_stime;
1237 	unsigned long nvcsw, nivcsw; /* context switch counts */
1238 	struct timespec start_time; 		/* monotonic time */
1239 	struct timespec real_start_time;	/* boot based time */
1240 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1241 	unsigned long min_flt, maj_flt;
1242 
1243 	struct task_cputime cputime_expires;
1244 	struct list_head cpu_timers[3];
1245 
1246 /* process credentials */
1247 	const struct cred *real_cred;	/* objective and real subjective task
1248 					 * credentials (COW) */
1249 	const struct cred *cred;	/* effective (overridable) subjective task
1250 					 * credentials (COW) */
1251 	struct mutex cred_exec_mutex;	/* execve vs ptrace cred calculation mutex */
1252 
1253 	char comm[TASK_COMM_LEN]; /* executable name excluding path
1254 				     - access with [gs]et_task_comm (which lock
1255 				       it with task_lock())
1256 				     - initialized normally by flush_old_exec */
1257 /* file system info */
1258 	int link_count, total_link_count;
1259 #ifdef CONFIG_SYSVIPC
1260 /* ipc stuff */
1261 	struct sysv_sem sysvsem;
1262 #endif
1263 #ifdef CONFIG_DETECT_SOFTLOCKUP
1264 /* hung task detection */
1265 	unsigned long last_switch_timestamp;
1266 	unsigned long last_switch_count;
1267 #endif
1268 /* CPU-specific state of this task */
1269 	struct thread_struct thread;
1270 /* filesystem information */
1271 	struct fs_struct *fs;
1272 /* open file information */
1273 	struct files_struct *files;
1274 /* namespaces */
1275 	struct nsproxy *nsproxy;
1276 /* signal handlers */
1277 	struct signal_struct *signal;
1278 	struct sighand_struct *sighand;
1279 
1280 	sigset_t blocked, real_blocked;
1281 	sigset_t saved_sigmask;	/* restored if set_restore_sigmask() was used */
1282 	struct sigpending pending;
1283 
1284 	unsigned long sas_ss_sp;
1285 	size_t sas_ss_size;
1286 	int (*notifier)(void *priv);
1287 	void *notifier_data;
1288 	sigset_t *notifier_mask;
1289 	struct audit_context *audit_context;
1290 #ifdef CONFIG_AUDITSYSCALL
1291 	uid_t loginuid;
1292 	unsigned int sessionid;
1293 #endif
1294 	seccomp_t seccomp;
1295 
1296 /* Thread group tracking */
1297    	u32 parent_exec_id;
1298    	u32 self_exec_id;
1299 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1300 	spinlock_t alloc_lock;
1301 
1302 	/* Protection of the PI data structures: */
1303 	spinlock_t pi_lock;
1304 
1305 #ifdef CONFIG_RT_MUTEXES
1306 	/* PI waiters blocked on a rt_mutex held by this task */
1307 	struct plist_head pi_waiters;
1308 	/* Deadlock detection and priority inheritance handling */
1309 	struct rt_mutex_waiter *pi_blocked_on;
1310 #endif
1311 
1312 #ifdef CONFIG_DEBUG_MUTEXES
1313 	/* mutex deadlock detection */
1314 	struct mutex_waiter *blocked_on;
1315 #endif
1316 #ifdef CONFIG_TRACE_IRQFLAGS
1317 	unsigned int irq_events;
1318 	int hardirqs_enabled;
1319 	unsigned long hardirq_enable_ip;
1320 	unsigned int hardirq_enable_event;
1321 	unsigned long hardirq_disable_ip;
1322 	unsigned int hardirq_disable_event;
1323 	int softirqs_enabled;
1324 	unsigned long softirq_disable_ip;
1325 	unsigned int softirq_disable_event;
1326 	unsigned long softirq_enable_ip;
1327 	unsigned int softirq_enable_event;
1328 	int hardirq_context;
1329 	int softirq_context;
1330 #endif
1331 #ifdef CONFIG_LOCKDEP
1332 # define MAX_LOCK_DEPTH 48UL
1333 	u64 curr_chain_key;
1334 	int lockdep_depth;
1335 	unsigned int lockdep_recursion;
1336 	struct held_lock held_locks[MAX_LOCK_DEPTH];
1337 #endif
1338 
1339 /* journalling filesystem info */
1340 	void *journal_info;
1341 
1342 /* stacked block device info */
1343 	struct bio *bio_list, **bio_tail;
1344 
1345 /* VM state */
1346 	struct reclaim_state *reclaim_state;
1347 
1348 	struct backing_dev_info *backing_dev_info;
1349 
1350 	struct io_context *io_context;
1351 
1352 	unsigned long ptrace_message;
1353 	siginfo_t *last_siginfo; /* For ptrace use.  */
1354 	struct task_io_accounting ioac;
1355 #if defined(CONFIG_TASK_XACCT)
1356 	u64 acct_rss_mem1;	/* accumulated rss usage */
1357 	u64 acct_vm_mem1;	/* accumulated virtual memory usage */
1358 	cputime_t acct_timexpd;	/* stime + utime since last update */
1359 #endif
1360 #ifdef CONFIG_CPUSETS
1361 	nodemask_t mems_allowed;
1362 	int cpuset_mems_generation;
