xref: /linux/include/linux/sched.h (revision f3d9478b2ce468c3115b02ecae7e975990697f15)
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3 
4 #include <linux/auxvec.h>	/* For AT_VECTOR_SIZE */
5 
6 /*
7  * cloning flags:
8  */
9 #define CSIGNAL		0x000000ff	/* signal mask to be sent at exit */
10 #define CLONE_VM	0x00000100	/* set if VM shared between processes */
11 #define CLONE_FS	0x00000200	/* set if fs info shared between processes */
12 #define CLONE_FILES	0x00000400	/* set if open files shared between processes */
13 #define CLONE_SIGHAND	0x00000800	/* set if signal handlers and blocked signals shared */
14 #define CLONE_PTRACE	0x00002000	/* set if we want to let tracing continue on the child too */
15 #define CLONE_VFORK	0x00004000	/* set if the parent wants the child to wake it up on mm_release */
16 #define CLONE_PARENT	0x00008000	/* set if we want to have the same parent as the cloner */
17 #define CLONE_THREAD	0x00010000	/* Same thread group? */
18 #define CLONE_NEWNS	0x00020000	/* New namespace group? */
19 #define CLONE_SYSVSEM	0x00040000	/* share system V SEM_UNDO semantics */
20 #define CLONE_SETTLS	0x00080000	/* create a new TLS for the child */
21 #define CLONE_PARENT_SETTID	0x00100000	/* set the TID in the parent */
22 #define CLONE_CHILD_CLEARTID	0x00200000	/* clear the TID in the child */
23 #define CLONE_DETACHED		0x00400000	/* Unused, ignored */
24 #define CLONE_UNTRACED		0x00800000	/* set if the tracing process can't force CLONE_PTRACE on this clone */
25 #define CLONE_CHILD_SETTID	0x01000000	/* set the TID in the child */
26 #define CLONE_STOPPED		0x02000000	/* Start in stopped state */
27 
28 /*
29  * Scheduling policies
30  */
31 #define SCHED_NORMAL		0
32 #define SCHED_FIFO		1
33 #define SCHED_RR		2
34 #define SCHED_BATCH		3
35 
36 #ifdef __KERNEL__
37 
38 struct sched_param {
39 	int sched_priority;
40 };
41 
42 #include <asm/param.h>	/* for HZ */
43 
44 #include <linux/capability.h>
45 #include <linux/threads.h>
46 #include <linux/kernel.h>
47 #include <linux/types.h>
48 #include <linux/timex.h>
49 #include <linux/jiffies.h>
50 #include <linux/rbtree.h>
51 #include <linux/thread_info.h>
52 #include <linux/cpumask.h>
53 #include <linux/errno.h>
54 #include <linux/nodemask.h>
55 
56 #include <asm/system.h>
57 #include <asm/semaphore.h>
58 #include <asm/page.h>
59 #include <asm/ptrace.h>
60 #include <asm/mmu.h>
61 #include <asm/cputime.h>
62 
63 #include <linux/smp.h>
64 #include <linux/sem.h>
65 #include <linux/signal.h>
66 #include <linux/securebits.h>
67 #include <linux/fs_struct.h>
68 #include <linux/compiler.h>
69 #include <linux/completion.h>
70 #include <linux/pid.h>
71 #include <linux/percpu.h>
72 #include <linux/topology.h>
73 #include <linux/seccomp.h>
74 #include <linux/rcupdate.h>
75 #include <linux/futex.h>
76 
77 #include <linux/time.h>
78 #include <linux/param.h>
79 #include <linux/resource.h>
80 #include <linux/timer.h>
81 #include <linux/hrtimer.h>
82 
83 #include <asm/processor.h>
84 
85 struct exec_domain;
86 
87 /*
88  * List of flags we want to share for kernel threads,
89  * if only because they are not used by them anyway.
90  */
91 #define CLONE_KERNEL	(CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
92 
93 /*
94  * These are the constant used to fake the fixed-point load-average
95  * counting. Some notes:
96  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
97  *    a load-average precision of 10 bits integer + 11 bits fractional
98  *  - if you want to count load-averages more often, you need more
99  *    precision, or rounding will get you. With 2-second counting freq,
100  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
101  *    11 bit fractions.
102  */
103 extern unsigned long avenrun[];		/* Load averages */
104 
105 #define FSHIFT		11		/* nr of bits of precision */
106 #define FIXED_1		(1<<FSHIFT)	/* 1.0 as fixed-point */
107 #define LOAD_FREQ	(5*HZ)		/* 5 sec intervals */
108 #define EXP_1		1884		/* 1/exp(5sec/1min) as fixed-point */
109 #define EXP_5		2014		/* 1/exp(5sec/5min) */
110 #define EXP_15		2037		/* 1/exp(5sec/15min) */
111 
112 #define CALC_LOAD(load,exp,n) \
113 	load *= exp; \
114 	load += n*(FIXED_1-exp); \
115 	load >>= FSHIFT;
116 
117 extern unsigned long total_forks;
118 extern int nr_threads;
119 extern int last_pid;
120 DECLARE_PER_CPU(unsigned long, process_counts);
121 extern int nr_processes(void);
122 extern unsigned long nr_running(void);
123 extern unsigned long nr_uninterruptible(void);
124 extern unsigned long nr_active(void);
125 extern unsigned long nr_iowait(void);
126 
127 
128 /*
129  * Task state bitmask. NOTE! These bits are also
130  * encoded in fs/proc/array.c: get_task_state().
131  *
132  * We have two separate sets of flags: task->state
133  * is about runnability, while task->exit_state are
134  * about the task exiting. Confusing, but this way
135  * modifying one set can't modify the other one by
136  * mistake.
137  */
138 #define TASK_RUNNING		0
139 #define TASK_INTERRUPTIBLE	1
140 #define TASK_UNINTERRUPTIBLE	2
141 #define TASK_STOPPED		4
142 #define TASK_TRACED		8
143 /* in tsk->exit_state */
144 #define EXIT_ZOMBIE		16
145 #define EXIT_DEAD		32
146 /* in tsk->state again */
147 #define TASK_NONINTERACTIVE	64
148 
149 #define __set_task_state(tsk, state_value)		\
150 	do { (tsk)->state = (state_value); } while (0)
151 #define set_task_state(tsk, state_value)		\
152 	set_mb((tsk)->state, (state_value))
153 
154 /*
155  * set_current_state() includes a barrier so that the write of current->state
156  * is correctly serialised wrt the caller's subsequent test of whether to
157  * actually sleep:
158  *
159  *	set_current_state(TASK_UNINTERRUPTIBLE);
160  *	if (do_i_need_to_sleep())
161  *		schedule();
162  *
163  * If the caller does not need such serialisation then use __set_current_state()
164  */
165 #define __set_current_state(state_value)			\
166 	do { current->state = (state_value); } while (0)
167 #define set_current_state(state_value)		\
168 	set_mb(current->state, (state_value))
169 
170 /* Task command name length */
171 #define TASK_COMM_LEN 16
172 
173 #include <linux/spinlock.h>
174 
175 /*
176  * This serializes "schedule()" and also protects
177  * the run-queue from deletions/modifications (but
178  * _adding_ to the beginning of the run-queue has
179  * a separate lock).
