xref: /linux/include/linux/sched.h (revision d8327c784b51b57dac2c26cfad87dce0d68dfd98)
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 #define SCHED_BATCH		3
164 
165 struct sched_param {
166 	int sched_priority;
167 };
168 
169 #ifdef __KERNEL__
170 
171 #include <linux/spinlock.h>
172 
173 /*
174  * This serializes "schedule()" and also protects
175  * the run-queue from deletions/modifications (but
176  * _adding_ to the beginning of the run-queue has
177  * a separate lock).
178  */
179 extern rwlock_t tasklist_lock;
180 extern spinlock_t mmlist_lock;
181 
182 typedef struct task_struct task_t;
183 
184 extern void sched_init(void);
185 extern void sched_init_smp(void);
186 extern void init_idle(task_t *idle, int cpu);
187 
188 extern cpumask_t nohz_cpu_mask;
189 
190 extern void show_state(void);
191 extern void show_regs(struct pt_regs *);
192 
193 /*
194  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
195  * task), SP is the stack pointer of the first frame that should be shown in the back
196  * trace (or NULL if the entire call-chain of the task should be shown).
197  */
198 extern void show_stack(struct task_struct *task, unsigned long *sp);
199 
200 void io_schedule(void);
201 long io_schedule_timeout(long timeout);
202 
203 extern void cpu_init (void);
204 extern void trap_init(void);
205 extern void update_process_times(int user);
206 extern void scheduler_tick(void);
207 
208 #ifdef CONFIG_DETECT_SOFTLOCKUP
209 extern void softlockup_tick(struct pt_regs *regs);
210 extern void spawn_softlockup_task(void);
211 extern void touch_softlockup_watchdog(void);
212 #else
213 static inline void softlockup_tick(struct pt_regs *regs)
214 {
215 }
216 static inline void spawn_softlockup_task(void)
217 {
218 }
219 static inline void touch_softlockup_watchdog(void)
220 {
221 }
222 #endif
223 
224 
225 /* Attach to any functions which should be ignored in wchan output. */
226 #define __sched		__attribute__((__section__(".sched.text")))
227 /* Is this address in the __sched functions? */
228 extern int in_sched_functions(unsigned long addr);
229 
230 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
231 extern signed long FASTCALL(schedule_timeout(signed long timeout));
232 extern signed long schedule_timeout_interruptible(signed long timeout);
233 extern signed long schedule_timeout_uninterruptible(signed long timeout);
234 asmlinkage void schedule(void);
235 
236 struct namespace;
237 
238 /* Maximum number of active map areas.. This is a random (large) number */
239 #define DEFAULT_MAX_MAP_COUNT	65536
240 
241 extern int sysctl_max_map_count;
242 
243 #include <linux/aio.h>
244 
245 extern unsigned long
246 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
247 		       unsigned long, unsigned long);
248 extern unsigned long
249 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
250 			  unsigned long len, unsigned long pgoff,
251 			  unsigned long flags);
252 extern void arch_unmap_area(struct mm_struct *, unsigned long);
253 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
254 
255 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
256 /*
257  * The mm counters are not protected by its page_table_lock,
258  * so must be incremented atomically.
259  */
260 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
261 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
262 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
263 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
264 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
265 typedef atomic_long_t mm_counter_t;
266 
267 #else  /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
268 /*
269  * The mm counters are protected by its page_table_lock,
270  * so can be incremented directly.