1363 	int cpuset_mem_spread_rotor;
1364 #endif
1365 #ifdef CONFIG_CGROUPS
1366 	/* Control Group info protected by css_set_lock */
1367 	struct css_set *cgroups;
1368 	/* cg_list protected by css_set_lock and tsk->alloc_lock */
1369 	struct list_head cg_list;
1370 #endif
1371 #ifdef CONFIG_FUTEX
1372 	struct robust_list_head __user *robust_list;
1373 #ifdef CONFIG_COMPAT
1374 	struct compat_robust_list_head __user *compat_robust_list;
1375 #endif
1376 	struct list_head pi_state_list;
1377 	struct futex_pi_state *pi_state_cache;
1378 #endif
1379 #ifdef CONFIG_NUMA
1380 	struct mempolicy *mempolicy;
1381 	short il_next;
1382 #endif
1383 	atomic_t fs_excl;	/* holding fs exclusive resources */
1384 	struct rcu_head rcu;
1385 
1386 	/*
1387 	 * cache last used pipe for splice
1388 	 */
1389 	struct pipe_inode_info *splice_pipe;
1390 #ifdef	CONFIG_TASK_DELAY_ACCT
1391 	struct task_delay_info *delays;
1392 #endif
1393 #ifdef CONFIG_FAULT_INJECTION
1394 	int make_it_fail;
1395 #endif
1396 	struct prop_local_single dirties;
1397 #ifdef CONFIG_LATENCYTOP
1398 	int latency_record_count;
1399 	struct latency_record latency_record[LT_SAVECOUNT];
1400 #endif
1401 	/*
1402 	 * time slack values; these are used to round up poll() and
1403 	 * select() etc timeout values. These are in nanoseconds.
1404 	 */
1405 	unsigned long timer_slack_ns;
1406 	unsigned long default_timer_slack_ns;
1407 
1408 	struct list_head	*scm_work_list;
1409 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1410 	/* Index of current stored adress in ret_stack */
1411 	int curr_ret_stack;
1412 	/* Stack of return addresses for return function tracing */
1413 	struct ftrace_ret_stack	*ret_stack;
1414 	/*
1415 	 * Number of functions that haven't been traced
1416 	 * because of depth overrun.
1417 	 */
1418 	atomic_t trace_overrun;
1419 	/* Pause for the tracing */
1420 	atomic_t tracing_graph_pause;
1421 #endif
1422 #ifdef CONFIG_TRACING
1423 	/* state flags for use by tracers */
1424 	unsigned long trace;
1425 #endif
1426 };
1427 
1428 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1429 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1430 
1431 /*
1432  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1433  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1434  * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1435  * values are inverted: lower p->prio value means higher priority.
1436  *
1437  * The MAX_USER_RT_PRIO value allows the actual maximum
1438  * RT priority to be separate from the value exported to
1439  * user-space.  This allows kernel threads to set their
1440  * priority to a value higher than any user task. Note:
1441  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1442  */
1443 
1444 #define MAX_USER_RT_PRIO	100
1445 #define MAX_RT_PRIO		MAX_USER_RT_PRIO
1446 
1447 #define MAX_PRIO		(MAX_RT_PRIO + 40)
1448 #define DEFAULT_PRIO		(MAX_RT_PRIO + 20)
1449 
1450 static inline int rt_prio(int prio)
1451 {
1452 	if (unlikely(prio < MAX_RT_PRIO))
1453 		return 1;
1454 	return 0;
1455 }
1456 
1457 static inline int rt_task(struct task_struct *p)
1458 {
1459 	return rt_prio(p->prio);
1460 }
1461 
1462 static inline void set_task_session(struct task_struct *tsk, pid_t session)
1463 {
1464 	tsk->signal->__session = session;
1465 }
1466 
1467 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1468 {
1469 	tsk->signal->__pgrp = pgrp;
1470 }
1471 
1472 static inline struct pid *task_pid(struct task_struct *task)
1473 {
1474 	return task->pids[PIDTYPE_PID].pid;
1475 }
1476 
1477 static inline struct pid *task_tgid(struct task_struct *task)
1478 {
1479 	return task->group_leader->pids[PIDTYPE_PID].pid;
1480 }
1481 
1482 static inline struct pid *task_pgrp(struct task_struct *task)
1483 {
1484 	return task->group_leader->pids[PIDTYPE_PGID].pid;
1485 }
1486 
1487 static inline struct pid *task_session(struct task_struct *task)
1488 {
1489 	return task->group_leader->pids[PIDTYPE_SID].pid;
1490 }
1491 
1492 struct pid_namespace;
1493 
1494 /*
1495  * the helpers to get the task's different pids as they are seen
1496  * from various namespaces
1497  *
1498  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1499  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1500  *                     current.
