xref: /linux-6.15/include/linux/sched.h (revision 87c2ce3b)
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3 
4 #include <asm/param.h>	/* for HZ */
5 
6 #include <linux/config.h>
7 #include <linux/capability.h>
8 #include <linux/threads.h>
9 #include <linux/kernel.h>
10 #include <linux/types.h>
11 #include <linux/timex.h>
12 #include <linux/jiffies.h>
13 #include <linux/rbtree.h>
14 #include <linux/thread_info.h>
15 #include <linux/cpumask.h>
16 #include <linux/errno.h>
17 #include <linux/nodemask.h>
18 
19 #include <asm/system.h>
20 #include <asm/semaphore.h>
21 #include <asm/page.h>
22 #include <asm/ptrace.h>
23 #include <asm/mmu.h>
24 #include <asm/cputime.h>
25 
26 #include <linux/smp.h>
27 #include <linux/sem.h>
28 #include <linux/signal.h>
29 #include <linux/securebits.h>
30 #include <linux/fs_struct.h>
31 #include <linux/compiler.h>
32 #include <linux/completion.h>
33 #include <linux/pid.h>
34 #include <linux/percpu.h>
35 #include <linux/topology.h>
36 #include <linux/seccomp.h>
37 #include <linux/rcupdate.h>
38 
39 #include <linux/auxvec.h>	/* For AT_VECTOR_SIZE */
40 
41 struct exec_domain;
42 
43 /*
44  * cloning flags:
45  */
46 #define CSIGNAL		0x000000ff	/* signal mask to be sent at exit */
47 #define CLONE_VM	0x00000100	/* set if VM shared between processes */
48 #define CLONE_FS	0x00000200	/* set if fs info shared between processes */
49 #define CLONE_FILES	0x00000400	/* set if open files shared between processes */
50 #define CLONE_SIGHAND	0x00000800	/* set if signal handlers and blocked signals shared */
51 #define CLONE_PTRACE	0x00002000	/* set if we want to let tracing continue on the child too */
52 #define CLONE_VFORK	0x00004000	/* set if the parent wants the child to wake it up on mm_release */
53 #define CLONE_PARENT	0x00008000	/* set if we want to have the same parent as the cloner */
54 #define CLONE_THREAD	0x00010000	/* Same thread group? */
55 #define CLONE_NEWNS	0x00020000	/* New namespace group? */
56 #define CLONE_SYSVSEM	0x00040000	/* share system V SEM_UNDO semantics */
57 #define CLONE_SETTLS	0x00080000	/* create a new TLS for the child */
58 #define CLONE_PARENT_SETTID	0x00100000	/* set the TID in the parent */
59 #define CLONE_CHILD_CLEARTID	0x00200000	/* clear the TID in the child */
60 #define CLONE_DETACHED		0x00400000	/* Unused, ignored */
61 #define CLONE_UNTRACED		0x00800000	/* set if the tracing process can't force CLONE_PTRACE on this clone */
62 #define CLONE_CHILD_SETTID	0x01000000	/* set the TID in the child */
63 #define CLONE_STOPPED		0x02000000	/* Start in stopped state */
64 
65 /*
66  * List of flags we want to share for kernel threads,
67  * if only because they are not used by them anyway.
68  */
69 #define CLONE_KERNEL	(CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
70 
71 /*
72  * These are the constant used to fake the fixed-point load-average
73  * counting. Some notes:
74  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
75  *    a load-average precision of 10 bits integer + 11 bits fractional
76  *  - if you want to count load-averages more often, you need more
77  *    precision, or rounding will get you. With 2-second counting freq,
78  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
79  *    11 bit fractions.
80  */
81 extern unsigned long avenrun[];		/* Load averages */
82 
83 #define FSHIFT		11		/* nr of bits of precision */
84 #define FIXED_1		(1<<FSHIFT)	/* 1.0 as fixed-point */
85 #define LOAD_FREQ	(5*HZ)		/* 5 sec intervals */
86 #define EXP_1		1884		/* 1/exp(5sec/1min) as fixed-point */
87 #define EXP_5		2014		/* 1/exp(5sec/5min) */
88 #define EXP_15		2037		/* 1/exp(5sec/15min) */
89 
90 #define CALC_LOAD(load,exp,n) \
91 	load *= exp; \
92 	load += n*(FIXED_1-exp); \
93 	load >>= FSHIFT;
94 
95 extern unsigned long total_forks;
96 extern int nr_threads;
97 extern int last_pid;
98 DECLARE_PER_CPU(unsigned long, process_counts);
99 extern int nr_processes(void);
100 extern unsigned long nr_running(void);
101 extern unsigned long nr_uninterruptible(void);
102 extern unsigned long nr_iowait(void);
103 
104 #include <linux/time.h>
105 #include <linux/param.h>
106 #include <linux/resource.h>
107 #include <linux/timer.h>
108 #include <linux/hrtimer.h>
109 
110 #include <asm/processor.h>
111 
112 /*
113  * Task state bitmask. NOTE! These bits are also
114  * encoded in fs/proc/array.c: get_task_state().
115  *
116  * We have two separate sets of flags: task->state
117  * is about runnability, while task->exit_state are
118  * about the task exiting. Confusing, but this way
119  * modifying one set can't modify the other one by
120  * mistake.
121  */
122 #define TASK_RUNNING		0
123 #define TASK_INTERRUPTIBLE	1
124 #define TASK_UNINTERRUPTIBLE	2
125 #define TASK_STOPPED		4
126 #define TASK_TRACED		8
127 /* in tsk->exit_state */
128 #define EXIT_ZOMBIE		16
129 #define EXIT_DEAD		32
130 /* in tsk->state again */
131 #define TASK_NONINTERACTIVE	64
132 
133 #define __set_task_state(tsk, state_value)		\
134 	do { (tsk)->state = (state_value); } while (0)
135 #define set_task_state(tsk, state_value)		\
136 	set_mb((tsk)->state, (state_value))
137 
138 /*
139  * set_current_state() includes a barrier so that the write of current->state
140  * is correctly serialised wrt the caller's subsequent test of whether to
141  * actually sleep:
142  *
143  *	set_current_state(TASK_UNINTERRUPTIBLE);
144  *	if (do_i_need_to_sleep())
145  *		schedule();
146  *
147  * If the caller does not need such serialisation then use __set_current_state()
148  */
149 #define __set_current_state(state_value)			\
150 	do { current->state = (state_value); } while (0)
151 #define set_current_state(state_value)		\
152 	set_mb(current->state, (state_value))
153 
154 /* Task command name length */
155 #define TASK_COMM_LEN 16
156 
157 /*
158  * Scheduling policies
159  */
160 #define SCHED_NORMAL		0
161 #define SCHED_FIFO		1
162 #define SCHED_RR		2
163 
164 struct sched_param {
165 	int sched_priority;
166 };
167 
168 #ifdef __KERNEL__
169 
170 #include <linux/spinlock.h>
171 
172 /*
173  * This serializes "schedule()" and also protects
174  * the run-queue from deletions/modifications (but
175  * _adding_ to the beginning of the run-queue has
176  * a separate lock).