180  */
181 extern rwlock_t tasklist_lock;
182 extern spinlock_t mmlist_lock;
183 
184 typedef struct task_struct task_t;
185 
186 extern void sched_init(void);
187 extern void sched_init_smp(void);
188 extern void init_idle(task_t *idle, int cpu);
189 
190 extern cpumask_t nohz_cpu_mask;
191 
192 extern void show_state(void);
193 extern void show_regs(struct pt_regs *);
194 
195 /*
196  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
197  * task), SP is the stack pointer of the first frame that should be shown in the back
198  * trace (or NULL if the entire call-chain of the task should be shown).
199  */
200 extern void show_stack(struct task_struct *task, unsigned long *sp);
201 
202 void io_schedule(void);
203 long io_schedule_timeout(long timeout);
204 
205 extern void cpu_init (void);
206 extern void trap_init(void);
207 extern void update_process_times(int user);
208 extern void scheduler_tick(void);
209 
210 #ifdef CONFIG_DETECT_SOFTLOCKUP
211 extern void softlockup_tick(void);
212 extern void spawn_softlockup_task(void);
213 extern void touch_softlockup_watchdog(void);
214 #else
215 static inline void softlockup_tick(void)
216 {
217 }
218 static inline void spawn_softlockup_task(void)
219 {
220 }
221 static inline void touch_softlockup_watchdog(void)
222 {
223 }
224 #endif
225 
226 
227 /* Attach to any functions which should be ignored in wchan output. */
228 #define __sched		__attribute__((__section__(".sched.text")))
229 /* Is this address in the __sched functions? */
230 extern int in_sched_functions(unsigned long addr);
231 
232 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
233 extern signed long FASTCALL(schedule_timeout(signed long timeout));
234 extern signed long schedule_timeout_interruptible(signed long timeout);
235 extern signed long schedule_timeout_uninterruptible(signed long timeout);
236 asmlinkage void schedule(void);
237 
238 struct namespace;
239 
240 /* Maximum number of active map areas.. This is a random (large) number */
241 #define DEFAULT_MAX_MAP_COUNT	65536
242 
243 extern int sysctl_max_map_count;
244 
245 #include <linux/aio.h>
246 
247 extern unsigned long
248 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
249 		       unsigned long, unsigned long);
250 extern unsigned long
251 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
252 			  unsigned long len, unsigned long pgoff,
253 			  unsigned long flags);
254 extern void arch_unmap_area(struct mm_struct *, unsigned long);
255 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
256 
257 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
258 /*
259  * The mm counters are not protected by its page_table_lock,
260  * so must be incremented atomically.
261  */
262 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
263 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
264 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
265 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
266 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
267 typedef atomic_long_t mm_counter_t;
268 
269 #else  /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
270 /*
271  * The mm counters are protected by its page_table_lock,
272  * so can be incremented directly.
273  */
274 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
275 #define get_mm_counter(mm, member) ((mm)->_##member)
276 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
277 #define inc_mm_counter(mm, member) (mm)->_##member++
278 #define dec_mm_counter(mm, member) (mm)->_##member--
279 typedef unsigned long mm_counter_t;
280 
281 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
282 
283 #define get_mm_rss(mm)					\
284 	(get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
285 #define update_hiwater_rss(mm)	do {			\
286 	unsigned long _rss = get_mm_rss(mm);		\
287 	if ((mm)->hiwater_rss < _rss)			\
288 		(mm)->hiwater_rss = _rss;		\
289 } while (0)
290 #define update_hiwater_vm(mm)	do {			\
291 	if ((mm)->hiwater_vm < (mm)->total_vm)		\
292 		(mm)->hiwater_vm = (mm)->total_vm;	\
293 } while (0)
294 
295 struct mm_struct {
296 	struct vm_area_struct * mmap;		/* list of VMAs */
297 	struct rb_root mm_rb;
298 	struct vm_area_struct * mmap_cache;	/* last find_vma result */
299 	unsigned long (*get_unmapped_area) (struct file *filp,
300 				unsigned long addr, unsigned long len,
301 				unsigned long pgoff, unsigned long flags);
302 	void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
303 	unsigned long mmap_base;		/* base of mmap area */
304 	unsigned long task_size;		/* size of task vm space */
305 	unsigned long cached_hole_size;         /* if non-zero, the largest hole below free_area_cache */
306 	unsigned long free_area_cache;		/* first hole of size cached_hole_size or larger */
307 	pgd_t * pgd;
308 	atomic_t mm_users;			/* How many users with user space? */
309 	atomic_t mm_count;			/* How many references to "struct mm_struct" (users count as 1) */
310 	int map_count;				/* number of VMAs */
311 	struct rw_semaphore mmap_sem;
312 	spinlock_t page_table_lock;		/* Protects page tables and some counters */
313 
314 	struct list_head mmlist;		/* List of maybe swapped mm's.  These are globally strung
315 						 * together off init_mm.mmlist, and are protected
316 						 * by mmlist_lock
317 						 */
318 
319 	/* Special counters, in some configurations protected by the
320 	 * page_table_lock, in other configurations by being atomic.
321 	 */
322 	mm_counter_t _file_rss;
323 	mm_counter_t _anon_rss;
324 
325 	unsigned long hiwater_rss;	/* High-watermark of RSS usage */
326 	unsigned long hiwater_vm;	/* High-water virtual memory usage */
327 
328 	unsigned long total_vm, locked_vm, shared_vm, exec_vm;
329 	unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
330 	unsigned long start_code, end_code, start_data, end_data;
331 	unsigned long start_brk, brk, start_stack;
332 	unsigned long arg_start, arg_end, env_start, env_end;
333 
334 	unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
335 
336 	unsigned dumpable:2;
337 	cpumask_t cpu_vm_mask;
338 
339 	/* Architecture-specific MM context */
340 	mm_context_t context;
341 
342 	/* Token based thrashing protection. */
343 	unsigned long swap_token_time;
344 	char recent_pagein;
345 
346 	/* coredumping support */
347 	int core_waiters;
348 	struct completion *core_startup_done, core_done;
349 
350 	/* aio bits */
351 	rwlock_t		ioctx_list_lock;
352 	struct kioctx		*ioctx_list;
353 };
354 
355 struct sighand_struct {
356 	atomic_t		count;
357 	struct k_sigaction	action[_NSIG];
358 	spinlock_t		siglock;
359 };
360 
361 /*
362  * NOTE! "signal_struct" does not have it's own
363  * locking, because a shared signal_struct always
364  * implies a shared sighand_struct, so locking
365  * sighand_struct is always a proper superset of
366  * the locking of signal_struct.