1501  * task_xid_nr_ns()  : id seen from the ns specified;
1502  *
1503  * set_task_vxid()   : assigns a virtual id to a task;
1504  *
1505  * see also pid_nr() etc in include/linux/pid.h
1506  */
1507 
1508 static inline pid_t task_pid_nr(struct task_struct *tsk)
1509 {
1510 	return tsk->pid;
1511 }
1512 
1513 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1514 
1515 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1516 {
1517 	return pid_vnr(task_pid(tsk));
1518 }
1519 
1520 
1521 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1522 {
1523 	return tsk->tgid;
1524 }
1525 
1526 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1527 
1528 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1529 {
1530 	return pid_vnr(task_tgid(tsk));
1531 }
1532 
1533 
1534 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1535 {
1536 	return tsk->signal->__pgrp;
1537 }
1538 
1539 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1540 
1541 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1542 {
1543 	return pid_vnr(task_pgrp(tsk));
1544 }
1545 
1546 
1547 static inline pid_t task_session_nr(struct task_struct *tsk)
1548 {
1549 	return tsk->signal->__session;
1550 }
1551 
1552 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1553 
1554 static inline pid_t task_session_vnr(struct task_struct *tsk)
1555 {
1556 	return pid_vnr(task_session(tsk));
1557 }
1558 
1559 
1560 /**
1561  * pid_alive - check that a task structure is not stale
1562  * @p: Task structure to be checked.
1563  *
1564  * Test if a process is not yet dead (at most zombie state)
1565  * If pid_alive fails, then pointers within the task structure
1566  * can be stale and must not be dereferenced.
1567  */
1568 static inline int pid_alive(struct task_struct *p)
1569 {
1570 	return p->pids[PIDTYPE_PID].pid != NULL;
1571 }
1572 
1573 /**
1574  * is_global_init - check if a task structure is init
1575  * @tsk: Task structure to be checked.
1576  *
1577  * Check if a task structure is the first user space task the kernel created.
1578  */
1579 static inline int is_global_init(struct task_struct *tsk)
1580 {
1581 	return tsk->pid == 1;
1582 }
1583 
1584 /*
1585  * is_container_init:
1586  * check whether in the task is init in its own pid namespace.
1587  */
1588 extern int is_container_init(struct task_struct *tsk);
1589 
1590 extern struct pid *cad_pid;
1591 
1592 extern void free_task(struct task_struct *tsk);
1593 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1594 
1595 extern void __put_task_struct(struct task_struct *t);
1596 
1597 static inline void put_task_struct(struct task_struct *t)
1598 {
1599 	if (atomic_dec_and_test(&t->usage))
1600 		__put_task_struct(t);
1601 }
1602 
1603 extern cputime_t task_utime(struct task_struct *p);
1604 extern cputime_t task_stime(struct task_struct *p);
1605 extern cputime_t task_gtime(struct task_struct *p);
1606 
1607 /*
1608  * Per process flags
1609  */
1610 #define PF_ALIGNWARN	0x00000001	/* Print alignment warning msgs */
1611 					/* Not implemented yet, only for 486*/
1612 #define PF_STARTING	0x00000002	/* being created */
1613 #define PF_EXITING	0x00000004	/* getting shut down */
1614 #define PF_EXITPIDONE	0x00000008	/* pi exit done on shut down */
1615 #define PF_VCPU		0x00000010	/* I'm a virtual CPU */
1616 #define PF_FORKNOEXEC	0x00000040	/* forked but didn't exec */
1617 #define PF_SUPERPRIV	0x00000100	/* used super-user privileges */
1618 #define PF_DUMPCORE	0x00000200	/* dumped core */
1619 #define PF_SIGNALED	0x00000400	/* killed by a signal */
1620 #define PF_MEMALLOC	0x00000800	/* Allocating memory */
1621 #define PF_FLUSHER	0x00001000	/* responsible for disk writeback */
1622 #define PF_USED_MATH	0x00002000	/* if unset the fpu must be initialized before use */
1623 #define PF_NOFREEZE	0x00008000	/* this thread should not be frozen */
1624 #define PF_FROZEN	0x00010000	/* frozen for system suspend */
1625 #define PF_FSTRANS	0x00020000	/* inside a filesystem transaction */
1626 #define PF_KSWAPD	0x00040000	/* I am kswapd */
1627 #define PF_SWAPOFF	0x00080000	/* I am in swapoff */
1628 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
1629 #define PF_KTHREAD	0x00200000	/* I am a kernel thread */
1630 #define PF_RANDOMIZE	0x00400000	/* randomize virtual address space */
1631 #define PF_SWAPWRITE	0x00800000	/* Allowed to write to swap */
1632 #define PF_SPREAD_PAGE	0x01000000	/* Spread page cache over cpuset */
1633 #define PF_SPREAD_SLAB	0x02000000	/* Spread some slab caches over cpuset */
1634 #define PF_THREAD_BOUND	0x04000000	/* Thread bound to specific cpu */
1635 #define PF_MEMPOLICY	0x10000000	/* Non-default NUMA mempolicy */
1636 #define PF_MUTEX_TESTER	0x20000000	/* Thread belongs to the rt mutex tester */
1637 #define PF_FREEZER_SKIP	0x40000000	/* Freezer should not count it as freezeable */
1638 #define PF_FREEZER_NOSIG 0x80000000	/* Freezer won't send signals to it */
1639 
1640 /*
1641  * Only the _current_ task can read/write to tsk->flags, but other
1642  * tasks can access tsk->flags in readonly mode for example
1643  * with tsk_used_math (like during threaded core dumping).