177  */
178 extern rwlock_t tasklist_lock;
179 extern spinlock_t mmlist_lock;
180 
181 typedef struct task_struct task_t;
182 
183 extern void sched_init(void);
184 extern void sched_init_smp(void);
185 extern void init_idle(task_t *idle, int cpu);
186 
187 extern cpumask_t nohz_cpu_mask;
188 
189 extern void show_state(void);
190 extern void show_regs(struct pt_regs *);
191 
192 /*
193  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
194  * task), SP is the stack pointer of the first frame that should be shown in the back
195  * trace (or NULL if the entire call-chain of the task should be shown).
196  */
197 extern void show_stack(struct task_struct *task, unsigned long *sp);
198 
199 void io_schedule(void);
200 long io_schedule_timeout(long timeout);
201 
202 extern void cpu_init (void);
203 extern void trap_init(void);
204 extern void update_process_times(int user);
205 extern void scheduler_tick(void);
206 
207 #ifdef CONFIG_DETECT_SOFTLOCKUP
208 extern void softlockup_tick(struct pt_regs *regs);
209 extern void spawn_softlockup_task(void);
210 extern void touch_softlockup_watchdog(void);
211 #else
212 static inline void softlockup_tick(struct pt_regs *regs)
213 {
214 }
215 static inline void spawn_softlockup_task(void)
216 {
217 }
218 static inline void touch_softlockup_watchdog(void)
219 {
220 }
221 #endif
222 
223 
224 /* Attach to any functions which should be ignored in wchan output. */
225 #define __sched		__attribute__((__section__(".sched.text")))
226 /* Is this address in the __sched functions? */
227 extern int in_sched_functions(unsigned long addr);
228 
229 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
230 extern signed long FASTCALL(schedule_timeout(signed long timeout));
231 extern signed long schedule_timeout_interruptible(signed long timeout);
232 extern signed long schedule_timeout_uninterruptible(signed long timeout);
233 asmlinkage void schedule(void);
234 
235 struct namespace;
236 
237 /* Maximum number of active map areas.. This is a random (large) number */
238 #define DEFAULT_MAX_MAP_COUNT	65536
239 
240 extern int sysctl_max_map_count;
241 
242 #include <linux/aio.h>
243 
244 extern unsigned long
245 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
246 		       unsigned long, unsigned long);
247 extern unsigned long
248 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
249 			  unsigned long len, unsigned long pgoff,
250 			  unsigned long flags);
251 extern void arch_unmap_area(struct mm_struct *, unsigned long);
252 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
253 
254 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
255 /*
256  * The mm counters are not protected by its page_table_lock,
257  * so must be incremented atomically.
258  */
259 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
260 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
261 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
262 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
263 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
264 typedef atomic_long_t mm_counter_t;
265 
266 #else  /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
267 /*
268  * The mm counters are protected by its page_table_lock,
269  * so can be incremented directly.
270  */
271 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
272 #define get_mm_counter(mm, member) ((mm)->_##member)
273 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
274 #define inc_mm_counter(mm, member) (mm)->_##member++
275 #define dec_mm_counter(mm, member) (mm)->_##member--
276 typedef unsigned long mm_counter_t;
277 
278 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
279 
280 #define get_mm_rss(mm)					\
281 	(get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
282 #define update_hiwater_rss(mm)	do {			\
283 	unsigned long _rss = get_mm_rss(mm);		\
284 	if ((mm)->hiwater_rss < _rss)			\
285 		(mm)->hiwater_rss = _rss;		\
286 } while (0)
287 #define update_hiwater_vm(mm)	do {			\
288 	if ((mm)->hiwater_vm < (mm)->total_vm)		\
289 		(mm)->hiwater_vm = (mm)->total_vm;	\
290 } while (0)
291 
292 struct mm_struct {
293 	struct vm_area_struct * mmap;		/* list of VMAs */
294 	struct rb_root mm_rb;
295 	struct vm_area_struct * mmap_cache;	/* last find_vma result */
296 	unsigned long (*get_unmapped_area) (struct file *filp,
297 				unsigned long addr, unsigned long len,
298 				unsigned long pgoff, unsigned long flags);
299 	void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
300         unsigned long mmap_base;		/* base of mmap area */
301         unsigned long cached_hole_size;         /* if non-zero, the largest hole below free_area_cache */
302 	unsigned long free_area_cache;		/* first hole of size cached_hole_size or larger */
303 	pgd_t * pgd;
304 	atomic_t mm_users;			/* How many users with user space? */
305 	atomic_t mm_count;			/* How many references to "struct mm_struct" (users count as 1) */
306 	int map_count;				/* number of VMAs */
307 	struct rw_semaphore mmap_sem;
308 	spinlock_t page_table_lock;		/* Protects page tables and some counters */
309 
310 	struct list_head mmlist;		/* List of maybe swapped mm's.  These are globally strung
311 						 * together off init_mm.mmlist, and are protected
312 						 * by mmlist_lock
313 						 */
314 
315 	/* Special counters, in some configurations protected by the
316 	 * page_table_lock, in other configurations by being atomic.
317 	 */
318 	mm_counter_t _file_rss;
319 	mm_counter_t _anon_rss;
320 
321 	unsigned long hiwater_rss;	/* High-watermark of RSS usage */
322 	unsigned long hiwater_vm;	/* High-water virtual memory usage */
323 
324 	unsigned long total_vm, locked_vm, shared_vm, exec_vm;
325 	unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
326 	unsigned long start_code, end_code, start_data, end_data;
327 	unsigned long start_brk, brk, start_stack;
328 	unsigned long arg_start, arg_end, env_start, env_end;
329 
330 	unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
331 
332 	unsigned dumpable:2;
333 	cpumask_t cpu_vm_mask;
334 
335 	/* Architecture-specific MM context */
336 	mm_context_t context;
337 
338 	/* Token based thrashing protection. */
339 	unsigned long swap_token_time;
340 	char recent_pagein;
341 
342 	/* coredumping support */
343 	int core_waiters;
344 	struct completion *core_startup_done, core_done;
345 
346 	/* aio bits */
347 	rwlock_t		ioctx_list_lock;
348 	struct kioctx		*ioctx_list;
349 };
350 
351 struct sighand_struct {
352 	atomic_t		count;
353 	struct k_sigaction	action[_NSIG];
354 	spinlock_t		siglock;
355 	struct rcu_head		rcu;
356 };
357 
358 extern void sighand_free_cb(struct rcu_head *rhp);
359 
360 static inline void sighand_free(struct sighand_struct *sp)
361 {
362 	call_rcu(&sp->rcu, sighand_free_cb);
363 }
364 
365 /*
366  * NOTE! "signal_struct" does not have it's own
367  * locking, because a shared signal_struct always
368  * implies a shared sighand_struct, so locking
369  * sighand_struct is always a proper superset of
370  * the locking of signal_struct.