367  */
368 struct signal_struct {
369 	atomic_t		count;
370 	atomic_t		live;
371 
372 	wait_queue_head_t	wait_chldexit;	/* for wait4() */
373 
374 	/* current thread group signal load-balancing target: */
375 	task_t			*curr_target;
376 
377 	/* shared signal handling: */
378 	struct sigpending	shared_pending;
379 
380 	/* thread group exit support */
381 	int			group_exit_code;
382 	/* overloaded:
383 	 * - notify group_exit_task when ->count is equal to notify_count
384 	 * - everyone except group_exit_task is stopped during signal delivery
385 	 *   of fatal signals, group_exit_task processes the signal.
386 	 */
387 	struct task_struct	*group_exit_task;
388 	int			notify_count;
389 
390 	/* thread group stop support, overloads group_exit_code too */
391 	int			group_stop_count;
392 	unsigned int		flags; /* see SIGNAL_* flags below */
393 
394 	/* POSIX.1b Interval Timers */
395 	struct list_head posix_timers;
396 
397 	/* ITIMER_REAL timer for the process */
398 	struct hrtimer real_timer;
399 	struct task_struct *tsk;
400 	ktime_t it_real_incr;
401 
402 	/* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
403 	cputime_t it_prof_expires, it_virt_expires;
404 	cputime_t it_prof_incr, it_virt_incr;
405 
406 	/* job control IDs */
407 	pid_t pgrp;
408 	pid_t tty_old_pgrp;
409 	pid_t session;
410 	/* boolean value for session group leader */
411 	int leader;
412 
413 	struct tty_struct *tty; /* NULL if no tty */
414 
415 	/*
416 	 * Cumulative resource counters for dead threads in the group,
417 	 * and for reaped dead child processes forked by this group.
418 	 * Live threads maintain their own counters and add to these
419 	 * in __exit_signal, except for the group leader.
420 	 */
421 	cputime_t utime, stime, cutime, cstime;
422 	unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
423 	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
424 
425 	/*
426 	 * Cumulative ns of scheduled CPU time for dead threads in the
427 	 * group, not including a zombie group leader.  (This only differs
428 	 * from jiffies_to_ns(utime + stime) if sched_clock uses something
429 	 * other than jiffies.)
430 	 */
431 	unsigned long long sched_time;
432 
433 	/*
434 	 * We don't bother to synchronize most readers of this at all,
435 	 * because there is no reader checking a limit that actually needs
436 	 * to get both rlim_cur and rlim_max atomically, and either one
437 	 * alone is a single word that can safely be read normally.
438 	 * getrlimit/setrlimit use task_lock(current->group_leader) to
439 	 * protect this instead of the siglock, because they really
440 	 * have no need to disable irqs.
441 	 */
442 	struct rlimit rlim[RLIM_NLIMITS];
443 
444 	struct list_head cpu_timers[3];
445 
446 	/* keep the process-shared keyrings here so that they do the right
447 	 * thing in threads created with CLONE_THREAD */
448 #ifdef CONFIG_KEYS
449 	struct key *session_keyring;	/* keyring inherited over fork */
450 	struct key *process_keyring;	/* keyring private to this process */
451 #endif
452 };
453 
454 /* Context switch must be unlocked if interrupts are to be enabled */
455 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
456 # define __ARCH_WANT_UNLOCKED_CTXSW
457 #endif
458 
459 /*
460  * Bits in flags field of signal_struct.
461  */
462 #define SIGNAL_STOP_STOPPED	0x00000001 /* job control stop in effect */
463 #define SIGNAL_STOP_DEQUEUED	0x00000002 /* stop signal dequeued */
464 #define SIGNAL_STOP_CONTINUED	0x00000004 /* SIGCONT since WCONTINUED reap */
465 #define SIGNAL_GROUP_EXIT	0x00000008 /* group exit in progress */
466 
467 
468 /*
469  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
470  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
471  * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
472  * values are inverted: lower p->prio value means higher priority.
473  *
474  * The MAX_USER_RT_PRIO value allows the actual maximum
475  * RT priority to be separate from the value exported to
476  * user-space.  This allows kernel threads to set their
477  * priority to a value higher than any user task. Note:
478  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
479  */
480 
481 #define MAX_USER_RT_PRIO	100
482 #define MAX_RT_PRIO		MAX_USER_RT_PRIO
483 
484 #define MAX_PRIO		(MAX_RT_PRIO + 40)
485 
486 #define rt_task(p)		(unlikely((p)->prio < MAX_RT_PRIO))
487 #define batch_task(p)		(unlikely((p)->policy == SCHED_BATCH))
488 
489 /*
490  * Some day this will be a full-fledged user tracking system..
491  */
492 struct user_struct {
493 	atomic_t __count;	/* reference count */
494 	atomic_t processes;	/* How many processes does this user have? */
495 	atomic_t files;		/* How many open files does this user have? */
496 	atomic_t sigpending;	/* How many pending signals does this user have? */
497 #ifdef CONFIG_INOTIFY_USER
498 	atomic_t inotify_watches; /* How many inotify watches does this user have? */
499 	atomic_t inotify_devs;	/* How many inotify devs does this user have opened? */
500 #endif
501 	/* protected by mq_lock	*/
502 	unsigned long mq_bytes;	/* How many bytes can be allocated to mqueue? */
503 	unsigned long locked_shm; /* How many pages of mlocked shm ? */
504 
505 #ifdef CONFIG_KEYS
506 	struct key *uid_keyring;	/* UID specific keyring */
507 	struct key *session_keyring;	/* UID's default session keyring */
508 #endif
509 
510 	/* Hash table maintenance information */
511 	struct list_head uidhash_list;
512 	uid_t uid;
513 };
514 
515 extern struct user_struct *find_user(uid_t);
516 
517 extern struct user_struct root_user;
518 #define INIT_USER (&root_user)
519 
520 typedef struct prio_array prio_array_t;
521 struct backing_dev_info;
522 struct reclaim_state;
523 
524 #ifdef CONFIG_SCHEDSTATS
525 struct sched_info {
526 	/* cumulative counters */
527 	unsigned long	cpu_time,	/* time spent on the cpu */
528 			run_delay,	/* time spent waiting on a runqueue */
529 			pcnt;		/* # of timeslices run on this cpu */
530 
531 	/* timestamps */
532 	unsigned long	last_arrival,	/* when we last ran on a cpu */
533 			last_queued;	/* when we were last queued to run */
534 };
535 
536 extern struct file_operations proc_schedstat_operations;
537 #endif
538 
539 enum idle_type
540 {
541 	SCHED_IDLE,
542 	NOT_IDLE,
543 	NEWLY_IDLE,
544 	MAX_IDLE_TYPES
545 };
546 
547 /*
548  * sched-domains (multiprocessor balancing) declarations:
549  */
550 #ifdef CONFIG_SMP
551 #define SCHED_LOAD_SCALE	128UL	/* increase resolution of load */
552 
553 #define SD_LOAD_BALANCE		1	/* Do load balancing on this domain. */
554 #define SD_BALANCE_NEWIDLE	2	/* Balance when about to become idle */
555 #define SD_BALANCE_EXEC		4	/* Balance on exec */
556 #define SD_BALANCE_FORK		8	/* Balance on fork, clone */
557 #define SD_WAKE_IDLE		16	/* Wake to idle CPU on task wakeup */
558 #define SD_WAKE_AFFINE		32	/* Wake task to waking CPU */
559 #define SD_WAKE_BALANCE		64	/* Perform balancing at task wakeup */
560 #define SD_SHARE_CPUPOWER	128	/* Domain members share cpu power */
561 
562 struct sched_group {
563 	struct sched_group *next;	/* Must be a circular list */
564 	cpumask_t cpumask;
565 
566 	/*
567 	 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
568 	 * single CPU. This is read only (except for setup, hotplug CPU).