271  */
272 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
273 #define get_mm_counter(mm, member) ((mm)->_##member)
274 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
275 #define inc_mm_counter(mm, member) (mm)->_##member++
276 #define dec_mm_counter(mm, member) (mm)->_##member--
277 typedef unsigned long mm_counter_t;
278 
279 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
280 
281 #define get_mm_rss(mm)					\
282 	(get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
283 #define update_hiwater_rss(mm)	do {			\
284 	unsigned long _rss = get_mm_rss(mm);		\
285 	if ((mm)->hiwater_rss < _rss)			\
286 		(mm)->hiwater_rss = _rss;		\
287 } while (0)
288 #define update_hiwater_vm(mm)	do {			\
289 	if ((mm)->hiwater_vm < (mm)->total_vm)		\
290 		(mm)->hiwater_vm = (mm)->total_vm;	\
291 } while (0)
292 
293 struct mm_struct {
294 	struct vm_area_struct * mmap;		/* list of VMAs */
295 	struct rb_root mm_rb;
296 	struct vm_area_struct * mmap_cache;	/* last find_vma result */
297 	unsigned long (*get_unmapped_area) (struct file *filp,
298 				unsigned long addr, unsigned long len,
299 				unsigned long pgoff, unsigned long flags);
300 	void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
301 	unsigned long mmap_base;		/* base of mmap area */
302 	unsigned long task_size;		/* size of task vm space */
303 	unsigned long cached_hole_size;         /* if non-zero, the largest hole below free_area_cache */
304 	unsigned long free_area_cache;		/* first hole of size cached_hole_size or larger */
305 	pgd_t * pgd;
306 	atomic_t mm_users;			/* How many users with user space? */
307 	atomic_t mm_count;			/* How many references to "struct mm_struct" (users count as 1) */
308 	int map_count;				/* number of VMAs */
309 	struct rw_semaphore mmap_sem;
310 	spinlock_t page_table_lock;		/* Protects page tables and some counters */
311 
312 	struct list_head mmlist;		/* List of maybe swapped mm's.  These are globally strung
313 						 * together off init_mm.mmlist, and are protected
314 						 * by mmlist_lock
315 						 */
316 
317 	/* Special counters, in some configurations protected by the
318 	 * page_table_lock, in other configurations by being atomic.
319 	 */
320 	mm_counter_t _file_rss;
321 	mm_counter_t _anon_rss;
322 
323 	unsigned long hiwater_rss;	/* High-watermark of RSS usage */
324 	unsigned long hiwater_vm;	/* High-water virtual memory usage */
325 
326 	unsigned long total_vm, locked_vm, shared_vm, exec_vm;
327 	unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
328 	unsigned long start_code, end_code, start_data, end_data;
329 	unsigned long start_brk, brk, start_stack;
330 	unsigned long arg_start, arg_end, env_start, env_end;
331 
332 	unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
333 
334 	unsigned dumpable:2;
335 	cpumask_t cpu_vm_mask;
336 
337 	/* Architecture-specific MM context */
338 	mm_context_t context;
339 
340 	/* Token based thrashing protection. */
341 	unsigned long swap_token_time;
342 	char recent_pagein;
343 
344 	/* coredumping support */
345 	int core_waiters;
346 	struct completion *core_startup_done, core_done;
347 
348 	/* aio bits */
349 	rwlock_t		ioctx_list_lock;
350 	struct kioctx		*ioctx_list;
351 };
352 
353 struct sighand_struct {
354 	atomic_t		count;
355 	struct k_sigaction	action[_NSIG];
356 	spinlock_t		siglock;
357 	struct rcu_head		rcu;
358 };
359 
360 extern void sighand_free_cb(struct rcu_head *rhp);
361 
362 static inline void sighand_free(struct sighand_struct *sp)
363 {
364 	call_rcu(&sp->rcu, sighand_free_cb);
365 }
366 
367 /*
368  * NOTE! "signal_struct" does not have it's own
369  * locking, because a shared signal_struct always
370  * implies a shared sighand_struct, so locking
371  * sighand_struct is always a proper superset of
372  * the locking of signal_struct.
373  */
374 struct signal_struct {
375 	atomic_t		count;
376 	atomic_t		live;
377 
378 	wait_queue_head_t	wait_chldexit;	/* for wait4() */
379 
380 	/* current thread group signal load-balancing target: */
381 	task_t			*curr_target;
382 
383 	/* shared signal handling: */
384 	struct sigpending	shared_pending;
385 
386 	/* thread group exit support */
387 	int			group_exit_code;
388 	/* overloaded:
389 	 * - notify group_exit_task when ->count is equal to notify_count
390 	 * - everyone except group_exit_task is stopped during signal delivery
391 	 *   of fatal signals, group_exit_task processes the signal.
392 	 */
393 	struct task_struct	*group_exit_task;
394 	int			notify_count;
395 
396 	/* thread group stop support, overloads group_exit_code too */
397 	int			group_stop_count;
398 	unsigned int		flags; /* see SIGNAL_* flags below */
399 
400 	/* POSIX.1b Interval Timers */
401 	struct list_head posix_timers;
402 
403 	/* ITIMER_REAL timer for the process */
404 	struct hrtimer real_timer;
405 	ktime_t it_real_incr;
406 
407 	/* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
408 	cputime_t it_prof_expires, it_virt_expires;
409 	cputime_t it_prof_incr, it_virt_incr;
410 
411 	/* job control IDs */
412 	pid_t pgrp;
413 	pid_t tty_old_pgrp;
414 	pid_t session;
415 	/* boolean value for session group leader */
416 	int leader;
417 
418 	struct tty_struct *tty; /* NULL if no tty */
419 
420 	/*
421 	 * Cumulative resource counters for dead threads in the group,
422 	 * and for reaped dead child processes forked by this group.