371  */
372 struct signal_struct {
373 	atomic_t		count;
374 	atomic_t		live;
375 
376 	wait_queue_head_t	wait_chldexit;	/* for wait4() */
377 
378 	/* current thread group signal load-balancing target: */
379 	task_t			*curr_target;
380 
381 	/* shared signal handling: */
382 	struct sigpending	shared_pending;
383 
384 	/* thread group exit support */
385 	int			group_exit_code;
386 	/* overloaded:
387 	 * - notify group_exit_task when ->count is equal to notify_count
388 	 * - everyone except group_exit_task is stopped during signal delivery
389 	 *   of fatal signals, group_exit_task processes the signal.
390 	 */
391 	struct task_struct	*group_exit_task;
392 	int			notify_count;
393 
394 	/* thread group stop support, overloads group_exit_code too */
395 	int			group_stop_count;
396 	unsigned int		flags; /* see SIGNAL_* flags below */
397 
398 	/* POSIX.1b Interval Timers */
399 	struct list_head posix_timers;
400 
401 	/* ITIMER_REAL timer for the process */
402 	struct hrtimer real_timer;
403 	ktime_t it_real_incr;
404 
405 	/* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
406 	cputime_t it_prof_expires, it_virt_expires;
407 	cputime_t it_prof_incr, it_virt_incr;
408 
409 	/* job control IDs */
410 	pid_t pgrp;
411 	pid_t tty_old_pgrp;
412 	pid_t session;
413 	/* boolean value for session group leader */
414 	int leader;
415 
416 	struct tty_struct *tty; /* NULL if no tty */
417 
418 	/*
419 	 * Cumulative resource counters for dead threads in the group,
420 	 * and for reaped dead child processes forked by this group.
421 	 * Live threads maintain their own counters and add to these
422 	 * in __exit_signal, except for the group leader.
423 	 */
424 	cputime_t utime, stime, cutime, cstime;
425 	unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
426 	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
427 
428 	/*
429 	 * Cumulative ns of scheduled CPU time for dead threads in the
430 	 * group, not including a zombie group leader.  (This only differs
431 	 * from jiffies_to_ns(utime + stime) if sched_clock uses something
432 	 * other than jiffies.)
433 	 */
434 	unsigned long long sched_time;
435 
436 	/*
437 	 * We don't bother to synchronize most readers of this at all,
438 	 * because there is no reader checking a limit that actually needs
439 	 * to get both rlim_cur and rlim_max atomically, and either one
440 	 * alone is a single word that can safely be read normally.
441 	 * getrlimit/setrlimit use task_lock(current->group_leader) to
442 	 * protect this instead of the siglock, because they really
443 	 * have no need to disable irqs.
444 	 */
445 	struct rlimit rlim[RLIM_NLIMITS];
446 
447 	struct list_head cpu_timers[3];
448 
449 	/* keep the process-shared keyrings here so that they do the right
450 	 * thing in threads created with CLONE_THREAD */
451 #ifdef CONFIG_KEYS
452 	struct key *session_keyring;	/* keyring inherited over fork */
453 	struct key *process_keyring;	/* keyring private to this process */
454 #endif
455 };
456 
457 /* Context switch must be unlocked if interrupts are to be enabled */
458 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
459 # define __ARCH_WANT_UNLOCKED_CTXSW
460 #endif
461 
462 /*
463  * Bits in flags field of signal_struct.
464  */
465 #define SIGNAL_STOP_STOPPED	0x00000001 /* job control stop in effect */
466 #define SIGNAL_STOP_DEQUEUED	0x00000002 /* stop signal dequeued */
467 #define SIGNAL_STOP_CONTINUED	0x00000004 /* SIGCONT since WCONTINUED reap */
468 #define SIGNAL_GROUP_EXIT	0x00000008 /* group exit in progress */
469 
470 
471 /*
472  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
473  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL tasks are
474  * in the range MAX_RT_PRIO..MAX_PRIO-1. Priority values
475  * are inverted: lower p->prio value means higher priority.
476  *
477  * The MAX_USER_RT_PRIO value allows the actual maximum
478  * RT priority to be separate from the value exported to
479  * user-space.  This allows kernel threads to set their
480  * priority to a value higher than any user task. Note:
481  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
482  */
483 
484 #define MAX_USER_RT_PRIO	100
485 #define MAX_RT_PRIO		MAX_USER_RT_PRIO
486 
487 #define MAX_PRIO		(MAX_RT_PRIO + 40)
488 
489 #define rt_task(p)		(unlikely((p)->prio < MAX_RT_PRIO))
490 
491 /*
492  * Some day this will be a full-fledged user tracking system..
493  */
494 struct user_struct {
495 	atomic_t __count;	/* reference count */
496 	atomic_t processes;	/* How many processes does this user have? */
497 	atomic_t files;		/* How many open files does this user have? */
498 	atomic_t sigpending;	/* How many pending signals does this user have? */
499 #ifdef CONFIG_INOTIFY
500 	atomic_t inotify_watches; /* How many inotify watches does this user have? */
501 	atomic_t inotify_devs;	/* How many inotify devs does this user have opened? */
502 #endif
503 	/* protected by mq_lock	*/
504 	unsigned long mq_bytes;	/* How many bytes can be allocated to mqueue? */
505 	unsigned long locked_shm; /* How many pages of mlocked shm ? */
506 
507 #ifdef CONFIG_KEYS
508 	struct key *uid_keyring;	/* UID specific keyring */
509 	struct key *session_keyring;	/* UID's default session keyring */
510 #endif
511 
512 	/* Hash table maintenance information */
513 	struct list_head uidhash_list;
514 	uid_t uid;
515 };
516 
517 extern struct user_struct *find_user(uid_t);
518 
519 extern struct user_struct root_user;
520 #define INIT_USER (&root_user)
521 
522 typedef struct prio_array prio_array_t;
523 struct backing_dev_info;
524 struct reclaim_state;
525 
526 #ifdef CONFIG_SCHEDSTATS
527 struct sched_info {
528 	/* cumulative counters */
529 	unsigned long	cpu_time,	/* time spent on the cpu */
530 			run_delay,	/* time spent waiting on a runqueue */
531 			pcnt;		/* # of timeslices run on this cpu */
532 
533 	/* timestamps */
534 	unsigned long	last_arrival,	/* when we last ran on a cpu */
535 			last_queued;	/* when we were last queued to run */
536 };
537 
538 extern struct file_operations proc_schedstat_operations;
539 #endif
540 
541 enum idle_type
542 {
543 	SCHED_IDLE,
544 	NOT_IDLE,
545 	NEWLY_IDLE,
546 	MAX_IDLE_TYPES
547 };
548 
549 /*
550  * sched-domains (multiprocessor balancing) declarations:
551  */
552 #ifdef CONFIG_SMP
553 #define SCHED_LOAD_SCALE	128UL	/* increase resolution of load */
554 
555 #define SD_LOAD_BALANCE		1	/* Do load balancing on this domain. */
556 #define SD_BALANCE_NEWIDLE	2	/* Balance when about to become idle */
557 #define SD_BALANCE_EXEC		4	/* Balance on exec */
558 #define SD_BALANCE_FORK		8	/* Balance on fork, clone */
559 #define SD_WAKE_IDLE		16	/* Wake to idle CPU on task wakeup */
560 #define SD_WAKE_AFFINE		32	/* Wake task to waking CPU */
561 #define SD_WAKE_BALANCE		64	/* Perform balancing at task wakeup */
562 #define SD_SHARE_CPUPOWER	128	/* Domain members share cpu power */
563 
564 struct sched_group {
565 	struct sched_group *next;	/* Must be a circular list */
566 	cpumask_t cpumask;
567 
568 	/*
569 	 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
570 	 * single CPU. This is read only (except for setup, hotplug CPU).
