xref: /linux-6.15/include/linux/sched.h (revision 643d1f7f)
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/semaphore.h>
65 #include <asm/page.h>
66 #include <asm/ptrace.h>
67 #include <asm/cputime.h>
68 
69 #include <linux/smp.h>
70 #include <linux/sem.h>
71 #include <linux/signal.h>
72 #include <linux/securebits.h>
73 #include <linux/fs_struct.h>
74 #include <linux/compiler.h>
75 #include <linux/completion.h>
76 #include <linux/pid.h>
77 #include <linux/percpu.h>
78 #include <linux/topology.h>
79 #include <linux/proportions.h>
80 #include <linux/seccomp.h>
81 #include <linux/rcupdate.h>
82 #include <linux/rtmutex.h>
83 
84 #include <linux/time.h>
85 #include <linux/param.h>
86 #include <linux/resource.h>
87 #include <linux/timer.h>
88 #include <linux/hrtimer.h>
89 #include <linux/task_io_accounting.h>
90 #include <linux/kobject.h>
91 #include <linux/latencytop.h>
92 
93 #include <asm/processor.h>
94 
95 struct exec_domain;
96 struct futex_pi_state;
97 struct robust_list_head;
98 struct bio;
99 
100 /*
101  * List of flags we want to share for kernel threads,
102  * if only because they are not used by them anyway.
103  */
104 #define CLONE_KERNEL	(CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
105 
106 /*
107  * These are the constant used to fake the fixed-point load-average
108  * counting. Some notes:
109  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
110  *    a load-average precision of 10 bits integer + 11 bits fractional
111  *  - if you want to count load-averages more often, you need more
112  *    precision, or rounding will get you. With 2-second counting freq,
113  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
114  *    11 bit fractions.
115  */
116 extern unsigned long avenrun[];		/* Load averages */
117 
118 #define FSHIFT		11		/* nr of bits of precision */
119 #define FIXED_1		(1<<FSHIFT)	/* 1.0 as fixed-point */
120 #define LOAD_FREQ	(5*HZ+1)	/* 5 sec intervals */
121 #define EXP_1		1884		/* 1/exp(5sec/1min) as fixed-point */
122 #define EXP_5		2014		/* 1/exp(5sec/5min) */
123 #define EXP_15		2037		/* 1/exp(5sec/15min) */
124 
125 #define CALC_LOAD(load,exp,n) \
126 	load *= exp; \
127 	load += n*(FIXED_1-exp); \
128 	load >>= FSHIFT;
129 
130 extern unsigned long total_forks;
131 extern int nr_threads;
132 DECLARE_PER_CPU(unsigned long, process_counts);
133 extern int nr_processes(void);
134 extern unsigned long nr_running(void);
135 extern unsigned long nr_uninterruptible(void);
136 extern unsigned long nr_active(void);
137 extern unsigned long nr_iowait(void);
138 extern unsigned long weighted_cpuload(const int cpu);
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 /*
163  * Task state bitmask. NOTE! These bits are also
164  * encoded in fs/proc/array.c: get_task_state().
165  *
166  * We have two separate sets of flags: task->state
167  * is about runnability, while task->exit_state are
168  * about the task exiting. Confusing, but this way
169  * modifying one set can't modify the other one by
170  * mistake.
171  */
172 #define TASK_RUNNING		0
173 #define TASK_INTERRUPTIBLE	1
174 #define TASK_UNINTERRUPTIBLE	2
175 #define __TASK_STOPPED		4
176 #define __TASK_TRACED		8
177 /* in tsk->exit_state */
178 #define EXIT_ZOMBIE		16
179 #define EXIT_DEAD		32
180 /* in tsk->state again */
181 #define TASK_DEAD		64
182 #define TASK_WAKEKILL		128
183 
184 /* Convenience macros for the sake of set_task_state */
185 #define TASK_KILLABLE		(TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
186 #define TASK_STOPPED		(TASK_WAKEKILL | __TASK_STOPPED)
187 #define TASK_TRACED		(TASK_WAKEKILL | __TASK_TRACED)
188 
189 /* Convenience macros for the sake of wake_up */
190 #define TASK_NORMAL		(TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
191 #define TASK_ALL		(TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
192 
193 /* get_task_state() */
194 #define TASK_REPORT		(TASK_RUNNING | TASK_INTERRUPTIBLE | \
195 				 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
196 				 __TASK_TRACED)
197 
198 #define task_is_traced(task)	((task->state & __TASK_TRACED) != 0)
199 #define task_is_stopped(task)	((task->state & __TASK_STOPPED) != 0)
200 #define task_is_stopped_or_traced(task)	\
201 			((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
202 #define task_contributes_to_load(task)	\
203 				((task->state & TASK_UNINTERRUPTIBLE) != 0)
204 
205 #define __set_task_state(tsk, state_value)		\
206 	do { (tsk)->state = (state_value); } while (0)
207 #define set_task_state(tsk, state_value)		\
208 	set_mb((tsk)->state, (state_value))
209 
210 /*
211  * set_current_state() includes a barrier so that the write of current->state
212  * is correctly serialised wrt the caller's subsequent test of whether to
213  * actually sleep:
214  *
215  *	set_current_state(TASK_UNINTERRUPTIBLE);
216  *	if (do_i_need_to_sleep())
217  *		schedule();
218  *
219  * If the caller does not need such serialisation then use __set_current_state()
220  */
221 #define __set_current_state(state_value)			\
222 	do { current->state = (state_value); } while (0)
223 #define set_current_state(state_value)		\
224 	set_mb(current->state, (state_value))
225 
226 /* Task command name length */
227 #define TASK_COMM_LEN 16
228 
229 #include <linux/spinlock.h>
230 
231 /*
232  * This serializes "schedule()" and also protects
233  * the run-queue from deletions/modifications (but
234  * _adding_ to the beginning of the run-queue has
235  * a separate lock).
236  */
237 extern rwlock_t tasklist_lock;
238 extern spinlock_t mmlist_lock;
239 
240 struct task_struct;
241 
242 extern void sched_init(void);
243 extern void sched_init_smp(void);
244 extern void init_idle(struct task_struct *idle, int cpu);
245 extern void init_idle_bootup_task(struct task_struct *idle);
246 
247 extern cpumask_t nohz_cpu_mask;
248 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
249 extern int select_nohz_load_balancer(int cpu);
250 #else
251 static inline int select_nohz_load_balancer(int cpu)
252 {
253 	return 0;
254 }
255 #endif
256 
257 extern unsigned long rt_needs_cpu(int cpu);
258 
259 /*
260  * Only dump TASK_* tasks. (0 for all tasks)
261  */
262 extern void show_state_filter(unsigned long state_filter);
263 
264 static inline void show_state(void)
265 {
266 	show_state_filter(0);
267 }
268 
269 extern void show_regs(struct pt_regs *);
270 
271 /*
272  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
273  * task), SP is the stack pointer of the first frame that should be shown in the back
274  * trace (or NULL if the entire call-chain of the task should be shown).
275  */
276 extern void show_stack(struct task_struct *task, unsigned long *sp);
277 
278 void io_schedule(void);
279 long io_schedule_timeout(long timeout);
280 
281 extern void cpu_init (void);
282 extern void trap_init(void);
283 extern void account_process_tick(struct task_struct *task, int user);
284 extern void update_process_times(int user);
285 extern void scheduler_tick(void);
286 extern void hrtick_resched(void);
287 
288 extern void sched_show_task(struct task_struct *p);
289 
290 #ifdef CONFIG_DETECT_SOFTLOCKUP
291 extern void softlockup_tick(void);
292 extern void spawn_softlockup_task(void);
293 extern void touch_softlockup_watchdog(void);
294 extern void touch_all_softlockup_watchdogs(void);
295 extern unsigned long  softlockup_thresh;
296 extern unsigned long sysctl_hung_task_check_count;
297 extern unsigned long sysctl_hung_task_timeout_secs;
298 extern unsigned long sysctl_hung_task_warnings;
299 #else
300 static inline void softlockup_tick(void)
301 {
302 }
303 static inline void spawn_softlockup_task(void)
304 {
305 }
306 static inline void touch_softlockup_watchdog(void)
307 {
308 }
309 static inline void touch_all_softlockup_watchdogs(void)
310 {
311 }
312 #endif
313 
314 
315 /* Attach to any functions which should be ignored in wchan output. */
316 #define __sched		__attribute__((__section__(".sched.text")))
317 
318 /* Linker adds these: start and end of __sched functions */
319 extern char __sched_text_start[], __sched_text_end[];
320 
321 /* Is this address in the __sched functions? */
322 extern int in_sched_functions(unsigned long addr);
323 
324 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
325 extern signed long FASTCALL(schedule_timeout(signed long timeout));
326 extern signed long schedule_timeout_interruptible(signed long timeout);
327 extern signed long schedule_timeout_killable(signed long timeout);
328 extern signed long schedule_timeout_uninterruptible(signed long timeout);
329 asmlinkage void schedule(void);
330 
331 struct nsproxy;
332 struct user_namespace;
333 
334 /* Maximum number of active map areas.. This is a random (large) number */
335 #define DEFAULT_MAX_MAP_COUNT	65536
336 
337 extern int sysctl_max_map_count;
338 
339 #include <linux/aio.h>
340 
341 extern unsigned long
342 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
343 		       unsigned long, unsigned long);
344 extern unsigned long
345 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
346 			  unsigned long len, unsigned long pgoff,
347 			  unsigned long flags);
348 extern void arch_unmap_area(struct mm_struct *, unsigned long);
349 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
350 
351 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
352 /*
353  * The mm counters are not protected by its page_table_lock,
354  * so must be incremented atomically.