423 	 * Live threads maintain their own counters and add to these
424 	 * in __exit_signal, except for the group leader.
425 	 */
426 	cputime_t utime, stime, cutime, cstime;
427 	unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
428 	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
429 
430 	/*
431 	 * Cumulative ns of scheduled CPU time for dead threads in the
432 	 * group, not including a zombie group leader.  (This only differs
433 	 * from jiffies_to_ns(utime + stime) if sched_clock uses something
434 	 * other than jiffies.)
435 	 */
436 	unsigned long long sched_time;
437 
438 	/*
439 	 * We don't bother to synchronize most readers of this at all,
440 	 * because there is no reader checking a limit that actually needs
441 	 * to get both rlim_cur and rlim_max atomically, and either one
442 	 * alone is a single word that can safely be read normally.
443 	 * getrlimit/setrlimit use task_lock(current->group_leader) to
444 	 * protect this instead of the siglock, because they really
445 	 * have no need to disable irqs.
446 	 */
447 	struct rlimit rlim[RLIM_NLIMITS];
448 
449 	struct list_head cpu_timers[3];
450 
451 	/* keep the process-shared keyrings here so that they do the right
452 	 * thing in threads created with CLONE_THREAD */
453 #ifdef CONFIG_KEYS
454 	struct key *session_keyring;	/* keyring inherited over fork */
455 	struct key *process_keyring;	/* keyring private to this process */
456 #endif
457 };
458 
459 /* Context switch must be unlocked if interrupts are to be enabled */
460 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
461 # define __ARCH_WANT_UNLOCKED_CTXSW
462 #endif
463 
464 /*
465  * Bits in flags field of signal_struct.
466  */
467 #define SIGNAL_STOP_STOPPED	0x00000001 /* job control stop in effect */
468 #define SIGNAL_STOP_DEQUEUED	0x00000002 /* stop signal dequeued */
469 #define SIGNAL_STOP_CONTINUED	0x00000004 /* SIGCONT since WCONTINUED reap */
470 #define SIGNAL_GROUP_EXIT	0x00000008 /* group exit in progress */
471 
472 
473 /*
474  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
475  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
476  * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
477  * values are inverted: lower p->prio value means higher priority.
478  *
479  * The MAX_USER_RT_PRIO value allows the actual maximum
480  * RT priority to be separate from the value exported to
481  * user-space.  This allows kernel threads to set their
482  * priority to a value higher than any user task. Note:
483  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
484  */
485 
486 #define MAX_USER_RT_PRIO	100
487 #define MAX_RT_PRIO		MAX_USER_RT_PRIO
488 
489 #define MAX_PRIO		(MAX_RT_PRIO + 40)
490 
491 #define rt_task(p)		(unlikely((p)->prio < MAX_RT_PRIO))
492 
493 /*
494  * Some day this will be a full-fledged user tracking system..