571 	 */
572 	unsigned long cpu_power;
573 };
574 
575 struct sched_domain {
576 	/* These fields must be setup */
577 	struct sched_domain *parent;	/* top domain must be null terminated */
578 	struct sched_group *groups;	/* the balancing groups of the domain */
579 	cpumask_t span;			/* span of all CPUs in this domain */
580 	unsigned long min_interval;	/* Minimum balance interval ms */
581 	unsigned long max_interval;	/* Maximum balance interval ms */
582 	unsigned int busy_factor;	/* less balancing by factor if busy */
583 	unsigned int imbalance_pct;	/* No balance until over watermark */
584 	unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
585 	unsigned int cache_nice_tries;	/* Leave cache hot tasks for # tries */
586 	unsigned int per_cpu_gain;	/* CPU % gained by adding domain cpus */
587 	unsigned int busy_idx;
588 	unsigned int idle_idx;
589 	unsigned int newidle_idx;
590 	unsigned int wake_idx;
591 	unsigned int forkexec_idx;
592 	int flags;			/* See SD_* */
593 
594 	/* Runtime fields. */
595 	unsigned long last_balance;	/* init to jiffies. units in jiffies */
596 	unsigned int balance_interval;	/* initialise to 1. units in ms. */
597 	unsigned int nr_balance_failed; /* initialise to 0 */
598 
599 #ifdef CONFIG_SCHEDSTATS
600 	/* load_balance() stats */
601 	unsigned long lb_cnt[MAX_IDLE_TYPES];
602 	unsigned long lb_failed[MAX_IDLE_TYPES];
603 	unsigned long lb_balanced[MAX_IDLE_TYPES];
604 	unsigned long lb_imbalance[MAX_IDLE_TYPES];
605 	unsigned long lb_gained[MAX_IDLE_TYPES];
606 	unsigned long lb_hot_gained[MAX_IDLE_TYPES];
607 	unsigned long lb_nobusyg[MAX_IDLE_TYPES];
608 	unsigned long lb_nobusyq[MAX_IDLE_TYPES];
609 
610 	/* Active load balancing */
611 	unsigned long alb_cnt;
612 	unsigned long alb_failed;
613 	unsigned long alb_pushed;
614 
615 	/* SD_BALANCE_EXEC stats */
616 	unsigned long sbe_cnt;
617 	unsigned long sbe_balanced;
618 	unsigned long sbe_pushed;
619 
620 	/* SD_BALANCE_FORK stats */
621 	unsigned long sbf_cnt;
622 	unsigned long sbf_balanced;
623 	unsigned long sbf_pushed;
624 
625 	/* try_to_wake_up() stats */
626 	unsigned long ttwu_wake_remote;
627 	unsigned long ttwu_move_affine;
628 	unsigned long ttwu_move_balance;
629 #endif
630 };
631 
632 extern void partition_sched_domains(cpumask_t *partition1,
633 				    cpumask_t *partition2);
634 #endif /* CONFIG_SMP */
635 
636 
637 struct io_context;			/* See blkdev.h */
638 void exit_io_context(void);
639 struct cpuset;
640 
641 #define NGROUPS_SMALL		32
642 #define NGROUPS_PER_BLOCK	((int)(PAGE_SIZE / sizeof(gid_t)))
643 struct group_info {
644 	int ngroups;
645 	atomic_t usage;
646 	gid_t small_block[NGROUPS_SMALL];
647 	int nblocks;
648 	gid_t *blocks[0];
649 };
650 
651 /*
652  * get_group_info() must be called with the owning task locked (via task_lock())
653  * when task != current.  The reason being that the vast majority of callers are
654  * looking at current->group_info, which can not be changed except by the
655  * current task.  Changing current->group_info requires the task lock, too.
656  */
657 #define get_group_info(group_info) do { \
658 	atomic_inc(&(group_info)->usage); \
659 } while (0)
660 
661 #define put_group_info(group_info) do { \
662 	if (atomic_dec_and_test(&(group_info)->usage)) \
663 		groups_free(group_info); \
664 } while (0)
665 
666 extern struct group_info *groups_alloc(int gidsetsize);
667 extern void groups_free(struct group_info *group_info);
668 extern int set_current_groups(struct group_info *group_info);
669 extern int groups_search(struct group_info *group_info, gid_t grp);
670 /* access the groups "array" with this macro */
671 #define GROUP_AT(gi, i) \
672     ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
673 
674 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
675 extern void prefetch_stack(struct task_struct*);
676 #else
677 static inline void prefetch_stack(struct task_struct *t) { }
678 #endif
679 
680 struct audit_context;		/* See audit.c */
681 struct mempolicy;
682 
683 struct task_struct {
684 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
685 	struct thread_info *thread_info;
686 	atomic_t usage;
687 	unsigned long flags;	/* per process flags, defined below */
688 	unsigned long ptrace;
689 
690 	int lock_depth;		/* BKL lock depth */
691 
692 #if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW)
693 	int oncpu;
694 #endif
695 	int prio, static_prio;
696 	struct list_head run_list;
697 	prio_array_t *array;
698 
699 	unsigned short ioprio;
700 
701 	unsigned long sleep_avg;
702 	unsigned long long timestamp, last_ran;
703 	unsigned long long sched_time; /* sched_clock time spent running */
704 	int activated;
705 
706 	unsigned long policy;
707 	cpumask_t cpus_allowed;
708 	unsigned int time_slice, first_time_slice;
709 
710 #ifdef CONFIG_SCHEDSTATS
711 	struct sched_info sched_info;
712 #endif
713 
714 	struct list_head tasks;
715 	/*
716 	 * ptrace_list/ptrace_children forms the list of my children
717 	 * that were stolen by a ptracer.