1644  * There is however an exception to this rule during ptrace
1645  * or during fork: the ptracer task is allowed to write to the
1646  * child->flags of its traced child (same goes for fork, the parent
1647  * can write to the child->flags), because we're guaranteed the
1648  * child is not running and in turn not changing child->flags
1649  * at the same time the parent does it.
1650  */
1651 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1652 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1653 #define clear_used_math() clear_stopped_child_used_math(current)
1654 #define set_used_math() set_stopped_child_used_math(current)
1655 #define conditional_stopped_child_used_math(condition, child) \
1656 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1657 #define conditional_used_math(condition) \
1658 	conditional_stopped_child_used_math(condition, current)
1659 #define copy_to_stopped_child_used_math(child) \
1660 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1661 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1662 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1663 #define used_math() tsk_used_math(current)
1664 
1665 #ifdef CONFIG_SMP
1666 extern int set_cpus_allowed_ptr(struct task_struct *p,
1667 				const struct cpumask *new_mask);
1668 #else
1669 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1670 				       const struct cpumask *new_mask)
1671 {
1672 	if (!cpumask_test_cpu(0, new_mask))
1673 		return -EINVAL;
1674 	return 0;
1675 }
1676 #endif
1677 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1678 {
1679 	return set_cpus_allowed_ptr(p, &new_mask);
1680 }
1681 
1682 /*
1683  * Architectures can set this to 1 if they have specified
1684  * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1685  * but then during bootup it turns out that sched_clock()
1686  * is reliable after all:
1687  */
1688 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1689 extern int sched_clock_stable;
1690 #endif
1691 
1692 extern unsigned long long sched_clock(void);
1693 
1694 extern void sched_clock_init(void);
1695 extern u64 sched_clock_cpu(int cpu);
1696 
1697 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1698 static inline void sched_clock_tick(void)
1699 {
1700 }
1701 
1702 static inline void sched_clock_idle_sleep_event(void)
1703 {
1704 }
1705 
1706 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1707 {
1708 }
1709 #else
1710 extern void sched_clock_tick(void);
1711 extern void sched_clock_idle_sleep_event(void);
1712 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1713 #endif
1714 
1715 /*
1716  * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1717  * clock constructed from sched_clock():
1718  */
1719 extern unsigned long long cpu_clock(int cpu);
1720 
1721 extern unsigned long long
1722 task_sched_runtime(struct task_struct *task);
1723 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1724 
1725 /* sched_exec is called by processes performing an exec */
1726 #ifdef CONFIG_SMP
1727 extern void sched_exec(void);
1728 #else
1729 #define sched_exec()   {}
1730 #endif
1731 
1732 extern void sched_clock_idle_sleep_event(void);
1733 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1734 
1735 #ifdef CONFIG_HOTPLUG_CPU
1736 extern void idle_task_exit(void);
1737 #else
1738 static inline void idle_task_exit(void) {}
1739 #endif
1740 
1741 extern void sched_idle_next(void);
1742 
1743 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1744 extern void wake_up_idle_cpu(int cpu);
1745 #else
1746 static inline void wake_up_idle_cpu(int cpu) { }
1747 #endif
1748 
1749 extern unsigned int sysctl_sched_latency;
1750 extern unsigned int sysctl_sched_min_granularity;
1751 extern unsigned int sysctl_sched_wakeup_granularity;
1752 extern unsigned int sysctl_sched_shares_ratelimit;
1753 extern unsigned int sysctl_sched_shares_thresh;
1754 #ifdef CONFIG_SCHED_DEBUG
1755 extern unsigned int sysctl_sched_child_runs_first;
1756 extern unsigned int sysctl_sched_features;
1757 extern unsigned int sysctl_sched_migration_cost;
1758 extern unsigned int sysctl_sched_nr_migrate;
1759 
1760 int sched_nr_latency_handler(struct ctl_table *table, int write,
1761 		struct file *file, void __user *buffer, size_t *length,
1762 		loff_t *ppos);
1763 #endif
1764 extern unsigned int sysctl_sched_rt_period;
1765 extern int sysctl_sched_rt_runtime;
1766 
1767 int sched_rt_handler(struct ctl_table *table, int write,
1768 		struct file *filp, void __user *buffer, size_t *lenp,
1769 		loff_t *ppos);
1770 
1771 extern unsigned int sysctl_sched_compat_yield;
1772 
1773 #ifdef CONFIG_RT_MUTEXES
1774 extern int rt_mutex_getprio(struct task_struct *p);
1775 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1776 extern void rt_mutex_adjust_pi(struct task_struct *p);
1777 #else
1778 static inline int rt_mutex_getprio(struct task_struct *p)
1779 {
1780 	return p->normal_prio;
1781 }
1782 # define rt_mutex_adjust_pi(p)		do { } while (0)
1783 #endif
1784 
1785 extern void set_user_nice(struct task_struct *p, long nice);
1786 extern int task_prio(const struct task_struct *p);
1787 extern int task_nice(const struct task_struct *p);
1788 extern int can_nice(const struct task_struct *p, const int nice);
1789 extern int task_curr(const struct task_struct *p);
1790 extern int idle_cpu(int cpu);
1791 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1792 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1793 				      struct sched_param *);
1794 extern struct task_struct *idle_task(int cpu);
1795 extern struct task_struct *curr_task(int cpu);
1796 extern void set_curr_task(int cpu, struct task_struct *p);
1797 
1798 void yield(void);
1799 
1800 /*
1801  * The default (Linux) execution domain.