355  */
356 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
357 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
358 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
359 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
360 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
361 
362 #else  /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
363 /*
364  * The mm counters are protected by its page_table_lock,
365  * so can be incremented directly.
366  */
367 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
368 #define get_mm_counter(mm, member) ((mm)->_##member)
369 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
370 #define inc_mm_counter(mm, member) (mm)->_##member++
371 #define dec_mm_counter(mm, member) (mm)->_##member--
372 
373 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
374 
375 #define get_mm_rss(mm)					\
376 	(get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
377 #define update_hiwater_rss(mm)	do {			\
378 	unsigned long _rss = get_mm_rss(mm);		\
379 	if ((mm)->hiwater_rss < _rss)			\
380 		(mm)->hiwater_rss = _rss;		\
381 } while (0)
382 #define update_hiwater_vm(mm)	do {			\
383 	if ((mm)->hiwater_vm < (mm)->total_vm)		\
384 		(mm)->hiwater_vm = (mm)->total_vm;	\
385 } while (0)
386 
387 extern void set_dumpable(struct mm_struct *mm, int value);
388 extern int get_dumpable(struct mm_struct *mm);
389 
390 /* mm flags */
391 /* dumpable bits */
392 #define MMF_DUMPABLE      0  /* core dump is permitted */
393 #define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
394 #define MMF_DUMPABLE_BITS 2
395 
396 /* coredump filter bits */
397 #define MMF_DUMP_ANON_PRIVATE	2
398 #define MMF_DUMP_ANON_SHARED	3
399 #define MMF_DUMP_MAPPED_PRIVATE	4
400 #define MMF_DUMP_MAPPED_SHARED	5
401 #define MMF_DUMP_ELF_HEADERS	6
402 #define MMF_DUMP_FILTER_SHIFT	MMF_DUMPABLE_BITS
403 #define MMF_DUMP_FILTER_BITS	5
404 #define MMF_DUMP_FILTER_MASK \
405 	(((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
406 #define MMF_DUMP_FILTER_DEFAULT \
407 	((1 << MMF_DUMP_ANON_PRIVATE) |	(1 << MMF_DUMP_ANON_SHARED))
408 
409 struct sighand_struct {
410 	atomic_t		count;
411 	struct k_sigaction	action[_NSIG];
412 	spinlock_t		siglock;
413 	wait_queue_head_t	signalfd_wqh;
414 };
415 
416 struct pacct_struct {
417 	int			ac_flag;
418 	long			ac_exitcode;
419 	unsigned long		ac_mem;
420 	cputime_t		ac_utime, ac_stime;
421 	unsigned long		ac_minflt, ac_majflt;
422 };
423 
424 /*
425  * NOTE! "signal_struct" does not have it's own
426  * locking, because a shared signal_struct always
427  * implies a shared sighand_struct, so locking
428  * sighand_struct is always a proper superset of
429  * the locking of signal_struct.
430  */
431 struct signal_struct {
432 	atomic_t		count;
433 	atomic_t		live;
434 
435 	wait_queue_head_t	wait_chldexit;	/* for wait4() */
436 
437 	/* current thread group signal load-balancing target: */
438 	struct task_struct	*curr_target;
439 
440 	/* shared signal handling: */
441 	struct sigpending	shared_pending;
442 
443 	/* thread group exit support */
444 	int			group_exit_code;
445 	/* overloaded:
446 	 * - notify group_exit_task when ->count is equal to notify_count
447 	 * - everyone except group_exit_task is stopped during signal delivery
448 	 *   of fatal signals, group_exit_task processes the signal.
449 	 */
450 	struct task_struct	*group_exit_task;
451 	int			notify_count;
452 
453 	/* thread group stop support, overloads group_exit_code too */
454 	int			group_stop_count;
455 	unsigned int		flags; /* see SIGNAL_* flags below */
456 
457 	/* POSIX.1b Interval Timers */
458 	struct list_head posix_timers;
459 
460 	/* ITIMER_REAL timer for the process */
461 	struct hrtimer real_timer;
462 	struct task_struct *tsk;
463 	ktime_t it_real_incr;
464 
465 	/* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
466 	cputime_t it_prof_expires, it_virt_expires;
467 	cputime_t it_prof_incr, it_virt_incr;
468 
469 	/* job control IDs */
470 
471 	/*
472 	 * pgrp and session fields are deprecated.
473 	 * use the task_session_Xnr and task_pgrp_Xnr routines below
474 	 */
475 
476 	union {
477 		pid_t pgrp __deprecated;
478 		pid_t __pgrp;
479 	};
480 
481 	struct pid *tty_old_pgrp;
482 
483 	union {
484 		pid_t session __deprecated;
485 		pid_t __session;
486 	};
487 
488 	/* boolean value for session group leader */
489 	int leader;
490 
491 	struct tty_struct *tty; /* NULL if no tty */
492 
493 	/*
494 	 * Cumulative resource counters for dead threads in the group,
495 	 * and for reaped dead child processes forked by this group.
496 	 * Live threads maintain their own counters and add to these
497 	 * in __exit_signal, except for the group leader.
498 	 */
499 	cputime_t utime, stime, cutime, cstime;
500 	cputime_t gtime;
501 	cputime_t cgtime;
502 	unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
503 	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
504 	unsigned long inblock, oublock, cinblock, coublock;
505 
506 	/*
507 	 * Cumulative ns of scheduled CPU time for dead threads in the
508 	 * group, not including a zombie group leader.  (This only differs
509 	 * from jiffies_to_ns(utime + stime) if sched_clock uses something
510 	 * other than jiffies.)
511 	 */
512 	unsigned long long sum_sched_runtime;
513 
514 	/*
515 	 * We don't bother to synchronize most readers of this at all,
516 	 * because there is no reader checking a limit that actually needs
517 	 * to get both rlim_cur and rlim_max atomically, and either one
518 	 * alone is a single word that can safely be read normally.
519 	 * getrlimit/setrlimit use task_lock(current->group_leader) to
520 	 * protect this instead of the siglock, because they really
521 	 * have no need to disable irqs.
522 	 */
523 	struct rlimit rlim[RLIM_NLIMITS];
524 
525 	struct list_head cpu_timers[3];
526 
527 	/* keep the process-shared keyrings here so that they do the right
528 	 * thing in threads created with CLONE_THREAD */
529 #ifdef CONFIG_KEYS
530 	struct key *session_keyring;	/* keyring inherited over fork */
531 	struct key *process_keyring;	/* keyring private to this process */
532 #endif
533 #ifdef CONFIG_BSD_PROCESS_ACCT
534 	struct pacct_struct pacct;	/* per-process accounting information */
535 #endif
536 #ifdef CONFIG_TASKSTATS
537 	struct taskstats *stats;
538 #endif
539 #ifdef CONFIG_AUDIT
540 	unsigned audit_tty;
541 	struct tty_audit_buf *tty_audit_buf;
542 #endif
543 };
544 
545 /* Context switch must be unlocked if interrupts are to be enabled */
546 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
547 # define __ARCH_WANT_UNLOCKED_CTXSW
548 #endif
549 
550 /*
551  * Bits in flags field of signal_struct.
552  */
553 #define SIGNAL_STOP_STOPPED	0x00000001 /* job control stop in effect */
554 #define SIGNAL_STOP_DEQUEUED	0x00000002 /* stop signal dequeued */
555 #define SIGNAL_STOP_CONTINUED	0x00000004 /* SIGCONT since WCONTINUED reap */
556 #define SIGNAL_GROUP_EXIT	0x00000008 /* group exit in progress */
557 
558 /*
559  * Some day this will be a full-fledged user tracking system..