495  */
496 struct user_struct {
497 	atomic_t __count;	/* reference count */
498 	atomic_t processes;	/* How many processes does this user have? */
499 	atomic_t files;		/* How many open files does this user have? */
500 	atomic_t sigpending;	/* How many pending signals does this user have? */
501 #ifdef CONFIG_INOTIFY
502 	atomic_t inotify_watches; /* How many inotify watches does this user have? */
503 	atomic_t inotify_devs;	/* How many inotify devs does this user have opened? */
504 #endif
505 	/* protected by mq_lock	*/
506 	unsigned long mq_bytes;	/* How many bytes can be allocated to mqueue? */
507 	unsigned long locked_shm; /* How many pages of mlocked shm ? */
508 
509 #ifdef CONFIG_KEYS
510 	struct key *uid_keyring;	/* UID specific keyring */
511 	struct key *session_keyring;	/* UID's default session keyring */
512 #endif
513 
514 	/* Hash table maintenance information */
515 	struct list_head uidhash_list;
516 	uid_t uid;
517 };
518 
519 extern struct user_struct *find_user(uid_t);
520 
521 extern struct user_struct root_user;
522 #define INIT_USER (&root_user)
523 
524 typedef struct prio_array prio_array_t;
525 struct backing_dev_info;
526 struct reclaim_state;
527 
528 #ifdef CONFIG_SCHEDSTATS
529 struct sched_info {
530 	/* cumulative counters */
531 	unsigned long	cpu_time,	/* time spent on the cpu */
532 			run_delay,	/* time spent waiting on a runqueue */
533 			pcnt;		/* # of timeslices run on this cpu */
534 
535 	/* timestamps */
536 	unsigned long	last_arrival,	/* when we last ran on a cpu */
537 			last_queued;	/* when we were last queued to run */
538 };
539 
540 extern struct file_operations proc_schedstat_operations;
541 #endif
542 
543 enum idle_type
544 {
545 	SCHED_IDLE,
546 	NOT_IDLE,
547 	NEWLY_IDLE,
548 	MAX_IDLE_TYPES
549 };
550 
551 /*
552  * sched-domains (multiprocessor balancing) declarations:
553  */
554 #ifdef CONFIG_SMP
555 #define SCHED_LOAD_SCALE	128UL	/* increase resolution of load */
556 
557 #define SD_LOAD_BALANCE		1	/* Do load balancing on this domain. */
558 #define SD_BALANCE_NEWIDLE	2	/* Balance when about to become idle */
559 #define SD_BALANCE_EXEC		4	/* Balance on exec */
560 #define SD_BALANCE_FORK		8	/* Balance on fork, clone */
561 #define SD_WAKE_IDLE		16	/* Wake to idle CPU on task wakeup */
562 #define SD_WAKE_AFFINE		32	/* Wake task to waking CPU */
563 #define SD_WAKE_BALANCE		64	/* Perform balancing at task wakeup */
564 #define SD_SHARE_CPUPOWER	128	/* Domain members share cpu power */
565 
566 struct sched_group {
567 	struct sched_group *next;	/* Must be a circular list */
568 	cpumask_t cpumask;
569 
570 	/*
571 	 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
572 	 * single CPU. This is read only (except for setup, hotplug CPU).
573 	 */
574 	unsigned long cpu_power;
575 };
576 
577 struct sched_domain {
578 	/* These fields must be setup */
579 	struct sched_domain *parent;	/* top domain must be null terminated */
580 	struct sched_group *groups;	/* the balancing groups of the domain */
581 	cpumask_t span;			/* span of all CPUs in this domain */
582 	unsigned long min_interval;	/* Minimum balance interval ms */
583 	unsigned long max_interval;	/* Maximum balance interval ms */
584 	unsigned int busy_factor;	/* less balancing by factor if busy */
585 	unsigned int imbalance_pct;	/* No balance until over watermark */
586 	unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
587 	unsigned int cache_nice_tries;	/* Leave cache hot tasks for # tries */
588 	unsigned int per_cpu_gain;	/* CPU % gained by adding domain cpus */
589 	unsigned int busy_idx;
590 	unsigned int idle_idx;
591 	unsigned int newidle_idx;
592 	unsigned int wake_idx;
593 	unsigned int forkexec_idx;
594 	int flags;			/* See SD_* */
595 
596 	/* Runtime fields. */
597 	unsigned long last_balance;	/* init to jiffies. units in jiffies */
598 	unsigned int balance_interval;	/* initialise to 1. units in ms. */
599 	unsigned int nr_balance_failed; /* initialise to 0 */
600 
601 #ifdef CONFIG_SCHEDSTATS
602 	/* load_balance() stats */
603 	unsigned long lb_cnt[MAX_IDLE_TYPES];
604 	unsigned long lb_failed[MAX_IDLE_TYPES];
605 	unsigned long lb_balanced[MAX_IDLE_TYPES];
606 	unsigned long lb_imbalance[MAX_IDLE_TYPES];
607 	unsigned long lb_gained[MAX_IDLE_TYPES];
608 	unsigned long lb_hot_gained[MAX_IDLE_TYPES];
609 	unsigned long lb_nobusyg[MAX_IDLE_TYPES];