569 	 */
570 	unsigned long cpu_power;
571 };
572 
573 struct sched_domain {
574 	/* These fields must be setup */
575 	struct sched_domain *parent;	/* top domain must be null terminated */
576 	struct sched_group *groups;	/* the balancing groups of the domain */
577 	cpumask_t span;			/* span of all CPUs in this domain */
578 	unsigned long min_interval;	/* Minimum balance interval ms */
579 	unsigned long max_interval;	/* Maximum balance interval ms */
580 	unsigned int busy_factor;	/* less balancing by factor if busy */
581 	unsigned int imbalance_pct;	/* No balance until over watermark */
582 	unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
583 	unsigned int cache_nice_tries;	/* Leave cache hot tasks for # tries */
584 	unsigned int per_cpu_gain;	/* CPU % gained by adding domain cpus */
585 	unsigned int busy_idx;
586 	unsigned int idle_idx;
587 	unsigned int newidle_idx;
588 	unsigned int wake_idx;
589 	unsigned int forkexec_idx;
590 	int flags;			/* See SD_* */
591 
592 	/* Runtime fields. */
593 	unsigned long last_balance;	/* init to jiffies. units in jiffies */
594 	unsigned int balance_interval;	/* initialise to 1. units in ms. */
595 	unsigned int nr_balance_failed; /* initialise to 0 */
596 
597 #ifdef CONFIG_SCHEDSTATS
598 	/* load_balance() stats */
599 	unsigned long lb_cnt[MAX_IDLE_TYPES];
600 	unsigned long lb_failed[MAX_IDLE_TYPES];
601 	unsigned long lb_balanced[MAX_IDLE_TYPES];
602 	unsigned long lb_imbalance[MAX_IDLE_TYPES];
603 	unsigned long lb_gained[MAX_IDLE_TYPES];
604 	unsigned long lb_hot_gained[MAX_IDLE_TYPES];
605 	unsigned long lb_nobusyg[MAX_IDLE_TYPES];
606 	unsigned long lb_nobusyq[MAX_IDLE_TYPES];
607 
608 	/* Active load balancing */
609 	unsigned long alb_cnt;
610 	unsigned long alb_failed;
611 	unsigned long alb_pushed;
612 
613 	/* SD_BALANCE_EXEC stats */
614 	unsigned long sbe_cnt;
615 	unsigned long sbe_balanced;
616 	unsigned long sbe_pushed;
617 
618 	/* SD_BALANCE_FORK stats */
619 	unsigned long sbf_cnt;
620 	unsigned long sbf_balanced;
621 	unsigned long sbf_pushed;
622 
623 	/* try_to_wake_up() stats */
624 	unsigned long ttwu_wake_remote;
625 	unsigned long ttwu_move_affine;
626 	unsigned long ttwu_move_balance;
627 #endif
628 };
629 
630 extern void partition_sched_domains(cpumask_t *partition1,
631 				    cpumask_t *partition2);
632 
633 /*
634  * Maximum cache size the migration-costs auto-tuning code will
635  * search from:
636  */
637 extern unsigned int max_cache_size;
638 
639 #endif	/* CONFIG_SMP */
640 
641 
642 struct io_context;			/* See blkdev.h */
643 void exit_io_context(void);
644 struct cpuset;
645 
646 #define NGROUPS_SMALL		32
647 #define NGROUPS_PER_BLOCK	((int)(PAGE_SIZE / sizeof(gid_t)))
648 struct group_info {
649 	int ngroups;
650 	atomic_t usage;
651 	gid_t small_block[NGROUPS_SMALL];
652 	int nblocks;
653 	gid_t *blocks[0];
654 };
655 
656 /*
657  * get_group_info() must be called with the owning task locked (via task_lock())
658  * when task != current.  The reason being that the vast majority of callers are
659  * looking at current->group_info, which can not be changed except by the
660  * current task.  Changing current->group_info requires the task lock, too.
661  */
662 #define get_group_info(group_info) do { \
663 	atomic_inc(&(group_info)->usage); \
664 } while (0)
665 
666 #define put_group_info(group_info) do { \
667 	if (atomic_dec_and_test(&(group_info)->usage)) \
668 		groups_free(group_info); \
669 } while (0)
670 
671 extern struct group_info *groups_alloc(int gidsetsize);
672 extern void groups_free(struct group_info *group_info);
673 extern int set_current_groups(struct group_info *group_info);
674 extern int groups_search(struct group_info *group_info, gid_t grp);
675 /* access the groups "array" with this macro */
676 #define GROUP_AT(gi, i) \
677     ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
678 
679 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
680 extern void prefetch_stack(struct task_struct*);
681 #else
682 static inline void prefetch_stack(struct task_struct *t) { }
683 #endif
684 
685 struct audit_context;		/* See audit.c */
686 struct mempolicy;
687 struct pipe_inode_info;
688 
689 enum sleep_type {
690 	SLEEP_NORMAL,
691 	SLEEP_NONINTERACTIVE,
692 	SLEEP_INTERACTIVE,
693 	SLEEP_INTERRUPTED,
694 };
695 
696 struct task_struct {
697 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
698 	struct thread_info *thread_info;
699 	atomic_t usage;
700 	unsigned long flags;	/* per process flags, defined below */
701 	unsigned long ptrace;
702 
703 	int lock_depth;		/* BKL lock depth */
704 
705 #if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW)
706 	int oncpu;
707 #endif
708 	int prio, static_prio;
709 	struct list_head run_list;
710 	prio_array_t *array;
711 
712 	unsigned short ioprio;
713 	unsigned int btrace_seq;
714 
715 	unsigned long sleep_avg;
716 	unsigned long long timestamp, last_ran;
717 	unsigned long long sched_time; /* sched_clock time spent running */
718 	enum sleep_type sleep_type;
719 
720 	unsigned long policy;
721 	cpumask_t cpus_allowed;
722 	unsigned int time_slice, first_time_slice;
723 
724 #ifdef CONFIG_SCHEDSTATS
725 	struct sched_info sched_info;
726 #endif
727 
728 	struct list_head tasks;
729 	/*
730 	 * ptrace_list/ptrace_children forms the list of my children
731 	 * that were stolen by a ptracer.