560  */
561 struct user_struct {
562 	atomic_t __count;	/* reference count */
563 	atomic_t processes;	/* How many processes does this user have? */
564 	atomic_t files;		/* How many open files does this user have? */
565 	atomic_t sigpending;	/* How many pending signals does this user have? */
566 #ifdef CONFIG_INOTIFY_USER
567 	atomic_t inotify_watches; /* How many inotify watches does this user have? */
568 	atomic_t inotify_devs;	/* How many inotify devs does this user have opened? */
569 #endif
570 #ifdef CONFIG_POSIX_MQUEUE
571 	/* protected by mq_lock	*/
572 	unsigned long mq_bytes;	/* How many bytes can be allocated to mqueue? */
573 #endif
574 	unsigned long locked_shm; /* How many pages of mlocked shm ? */
575 
576 #ifdef CONFIG_KEYS
577 	struct key *uid_keyring;	/* UID specific keyring */
578 	struct key *session_keyring;	/* UID's default session keyring */
579 #endif
580 
581 	/* Hash table maintenance information */
582 	struct hlist_node uidhash_node;
583 	uid_t uid;
584 
585 #ifdef CONFIG_FAIR_USER_SCHED
586 	struct task_group *tg;
587 #ifdef CONFIG_SYSFS
588 	struct kobject kobj;
589 	struct work_struct work;
590 #endif
591 #endif
592 };
593 
594 extern int uids_sysfs_init(void);
595 
596 extern struct user_struct *find_user(uid_t);
597 
598 extern struct user_struct root_user;
599 #define INIT_USER (&root_user)
600 
601 struct backing_dev_info;
602 struct reclaim_state;
603 
604 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
605 struct sched_info {
606 	/* cumulative counters */
607 	unsigned long pcount;	      /* # of times run on this cpu */
608 	unsigned long long cpu_time,  /* time spent on the cpu */
609 			   run_delay; /* time spent waiting on a runqueue */
610 
611 	/* timestamps */
612 	unsigned long long last_arrival,/* when we last ran on a cpu */
613 			   last_queued;	/* when we were last queued to run */
614 #ifdef CONFIG_SCHEDSTATS
615 	/* BKL stats */
616 	unsigned int bkl_count;
617 #endif
618 };
619 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
620 
621 #ifdef CONFIG_SCHEDSTATS
622 extern const struct file_operations proc_schedstat_operations;
623 #endif /* CONFIG_SCHEDSTATS */
624 
625 #ifdef CONFIG_TASK_DELAY_ACCT
626 struct task_delay_info {
627 	spinlock_t	lock;
628 	unsigned int	flags;	/* Private per-task flags */
629 
630 	/* For each stat XXX, add following, aligned appropriately
631 	 *
632 	 * struct timespec XXX_start, XXX_end;
633 	 * u64 XXX_delay;
634 	 * u32 XXX_count;
635 	 *
636 	 * Atomicity of updates to XXX_delay, XXX_count protected by
637 	 * single lock above (split into XXX_lock if contention is an issue).
638 	 */
639 
640 	/*
641 	 * XXX_count is incremented on every XXX operation, the delay
642 	 * associated with the operation is added to XXX_delay.
643 	 * XXX_delay contains the accumulated delay time in nanoseconds.
644 	 */
645 	struct timespec blkio_start, blkio_end;	/* Shared by blkio, swapin */
646 	u64 blkio_delay;	/* wait for sync block io completion */
647 	u64 swapin_delay;	/* wait for swapin block io completion */
648 	u32 blkio_count;	/* total count of the number of sync block */
649 				/* io operations performed */
650 	u32 swapin_count;	/* total count of the number of swapin block */
651 				/* io operations performed */
652 };
653 #endif	/* CONFIG_TASK_DELAY_ACCT */
654 
655 static inline int sched_info_on(void)
656 {
657 #ifdef CONFIG_SCHEDSTATS
658 	return 1;
659 #elif defined(CONFIG_TASK_DELAY_ACCT)
660 	extern int delayacct_on;
661 	return delayacct_on;
662 #else
663 	return 0;
664 #endif
665 }
666 
667 enum cpu_idle_type {
668 	CPU_IDLE,
669 	CPU_NOT_IDLE,
670 	CPU_NEWLY_IDLE,
671 	CPU_MAX_IDLE_TYPES
672 };
673 
674 /*
675  * sched-domains (multiprocessor balancing) declarations:
676  */
677 
678 /*
679  * Increase resolution of nice-level calculations:
680  */
681 #define SCHED_LOAD_SHIFT	10
682 #define SCHED_LOAD_SCALE	(1L << SCHED_LOAD_SHIFT)
683 
684 #define SCHED_LOAD_SCALE_FUZZ	SCHED_LOAD_SCALE
685 
686 #ifdef CONFIG_SMP
687 #define SD_LOAD_BALANCE		1	/* Do load balancing on this domain. */
688 #define SD_BALANCE_NEWIDLE	2	/* Balance when about to become idle */
689 #define SD_BALANCE_EXEC		4	/* Balance on exec */
690 #define SD_BALANCE_FORK		8	/* Balance on fork, clone */
691 #define SD_WAKE_IDLE		16	/* Wake to idle CPU on task wakeup */
692 #define SD_WAKE_AFFINE		32	/* Wake task to waking CPU */
693 #define SD_WAKE_BALANCE		64	/* Perform balancing at task wakeup */
694 #define SD_SHARE_CPUPOWER	128	/* Domain members share cpu power */
695 #define SD_POWERSAVINGS_BALANCE	256	/* Balance for power savings */
696 #define SD_SHARE_PKG_RESOURCES	512	/* Domain members share cpu pkg resources */
697 #define SD_SERIALIZE		1024	/* Only a single load balancing instance */
698 
699 #define BALANCE_FOR_MC_POWER	\
700 	(sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
701 
702 #define BALANCE_FOR_PKG_POWER	\
703 	((sched_mc_power_savings || sched_smt_power_savings) ?	\
704 	 SD_POWERSAVINGS_BALANCE : 0)
705 
706 #define test_sd_parent(sd, flag)	((sd->parent &&		\
707 					 (sd->parent->flags & flag)) ? 1 : 0)
708 
709 
710 struct sched_group {
711 	struct sched_group *next;	/* Must be a circular list */
712 	cpumask_t cpumask;
713 
714 	/*
715 	 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
716 	 * single CPU. This is read only (except for setup, hotplug CPU).
717 	 * Note : Never change cpu_power without recompute its reciprocal
718 	 */
719 	unsigned int __cpu_power;
720 	/*
721 	 * reciprocal value of cpu_power to avoid expensive divides
722 	 * (see include/linux/reciprocal_div.h)
723 	 */
724 	u32 reciprocal_cpu_power;
725 };
726 
727 struct sched_domain {
728 	/* These fields must be setup */
729 	struct sched_domain *parent;	/* top domain must be null terminated */
730 	struct sched_domain *child;	/* bottom domain must be null terminated */
731 	struct sched_group *groups;	/* the balancing groups of the domain */
732 	cpumask_t span;			/* span of all CPUs in this domain */
733 	unsigned long min_interval;	/* Minimum balance interval ms */
734 	unsigned long max_interval;	/* Maximum balance interval ms */
735 	unsigned int busy_factor;	/* less balancing by factor if busy */
736 	unsigned int imbalance_pct;	/* No balance until over watermark */
737 	unsigned int cache_nice_tries;	/* Leave cache hot tasks for # tries */
738 	unsigned int busy_idx;
739 	unsigned int idle_idx;
740 	unsigned int newidle_idx;
741 	unsigned int wake_idx;
742 	unsigned int forkexec_idx;
743 	int flags;			/* See SD_* */
744 
745 	/* Runtime fields. */
746 	unsigned long last_balance;	/* init to jiffies. units in jiffies */
747 	unsigned int balance_interval;	/* initialise to 1. units in ms. */
748 	unsigned int nr_balance_failed; /* initialise to 0 */
749 
750 #ifdef CONFIG_SCHEDSTATS
751 	/* load_balance() stats */
752 	unsigned int lb_count[CPU_MAX_IDLE_TYPES];
753 	unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
754 	unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
755 	unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
756 	unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
757 	unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
758 	unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
759 	unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
760 
761 	/* Active load balancing */
762 	unsigned int alb_count;
763 	unsigned int alb_failed;
764 	unsigned int alb_pushed;
765 
766 	/* SD_BALANCE_EXEC stats */
767 	unsigned int sbe_count;
768 	unsigned int sbe_balanced;
769 	unsigned int sbe_pushed;
770 
771 	/* SD_BALANCE_FORK stats */
772 	unsigned int sbf_count;
773 	unsigned int sbf_balanced;
774 	unsigned int sbf_pushed;
775 
776 	/* try_to_wake_up() stats */
777 	unsigned int ttwu_wake_remote;
778 	unsigned int ttwu_move_affine;
779 	unsigned int ttwu_move_balance;
780 #endif
781 };
782 
783 extern void partition_sched_domains(int ndoms_new, cpumask_t *doms_new);
784 
785 #endif	/* CONFIG_SMP */
786 
787 /*
788  * A runqueue laden with a single nice 0 task scores a weighted_cpuload of
789  * SCHED_LOAD_SCALE. This function returns 1 if any cpu is laden with a
790  * task of nice 0 or enough lower priority tasks to bring up the
791  * weighted_cpuload
792  */
793 static inline int above_background_load(void)
794 {
795 	unsigned long cpu;
796 
797 	for_each_online_cpu(cpu) {
798 		if (weighted_cpuload(cpu) >= SCHED_LOAD_SCALE)
799 			return 1;
800 	}
801 	return 0;
802 }
803 
804 struct io_context;			/* See blkdev.h */
805 #define NGROUPS_SMALL		32
806 #define NGROUPS_PER_BLOCK	((int)(PAGE_SIZE / sizeof(gid_t)))
807 struct group_info {
808 	int ngroups;
809 	atomic_t usage;
810 	gid_t small_block[NGROUPS_SMALL];
811 	int nblocks;
812 	gid_t *blocks[0];
813 };
814 
815 /*
816  * get_group_info() must be called with the owning task locked (via task_lock())
817  * when task != current.  The reason being that the vast majority of callers are
818  * looking at current->group_info, which can not be changed except by the
819  * current task.  Changing current->group_info requires the task lock, too.