1802  */
1803 extern struct exec_domain	default_exec_domain;
1804 
1805 union thread_union {
1806 	struct thread_info thread_info;
1807 	unsigned long stack[THREAD_SIZE/sizeof(long)];
1808 };
1809 
1810 #ifndef __HAVE_ARCH_KSTACK_END
1811 static inline int kstack_end(void *addr)
1812 {
1813 	/* Reliable end of stack detection:
1814 	 * Some APM bios versions misalign the stack
1815 	 */
1816 	return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1817 }
1818 #endif
1819 
1820 extern union thread_union init_thread_union;
1821 extern struct task_struct init_task;
1822 
1823 extern struct   mm_struct init_mm;
1824 
1825 extern struct pid_namespace init_pid_ns;
1826 
1827 /*
1828  * find a task by one of its numerical ids
1829  *
1830  * find_task_by_pid_type_ns():
1831  *      it is the most generic call - it finds a task by all id,
1832  *      type and namespace specified
1833  * find_task_by_pid_ns():
1834  *      finds a task by its pid in the specified namespace
1835  * find_task_by_vpid():
1836  *      finds a task by its virtual pid
1837  *
1838  * see also find_vpid() etc in include/linux/pid.h
1839  */
1840 
1841 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1842 		struct pid_namespace *ns);
1843 
1844 extern struct task_struct *find_task_by_vpid(pid_t nr);
1845 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1846 		struct pid_namespace *ns);
1847 
1848 extern void __set_special_pids(struct pid *pid);
1849 
1850 /* per-UID process charging. */
1851 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1852 static inline struct user_struct *get_uid(struct user_struct *u)
1853 {
1854 	atomic_inc(&u->__count);
1855 	return u;
1856 }
1857 extern void free_uid(struct user_struct *);
1858 extern void release_uids(struct user_namespace *ns);
1859 
1860 #include <asm/current.h>
1861 
1862 extern void do_timer(unsigned long ticks);
1863 
1864 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1865 extern int wake_up_process(struct task_struct *tsk);
1866 extern void wake_up_new_task(struct task_struct *tsk,
1867 				unsigned long clone_flags);
1868 #ifdef CONFIG_SMP
1869  extern void kick_process(struct task_struct *tsk);
1870 #else
1871  static inline void kick_process(struct task_struct *tsk) { }
1872 #endif
1873 extern void sched_fork(struct task_struct *p, int clone_flags);
1874 extern void sched_dead(struct task_struct *p);
1875 
1876 extern void proc_caches_init(void);
1877 extern void flush_signals(struct task_struct *);
1878 extern void ignore_signals(struct task_struct *);
1879 extern void flush_signal_handlers(struct task_struct *, int force_default);
1880 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1881 
1882 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1883 {
1884 	unsigned long flags;
1885 	int ret;
1886 
1887 	spin_lock_irqsave(&tsk->sighand->siglock, flags);
1888 	ret = dequeue_signal(tsk, mask, info);
1889 	spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1890 
1891 	return ret;
1892 }
1893 
1894 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1895 			      sigset_t *mask);
1896 extern void unblock_all_signals(void);
1897 extern void release_task(struct task_struct * p);
1898 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1899 extern int force_sigsegv(int, struct task_struct *);
1900 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1901 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1902 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1903 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1904 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1905 extern int kill_pid(struct pid *pid, int sig, int priv);
1906 extern int kill_proc_info(int, struct siginfo *, pid_t);
1907 extern int do_notify_parent(struct task_struct *, int);
1908 extern void force_sig(int, struct task_struct *);
1909 extern void force_sig_specific(int, struct task_struct *);
1910 extern int send_sig(int, struct task_struct *, int);
1911 extern void zap_other_threads(struct task_struct *p);
1912 extern struct sigqueue *sigqueue_alloc(void);
1913 extern void sigqueue_free(struct sigqueue *);
1914 extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
1915 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1916 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1917 
1918 static inline int kill_cad_pid(int sig, int priv)
1919 {
1920 	return kill_pid(cad_pid, sig, priv);
1921 }
1922 
1923 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
1924 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1925 #define SEND_SIG_PRIV	((struct siginfo *) 1)
1926 #define SEND_SIG_FORCED	((struct siginfo *) 2)