718 	 */
719 	struct list_head ptrace_children;
720 	struct list_head ptrace_list;
721 
722 	struct mm_struct *mm, *active_mm;
723 
724 /* task state */
725 	struct linux_binfmt *binfmt;
726 	long exit_state;
727 	int exit_code, exit_signal;
728 	int pdeath_signal;  /*  The signal sent when the parent dies  */
729 	/* ??? */
730 	unsigned long personality;
731 	unsigned did_exec:1;
732 	pid_t pid;
733 	pid_t tgid;
734 	/*
735 	 * pointers to (original) parent process, youngest child, younger sibling,
736 	 * older sibling, respectively.  (p->father can be replaced with
737 	 * p->parent->pid)
738 	 */
739 	struct task_struct *real_parent; /* real parent process (when being debugged) */
740 	struct task_struct *parent;	/* parent process */
741 	/*
742 	 * children/sibling forms the list of my children plus the
743 	 * tasks I'm ptracing.
744 	 */
745 	struct list_head children;	/* list of my children */
746 	struct list_head sibling;	/* linkage in my parent's children list */
747 	struct task_struct *group_leader;	/* threadgroup leader */
748 
749 	/* PID/PID hash table linkage. */
750 	struct pid pids[PIDTYPE_MAX];
751 
752 	struct completion *vfork_done;		/* for vfork() */
753 	int __user *set_child_tid;		/* CLONE_CHILD_SETTID */
754 	int __user *clear_child_tid;		/* CLONE_CHILD_CLEARTID */
755 
756 	unsigned long rt_priority;
757 	cputime_t utime, stime;
758 	unsigned long nvcsw, nivcsw; /* context switch counts */
759 	struct timespec start_time;
760 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
761 	unsigned long min_flt, maj_flt;
762 
763   	cputime_t it_prof_expires, it_virt_expires;
764 	unsigned long long it_sched_expires;
765 	struct list_head cpu_timers[3];
766 
767 /* process credentials */
768 	uid_t uid,euid,suid,fsuid;
769 	gid_t gid,egid,sgid,fsgid;
770 	struct group_info *group_info;
771 	kernel_cap_t   cap_effective, cap_inheritable, cap_permitted;
772 	unsigned keep_capabilities:1;
773 	struct user_struct *user;
774 #ifdef CONFIG_KEYS
775 	struct key *request_key_auth;	/* assumed request_key authority */
776 	struct key *thread_keyring;	/* keyring private to this thread */
777 	unsigned char jit_keyring;	/* default keyring to attach requested keys to */
778 #endif
779 	int oomkilladj; /* OOM kill score adjustment (bit shift). */
780 	char comm[TASK_COMM_LEN]; /* executable name excluding path
781 				     - access with [gs]et_task_comm (which lock
782 				       it with task_lock())
783 				     - initialized normally by flush_old_exec */
784 /* file system info */
785 	int link_count, total_link_count;
786 /* ipc stuff */
787 	struct sysv_sem sysvsem;
788 /* CPU-specific state of this task */
789 	struct thread_struct thread;
790 /* filesystem information */
791 	struct fs_struct *fs;
792 /* open file information */
793 	struct files_struct *files;
794 /* namespace */
795 	struct namespace *namespace;
796 /* signal handlers */
797 	struct signal_struct *signal;
798 	struct sighand_struct *sighand;
799 
800 	sigset_t blocked, real_blocked;
801 	struct sigpending pending;
802 
803 	unsigned long sas_ss_sp;
804 	size_t sas_ss_size;
805 	int (*notifier)(void *priv);
806 	void *notifier_data;
807 	sigset_t *notifier_mask;
808 
809 	void *security;
810 	struct audit_context *audit_context;
811 	seccomp_t seccomp;
812 
813 /* Thread group tracking */
814    	u32 parent_exec_id;
815    	u32 self_exec_id;
816 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
817 	spinlock_t alloc_lock;
818 /* Protection of proc_dentry: nesting proc_lock, dcache_lock, write_lock_irq(&tasklist_lock); */
819 	spinlock_t proc_lock;
820 
821 #ifdef CONFIG_DEBUG_MUTEXES
822 	/* mutex deadlock detection */
823 	struct mutex_waiter *blocked_on;
824 #endif
825 
826 /* journalling filesystem info */
827 	void *journal_info;
828 
829 /* VM state */
830 	struct reclaim_state *reclaim_state;
831 
832 	struct dentry *proc_dentry;
833 	struct backing_dev_info *backing_dev_info;
834 
835 	struct io_context *io_context;
836 
837 	unsigned long ptrace_message;
838 	siginfo_t *last_siginfo; /* For ptrace use.  */
839 /*
840  * current io wait handle: wait queue entry to use for io waits
841  * If this thread is processing aio, this points at the waitqueue
842  * inside the currently handled kiocb. It may be NULL (i.e. default
843  * to a stack based synchronous wait) if its doing sync IO.
844  */
845 	wait_queue_t *io_wait;
846 /* i/o counters(bytes read/written, #syscalls */
847 	u64 rchar, wchar, syscr, syscw;
848 #if defined(CONFIG_BSD_PROCESS_ACCT)
849 	u64 acct_rss_mem1;	/* accumulated rss usage */
850 	u64 acct_vm_mem1;	/* accumulated virtual memory usage */
851 	clock_t acct_stimexpd;	/* clock_t-converted stime since last update */
852 #endif
853 #ifdef CONFIG_NUMA
854   	struct mempolicy *mempolicy;
855 	short il_next;
856 #endif
857 #ifdef CONFIG_CPUSETS
858 	struct cpuset *cpuset;
859 	nodemask_t mems_allowed;
860 	int cpuset_mems_generation;
861 #endif
862 	atomic_t fs_excl;	/* holding fs exclusive resources */
863 	struct rcu_head rcu;
864 };
865 
866 static inline pid_t process_group(struct task_struct *tsk)
867 {
868 	return tsk->signal->pgrp;
869 }
870 
871 /**
872  * pid_alive - check that a task structure is not stale
873  * @p: Task structure to be checked.
874  *
875  * Test if a process is not yet dead (at most zombie state)
876  * If pid_alive fails, then pointers within the task structure
877  * can be stale and must not be dereferenced.