820  */
821 #define get_group_info(group_info) do { \
822 	atomic_inc(&(group_info)->usage); \
823 } while (0)
824 
825 #define put_group_info(group_info) do { \
826 	if (atomic_dec_and_test(&(group_info)->usage)) \
827 		groups_free(group_info); \
828 } while (0)
829 
830 extern struct group_info *groups_alloc(int gidsetsize);
831 extern void groups_free(struct group_info *group_info);
832 extern int set_current_groups(struct group_info *group_info);
833 extern int groups_search(struct group_info *group_info, gid_t grp);
834 /* access the groups "array" with this macro */
835 #define GROUP_AT(gi, i) \
836     ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
837 
838 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
839 extern void prefetch_stack(struct task_struct *t);
840 #else
841 static inline void prefetch_stack(struct task_struct *t) { }
842 #endif
843 
844 struct audit_context;		/* See audit.c */
845 struct mempolicy;
846 struct pipe_inode_info;
847 struct uts_namespace;
848 
849 struct rq;
850 struct sched_domain;
851 
852 struct sched_class {
853 	const struct sched_class *next;
854 
855 	void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
856 	void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
857 	void (*yield_task) (struct rq *rq);
858 	int  (*select_task_rq)(struct task_struct *p, int sync);
859 
860 	void (*check_preempt_curr) (struct rq *rq, struct task_struct *p);
861 
862 	struct task_struct * (*pick_next_task) (struct rq *rq);
863 	void (*put_prev_task) (struct rq *rq, struct task_struct *p);
864 
865 #ifdef CONFIG_SMP
866 	unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
867 			struct rq *busiest, unsigned long max_load_move,
868 			struct sched_domain *sd, enum cpu_idle_type idle,
869 			int *all_pinned, int *this_best_prio);
870 
871 	int (*move_one_task) (struct rq *this_rq, int this_cpu,
872 			      struct rq *busiest, struct sched_domain *sd,
873 			      enum cpu_idle_type idle);
874 	void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
875 	void (*post_schedule) (struct rq *this_rq);
876 	void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
877 #endif
878 
879 	void (*set_curr_task) (struct rq *rq);
880 	void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
881 	void (*task_new) (struct rq *rq, struct task_struct *p);
882 	void (*set_cpus_allowed)(struct task_struct *p, cpumask_t *newmask);
883 
884 	void (*join_domain)(struct rq *rq);
885 	void (*leave_domain)(struct rq *rq);
886 
887 	void (*switched_from) (struct rq *this_rq, struct task_struct *task,
888 			       int running);
889 	void (*switched_to) (struct rq *this_rq, struct task_struct *task,
890 			     int running);
891 	void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
892 			     int oldprio, int running);
893 };
894 
895 struct load_weight {
896 	unsigned long weight, inv_weight;
897 };
898 
899 /*
900  * CFS stats for a schedulable entity (task, task-group etc)
901  *
902  * Current field usage histogram:
903  *
904  *     4 se->block_start
905  *     4 se->run_node
906  *     4 se->sleep_start
907  *     6 se->load.weight
908  */
909 struct sched_entity {
910 	struct load_weight	load;		/* for load-balancing */
911 	struct rb_node		run_node;
912 	unsigned int		on_rq;
913 
914 	u64			exec_start;
915 	u64			sum_exec_runtime;
916 	u64			vruntime;
917 	u64			prev_sum_exec_runtime;
918 
919 #ifdef CONFIG_SCHEDSTATS
920 	u64			wait_start;
921 	u64			wait_max;
922 	u64			wait_count;
923 	u64			wait_sum;
924 
925 	u64			sleep_start;
926 	u64			sleep_max;
927 	s64			sum_sleep_runtime;
928 
929 	u64			block_start;
930 	u64			block_max;
931 	u64			exec_max;
932 	u64			slice_max;
933 
934 	u64			nr_migrations;
935 	u64			nr_migrations_cold;
936 	u64			nr_failed_migrations_affine;
937 	u64			nr_failed_migrations_running;
938 	u64			nr_failed_migrations_hot;
939 	u64			nr_forced_migrations;
940 	u64			nr_forced2_migrations;
941 
942 	u64			nr_wakeups;
943 	u64			nr_wakeups_sync;
944 	u64			nr_wakeups_migrate;
945 	u64			nr_wakeups_local;
946 	u64			nr_wakeups_remote;
947 	u64			nr_wakeups_affine;
948 	u64			nr_wakeups_affine_attempts;
949 	u64			nr_wakeups_passive;
950 	u64			nr_wakeups_idle;
951 #endif
952 
953 #ifdef CONFIG_FAIR_GROUP_SCHED
954 	struct sched_entity	*parent;
955 	/* rq on which this entity is (to be) queued: */
956 	struct cfs_rq		*cfs_rq;
957 	/* rq "owned" by this entity/group: */
958 	struct cfs_rq		*my_q;
959 #endif
960 };
961 
962 struct sched_rt_entity {
963 	struct list_head run_list;
964 	unsigned int time_slice;
965 	unsigned long timeout;
966 	int nr_cpus_allowed;
967 
968 #ifdef CONFIG_FAIR_GROUP_SCHED
969 	struct sched_rt_entity	*parent;
970 	/* rq on which this entity is (to be) queued: */
971 	struct rt_rq		*rt_rq;
972 	/* rq "owned" by this entity/group: */
973 	struct rt_rq		*my_q;
974 #endif
975 };
976 
977 struct task_struct {
978 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
979 	void *stack;
980 	atomic_t usage;
981 	unsigned int flags;	/* per process flags, defined below */
982 	unsigned int ptrace;
983 
984 	int lock_depth;		/* BKL lock depth */
985 
986 #ifdef CONFIG_SMP
987 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
988 	int oncpu;
989 #endif
990 #endif
991 
992 	int prio, static_prio, normal_prio;
993 	const struct sched_class *sched_class;
994 	struct sched_entity se;
995 	struct sched_rt_entity rt;
996 
997 #ifdef CONFIG_PREEMPT_NOTIFIERS
998 	/* list of struct preempt_notifier: */
999 	struct hlist_head preempt_notifiers;
1000 #endif
1001 
1002 	/*
1003 	 * fpu_counter contains the number of consecutive context switches
1004 	 * that the FPU is used. If this is over a threshold, the lazy fpu
1005 	 * saving becomes unlazy to save the trap. This is an unsigned char
1006 	 * so that after 256 times the counter wraps and the behavior turns
1007 	 * lazy again; this to deal with bursty apps that only use FPU for
1008 	 * a short time
1009 	 */
1010 	unsigned char fpu_counter;
1011 	s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1012 #ifdef CONFIG_BLK_DEV_IO_TRACE
1013 	unsigned int btrace_seq;
1014 #endif
1015 
1016 	unsigned int policy;
1017 	cpumask_t cpus_allowed;
1018 
1019 #ifdef CONFIG_PREEMPT_RCU
1020 	int rcu_read_lock_nesting;
1021 	int rcu_flipctr_idx;
1022 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1023 
1024 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1025 	struct sched_info sched_info;
1026 #endif
1027 
1028 	struct list_head tasks;
1029 	/*
1030 	 * ptrace_list/ptrace_children forms the list of my children
1031 	 * that were stolen by a ptracer.
1032 	 */
1033 	struct list_head ptrace_children;
1034 	struct list_head ptrace_list;
1035 
1036 	struct mm_struct *mm, *active_mm;
1037 
1038 /* task state */
1039 	struct linux_binfmt *binfmt;
1040 	int exit_state;
1041 	int exit_code, exit_signal;
1042 	int pdeath_signal;  /*  The signal sent when the parent dies  */
1043 	/* ??? */
1044 	unsigned int personality;
1045 	unsigned did_exec:1;
1046 	pid_t pid;
1047 	pid_t tgid;
1048 
1049 #ifdef CONFIG_CC_STACKPROTECTOR
1050 	/* Canary value for the -fstack-protector gcc feature */
1051 	unsigned long stack_canary;
1052 #endif
1053 	/*
1054 	 * pointers to (original) parent process, youngest child, younger sibling,
1055 	 * older sibling, respectively.  (p->father can be replaced with
1056 	 * p->parent->pid)
1057 	 */
1058 	struct task_struct *real_parent; /* real parent process (when being debugged) */
1059 	struct task_struct *parent;	/* parent process */
1060 	/*
1061 	 * children/sibling forms the list of my children plus the
1062 	 * tasks I'm ptracing.