610 	unsigned long lb_nobusyq[MAX_IDLE_TYPES];
611 
612 	/* Active load balancing */
613 	unsigned long alb_cnt;
614 	unsigned long alb_failed;
615 	unsigned long alb_pushed;
616 
617 	/* SD_BALANCE_EXEC stats */
618 	unsigned long sbe_cnt;
619 	unsigned long sbe_balanced;
620 	unsigned long sbe_pushed;
621 
622 	/* SD_BALANCE_FORK stats */
623 	unsigned long sbf_cnt;
624 	unsigned long sbf_balanced;
625 	unsigned long sbf_pushed;
626 
627 	/* try_to_wake_up() stats */
628 	unsigned long ttwu_wake_remote;
629 	unsigned long ttwu_move_affine;
630 	unsigned long ttwu_move_balance;
631 #endif
632 };
633 
634 extern void partition_sched_domains(cpumask_t *partition1,
635 				    cpumask_t *partition2);
636 
637 /*
638  * Maximum cache size the migration-costs auto-tuning code will
639  * search from:
640  */
641 extern unsigned int max_cache_size;
642 
643 #endif	/* CONFIG_SMP */
644 
645 
646 struct io_context;			/* See blkdev.h */
647 void exit_io_context(void);
648 struct cpuset;
649 
650 #define NGROUPS_SMALL		32
651 #define NGROUPS_PER_BLOCK	((int)(PAGE_SIZE / sizeof(gid_t)))
652 struct group_info {
653 	int ngroups;
654 	atomic_t usage;
655 	gid_t small_block[NGROUPS_SMALL];
656 	int nblocks;
657 	gid_t *blocks[0];
658 };
659 
660 /*
661  * get_group_info() must be called with the owning task locked (via task_lock())
662  * when task != current.  The reason being that the vast majority of callers are
663  * looking at current->group_info, which can not be changed except by the
664  * current task.  Changing current->group_info requires the task lock, too.
665  */
666 #define get_group_info(group_info) do { \
667 	atomic_inc(&(group_info)->usage); \
668 } while (0)
669 
670 #define put_group_info(group_info) do { \
671 	if (atomic_dec_and_test(&(group_info)->usage)) \
672 		groups_free(group_info); \
673 } while (0)
674 
675 extern struct group_info *groups_alloc(int gidsetsize);
676 extern void groups_free(struct group_info *group_info);
677 extern int set_current_groups(struct group_info *group_info);
678 extern int groups_search(struct group_info *group_info, gid_t grp);
679 /* access the groups "array" with this macro */
680 #define GROUP_AT(gi, i) \
681     ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
682 
683 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
684 extern void prefetch_stack(struct task_struct*);
685 #else
686 static inline void prefetch_stack(struct task_struct *t) { }
687 #endif
688 
689 struct audit_context;		/* See audit.c */
690 struct mempolicy;
691 
692 struct task_struct {
693 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
694 	struct thread_info *thread_info;
695 	atomic_t usage;
696 	unsigned long flags;	/* per process flags, defined below */
697 	unsigned long ptrace;
698 
699 	int lock_depth;		/* BKL lock depth */
700 
701 #if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW)
702 	int oncpu;
703 #endif
704 	int prio, static_prio;
705 	struct list_head run_list;
706 	prio_array_t *array;
707 
708 	unsigned short ioprio;
709 
710 	unsigned long sleep_avg;
711 	unsigned long long timestamp, last_ran;
712 	unsigned long long sched_time; /* sched_clock time spent running */
713 	int activated;
714 
715 	unsigned long policy;
716 	cpumask_t cpus_allowed;
717 	unsigned int time_slice, first_time_slice;
718 
719 #ifdef CONFIG_SCHEDSTATS
720 	struct sched_info sched_info;
721 #endif
722 
723 	struct list_head tasks;
724 	/*
725 	 * ptrace_list/ptrace_children forms the list of my children
726 	 * that were stolen by a ptracer.
727 	 */
728 	struct list_head ptrace_children;
729 	struct list_head ptrace_list;
730 
731 	struct mm_struct *mm, *active_mm;
732 
733 /* task state */
734 	struct linux_binfmt *binfmt;
735 	long exit_state;
736 	int exit_code, exit_signal;
737 	int pdeath_signal;  /*  The signal sent when the parent dies  */
738 	/* ??? */
739 	unsigned long personality;
740 	unsigned did_exec:1;
741 	pid_t pid;
742 	pid_t tgid;
743 	/*
744 	 * pointers to (original) parent process, youngest child, younger sibling,
745 	 * older sibling, respectively.  (p->father can be replaced with
746 	 * p->parent->pid)
747 	 */
748 	struct task_struct *real_parent; /* real parent process (when being debugged) */
749 	struct task_struct *parent;	/* parent process */
750 	/*
751 	 * children/sibling forms the list of my children plus the
752 	 * tasks I'm ptracing.