1927 
1928 static inline int is_si_special(const struct siginfo *info)
1929 {
1930 	return info <= SEND_SIG_FORCED;
1931 }
1932 
1933 /* True if we are on the alternate signal stack.  */
1934 
1935 static inline int on_sig_stack(unsigned long sp)
1936 {
1937 	return (sp - current->sas_ss_sp < current->sas_ss_size);
1938 }
1939 
1940 static inline int sas_ss_flags(unsigned long sp)
1941 {
1942 	return (current->sas_ss_size == 0 ? SS_DISABLE
1943 		: on_sig_stack(sp) ? SS_ONSTACK : 0);
1944 }
1945 
1946 /*
1947  * Routines for handling mm_structs
1948  */
1949 extern struct mm_struct * mm_alloc(void);
1950 
1951 /* mmdrop drops the mm and the page tables */
1952 extern void __mmdrop(struct mm_struct *);
1953 static inline void mmdrop(struct mm_struct * mm)
1954 {
1955 	if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1956 		__mmdrop(mm);
1957 }
1958 
1959 /* mmput gets rid of the mappings and all user-space */
1960 extern void mmput(struct mm_struct *);
1961 /* Grab a reference to a task's mm, if it is not already going away */
1962 extern struct mm_struct *get_task_mm(struct task_struct *task);
1963 /* Remove the current tasks stale references to the old mm_struct */
1964 extern void mm_release(struct task_struct *, struct mm_struct *);
1965 /* Allocate a new mm structure and copy contents from tsk->mm */
1966 extern struct mm_struct *dup_mm(struct task_struct *tsk);
1967 
1968 extern int  copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1969 extern void flush_thread(void);
1970 extern void exit_thread(void);
1971 
1972 extern void exit_files(struct task_struct *);
1973 extern void __cleanup_signal(struct signal_struct *);
1974 extern void __cleanup_sighand(struct sighand_struct *);
1975 
1976 extern void exit_itimers(struct signal_struct *);
1977 extern void flush_itimer_signals(void);
1978 
1979 extern NORET_TYPE void do_group_exit(int);
1980 
1981 extern void daemonize(const char *, ...);
1982 extern int allow_signal(int);
1983 extern int disallow_signal(int);
1984 
1985 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1986 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1987 struct task_struct *fork_idle(int);
1988 
1989 extern void set_task_comm(struct task_struct *tsk, char *from);
1990 extern char *get_task_comm(char *to, struct task_struct *tsk);
1991 
1992 #ifdef CONFIG_SMP
1993 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1994 #else
1995 static inline unsigned long wait_task_inactive(struct task_struct *p,
1996 					       long match_state)
1997 {
1998 	return 1;
1999 }
2000 #endif
2001 
2002 #define next_task(p)	list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
2003 
2004 #define for_each_process(p) \
2005 	for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2006 
2007 extern bool is_single_threaded(struct task_struct *);
2008 
2009 /*
2010  * Careful: do_each_thread/while_each_thread is a double loop so
2011  *          'break' will not work as expected - use goto instead.
2012  */
2013 #define do_each_thread(g, t) \
2014 	for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2015 
2016 #define while_each_thread(g, t) \
2017 	while ((t = next_thread(t)) != g)
2018 
2019 /* de_thread depends on thread_group_leader not being a pid based check */
2020 #define thread_group_leader(p)	(p == p->group_leader)
2021 
2022 /* Do to the insanities of de_thread it is possible for a process
2023  * to have the pid of the thread group leader without actually being
2024  * the thread group leader.  For iteration through the pids in proc
2025  * all we care about is that we have a task with the appropriate
2026  * pid, we don't actually care if we have the right task.
2027  */
2028 static inline int has_group_leader_pid(struct task_struct *p)
2029 {
2030 	return p->pid == p->tgid;
2031 }
2032 
2033 static inline
2034 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2035 {
2036 	return p1->tgid == p2->tgid;
2037 }
2038 
2039 static inline struct task_struct *next_thread(const struct task_struct *p)
2040 {
2041 	return list_entry(rcu_dereference(p->thread_group.next),
2042 			  struct task_struct, thread_group);
2043 }
2044 
2045 static inline int thread_group_empty(struct task_struct *p)
2046 {
2047 	return list_empty(&p->thread_group);
2048 }
2049 
2050 #define delay_group_leader(p) \
2051 		(thread_group_leader(p) && !thread_group_empty(p))
2052 
2053 /*
2054  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2055  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2056  * pins the final release of task.io_context.  Also protects ->cpuset and
2057  * ->cgroup.subsys[].