1063 	 */
1064 	struct list_head children;	/* list of my children */
1065 	struct list_head sibling;	/* linkage in my parent's children list */
1066 	struct task_struct *group_leader;	/* threadgroup leader */
1067 
1068 	/* PID/PID hash table linkage. */
1069 	struct pid_link pids[PIDTYPE_MAX];
1070 	struct list_head thread_group;
1071 
1072 	struct completion *vfork_done;		/* for vfork() */
1073 	int __user *set_child_tid;		/* CLONE_CHILD_SETTID */
1074 	int __user *clear_child_tid;		/* CLONE_CHILD_CLEARTID */
1075 
1076 	unsigned int rt_priority;
1077 	cputime_t utime, stime, utimescaled, stimescaled;
1078 	cputime_t gtime;
1079 	cputime_t prev_utime, prev_stime;
1080 	unsigned long nvcsw, nivcsw; /* context switch counts */
1081 	struct timespec start_time; 		/* monotonic time */
1082 	struct timespec real_start_time;	/* boot based time */
1083 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1084 	unsigned long min_flt, maj_flt;
1085 
1086   	cputime_t it_prof_expires, it_virt_expires;
1087 	unsigned long long it_sched_expires;
1088 	struct list_head cpu_timers[3];
1089 
1090 /* process credentials */
1091 	uid_t uid,euid,suid,fsuid;
1092 	gid_t gid,egid,sgid,fsgid;
1093 	struct group_info *group_info;
1094 	kernel_cap_t   cap_effective, cap_inheritable, cap_permitted;
1095 	unsigned keep_capabilities:1;
1096 	struct user_struct *user;
1097 #ifdef CONFIG_KEYS
1098 	struct key *request_key_auth;	/* assumed request_key authority */
1099 	struct key *thread_keyring;	/* keyring private to this thread */
1100 	unsigned char jit_keyring;	/* default keyring to attach requested keys to */
1101 #endif
1102 	char comm[TASK_COMM_LEN]; /* executable name excluding path
1103 				     - access with [gs]et_task_comm (which lock
1104 				       it with task_lock())
1105 				     - initialized normally by flush_old_exec */
1106 /* file system info */
1107 	int link_count, total_link_count;
1108 #ifdef CONFIG_SYSVIPC
1109 /* ipc stuff */
1110 	struct sysv_sem sysvsem;
1111 #endif
1112 #ifdef CONFIG_DETECT_SOFTLOCKUP
1113 /* hung task detection */
1114 	unsigned long last_switch_timestamp;
1115 	unsigned long last_switch_count;
1116 #endif
1117 /* CPU-specific state of this task */
1118 	struct thread_struct thread;
1119 /* filesystem information */
1120 	struct fs_struct *fs;
1121 /* open file information */
1122 	struct files_struct *files;
1123 /* namespaces */
1124 	struct nsproxy *nsproxy;
1125 /* signal handlers */
1126 	struct signal_struct *signal;
1127 	struct sighand_struct *sighand;
1128 
1129 	sigset_t blocked, real_blocked;
1130 	sigset_t saved_sigmask;		/* To be restored with TIF_RESTORE_SIGMASK */
1131 	struct sigpending pending;
1132 
1133 	unsigned long sas_ss_sp;
1134 	size_t sas_ss_size;
1135 	int (*notifier)(void *priv);
1136 	void *notifier_data;
1137 	sigset_t *notifier_mask;
1138 #ifdef CONFIG_SECURITY
1139 	void *security;
1140 #endif
1141 	struct audit_context *audit_context;
1142 #ifdef CONFIG_AUDITSYSCALL
1143 	uid_t loginuid;
1144 	unsigned int sessionid;
1145 #endif
1146 	seccomp_t seccomp;
1147 
1148 /* Thread group tracking */
1149    	u32 parent_exec_id;
1150    	u32 self_exec_id;
1151 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1152 	spinlock_t alloc_lock;
1153 
1154 	/* Protection of the PI data structures: */
1155 	spinlock_t pi_lock;
1156 
1157 #ifdef CONFIG_RT_MUTEXES
1158 	/* PI waiters blocked on a rt_mutex held by this task */
1159 	struct plist_head pi_waiters;
1160 	/* Deadlock detection and priority inheritance handling */
1161 	struct rt_mutex_waiter *pi_blocked_on;
1162 #endif
1163 
1164 #ifdef CONFIG_DEBUG_MUTEXES
1165 	/* mutex deadlock detection */
1166 	struct mutex_waiter *blocked_on;
1167 #endif
1168 #ifdef CONFIG_TRACE_IRQFLAGS
1169 	unsigned int irq_events;
1170 	int hardirqs_enabled;
1171 	unsigned long hardirq_enable_ip;
1172 	unsigned int hardirq_enable_event;
1173 	unsigned long hardirq_disable_ip;
1174 	unsigned int hardirq_disable_event;
1175 	int softirqs_enabled;
1176 	unsigned long softirq_disable_ip;
1177 	unsigned int softirq_disable_event;
1178 	unsigned long softirq_enable_ip;
1179 	unsigned int softirq_enable_event;
1180 	int hardirq_context;
1181 	int softirq_context;
1182 #endif
1183 #ifdef CONFIG_LOCKDEP
1184 # define MAX_LOCK_DEPTH 30UL
1185 	u64 curr_chain_key;
1186 	int lockdep_depth;
1187 	struct held_lock held_locks[MAX_LOCK_DEPTH];
1188 	unsigned int lockdep_recursion;
1189 #endif
1190 
1191 /* journalling filesystem info */
1192 	void *journal_info;
1193 
1194 /* stacked block device info */
1195 	struct bio *bio_list, **bio_tail;
1196 
1197 /* VM state */
1198 	struct reclaim_state *reclaim_state;
1199 
1200 	struct backing_dev_info *backing_dev_info;
1201 
1202 	struct io_context *io_context;
1203 
1204 	unsigned long ptrace_message;
1205 	siginfo_t *last_siginfo; /* For ptrace use.  */
1206 #ifdef CONFIG_TASK_XACCT
1207 /* i/o counters(bytes read/written, #syscalls */
1208 	u64 rchar, wchar, syscr, syscw;
1209 #endif
1210 	struct task_io_accounting ioac;
1211 #if defined(CONFIG_TASK_XACCT)
1212 	u64 acct_rss_mem1;	/* accumulated rss usage */
1213 	u64 acct_vm_mem1;	/* accumulated virtual memory usage */
1214 	cputime_t acct_stimexpd;/* stime since last update */
1215 #endif
1216 #ifdef CONFIG_NUMA
1217   	struct mempolicy *mempolicy;
1218 	short il_next;
1219 #endif
1220 #ifdef CONFIG_CPUSETS
1221 	nodemask_t mems_allowed;
1222 	int cpuset_mems_generation;
1223 	int cpuset_mem_spread_rotor;
1224 #endif
1225 #ifdef CONFIG_CGROUPS
1226 	/* Control Group info protected by css_set_lock */
1227 	struct css_set *cgroups;
1228 	/* cg_list protected by css_set_lock and tsk->alloc_lock */
1229 	struct list_head cg_list;
1230 #endif
1231 #ifdef CONFIG_FUTEX
1232 	struct robust_list_head __user *robust_list;
1233 #ifdef CONFIG_COMPAT
1234 	struct compat_robust_list_head __user *compat_robust_list;
1235 #endif
1236 	struct list_head pi_state_list;
1237 	struct futex_pi_state *pi_state_cache;
1238 #endif
1239 	atomic_t fs_excl;	/* holding fs exclusive resources */
1240 	struct rcu_head rcu;
1241 
1242 	/*
1243 	 * cache last used pipe for splice
1244 	 */
1245 	struct pipe_inode_info *splice_pipe;
1246 #ifdef	CONFIG_TASK_DELAY_ACCT
1247 	struct task_delay_info *delays;
1248 #endif
1249 #ifdef CONFIG_FAULT_INJECTION
1250 	int make_it_fail;
1251 #endif
1252 	struct prop_local_single dirties;
1253 #ifdef CONFIG_LATENCYTOP
1254 	int latency_record_count;
1255 	struct latency_record latency_record[LT_SAVECOUNT];
1256 #endif
1257 };
1258 
1259 /*
1260  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1261  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1262  * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1263  * values are inverted: lower p->prio value means higher priority.