878  */
879 static inline int pid_alive(struct task_struct *p)
880 {
881 	return p->pids[PIDTYPE_PID].nr != 0;
882 }
883 
884 extern void free_task(struct task_struct *tsk);
885 extern void __put_task_struct(struct task_struct *tsk);
886 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
887 
888 extern void __put_task_struct_cb(struct rcu_head *rhp);
889 
890 static inline void put_task_struct(struct task_struct *t)
891 {
892 	if (atomic_dec_and_test(&t->usage))
893 		call_rcu(&t->rcu, __put_task_struct_cb);
894 }
895 
896 /*
897  * Per process flags
898  */
899 #define PF_ALIGNWARN	0x00000001	/* Print alignment warning msgs */
900 					/* Not implemented yet, only for 486*/
901 #define PF_STARTING	0x00000002	/* being created */
902 #define PF_EXITING	0x00000004	/* getting shut down */
903 #define PF_DEAD		0x00000008	/* Dead */
904 #define PF_FORKNOEXEC	0x00000040	/* forked but didn't exec */
905 #define PF_SUPERPRIV	0x00000100	/* used super-user privileges */
906 #define PF_DUMPCORE	0x00000200	/* dumped core */
907 #define PF_SIGNALED	0x00000400	/* killed by a signal */
908 #define PF_MEMALLOC	0x00000800	/* Allocating memory */
909 #define PF_FLUSHER	0x00001000	/* responsible for disk writeback */
910 #define PF_USED_MATH	0x00002000	/* if unset the fpu must be initialized before use */
911 #define PF_FREEZE	0x00004000	/* this task is being frozen for suspend now */
912 #define PF_NOFREEZE	0x00008000	/* this thread should not be frozen */
913 #define PF_FROZEN	0x00010000	/* frozen for system suspend */
914 #define PF_FSTRANS	0x00020000	/* inside a filesystem transaction */
915 #define PF_KSWAPD	0x00040000	/* I am kswapd */
916 #define PF_SWAPOFF	0x00080000	/* I am in swapoff */
917 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
918 #define PF_SYNCWRITE	0x00200000	/* I am doing a sync write */
919 #define PF_BORROWED_MM	0x00400000	/* I am a kthread doing use_mm */
920 #define PF_RANDOMIZE	0x00800000	/* randomize virtual address space */
921 #define PF_SWAPWRITE	0x01000000	/* Allowed to write to swap */
922 
923 /*
924  * Only the _current_ task can read/write to tsk->flags, but other
925  * tasks can access tsk->flags in readonly mode for example
926  * with tsk_used_math (like during threaded core dumping).
927  * There is however an exception to this rule during ptrace
928  * or during fork: the ptracer task is allowed to write to the
929  * child->flags of its traced child (same goes for fork, the parent
930  * can write to the child->flags), because we're guaranteed the
931  * child is not running and in turn not changing child->flags
932  * at the same time the parent does it.
933  */
934 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
935 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
936 #define clear_used_math() clear_stopped_child_used_math(current)
937 #define set_used_math() set_stopped_child_used_math(current)
938 #define conditional_stopped_child_used_math(condition, child) \
939 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
940 #define conditional_used_math(condition) \
941 	conditional_stopped_child_used_math(condition, current)
942 #define copy_to_stopped_child_used_math(child) \
943 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
944 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
945 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
946 #define used_math() tsk_used_math(current)
947 
948 #ifdef CONFIG_SMP
949 extern int set_cpus_allowed(task_t *p, cpumask_t new_mask);
950 #else
951 static inline int set_cpus_allowed(task_t *p, cpumask_t new_mask)
952 {
953 	if (!cpu_isset(0, new_mask))
954 		return -EINVAL;
955 	return 0;
956 }
957 #endif
958 
959 extern unsigned long long sched_clock(void);
960 extern unsigned long long current_sched_time(const task_t *current_task);
961 
962 /* sched_exec is called by processes performing an exec */
963 #ifdef CONFIG_SMP
964 extern void sched_exec(void);
965 #else
966 #define sched_exec()   {}
967 #endif
968 
969 #ifdef CONFIG_HOTPLUG_CPU
970 extern void idle_task_exit(void);
971 #else
972 static inline void idle_task_exit(void) {}
973 #endif
974 
975 extern void sched_idle_next(void);
976 extern void set_user_nice(task_t *p, long nice);
977 extern int task_prio(const task_t *p);
978 extern int task_nice(const task_t *p);
979 extern int can_nice(const task_t *p, const int nice);
980 extern int task_curr(const task_t *p);
981 extern int idle_cpu(int cpu);
982 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
983 extern task_t *idle_task(int cpu);
984 extern task_t *curr_task(int cpu);
985 extern void set_curr_task(int cpu, task_t *p);
986 
987 void yield(void);
988 
989 /*
990  * The default (Linux) execution domain.
991  */
992 extern struct exec_domain	default_exec_domain;
993 
994 union thread_union {
995 	struct thread_info thread_info;
996 	unsigned long stack[THREAD_SIZE/sizeof(long)];
997 };
998 
999 #ifndef __HAVE_ARCH_KSTACK_END
1000 static inline int kstack_end(void *addr)
1001 {
1002 	/* Reliable end of stack detection:
1003 	 * Some APM bios versions misalign the stack
1004 	 */
1005 	return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1006 }
1007 #endif
1008 
1009 extern union thread_union init_thread_union;
1010 extern struct task_struct init_task;
1011 
1012 extern struct   mm_struct init_mm;
1013 
1014 #define find_task_by_pid(nr)	find_task_by_pid_type(PIDTYPE_PID, nr)
1015 extern struct task_struct *find_task_by_pid_type(int type, int pid);
1016 extern void set_special_pids(pid_t session, pid_t pgrp);
1017 extern void __set_special_pids(pid_t session, pid_t pgrp);
1018 
1019 /* per-UID process charging. */
1020 extern struct user_struct * alloc_uid(uid_t);
1021 static inline struct user_struct *get_uid(struct user_struct *u)
1022 {
1023 	atomic_inc(&u->__count);
1024 	return u;
1025 }
1026 extern void free_uid(struct user_struct *);
1027 extern void switch_uid(struct user_struct *);
1028 
1029 #include <asm/current.h>
1030 
1031 extern void do_timer(struct pt_regs *);
1032 
1033 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1034 extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1035 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1036 						unsigned long clone_flags));
1037 #ifdef CONFIG_SMP
1038  extern void kick_process(struct task_struct *tsk);
1039 #else
1040  static inline void kick_process(struct task_struct *tsk) { }
1041 #endif
1042 extern void FASTCALL(sched_fork(task_t * p, int clone_flags));
1043 extern void FASTCALL(sched_exit(task_t * p));
1044 
1045 extern int in_group_p(gid_t);
1046 extern int in_egroup_p(gid_t);
1047 
1048 extern void proc_caches_init(void);
1049 extern void flush_signals(struct task_struct *);