2058  *
2059  * Nests both inside and outside of read_lock(&tasklist_lock).
2060  * It must not be nested with write_lock_irq(&tasklist_lock),
2061  * neither inside nor outside.
2062  */
2063 static inline void task_lock(struct task_struct *p)
2064 {
2065 	spin_lock(&p->alloc_lock);
2066 }
2067 
2068 static inline void task_unlock(struct task_struct *p)
2069 {
2070 	spin_unlock(&p->alloc_lock);
2071 }
2072 
2073 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2074 							unsigned long *flags);
2075 
2076 static inline void unlock_task_sighand(struct task_struct *tsk,
2077 						unsigned long *flags)
2078 {
2079 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2080 }
2081 
2082 #ifndef __HAVE_THREAD_FUNCTIONS
2083 
2084 #define task_thread_info(task)	((struct thread_info *)(task)->stack)
2085 #define task_stack_page(task)	((task)->stack)
2086 
2087 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2088 {
2089 	*task_thread_info(p) = *task_thread_info(org);
2090 	task_thread_info(p)->task = p;
2091 }
2092 
2093 static inline unsigned long *end_of_stack(struct task_struct *p)
2094 {
2095 	return (unsigned long *)(task_thread_info(p) + 1);
2096 }
2097 
2098 #endif
2099 
2100 static inline int object_is_on_stack(void *obj)
2101 {
2102 	void *stack = task_stack_page(current);
2103 
2104 	return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2105 }
2106 
2107 extern void thread_info_cache_init(void);
2108 
2109 #ifdef CONFIG_DEBUG_STACK_USAGE
2110 static inline unsigned long stack_not_used(struct task_struct *p)
2111 {
2112 	unsigned long *n = end_of_stack(p);
2113 
2114 	do { 	/* Skip over canary */
2115 		n++;
2116 	} while (!*n);
2117 
2118 	return (unsigned long)n - (unsigned long)end_of_stack(p);
2119 }
2120 #endif
2121 
2122 /* set thread flags in other task's structures
2123  * - see asm/thread_info.h for TIF_xxxx flags available
2124  */
2125 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2126 {
2127 	set_ti_thread_flag(task_thread_info(tsk), flag);
2128 }
2129 
2130 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2131 {
2132 	clear_ti_thread_flag(task_thread_info(tsk), flag);
2133 }
2134 
2135 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2136 {
2137 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2138 }
2139 
2140 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2141 {
2142 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2143 }
2144 
2145 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2146 {
2147 	return test_ti_thread_flag(task_thread_info(tsk), flag);
2148 }
2149 
2150 static inline void set_tsk_need_resched(struct task_struct *tsk)
2151 {
2152 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2153 }
2154 
2155 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2156 {
2157 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2158 }
2159 
2160 static inline int test_tsk_need_resched(struct task_struct *tsk)
2161 {
2162 	return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2163 }
2164 
2165 static inline int signal_pending(struct task_struct *p)
2166 {
2167 	return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2168 }
2169 
2170 extern int __fatal_signal_pending(struct task_struct *p);
2171 
2172 static inline int fatal_signal_pending(struct task_struct *p)
2173 {
2174 	return signal_pending(p) && __fatal_signal_pending(p);
2175 }
2176 
2177 static inline int signal_pending_state(long state, struct task_struct *p)
2178 {
2179 	if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2180 		return 0;
2181 	if (!signal_pending(p))
2182 		return 0;
2183 
2184 	return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2185 }
2186 
2187 static inline int need_resched(void)
2188 {
2189 	return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2190 }
2191 
2192 /*
2193  * cond_resched() and cond_resched_lock(): latency reduction via
2194  * explicit rescheduling in places that are safe. The return
2195  * value indicates whether a reschedule was done in fact.
2196  * cond_resched_lock() will drop the spinlock before scheduling,
2197  * cond_resched_softirq() will enable bhs before scheduling.
2198  */
2199 extern int _cond_resched(void);
2200 #ifdef CONFIG_PREEMPT_BKL
2201 static inline int cond_resched(void)
2202 {
2203 	return 0;
2204 }
2205 #else
2206 static inline int cond_resched(void)
2207 {
2208 	return _cond_resched();
2209 }
2210 #endif
2211 extern int cond_resched_lock(spinlock_t * lock);
2212 extern int cond_resched_softirq(void);
2213 static inline int cond_resched_bkl(void)
2214 {
2215 	return _cond_resched();
2216 }
2217 
2218 /*
2219  * Does a critical section need to be broken due to another
2220  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2221  * but a general need for low latency)
2222  */
2223 static inline int spin_needbreak(spinlock_t *lock)
2224 {
2225 #ifdef CONFIG_PREEMPT
2226 	return spin_is_contended(lock);
2227 #else
2228 	return 0;
2229 #endif
2230 }
2231 
2232 /*
2233  * Thread group CPU time accounting.