1264  *
1265  * The MAX_USER_RT_PRIO value allows the actual maximum
1266  * RT priority to be separate from the value exported to
1267  * user-space.  This allows kernel threads to set their
1268  * priority to a value higher than any user task. Note:
1269  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1270  */
1271 
1272 #define MAX_USER_RT_PRIO	100
1273 #define MAX_RT_PRIO		MAX_USER_RT_PRIO
1274 
1275 #define MAX_PRIO		(MAX_RT_PRIO + 40)
1276 #define DEFAULT_PRIO		(MAX_RT_PRIO + 20)
1277 
1278 static inline int rt_prio(int prio)
1279 {
1280 	if (unlikely(prio < MAX_RT_PRIO))
1281 		return 1;
1282 	return 0;
1283 }
1284 
1285 static inline int rt_task(struct task_struct *p)
1286 {
1287 	return rt_prio(p->prio);
1288 }
1289 
1290 static inline void set_task_session(struct task_struct *tsk, pid_t session)
1291 {
1292 	tsk->signal->__session = session;
1293 }
1294 
1295 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1296 {
1297 	tsk->signal->__pgrp = pgrp;
1298 }
1299 
1300 static inline struct pid *task_pid(struct task_struct *task)
1301 {
1302 	return task->pids[PIDTYPE_PID].pid;
1303 }
1304 
1305 static inline struct pid *task_tgid(struct task_struct *task)
1306 {
1307 	return task->group_leader->pids[PIDTYPE_PID].pid;
1308 }
1309 
1310 static inline struct pid *task_pgrp(struct task_struct *task)
1311 {
1312 	return task->group_leader->pids[PIDTYPE_PGID].pid;
1313 }
1314 
1315 static inline struct pid *task_session(struct task_struct *task)
1316 {
1317 	return task->group_leader->pids[PIDTYPE_SID].pid;
1318 }
1319 
1320 struct pid_namespace;
1321 
1322 /*
1323  * the helpers to get the task's different pids as they are seen
1324  * from various namespaces
1325  *
1326  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1327  * task_xid_vnr()    : virtual id, i.e. the id seen from the namespace the task
1328  *                     belongs to. this only makes sence when called in the
1329  *                     context of the task that belongs to the same namespace;
1330  * task_xid_nr_ns()  : id seen from the ns specified;
1331  *
1332  * set_task_vxid()   : assigns a virtual id to a task;
1333  *
1334  * see also pid_nr() etc in include/linux/pid.h
1335  */
1336 
1337 static inline pid_t task_pid_nr(struct task_struct *tsk)
1338 {
1339 	return tsk->pid;
1340 }
1341 
1342 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1343 
1344 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1345 {
1346 	return pid_vnr(task_pid(tsk));
1347 }
1348 
1349 
1350 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1351 {
1352 	return tsk->tgid;
1353 }
1354 
1355 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1356 
1357 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1358 {
1359 	return pid_vnr(task_tgid(tsk));
1360 }
1361 
1362 
1363 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1364 {
1365 	return tsk->signal->__pgrp;
1366 }
1367 
1368 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1369 
1370 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1371 {
1372 	return pid_vnr(task_pgrp(tsk));
1373 }
1374 
1375 
1376 static inline pid_t task_session_nr(struct task_struct *tsk)
1377 {
1378 	return tsk->signal->__session;
1379 }
1380 
1381 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1382 
1383 static inline pid_t task_session_vnr(struct task_struct *tsk)
1384 {
1385 	return pid_vnr(task_session(tsk));
1386 }
1387 
1388 
1389 /**
1390  * pid_alive - check that a task structure is not stale
1391  * @p: Task structure to be checked.
1392  *
1393  * Test if a process is not yet dead (at most zombie state)
1394  * If pid_alive fails, then pointers within the task structure
1395  * can be stale and must not be dereferenced.
1396  */
1397 static inline int pid_alive(struct task_struct *p)
1398 {
1399 	return p->pids[PIDTYPE_PID].pid != NULL;
1400 }
1401 
1402 /**
1403  * is_global_init - check if a task structure is init
1404  * @tsk: Task structure to be checked.
1405  *
1406  * Check if a task structure is the first user space task the kernel created.
1407  */
1408 static inline int is_global_init(struct task_struct *tsk)
1409 {
1410 	return tsk->pid == 1;
1411 }
1412 
1413 /*
1414  * is_container_init:
1415  * check whether in the task is init in its own pid namespace.
1416  */
1417 extern int is_container_init(struct task_struct *tsk);
1418 
1419 extern struct pid *cad_pid;
1420 
1421 extern void free_task(struct task_struct *tsk);
1422 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1423 
1424 extern void __put_task_struct(struct task_struct *t);
1425 
1426 static inline void put_task_struct(struct task_struct *t)
1427 {
1428 	if (atomic_dec_and_test(&t->usage))
1429 		__put_task_struct(t);
1430 }
1431 
1432 /*
1433  * Per process flags
1434  */
1435 #define PF_ALIGNWARN	0x00000001	/* Print alignment warning msgs */
1436 					/* Not implemented yet, only for 486*/
1437 #define PF_STARTING	0x00000002	/* being created */
1438 #define PF_EXITING	0x00000004	/* getting shut down */
1439 #define PF_EXITPIDONE	0x00000008	/* pi exit done on shut down */
1440 #define PF_VCPU		0x00000010	/* I'm a virtual CPU */
1441 #define PF_FORKNOEXEC	0x00000040	/* forked but didn't exec */
1442 #define PF_SUPERPRIV	0x00000100	/* used super-user privileges */
1443 #define PF_DUMPCORE	0x00000200	/* dumped core */
1444 #define PF_SIGNALED	0x00000400	/* killed by a signal */
1445 #define PF_MEMALLOC	0x00000800	/* Allocating memory */
1446 #define PF_FLUSHER	0x00001000	/* responsible for disk writeback */
1447 #define PF_USED_MATH	0x00002000	/* if unset the fpu must be initialized before use */
1448 #define PF_NOFREEZE	0x00008000	/* this thread should not be frozen */
1449 #define PF_FROZEN	0x00010000	/* frozen for system suspend */
1450 #define PF_FSTRANS	0x00020000	/* inside a filesystem transaction */
1451 #define PF_KSWAPD	0x00040000	/* I am kswapd */
1452 #define PF_SWAPOFF	0x00080000	/* I am in swapoff */
1453 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
1454 #define PF_BORROWED_MM	0x00200000	/* I am a kthread doing use_mm */
1455 #define PF_RANDOMIZE	0x00400000	/* randomize virtual address space */
1456 #define PF_SWAPWRITE	0x00800000	/* Allowed to write to swap */
1457 #define PF_SPREAD_PAGE	0x01000000	/* Spread page cache over cpuset */
1458 #define PF_SPREAD_SLAB	0x02000000	/* Spread some slab caches over cpuset */
1459 #define PF_MEMPOLICY	0x10000000	/* Non-default NUMA mempolicy */
1460 #define PF_MUTEX_TESTER	0x20000000	/* Thread belongs to the rt mutex tester */
1461 #define PF_FREEZER_SKIP	0x40000000	/* Freezer should not count it as freezeable */
1462 
1463 /*
1464  * Only the _current_ task can read/write to tsk->flags, but other
1465  * tasks can access tsk->flags in readonly mode for example
1466  * with tsk_used_math (like during threaded core dumping).
1467  * There is however an exception to this rule during ptrace
1468  * or during fork: the ptracer task is allowed to write to the
1469  * child->flags of its traced child (same goes for fork, the parent
1470  * can write to the child->flags), because we're guaranteed the
1471  * child is not running and in turn not changing child->flags
1472  * at the same time the parent does it.