1050 extern void flush_signal_handlers(struct task_struct *, int force_default);
1051 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1052 
1053 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1054 {
1055 	unsigned long flags;
1056 	int ret;
1057 
1058 	spin_lock_irqsave(&tsk->sighand->siglock, flags);
1059 	ret = dequeue_signal(tsk, mask, info);
1060 	spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1061 
1062 	return ret;
1063 }
1064 
1065 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1066 			      sigset_t *mask);
1067 extern void unblock_all_signals(void);
1068 extern void release_task(struct task_struct * p);
1069 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1070 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1071 extern int force_sigsegv(int, struct task_struct *);
1072 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1073 extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp);
1074 extern int kill_pg_info(int, struct siginfo *, pid_t);
1075 extern int kill_proc_info(int, struct siginfo *, pid_t);
1076 extern int kill_proc_info_as_uid(int, struct siginfo *, pid_t, uid_t, uid_t);
1077 extern void do_notify_parent(struct task_struct *, int);
1078 extern void force_sig(int, struct task_struct *);
1079 extern void force_sig_specific(int, struct task_struct *);
1080 extern int send_sig(int, struct task_struct *, int);
1081 extern void zap_other_threads(struct task_struct *p);
1082 extern int kill_pg(pid_t, int, int);
1083 extern int kill_sl(pid_t, int, int);
1084 extern int kill_proc(pid_t, int, int);
1085 extern struct sigqueue *sigqueue_alloc(void);
1086 extern void sigqueue_free(struct sigqueue *);
1087 extern int send_sigqueue(int, struct sigqueue *,  struct task_struct *);
1088 extern int send_group_sigqueue(int, struct sigqueue *,  struct task_struct *);
1089 extern int do_sigaction(int, const struct k_sigaction *, struct k_sigaction *);
1090 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1091 
1092 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
1093 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1094 #define SEND_SIG_PRIV	((struct siginfo *) 1)
1095 #define SEND_SIG_FORCED	((struct siginfo *) 2)
1096 
1097 static inline int is_si_special(const struct siginfo *info)
1098 {
1099 	return info <= SEND_SIG_FORCED;
1100 }
1101 
1102 /* True if we are on the alternate signal stack.  */
1103 
1104 static inline int on_sig_stack(unsigned long sp)
1105 {
1106 	return (sp - current->sas_ss_sp < current->sas_ss_size);
1107 }
1108 
1109 static inline int sas_ss_flags(unsigned long sp)
1110 {
1111 	return (current->sas_ss_size == 0 ? SS_DISABLE
1112 		: on_sig_stack(sp) ? SS_ONSTACK : 0);
1113 }
1114 
1115 
1116 #ifdef CONFIG_SECURITY
1117 /* code is in security.c */
1118 extern int capable(int cap);
1119 #else
1120 static inline int capable(int cap)
1121 {
1122 	if (cap_raised(current->cap_effective, cap)) {
1123 		current->flags |= PF_SUPERPRIV;
1124 		return 1;
1125 	}
1126 	return 0;
1127 }
1128 #endif
1129 
1130 /*
1131  * Routines for handling mm_structs
1132  */
1133 extern struct mm_struct * mm_alloc(void);
1134 
1135 /* mmdrop drops the mm and the page tables */
1136 extern void FASTCALL(__mmdrop(struct mm_struct *));
1137 static inline void mmdrop(struct mm_struct * mm)
1138 {
1139 	if (atomic_dec_and_test(&mm->mm_count))
1140 		__mmdrop(mm);
1141 }
1142 
1143 /* mmput gets rid of the mappings and all user-space */
1144 extern void mmput(struct mm_struct *);
1145 /* Grab a reference to a task's mm, if it is not already going away */
1146 extern struct mm_struct *get_task_mm(struct task_struct *task);
1147 /* Remove the current tasks stale references to the old mm_struct */
1148 extern void mm_release(struct task_struct *, struct mm_struct *);
1149 
1150 extern int  copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1151 extern void flush_thread(void);
1152 extern void exit_thread(void);
1153 
1154 extern void exit_files(struct task_struct *);
1155 extern void exit_signal(struct task_struct *);
1156 extern void __exit_signal(struct task_struct *);
1157 extern void exit_sighand(struct task_struct *);
1158 extern void __exit_sighand(struct task_struct *);
1159 extern void exit_itimers(struct signal_struct *);
1160 
1161 extern NORET_TYPE void do_group_exit(int);
1162 
1163 extern void daemonize(const char *, ...);
1164 extern int allow_signal(int);
1165 extern int disallow_signal(int);
1166 extern task_t *child_reaper;
1167 
1168 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1169 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1170 task_t *fork_idle(int);
1171 
1172 extern void set_task_comm(struct task_struct *tsk, char *from);
1173 extern void get_task_comm(char *to, struct task_struct *tsk);
1174 
1175 #ifdef CONFIG_SMP
1176 extern void wait_task_inactive(task_t * p);
1177 #else
1178 #define wait_task_inactive(p)	do { } while (0)
1179 #endif
1180 
1181 #define remove_parent(p)	list_del_init(&(p)->sibling)
1182 #define add_parent(p, parent)	list_add_tail(&(p)->sibling,&(parent)->children)
1183 
1184 #define REMOVE_LINKS(p) do {					\
1185 	if (thread_group_leader(p))				\
1186 		list_del_init(&(p)->tasks);			\
1187 	remove_parent(p);					\
1188 	} while (0)
1189 
1190 #define SET_LINKS(p) do {					\
1191 	if (thread_group_leader(p))				\
1192 		list_add_tail(&(p)->tasks,&init_task.tasks);	\
1193 	add_parent(p, (p)->parent);				\
1194 	} while (0)
1195 
1196 #define next_task(p)	list_entry((p)->tasks.next, struct task_struct, tasks)
1197 #define prev_task(p)	list_entry((p)->tasks.prev, struct task_struct, tasks)
1198 
1199 #define for_each_process(p) \
1200 	for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1201 
1202 /*
1203  * Careful: do_each_thread/while_each_thread is a double loop so
1204  *          'break' will not work as expected - use goto instead.
1205  */
1206 #define do_each_thread(g, t) \
1207 	for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1208 
1209 #define while_each_thread(g, t) \
1210 	while ((t = next_thread(t)) != g)
1211 
1212 extern task_t * FASTCALL(next_thread(const task_t *p));
1213 
1214 #define thread_group_leader(p)	(p->pid == p->tgid)
1215 
1216 static inline int thread_group_empty(task_t *p)
1217 {
1218 	return list_empty(&p->pids[PIDTYPE_TGID].pid_list);
1219 }
1220 
1221 #define delay_group_leader(p) \
1222 		(thread_group_leader(p) && !thread_group_empty(p))
1223 
1224 extern void unhash_process(struct task_struct *p);
1225 
1226 /*
1227  * Protects ->fs, ->files, ->mm, ->ptrace, ->group_info, ->comm, keyring
1228  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
1229  * pins the final release of task.io_context.  Also protects ->cpuset.
1230  *
1231  * Nests both inside and outside of read_lock(&tasklist_lock).