732 	 */
733 	struct list_head ptrace_children;
734 	struct list_head ptrace_list;
735 
736 	struct mm_struct *mm, *active_mm;
737 
738 /* task state */
739 	struct linux_binfmt *binfmt;
740 	long exit_state;
741 	int exit_code, exit_signal;
742 	int pdeath_signal;  /*  The signal sent when the parent dies  */
743 	/* ??? */
744 	unsigned long personality;
745 	unsigned did_exec:1;
746 	pid_t pid;
747 	pid_t tgid;
748 	/*
749 	 * pointers to (original) parent process, youngest child, younger sibling,
750 	 * older sibling, respectively.  (p->father can be replaced with
751 	 * p->parent->pid)
752 	 */
753 	struct task_struct *real_parent; /* real parent process (when being debugged) */
754 	struct task_struct *parent;	/* parent process */
755 	/*
756 	 * children/sibling forms the list of my children plus the
757 	 * tasks I'm ptracing.
758 	 */
759 	struct list_head children;	/* list of my children */
760 	struct list_head sibling;	/* linkage in my parent's children list */
761 	struct task_struct *group_leader;	/* threadgroup leader */
762 
763 	/* PID/PID hash table linkage. */
764 	struct pid_link pids[PIDTYPE_MAX];
765 	struct list_head thread_group;
766 
767 	struct completion *vfork_done;		/* for vfork() */
768 	int __user *set_child_tid;		/* CLONE_CHILD_SETTID */
769 	int __user *clear_child_tid;		/* CLONE_CHILD_CLEARTID */
770 
771 	unsigned long rt_priority;
772 	cputime_t utime, stime;
773 	unsigned long nvcsw, nivcsw; /* context switch counts */
774 	struct timespec start_time;
775 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
776 	unsigned long min_flt, maj_flt;
777 
778   	cputime_t it_prof_expires, it_virt_expires;
779 	unsigned long long it_sched_expires;
780 	struct list_head cpu_timers[3];
781 
782 /* process credentials */
783 	uid_t uid,euid,suid,fsuid;
784 	gid_t gid,egid,sgid,fsgid;
785 	struct group_info *group_info;
786 	kernel_cap_t   cap_effective, cap_inheritable, cap_permitted;
787 	unsigned keep_capabilities:1;
788 	struct user_struct *user;
789 #ifdef CONFIG_KEYS
790 	struct key *request_key_auth;	/* assumed request_key authority */
791 	struct key *thread_keyring;	/* keyring private to this thread */
792 	unsigned char jit_keyring;	/* default keyring to attach requested keys to */
793 #endif
794 	int oomkilladj; /* OOM kill score adjustment (bit shift). */
795 	char comm[TASK_COMM_LEN]; /* executable name excluding path
796 				     - access with [gs]et_task_comm (which lock
797 				       it with task_lock())
798 				     - initialized normally by flush_old_exec */
799 /* file system info */
800 	int link_count, total_link_count;
801 /* ipc stuff */
802 	struct sysv_sem sysvsem;
803 /* CPU-specific state of this task */
804 	struct thread_struct thread;
805 /* filesystem information */
806 	struct fs_struct *fs;
807 /* open file information */
808 	struct files_struct *files;
809 /* namespace */
810 	struct namespace *namespace;
811 /* signal handlers */
812 	struct signal_struct *signal;
813 	struct sighand_struct *sighand;
814 
815 	sigset_t blocked, real_blocked;
816 	sigset_t saved_sigmask;		/* To be restored with TIF_RESTORE_SIGMASK */
817 	struct sigpending pending;
818 
819 	unsigned long sas_ss_sp;
820 	size_t sas_ss_size;
821 	int (*notifier)(void *priv);
822 	void *notifier_data;
823 	sigset_t *notifier_mask;
824 
825 	void *security;
826 	struct audit_context *audit_context;
827 	seccomp_t seccomp;
828 
829 /* Thread group tracking */
830    	u32 parent_exec_id;
831    	u32 self_exec_id;
832 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
833 	spinlock_t alloc_lock;
834 /* Protection of proc_dentry: nesting proc_lock, dcache_lock, write_lock_irq(&tasklist_lock); */
835 	spinlock_t proc_lock;
836 
837 #ifdef CONFIG_DEBUG_MUTEXES
838 	/* mutex deadlock detection */
839 	struct mutex_waiter *blocked_on;
840 #endif
841 
842 /* journalling filesystem info */
843 	void *journal_info;
844 
845 /* VM state */
846 	struct reclaim_state *reclaim_state;
847 
848 	struct dentry *proc_dentry;
849 	struct backing_dev_info *backing_dev_info;
850 
851 	struct io_context *io_context;
852 
853 	unsigned long ptrace_message;
854 	siginfo_t *last_siginfo; /* For ptrace use.  */
855 /*
856  * current io wait handle: wait queue entry to use for io waits
857  * If this thread is processing aio, this points at the waitqueue
858  * inside the currently handled kiocb. It may be NULL (i.e. default
859  * to a stack based synchronous wait) if its doing sync IO.
860  */
861 	wait_queue_t *io_wait;
862 /* i/o counters(bytes read/written, #syscalls */
863 	u64 rchar, wchar, syscr, syscw;
864 #if defined(CONFIG_BSD_PROCESS_ACCT)
865 	u64 acct_rss_mem1;	/* accumulated rss usage */
866 	u64 acct_vm_mem1;	/* accumulated virtual memory usage */
867 	clock_t acct_stimexpd;	/* clock_t-converted stime since last update */
868 #endif
869 #ifdef CONFIG_NUMA
870   	struct mempolicy *mempolicy;
871 	short il_next;
872 #endif
873 #ifdef CONFIG_CPUSETS
874 	struct cpuset *cpuset;
875 	nodemask_t mems_allowed;
876 	int cpuset_mems_generation;
877 	int cpuset_mem_spread_rotor;
878 #endif
879 	struct robust_list_head __user *robust_list;
880 #ifdef CONFIG_COMPAT
881 	struct compat_robust_list_head __user *compat_robust_list;
882 #endif
883 
884 	atomic_t fs_excl;	/* holding fs exclusive resources */
885 	struct rcu_head rcu;
886 
887 	/*
888 	 * cache last used pipe for splice
889 	 */
890 	struct pipe_inode_info *splice_pipe;
891 };
892 
893 static inline pid_t process_group(struct task_struct *tsk)
894 {
895 	return tsk->signal->pgrp;
896 }
897 
898 /**
899  * pid_alive - check that a task structure is not stale
900  * @p: Task structure to be checked.