1473  */
1474 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1475 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1476 #define clear_used_math() clear_stopped_child_used_math(current)
1477 #define set_used_math() set_stopped_child_used_math(current)
1478 #define conditional_stopped_child_used_math(condition, child) \
1479 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1480 #define conditional_used_math(condition) \
1481 	conditional_stopped_child_used_math(condition, current)
1482 #define copy_to_stopped_child_used_math(child) \
1483 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1484 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1485 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1486 #define used_math() tsk_used_math(current)
1487 
1488 #ifdef CONFIG_SMP
1489 extern int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask);
1490 #else
1491 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1492 {
1493 	if (!cpu_isset(0, new_mask))
1494 		return -EINVAL;
1495 	return 0;
1496 }
1497 #endif
1498 
1499 extern unsigned long long sched_clock(void);
1500 
1501 /*
1502  * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1503  * clock constructed from sched_clock():
1504  */
1505 extern unsigned long long cpu_clock(int cpu);
1506 
1507 extern unsigned long long
1508 task_sched_runtime(struct task_struct *task);
1509 
1510 /* sched_exec is called by processes performing an exec */
1511 #ifdef CONFIG_SMP
1512 extern void sched_exec(void);
1513 #else
1514 #define sched_exec()   {}
1515 #endif
1516 
1517 extern void sched_clock_idle_sleep_event(void);
1518 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1519 
1520 #ifdef CONFIG_HOTPLUG_CPU
1521 extern void idle_task_exit(void);
1522 #else
1523 static inline void idle_task_exit(void) {}
1524 #endif
1525 
1526 extern void sched_idle_next(void);
1527 
1528 #ifdef CONFIG_SCHED_DEBUG
1529 extern unsigned int sysctl_sched_latency;
1530 extern unsigned int sysctl_sched_min_granularity;
1531 extern unsigned int sysctl_sched_wakeup_granularity;
1532 extern unsigned int sysctl_sched_batch_wakeup_granularity;
1533 extern unsigned int sysctl_sched_child_runs_first;
1534 extern unsigned int sysctl_sched_features;
1535 extern unsigned int sysctl_sched_migration_cost;
1536 extern unsigned int sysctl_sched_nr_migrate;
1537 extern unsigned int sysctl_sched_rt_period;
1538 extern unsigned int sysctl_sched_rt_ratio;
1539 #if defined(CONFIG_FAIR_GROUP_SCHED) && defined(CONFIG_SMP)
1540 extern unsigned int sysctl_sched_min_bal_int_shares;
1541 extern unsigned int sysctl_sched_max_bal_int_shares;
1542 #endif
1543 
1544 int sched_nr_latency_handler(struct ctl_table *table, int write,
1545 		struct file *file, void __user *buffer, size_t *length,
1546 		loff_t *ppos);
1547 #endif
1548 
1549 extern unsigned int sysctl_sched_compat_yield;
1550 
1551 #ifdef CONFIG_RT_MUTEXES
1552 extern int rt_mutex_getprio(struct task_struct *p);
1553 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1554 extern void rt_mutex_adjust_pi(struct task_struct *p);
1555 #else
1556 static inline int rt_mutex_getprio(struct task_struct *p)
1557 {
1558 	return p->normal_prio;
1559 }
1560 # define rt_mutex_adjust_pi(p)		do { } while (0)
1561 #endif
1562 
1563 extern void set_user_nice(struct task_struct *p, long nice);
1564 extern int task_prio(const struct task_struct *p);
1565 extern int task_nice(const struct task_struct *p);
1566 extern int can_nice(const struct task_struct *p, const int nice);
1567 extern int task_curr(const struct task_struct *p);
1568 extern int idle_cpu(int cpu);
1569 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1570 extern struct task_struct *idle_task(int cpu);
1571 extern struct task_struct *curr_task(int cpu);
1572 extern void set_curr_task(int cpu, struct task_struct *p);
1573 
1574 void yield(void);
1575 
1576 /*
1577  * The default (Linux) execution domain.
1578  */
1579 extern struct exec_domain	default_exec_domain;
1580 
1581 union thread_union {
1582 	struct thread_info thread_info;
1583 	unsigned long stack[THREAD_SIZE/sizeof(long)];
1584 };
1585 
1586 #ifndef __HAVE_ARCH_KSTACK_END
1587 static inline int kstack_end(void *addr)
1588 {
1589 	/* Reliable end of stack detection:
1590 	 * Some APM bios versions misalign the stack
1591 	 */
1592 	return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1593 }
1594 #endif
1595 
1596 extern union thread_union init_thread_union;
1597 extern struct task_struct init_task;
1598 
1599 extern struct   mm_struct init_mm;
1600 
1601 extern struct pid_namespace init_pid_ns;
1602 
1603 /*
1604  * find a task by one of its numerical ids
1605  *
1606  * find_task_by_pid_type_ns():
1607  *      it is the most generic call - it finds a task by all id,
1608  *      type and namespace specified
1609  * find_task_by_pid_ns():
1610  *      finds a task by its pid in the specified namespace
1611  * find_task_by_vpid():
1612  *      finds a task by its virtual pid
1613  * find_task_by_pid():
1614  *      finds a task by its global pid
1615  *
1616  * see also find_pid() etc in include/linux/pid.h
1617  */
1618 
1619 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1620 		struct pid_namespace *ns);
1621 
1622 extern struct task_struct *find_task_by_pid(pid_t nr);
1623 extern struct task_struct *find_task_by_vpid(pid_t nr);
1624 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1625 		struct pid_namespace *ns);
1626 
1627 extern void __set_special_pids(pid_t session, pid_t pgrp);
1628 
1629 /* per-UID process charging. */
1630 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
1631 static inline struct user_struct *get_uid(struct user_struct *u)
1632 {
1633 	atomic_inc(&u->__count);
1634 	return u;
1635 }
1636 extern void free_uid(struct user_struct *);
1637 extern void switch_uid(struct user_struct *);
1638 extern void release_uids(struct user_namespace *ns);
1639 
1640 #include <asm/current.h>
1641 
1642 extern void do_timer(unsigned long ticks);
1643 
1644 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1645 extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1646 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1647 						unsigned long clone_flags));
1648 #ifdef CONFIG_SMP
1649  extern void kick_process(struct task_struct *tsk);
1650 #else
1651  static inline void kick_process(struct task_struct *tsk) { }
1652 #endif
1653 extern void sched_fork(struct task_struct *p, int clone_flags);
1654 extern void sched_dead(struct task_struct *p);
1655 
1656 extern int in_group_p(gid_t);
1657 extern int in_egroup_p(gid_t);
1658 
1659 extern void proc_caches_init(void);
1660 extern void flush_signals(struct task_struct *);
1661 extern void ignore_signals(struct task_struct *);
1662 extern void flush_signal_handlers(struct task_struct *, int force_default);
1663 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1664 
1665 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1666 {
1667 	unsigned long flags;
1668 	int ret;
1669 
1670 	spin_lock_irqsave(&tsk->sighand->siglock, flags);
1671 	ret = dequeue_signal(tsk, mask, info);
1672 	spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1673 
1674 	return ret;
1675 }
1676 
1677 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1678 			      sigset_t *mask);
1679 extern void unblock_all_signals(void);
1680 extern void release_task(struct task_struct * p);
1681 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1682 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1683 extern int force_sigsegv(int, struct task_struct *);
1684 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1685 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1686 extern int kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1687 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1688 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1689 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1690 extern int kill_pid(struct pid *pid, int sig, int priv);
1691 extern int kill_proc_info(int, struct siginfo *, pid_t);
1692 extern void do_notify_parent(struct task_struct *, int);
1693 extern void force_sig(int, struct task_struct *);
1694 extern void force_sig_specific(int, struct task_struct *);
1695 extern int send_sig(int, struct task_struct *, int);
1696 extern void zap_other_threads(struct task_struct *p);
1697 extern int kill_proc(pid_t, int, int);
1698 extern struct sigqueue *sigqueue_alloc(void);
1699 extern void sigqueue_free(struct sigqueue *);
1700 extern int send_sigqueue(int, struct sigqueue *,  struct task_struct *);
1701 extern int send_group_sigqueue(int, struct sigqueue *,  struct task_struct *);
1702 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1703 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1704 
1705 static inline int kill_cad_pid(int sig, int priv)
1706 {
1707 	return kill_pid(cad_pid, sig, priv);
1708 }
1709 
1710 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
1711 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1712 #define SEND_SIG_PRIV	((struct siginfo *) 1)
1713 #define SEND_SIG_FORCED	((struct siginfo *) 2)
1714 
1715 static inline int is_si_special(const struct siginfo *info)
1716 {
1717 	return info <= SEND_SIG_FORCED;
1718 }
1719 
1720 /* True if we are on the alternate signal stack.  */
1721 
1722 static inline int on_sig_stack(unsigned long sp)
1723 {
1724 	return (sp - current->sas_ss_sp < current->sas_ss_size);
1725 }
1726 
1727 static inline int sas_ss_flags(unsigned long sp)
1728 {
1729 	return (current->sas_ss_size == 0 ? SS_DISABLE
1730 		: on_sig_stack(sp) ? SS_ONSTACK : 0);
1731 }
1732 
1733 /*
1734  * Routines for handling mm_structs
1735  */
1736 extern struct mm_struct * mm_alloc(void);
1737 
1738 /* mmdrop drops the mm and the page tables */
1739 extern void FASTCALL(__mmdrop(struct mm_struct *));
1740 static inline void mmdrop(struct mm_struct * mm)
1741 {
1742 	if (unlikely(atomic_dec_and_test(&mm->mm_count)))
1743 		__mmdrop(mm);
1744 }
1745 
1746 /* mmput gets rid of the mappings and all user-space */
1747 extern void mmput(struct mm_struct *);
1748 /* Grab a reference to a task's mm, if it is not already going away */
1749 extern struct mm_struct *get_task_mm(struct task_struct *task);
1750 /* Remove the current tasks stale references to the old mm_struct */
1751 extern void mm_release(struct task_struct *, struct mm_struct *);
1752 
1753 extern int  copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1754 extern void flush_thread(void);
1755 extern void exit_thread(void);
1756 
1757 extern void exit_files(struct task_struct *);
1758 extern void __cleanup_signal(struct signal_struct *);
1759 extern void __cleanup_sighand(struct sighand_struct *);
1760 extern void exit_itimers(struct signal_struct *);
1761 
1762 extern NORET_TYPE void do_group_exit(int);
1763 
1764 extern void daemonize(const char *, ...);
1765 extern int allow_signal(int);
1766 extern int disallow_signal(int);
1767 
1768 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1769 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1770 struct task_struct *fork_idle(int);
1771 
1772 extern void set_task_comm(struct task_struct *tsk, char *from);
1773 extern void get_task_comm(char *to, struct task_struct *tsk);
1774 
1775 #ifdef CONFIG_SMP
1776 extern void wait_task_inactive(struct task_struct * p);
1777 #else
1778 #define wait_task_inactive(p)	do { } while (0)
1779 #endif
1780 
1781 #define remove_parent(p)	list_del_init(&(p)->sibling)
1782 #define add_parent(p)		list_add_tail(&(p)->sibling,&(p)->parent->children)
1783 
1784 #define next_task(p)	list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1785 
1786 #define for_each_process(p) \
1787 	for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1788 
1789 /*
1790  * Careful: do_each_thread/while_each_thread is a double loop so
1791  *          'break' will not work as expected - use goto instead.