1232  * It must not be nested with write_lock_irq(&tasklist_lock),
1233  * neither inside nor outside.
1234  */
1235 static inline void task_lock(struct task_struct *p)
1236 {
1237 	spin_lock(&p->alloc_lock);
1238 }
1239 
1240 static inline void task_unlock(struct task_struct *p)
1241 {
1242 	spin_unlock(&p->alloc_lock);
1243 }
1244 
1245 #ifndef __HAVE_THREAD_FUNCTIONS
1246 
1247 #define task_thread_info(task) (task)->thread_info
1248 
1249 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1250 {
1251 	*task_thread_info(p) = *task_thread_info(org);
1252 	task_thread_info(p)->task = p;
1253 }
1254 
1255 static inline unsigned long *end_of_stack(struct task_struct *p)
1256 {
1257 	return (unsigned long *)(p->thread_info + 1);
1258 }
1259 
1260 #endif
1261 
1262 /* set thread flags in other task's structures
1263  * - see asm/thread_info.h for TIF_xxxx flags available
1264  */
1265 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1266 {
1267 	set_ti_thread_flag(task_thread_info(tsk), flag);
1268 }
1269 
1270 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1271 {
1272 	clear_ti_thread_flag(task_thread_info(tsk), flag);
1273 }
1274 
1275 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1276 {
1277 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1278 }
1279 
1280 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1281 {
1282 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1283 }
1284 
1285 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1286 {
1287 	return test_ti_thread_flag(task_thread_info(tsk), flag);
1288 }
1289 
1290 static inline void set_tsk_need_resched(struct task_struct *tsk)
1291 {
1292 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1293 }
1294 
1295 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1296 {
1297 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1298 }
1299 
1300 static inline int signal_pending(struct task_struct *p)
1301 {
1302 	return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1303 }
1304 
1305 static inline int need_resched(void)
1306 {
1307 	return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1308 }
1309 
1310 /*
1311  * cond_resched() and cond_resched_lock(): latency reduction via
1312  * explicit rescheduling in places that are safe. The return
1313  * value indicates whether a reschedule was done in fact.
1314  * cond_resched_lock() will drop the spinlock before scheduling,
1315  * cond_resched_softirq() will enable bhs before scheduling.
1316  */
1317 extern int cond_resched(void);
1318 extern int cond_resched_lock(spinlock_t * lock);
1319 extern int cond_resched_softirq(void);
1320 
1321 /*
1322  * Does a critical section need to be broken due to another
1323  * task waiting?:
1324  */
1325 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1326 # define need_lockbreak(lock) ((lock)->break_lock)
1327 #else
1328 # define need_lockbreak(lock) 0
1329 #endif
1330 
1331 /*
1332  * Does a critical section need to be broken due to another
1333  * task waiting or preemption being signalled:
1334  */
1335 static inline int lock_need_resched(spinlock_t *lock)
1336 {
1337 	if (need_lockbreak(lock) || need_resched())
1338 		return 1;
1339 	return 0;
1340 }
1341 
1342 /* Reevaluate whether the task has signals pending delivery.
1343    This is required every time the blocked sigset_t changes.
1344    callers must hold sighand->siglock.  */
1345 
1346 extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1347 extern void recalc_sigpending(void);
1348 
1349 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1350 
1351 /*
1352  * Wrappers for p->thread_info->cpu access. No-op on UP.
1353  */
1354 #ifdef CONFIG_SMP
1355 
1356 static inline unsigned int task_cpu(const struct task_struct *p)
1357 {
1358 	return task_thread_info(p)->cpu;
1359 }
1360 
1361 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1362 {
1363 	task_thread_info(p)->cpu = cpu;
1364 }
1365 
1366 #else
1367 
1368 static inline unsigned int task_cpu(const struct task_struct *p)
1369 {
1370 	return 0;
1371 }
1372 
1373 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1374 {
1375 }
1376 
1377 #endif /* CONFIG_SMP */
1378 
1379 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1380 extern void arch_pick_mmap_layout(struct mm_struct *mm);
1381 #else
1382 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1383 {
1384 	mm->mmap_base = TASK_UNMAPPED_BASE;
1385 	mm->get_unmapped_area = arch_get_unmapped_area;
1386 	mm->unmap_area = arch_unmap_area;
1387 }
1388 #endif
1389 
1390 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1391 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1392 
1393 #ifdef CONFIG_MAGIC_SYSRQ
1394 
1395 extern void normalize_rt_tasks(void);
1396 
1397 #endif
1398 
1399 #ifdef CONFIG_PM
1400 /*
1401  * Check if a process has been frozen
1402  */
1403 static inline int frozen(struct task_struct *p)
1404 {
1405 	return p->flags & PF_FROZEN;
1406 }
1407 
1408 /*
1409  * Check if there is a request to freeze a process
1410  */
1411 static inline int freezing(struct task_struct *p)
1412 {
1413 	return p->flags & PF_FREEZE;
1414 }
1415 
1416 /*
1417  * Request that a process be frozen
1418  * FIXME: SMP problem. We may not modify other process' flags!
1419  */
1420 static inline void freeze(struct task_struct *p)
1421 {
1422 	p->flags |= PF_FREEZE;
1423 }
1424 
1425 /*
1426  * Wake up a frozen process
1427  */
1428 static inline int thaw_process(struct task_struct *p)
1429 {
1430 	if (frozen(p)) {
1431 		p->flags &= ~PF_FROZEN;
1432 		wake_up_process(p);
1433 		return 1;
1434 	}
1435 	return 0;
1436 }
1437 
1438 /*
1439  * freezing is complete, mark process as frozen
1440  */
1441 static inline void frozen_process(struct task_struct *p)
1442 {
1443 	p->flags = (p->flags & ~PF_FREEZE) | PF_FROZEN;
1444 }
1445 
1446 extern void refrigerator(void);
1447 extern int freeze_processes(void);
1448 extern void thaw_processes(void);
1449 
1450 static inline int try_to_freeze(void)
1451 {
1452 	if (freezing(current)) {
1453 		refrigerator();
1454 		return 1;
1455 	} else
1456 		return 0;
1457 }
1458 #else
1459 static inline int frozen(struct task_struct *p) { return 0; }
1460 static inline int freezing(struct task_struct *p) { return 0; }
1461 static inline void freeze(struct task_struct *p) { BUG(); }
1462 static inline int thaw_process(struct task_struct *p) { return 1; }
1463 static inline void frozen_process(struct task_struct *p) { BUG(); }
1464 
1465 static inline void refrigerator(void) {}
1466 static inline int freeze_processes(void) { BUG(); return 0; }
1467 static inline void thaw_processes(void) {}
1468 
1469 static inline int try_to_freeze(void) { return 0; }
1470 
1471 #endif /* CONFIG_PM */
1472 #endif /* __KERNEL__ */
1473 
1474 #endif
1475