901  *
902  * Test if a process is not yet dead (at most zombie state)
903  * If pid_alive fails, then pointers within the task structure
904  * can be stale and must not be dereferenced.
905  */
906 static inline int pid_alive(struct task_struct *p)
907 {
908 	return p->pids[PIDTYPE_PID].pid != NULL;
909 }
910 
911 extern void free_task(struct task_struct *tsk);
912 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
913 
914 extern void __put_task_struct(struct task_struct *t);
915 
916 static inline void put_task_struct(struct task_struct *t)
917 {
918 	if (atomic_dec_and_test(&t->usage))
919 		__put_task_struct(t);
920 }
921 
922 /*
923  * Per process flags
924  */
925 #define PF_ALIGNWARN	0x00000001	/* Print alignment warning msgs */
926 					/* Not implemented yet, only for 486*/
927 #define PF_STARTING	0x00000002	/* being created */
928 #define PF_EXITING	0x00000004	/* getting shut down */
929 #define PF_DEAD		0x00000008	/* Dead */
930 #define PF_FORKNOEXEC	0x00000040	/* forked but didn't exec */
931 #define PF_SUPERPRIV	0x00000100	/* used super-user privileges */
932 #define PF_DUMPCORE	0x00000200	/* dumped core */
933 #define PF_SIGNALED	0x00000400	/* killed by a signal */
934 #define PF_MEMALLOC	0x00000800	/* Allocating memory */
935 #define PF_FLUSHER	0x00001000	/* responsible for disk writeback */
936 #define PF_USED_MATH	0x00002000	/* if unset the fpu must be initialized before use */
937 #define PF_FREEZE	0x00004000	/* this task is being frozen for suspend now */
938 #define PF_NOFREEZE	0x00008000	/* this thread should not be frozen */
939 #define PF_FROZEN	0x00010000	/* frozen for system suspend */
940 #define PF_FSTRANS	0x00020000	/* inside a filesystem transaction */
941 #define PF_KSWAPD	0x00040000	/* I am kswapd */
942 #define PF_SWAPOFF	0x00080000	/* I am in swapoff */
943 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
944 #define PF_SYNCWRITE	0x00200000	/* I am doing a sync write */
945 #define PF_BORROWED_MM	0x00400000	/* I am a kthread doing use_mm */
946 #define PF_RANDOMIZE	0x00800000	/* randomize virtual address space */
947 #define PF_SWAPWRITE	0x01000000	/* Allowed to write to swap */
948 #define PF_SPREAD_PAGE	0x04000000	/* Spread page cache over cpuset */
949 #define PF_SPREAD_SLAB	0x08000000	/* Spread some slab caches over cpuset */
950 #define PF_MEMPOLICY	0x10000000	/* Non-default NUMA mempolicy */
951 
952 /*
953  * Only the _current_ task can read/write to tsk->flags, but other
954  * tasks can access tsk->flags in readonly mode for example
955  * with tsk_used_math (like during threaded core dumping).
956  * There is however an exception to this rule during ptrace
957  * or during fork: the ptracer task is allowed to write to the
958  * child->flags of its traced child (same goes for fork, the parent
959  * can write to the child->flags), because we're guaranteed the
960  * child is not running and in turn not changing child->flags
961  * at the same time the parent does it.
962  */
963 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
964 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
965 #define clear_used_math() clear_stopped_child_used_math(current)
966 #define set_used_math() set_stopped_child_used_math(current)
967 #define conditional_stopped_child_used_math(condition, child) \
968 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
969 #define conditional_used_math(condition) \
970 	conditional_stopped_child_used_math(condition, current)
971 #define copy_to_stopped_child_used_math(child) \
972 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
973 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
974 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
975 #define used_math() tsk_used_math(current)
976 
977 #ifdef CONFIG_SMP
978 extern int set_cpus_allowed(task_t *p, cpumask_t new_mask);
979 #else
980 static inline int set_cpus_allowed(task_t *p, cpumask_t new_mask)
981 {
982 	if (!cpu_isset(0, new_mask))
983 		return -EINVAL;
984 	return 0;
985 }
986 #endif
987 
988 extern unsigned long long sched_clock(void);
989 extern unsigned long long current_sched_time(const task_t *current_task);
990 
991 /* sched_exec is called by processes performing an exec */
992 #ifdef CONFIG_SMP
993 extern void sched_exec(void);
994 #else
995 #define sched_exec()   {}
996 #endif
997 
998 #ifdef CONFIG_HOTPLUG_CPU
999 extern void idle_task_exit(void);
1000 #else
1001 static inline void idle_task_exit(void) {}
1002 #endif
1003 
1004 extern void sched_idle_next(void);
1005 extern void set_user_nice(task_t *p, long nice);
1006 extern int task_prio(const task_t *p);
1007 extern int task_nice(const task_t *p);
1008 extern int can_nice(const task_t *p, const int nice);
1009 extern int task_curr(const task_t *p);
1010 extern int idle_cpu(int cpu);
1011 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1012 extern task_t *idle_task(int cpu);
1013 extern task_t *curr_task(int cpu);
1014 extern void set_curr_task(int cpu, task_t *p);
1015 
1016 void yield(void);
1017 
1018 /*
1019  * The default (Linux) execution domain.