1792  */
1793 #define do_each_thread(g, t) \
1794 	for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1795 
1796 #define while_each_thread(g, t) \
1797 	while ((t = next_thread(t)) != g)
1798 
1799 /* de_thread depends on thread_group_leader not being a pid based check */
1800 #define thread_group_leader(p)	(p == p->group_leader)
1801 
1802 /* Do to the insanities of de_thread it is possible for a process
1803  * to have the pid of the thread group leader without actually being
1804  * the thread group leader.  For iteration through the pids in proc
1805  * all we care about is that we have a task with the appropriate
1806  * pid, we don't actually care if we have the right task.
1807  */
1808 static inline int has_group_leader_pid(struct task_struct *p)
1809 {
1810 	return p->pid == p->tgid;
1811 }
1812 
1813 static inline
1814 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
1815 {
1816 	return p1->tgid == p2->tgid;
1817 }
1818 
1819 static inline struct task_struct *next_thread(const struct task_struct *p)
1820 {
1821 	return list_entry(rcu_dereference(p->thread_group.next),
1822 			  struct task_struct, thread_group);
1823 }
1824 
1825 static inline int thread_group_empty(struct task_struct *p)
1826 {
1827 	return list_empty(&p->thread_group);
1828 }
1829 
1830 #define delay_group_leader(p) \
1831 		(thread_group_leader(p) && !thread_group_empty(p))
1832 
1833 /*
1834  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1835  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
1836  * pins the final release of task.io_context.  Also protects ->cpuset and
1837  * ->cgroup.subsys[].
1838  *
1839  * Nests both inside and outside of read_lock(&tasklist_lock).
1840  * It must not be nested with write_lock_irq(&tasklist_lock),
1841  * neither inside nor outside.
1842  */
1843 static inline void task_lock(struct task_struct *p)
1844 {
1845 	spin_lock(&p->alloc_lock);
1846 }
1847 
1848 static inline void task_unlock(struct task_struct *p)
1849 {
1850 	spin_unlock(&p->alloc_lock);
1851 }
1852 
1853 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1854 							unsigned long *flags);
1855 
1856 static inline void unlock_task_sighand(struct task_struct *tsk,
1857 						unsigned long *flags)
1858 {
1859 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1860 }
1861 
1862 #ifndef __HAVE_THREAD_FUNCTIONS
1863 
1864 #define task_thread_info(task)	((struct thread_info *)(task)->stack)
1865 #define task_stack_page(task)	((task)->stack)
1866 
1867 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1868 {
1869 	*task_thread_info(p) = *task_thread_info(org);
1870 	task_thread_info(p)->task = p;
1871 }
1872 
1873 static inline unsigned long *end_of_stack(struct task_struct *p)
1874 {
1875 	return (unsigned long *)(task_thread_info(p) + 1);
1876 }
1877 
1878 #endif
1879 
1880 /* set thread flags in other task's structures
1881  * - see asm/thread_info.h for TIF_xxxx flags available
1882  */
1883 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1884 {
1885 	set_ti_thread_flag(task_thread_info(tsk), flag);
1886 }
1887 
1888 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1889 {
1890 	clear_ti_thread_flag(task_thread_info(tsk), flag);
1891 }
1892 
1893 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1894 {
1895 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1896 }
1897 
1898 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1899 {
1900 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1901 }
1902 
1903 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1904 {
1905 	return test_ti_thread_flag(task_thread_info(tsk), flag);
1906 }
1907 
1908 static inline void set_tsk_need_resched(struct task_struct *tsk)
1909 {
1910 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1911 }
1912 
1913 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1914 {
1915 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1916 }
1917 
1918 static inline int signal_pending(struct task_struct *p)
1919 {
1920 	return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1921 }
1922 
1923 extern int FASTCALL(__fatal_signal_pending(struct task_struct *p));
1924 
1925 static inline int fatal_signal_pending(struct task_struct *p)
1926 {
1927 	return signal_pending(p) && __fatal_signal_pending(p);
1928 }
1929 
1930 static inline int need_resched(void)
1931 {
1932 	return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1933 }
1934 
1935 /*
1936  * cond_resched() and cond_resched_lock(): latency reduction via
1937  * explicit rescheduling in places that are safe. The return
1938  * value indicates whether a reschedule was done in fact.
1939  * cond_resched_lock() will drop the spinlock before scheduling,
1940  * cond_resched_softirq() will enable bhs before scheduling.
1941  */
1942 #ifdef CONFIG_PREEMPT
1943 static inline int cond_resched(void)
1944 {
1945 	return 0;
1946 }
1947 #else
1948 extern int _cond_resched(void);
1949 static inline int cond_resched(void)
1950 {
1951 	return _cond_resched();
1952 }
1953 #endif
1954 extern int cond_resched_lock(spinlock_t * lock);
1955 extern int cond_resched_softirq(void);
1956 
1957 /*
1958  * Does a critical section need to be broken due to another
1959  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
1960  * but a general need for low latency)
1961  */
1962 static inline int spin_needbreak(spinlock_t *lock)
1963 {
1964 #ifdef CONFIG_PREEMPT
1965 	return spin_is_contended(lock);
1966 #else
1967 	return 0;
1968 #endif
1969 }
1970 
1971 /*
1972  * Reevaluate whether the task has signals pending delivery.
1973  * Wake the task if so.
1974  * This is required every time the blocked sigset_t changes.
1975  * callers must hold sighand->siglock.
1976  */
1977 extern void recalc_sigpending_and_wake(struct task_struct *t);
1978 extern void recalc_sigpending(void);
1979 
1980 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1981 
1982 /*
1983  * Wrappers for p->thread_info->cpu access. No-op on UP.
1984  */
1985 #ifdef CONFIG_SMP
1986 
1987 static inline unsigned int task_cpu(const struct task_struct *p)
1988 {
1989 	return task_thread_info(p)->cpu;
1990 }
1991 
1992 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1993 
1994 #else
1995 
1996 static inline unsigned int task_cpu(const struct task_struct *p)
1997 {
1998 	return 0;
1999 }
2000 
2001 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2002 {
2003 }
2004 
2005 #endif /* CONFIG_SMP */
2006 
2007 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
2008 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2009 #else
2010 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
2011 {
2012 	mm->mmap_base = TASK_UNMAPPED_BASE;
2013 	mm->get_unmapped_area = arch_get_unmapped_area;
2014 	mm->unmap_area = arch_unmap_area;
2015 }
2016 #endif
2017 
2018 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
2019 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
2020 
2021 extern int sched_mc_power_savings, sched_smt_power_savings;
2022 
2023 extern void normalize_rt_tasks(void);
2024 
2025 #ifdef CONFIG_FAIR_GROUP_SCHED
2026 
2027 extern struct task_group init_task_group;
2028 
2029 extern struct task_group *sched_create_group(void);
2030 extern void sched_destroy_group(struct task_group *tg);
2031 extern void sched_move_task(struct task_struct *tsk);
2032 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2033 extern unsigned long sched_group_shares(struct task_group *tg);
2034 
2035 #endif
2036 
2037 #ifdef CONFIG_TASK_XACCT
2038 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2039 {
2040 	tsk->rchar += amt;
2041 }
2042 
2043 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2044 {
2045 	tsk->wchar += amt;
2046 }
2047 
2048 static inline void inc_syscr(struct task_struct *tsk)
2049 {
2050 	tsk->syscr++;
2051 }
2052 
2053 static inline void inc_syscw(struct task_struct *tsk)
2054 {
2055 	tsk->syscw++;
2056 }
2057 #else
2058 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2059 {
2060 }
2061 
2062 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2063 {
2064 }
2065 
2066 static inline void inc_syscr(struct task_struct *tsk)
2067 {
2068 }
2069 
2070 static inline void inc_syscw(struct task_struct *tsk)
2071 {
2072 }
2073 #endif
2074 
2075 #ifdef CONFIG_SMP
2076 void migration_init(void);
2077 #else
2078 static inline void migration_init(void)
2079 {
2080 }
2081 #endif
2082 
2083 #endif /* __KERNEL__ */
2084 
2085 #endif
2086