2234  */
2235 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2236 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2237 
2238 static inline void thread_group_cputime_init(struct signal_struct *sig)
2239 {
2240 	sig->cputimer.cputime = INIT_CPUTIME;
2241 	spin_lock_init(&sig->cputimer.lock);
2242 	sig->cputimer.running = 0;
2243 }
2244 
2245 static inline void thread_group_cputime_free(struct signal_struct *sig)
2246 {
2247 }
2248 
2249 /*
2250  * Reevaluate whether the task has signals pending delivery.
2251  * Wake the task if so.
2252  * This is required every time the blocked sigset_t changes.
2253  * callers must hold sighand->siglock.
2254  */
2255 extern void recalc_sigpending_and_wake(struct task_struct *t);
2256 extern void recalc_sigpending(void);
2257 
2258 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2259 
2260 /*
2261  * Wrappers for p->thread_info->cpu access. No-op on UP.
2262  */
2263 #ifdef CONFIG_SMP
2264 
2265 static inline unsigned int task_cpu(const struct task_struct *p)
2266 {
2267 	return task_thread_info(p)->cpu;
2268 }
2269 
2270 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2271 
2272 #else
2273 
2274 static inline unsigned int task_cpu(const struct task_struct *p)
2275 {
2276 	return 0;
2277 }
2278 
2279 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2280 {
2281 }
2282 
2283 #endif /* CONFIG_SMP */
2284 
2285 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2286 
2287 #ifdef CONFIG_TRACING
2288 extern void
2289 __trace_special(void *__tr, void *__data,
2290 		unsigned long arg1, unsigned long arg2, unsigned long arg3);
2291 #else
2292 static inline void
2293 __trace_special(void *__tr, void *__data,
2294 		unsigned long arg1, unsigned long arg2, unsigned long arg3)
2295 {
2296 }
2297 #endif
2298 
2299 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2300 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2301 
2302 extern void normalize_rt_tasks(void);
2303 
2304 #ifdef CONFIG_GROUP_SCHED
2305 
2306 extern struct task_group init_task_group;
2307 #ifdef CONFIG_USER_SCHED
2308 extern struct task_group root_task_group;
2309 extern void set_tg_uid(struct user_struct *user);
2310 #endif
2311 
2312 extern struct task_group *sched_create_group(struct task_group *parent);
2313 extern void sched_destroy_group(struct task_group *tg);
2314 extern void sched_move_task(struct task_struct *tsk);
2315 #ifdef CONFIG_FAIR_GROUP_SCHED
2316 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2317 extern unsigned long sched_group_shares(struct task_group *tg);
2318 #endif
2319 #ifdef CONFIG_RT_GROUP_SCHED
2320 extern int sched_group_set_rt_runtime(struct task_group *tg,
2321 				      long rt_runtime_us);
2322 extern long sched_group_rt_runtime(struct task_group *tg);
2323 extern int sched_group_set_rt_period(struct task_group *tg,
2324 				      long rt_period_us);
2325 extern long sched_group_rt_period(struct task_group *tg);
2326 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2327 #endif
2328 #endif
2329 
2330 extern int task_can_switch_user(struct user_struct *up,
2331 					struct task_struct *tsk);
2332 
2333 #ifdef CONFIG_TASK_XACCT
2334 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2335 {
2336 	tsk->ioac.rchar += amt;
2337 }
2338 
2339 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2340 {
2341 	tsk->ioac.wchar += amt;
2342 }
2343 
2344 static inline void inc_syscr(struct task_struct *tsk)
2345 {
2346 	tsk->ioac.syscr++;
2347 }
2348 
2349 static inline void inc_syscw(struct task_struct *tsk)
2350 {
2351 	tsk->ioac.syscw++;
2352 }
2353 #else
2354 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2355 {
2356 }
2357 
2358 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2359 {
2360 }
2361 
2362 static inline void inc_syscr(struct task_struct *tsk)
2363 {
2364 }
2365 
2366 static inline void inc_syscw(struct task_struct *tsk)
2367 {
2368 }
2369 #endif
2370 
2371 #ifndef TASK_SIZE_OF
2372 #define TASK_SIZE_OF(tsk)	TASK_SIZE
2373 #endif
2374 
2375 #ifdef CONFIG_MM_OWNER
2376 extern void mm_update_next_owner(struct mm_struct *mm);
2377 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2378 #else
2379 static inline void mm_update_next_owner(struct mm_struct *mm)
2380 {
2381 }
2382 
2383 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2384 {
2385 }
2386 #endif /* CONFIG_MM_OWNER */
2387 
2388 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2389 
2390 #endif /* __KERNEL__ */
2391 
2392 #endif
2393