1020  */
1021 extern struct exec_domain	default_exec_domain;
1022 
1023 union thread_union {
1024 	struct thread_info thread_info;
1025 	unsigned long stack[THREAD_SIZE/sizeof(long)];
1026 };
1027 
1028 #ifndef __HAVE_ARCH_KSTACK_END
1029 static inline int kstack_end(void *addr)
1030 {
1031 	/* Reliable end of stack detection:
1032 	 * Some APM bios versions misalign the stack
1033 	 */
1034 	return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1035 }
1036 #endif
1037 
1038 extern union thread_union init_thread_union;
1039 extern struct task_struct init_task;
1040 
1041 extern struct   mm_struct init_mm;
1042 
1043 #define find_task_by_pid(nr)	find_task_by_pid_type(PIDTYPE_PID, nr)
1044 extern struct task_struct *find_task_by_pid_type(int type, int pid);
1045 extern void set_special_pids(pid_t session, pid_t pgrp);
1046 extern void __set_special_pids(pid_t session, pid_t pgrp);
1047 
1048 /* per-UID process charging. */
1049 extern struct user_struct * alloc_uid(uid_t);
1050 static inline struct user_struct *get_uid(struct user_struct *u)
1051 {
1052 	atomic_inc(&u->__count);
1053 	return u;
1054 }
1055 extern void free_uid(struct user_struct *);
1056 extern void switch_uid(struct user_struct *);
1057 
1058 #include <asm/current.h>
1059 
1060 extern void do_timer(struct pt_regs *);
1061 
1062 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1063 extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1064 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1065 						unsigned long clone_flags));
1066 #ifdef CONFIG_SMP
1067  extern void kick_process(struct task_struct *tsk);
1068 #else
1069  static inline void kick_process(struct task_struct *tsk) { }
1070 #endif
1071 extern void FASTCALL(sched_fork(task_t * p, int clone_flags));
1072 extern void FASTCALL(sched_exit(task_t * p));
1073 
1074 extern int in_group_p(gid_t);
1075 extern int in_egroup_p(gid_t);
1076 
1077 extern void proc_caches_init(void);
1078 extern void flush_signals(struct task_struct *);
1079 extern void flush_signal_handlers(struct task_struct *, int force_default);
1080 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1081 
1082 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1083 {
1084 	unsigned long flags;
1085 	int ret;
1086 
1087 	spin_lock_irqsave(&tsk->sighand->siglock, flags);
1088 	ret = dequeue_signal(tsk, mask, info);
1089 	spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1090 
1091 	return ret;
1092 }
1093 
1094 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1095 			      sigset_t *mask);
1096 extern void unblock_all_signals(void);
1097 extern void release_task(struct task_struct * p);
1098 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1099 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1100 extern int force_sigsegv(int, struct task_struct *);
1101 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1102 extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp);
1103 extern int kill_pg_info(int, struct siginfo *, pid_t);
1104 extern int kill_proc_info(int, struct siginfo *, pid_t);
1105 extern int kill_proc_info_as_uid(int, struct siginfo *, pid_t, uid_t, uid_t);
1106 extern void do_notify_parent(struct task_struct *, int);
1107 extern void force_sig(int, struct task_struct *);
1108 extern void force_sig_specific(int, struct task_struct *);
1109 extern int send_sig(int, struct task_struct *, int);
1110 extern void zap_other_threads(struct task_struct *p);
1111 extern int kill_pg(pid_t, int, int);
1112 extern int kill_proc(pid_t, int, int);
1113 extern struct sigqueue *sigqueue_alloc(void);
1114 extern void sigqueue_free(struct sigqueue *);
1115 extern int send_sigqueue(int, struct sigqueue *,  struct task_struct *);
1116 extern int send_group_sigqueue(int, struct sigqueue *,  struct task_struct *);
1117 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1118 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1119 
1120 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
1121 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1122 #define SEND_SIG_PRIV	((struct siginfo *) 1)
1123 #define SEND_SIG_FORCED	((struct siginfo *) 2)
1124 
1125 static inline int is_si_special(const struct siginfo *info)
1126 {
1127 	return info <= SEND_SIG_FORCED;
1128 }
1129 
1130 /* True if we are on the alternate signal stack.  */
1131 
1132 static inline int on_sig_stack(unsigned long sp)
1133 {
1134 	return (sp - current->sas_ss_sp < current->sas_ss_size);
1135 }
1136 
1137 static inline int sas_ss_flags(unsigned long sp)
1138 {
1139 	return (current->sas_ss_size == 0 ? SS_DISABLE
1140 		: on_sig_stack(sp) ? SS_ONSTACK : 0);
1141 }
1142 
1143 /*
1144  * Routines for handling mm_structs
1145  */
1146 extern struct mm_struct * mm_alloc(void);
1147 
1148 /* mmdrop drops the mm and the page tables */
1149 extern void FASTCALL(__mmdrop(struct mm_struct *));
1150 static inline void mmdrop(struct mm_struct * mm)
1151 {
1152 	if (atomic_dec_and_test(&mm->mm_count))
1153 		__mmdrop(mm);
1154 }
1155 
1156 /* mmput gets rid of the mappings and all user-space */
1157 extern void mmput(struct mm_struct *);
1158 /* Grab a reference to a task's mm, if it is not already going away */
1159 extern struct mm_struct *get_task_mm(struct task_struct *task);
1160 /* Remove the current tasks stale references to the old mm_struct */
1161 extern void mm_release(struct task_struct *, struct mm_struct *);
1162 
1163 extern int  copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1164 extern void flush_thread(void);
1165 extern void exit_thread(void);
1166 
1167 extern void exit_files(struct task_struct *);
1168 extern void __cleanup_signal(struct signal_struct *);
1169 extern void __cleanup_sighand(struct sighand_struct *);
1170 extern void exit_itimers(struct signal_struct *);
1171 
1172 extern NORET_TYPE void do_group_exit(int);
1173 
1174 extern void daemonize(const char *, ...);
1175 extern int allow_signal(int);
1176 extern int disallow_signal(int);
1177 extern task_t *child_reaper;
1178 
1179 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1180 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1181 task_t *fork_idle(int);
1182 
1183 extern void set_task_comm(struct task_struct *tsk, char *from);
1184 extern void get_task_comm(char *to, struct task_struct *tsk);
1185 
1186 #ifdef CONFIG_SMP
1187 extern void wait_task_inactive(task_t * p);
1188 #else
1189 #define wait_task_inactive(p)	do { } while (0)
1190 #endif
1191 
1192 #define remove_parent(p)	list_del_init(&(p)->sibling)
1193 #define add_parent(p)		list_add_tail(&(p)->sibling,&(p)->parent->children)
1194 
1195 #define next_task(p)	list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1196 
1197 #define for_each_process(p) \
1198 	for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1199 
1200 /*
1201  * Careful: do_each_thread/while_each_thread is a double loop so
1202  *          'break' will not work as expected - use goto instead.
1203  */
1204 #define do_each_thread(g, t) \
1205 	for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1206 
1207 #define while_each_thread(g, t) \
1208 	while ((t = next_thread(t)) != g)
1209 
1210 /* de_thread depends on thread_group_leader not being a pid based check */
1211 #define thread_group_leader(p)	(p == p->group_leader)
1212 
1213 static inline task_t *next_thread(const task_t *p)
1214 {
1215 	return list_entry(rcu_dereference(p->thread_group.next),
1216 				task_t, thread_group);
1217 }
1218 
1219 static inline int thread_group_empty(task_t *p)
1220 {
1221 	return list_empty(&p->thread_group);
1222 }
1223 
1224 #define delay_group_leader(p) \
1225 		(thread_group_leader(p) && !thread_group_empty(p))
1226 
1227 /*
1228  * Protects ->fs, ->files, ->mm, ->ptrace, ->group_info, ->comm, keyring
1229  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
1230  * pins the final release of task.io_context.  Also protects ->cpuset.
1231  *
1232  * Nests both inside and outside of read_lock(&tasklist_lock).
1233  * It must not be nested with write_lock_irq(&tasklist_lock),
1234  * neither inside nor outside.
1235  */
1236 static inline void task_lock(struct task_struct *p)
1237 {
1238 	spin_lock(&p->alloc_lock);
1239 }
1240 
1241 static inline void task_unlock(struct task_struct *p)
1242 {
1243 	spin_unlock(&p->alloc_lock);
1244 }
1245 
1246 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1247 							unsigned long *flags);
1248 
1249 static inline void unlock_task_sighand(struct task_struct *tsk,
1250 						unsigned long *flags)
1251 {
1252 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1253 }
1254 
1255 #ifndef __HAVE_THREAD_FUNCTIONS
1256 
1257 #define task_thread_info(task) (task)->thread_info
1258 #define task_stack_page(task) ((void*)((task)->thread_info))
1259 
1260 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1261 {
1262 	*task_thread_info(p) = *task_thread_info(org);
1263 	task_thread_info(p)->task = p;
1264 }
1265 
1266 static inline unsigned long *end_of_stack(struct task_struct *p)
1267 {
1268 	return (unsigned long *)(p->thread_info + 1);
1269 }
1270 
1271 #endif
1272 
1273 /* set thread flags in other task's structures
1274  * - see asm/thread_info.h for TIF_xxxx flags available
1275  */
1276 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1277 {
1278 	set_ti_thread_flag(task_thread_info(tsk), flag);
1279 }
1280 
1281 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1282 {
1283 	clear_ti_thread_flag(task_thread_info(tsk), flag);
1284 }
1285 
1286 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1287 {
1288 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1289 }
1290 
1291 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1292 {
1293 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1294 }
1295 
1296 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1297 {
1298 	return test_ti_thread_flag(task_thread_info(tsk), flag);
1299 }
1300 
1301 static inline void set_tsk_need_resched(struct task_struct *tsk)
1302 {
1303 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1304 }
1305 
1306 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1307 {
1308 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1309 }
1310 
1311 static inline int signal_pending(struct task_struct *p)
1312 {
1313 	return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1314 }
1315 
1316 static inline int need_resched(void)
1317 {
1318 	return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1319 }
1320 
1321 /*
1322  * cond_resched() and cond_resched_lock(): latency reduction via
1323  * explicit rescheduling in places that are safe. The return
1324  * value indicates whether a reschedule was done in fact.
1325  * cond_resched_lock() will drop the spinlock before scheduling,
1326  * cond_resched_softirq() will enable bhs before scheduling.
1327  */
1328 extern int cond_resched(void);
1329 extern int cond_resched_lock(spinlock_t * lock);
1330 extern int cond_resched_softirq(void);
1331 
1332 /*
1333  * Does a critical section need to be broken due to another
1334  * task waiting?:
1335  */
1336 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1337 # define need_lockbreak(lock) ((lock)->break_lock)
1338 #else
1339 # define need_lockbreak(lock) 0
1340 #endif
1341 
1342 /*
1343  * Does a critical section need to be broken due to another
1344  * task waiting or preemption being signalled:
1345  */
1346 static inline int lock_need_resched(spinlock_t *lock)
1347 {
1348 	if (need_lockbreak(lock) || need_resched())
1349 		return 1;
1350 	return 0;
1351 }
1352 
1353 /* Reevaluate whether the task has signals pending delivery.
1354    This is required every time the blocked sigset_t changes.
1355    callers must hold sighand->siglock.  */
1356 
1357 extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1358 extern void recalc_sigpending(void);
1359 
1360 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1361 
1362 /*
1363  * Wrappers for p->thread_info->cpu access. No-op on UP.
1364  */
1365 #ifdef CONFIG_SMP
1366 
1367 static inline unsigned int task_cpu(const struct task_struct *p)
1368 {
1369 	return task_thread_info(p)->cpu;
1370 }
1371 
1372 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1373 {
1374 	task_thread_info(p)->cpu = cpu;
1375 }
1376 
1377 #else
1378 
1379 static inline unsigned int task_cpu(const struct task_struct *p)
1380 {
1381 	return 0;
1382 }
1383 
1384 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1385 {
1386 }
1387 
1388 #endif /* CONFIG_SMP */
1389 
1390 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1391 extern void arch_pick_mmap_layout(struct mm_struct *mm);
1392 #else
1393 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1394 {
1395 	mm->mmap_base = TASK_UNMAPPED_BASE;
1396 	mm->get_unmapped_area = arch_get_unmapped_area;
1397 	mm->unmap_area = arch_unmap_area;
1398 }
1399 #endif
1400 
1401 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1402 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1403 
1404 extern void normalize_rt_tasks(void);
1405 
1406 #ifdef CONFIG_PM
1407 /*
1408  * Check if a process has been frozen
1409  */
1410 static inline int frozen(struct task_struct *p)
1411 {
1412 	return p->flags & PF_FROZEN;
1413 }
1414 
1415 /*
1416  * Check if there is a request to freeze a process
1417  */
1418 static inline int freezing(struct task_struct *p)
1419 {
1420 	return p->flags & PF_FREEZE;
1421 }
1422 
1423 /*
1424  * Request that a process be frozen
1425  * FIXME: SMP problem. We may not modify other process' flags!
1426  */
1427 static inline void freeze(struct task_struct *p)
1428 {
1429 	p->flags |= PF_FREEZE;
1430 }
1431 
1432 /*
1433  * Wake up a frozen process
1434  */
1435 static inline int thaw_process(struct task_struct *p)
1436 {
1437 	if (frozen(p)) {
1438 		p->flags &= ~PF_FROZEN;
1439 		wake_up_process(p);
1440 		return 1;
1441 	}
1442 	return 0;
1443 }
1444 
1445 /*
1446  * freezing is complete, mark process as frozen
1447  */
1448 static inline void frozen_process(struct task_struct *p)
1449 {
1450 	p->flags = (p->flags & ~PF_FREEZE) | PF_FROZEN;
1451 }
1452 
1453 extern void refrigerator(void);
1454 extern int freeze_processes(void);
1455 extern void thaw_processes(void);
1456 
1457 static inline int try_to_freeze(void)
1458 {
1459 	if (freezing(current)) {
1460 		refrigerator();
1461 		return 1;
1462 	} else
1463 		return 0;
1464 }
1465 #else
1466 static inline int frozen(struct task_struct *p) { return 0; }
1467 static inline int freezing(struct task_struct *p) { return 0; }
1468 static inline void freeze(struct task_struct *p) { BUG(); }
1469 static inline int thaw_process(struct task_struct *p) { return 1; }
1470 static inline void frozen_process(struct task_struct *p) { BUG(); }
1471 
1472 static inline void refrigerator(void) {}
1473 static inline int freeze_processes(void) { BUG(); return 0; }
1474 static inline void thaw_processes(void) {}
1475 
1476 static inline int try_to_freeze(void) { return 0; }
1477 
1478 #endif /* CONFIG_PM */
1479 #endif /* __KERNEL__ */
1480 
1481 #endif
1482