753 	 */
754 	struct list_head children;	/* list of my children */
755 	struct list_head sibling;	/* linkage in my parent's children list */
756 	struct task_struct *group_leader;	/* threadgroup leader */
757 
758 	/* PID/PID hash table linkage. */
759 	struct pid pids[PIDTYPE_MAX];
760 
761 	struct completion *vfork_done;		/* for vfork() */
762 	int __user *set_child_tid;		/* CLONE_CHILD_SETTID */
763 	int __user *clear_child_tid;		/* CLONE_CHILD_CLEARTID */
764 
765 	unsigned long rt_priority;
766 	cputime_t utime, stime;
767 	unsigned long nvcsw, nivcsw; /* context switch counts */
768 	struct timespec start_time;
769 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
770 	unsigned long min_flt, maj_flt;
771 
772   	cputime_t it_prof_expires, it_virt_expires;
773 	unsigned long long it_sched_expires;
774 	struct list_head cpu_timers[3];
775 
776 /* process credentials */
777 	uid_t uid,euid,suid,fsuid;
778 	gid_t gid,egid,sgid,fsgid;
779 	struct group_info *group_info;
780 	kernel_cap_t   cap_effective, cap_inheritable, cap_permitted;
781 	unsigned keep_capabilities:1;
782 	struct user_struct *user;
783 #ifdef CONFIG_KEYS
784 	struct key *request_key_auth;	/* assumed request_key authority */
785 	struct key *thread_keyring;	/* keyring private to this thread */
786 	unsigned char jit_keyring;	/* default keyring to attach requested keys to */
787 #endif
788 	int oomkilladj; /* OOM kill score adjustment (bit shift). */
789 	char comm[TASK_COMM_LEN]; /* executable name excluding path
790 				     - access with [gs]et_task_comm (which lock
791 				       it with task_lock())
792 				     - initialized normally by flush_old_exec */
793 /* file system info */
794 	int link_count, total_link_count;
795 /* ipc stuff */
796 	struct sysv_sem sysvsem;
797 /* CPU-specific state of this task */
798 	struct thread_struct thread;
799 /* filesystem information */
800 	struct fs_struct *fs;
801 /* open file information */
802 	struct files_struct *files;
803 /* namespace */
804 	struct namespace *namespace;
805 /* signal handlers */
806 	struct signal_struct *signal;
807 	struct sighand_struct *sighand;
808 
809 	sigset_t blocked, real_blocked;
810 	sigset_t saved_sigmask;		/* To be restored with TIF_RESTORE_SIGMASK */
811 	struct sigpending pending;
812 
813 	unsigned long sas_ss_sp;
814 	size_t sas_ss_size;
815 	int (*notifier)(void *priv);
816 	void *notifier_data;
817 	sigset_t *notifier_mask;
818 
819 	void *security;
820 	struct audit_context *audit_context;
821 	seccomp_t seccomp;
822 
823 /* Thread group tracking */
824    	u32 parent_exec_id;
825    	u32 self_exec_id;
826 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
827 	spinlock_t alloc_lock;
828 /* Protection of proc_dentry: nesting proc_lock, dcache_lock, write_lock_irq(&tasklist_lock); */
829 	spinlock_t proc_lock;
830 
831 #ifdef CONFIG_DEBUG_MUTEXES
832 	/* mutex deadlock detection */
833 	struct mutex_waiter *blocked_on;
834 #endif
835 
836 /* journalling filesystem info */
837 	void *journal_info;
838 
839 /* VM state */
840 	struct reclaim_state *reclaim_state;
841 
842 	struct dentry *proc_dentry;
843 	struct backing_dev_info *backing_dev_info;
844 
845 	struct io_context *io_context;
846 
847 	unsigned long ptrace_message;
848 	siginfo_t *last_siginfo; /* For ptrace use.  */
849 /*
850  * current io wait handle: wait queue entry to use for io waits
851  * If this thread is processing aio, this points at the waitqueue
852  * inside the currently handled kiocb. It may be NULL (i.e. default
853  * to a stack based synchronous wait) if its doing sync IO.
854  */
855 	wait_queue_t *io_wait;
856 /* i/o counters(bytes read/written, #syscalls */
857 	u64 rchar, wchar, syscr, syscw;
858 #if defined(CONFIG_BSD_PROCESS_ACCT)
859 	u64 acct_rss_mem1;	/* accumulated rss usage */
860 	u64 acct_vm_mem1;	/* accumulated virtual memory usage */
861 	clock_t acct_stimexpd;	/* clock_t-converted stime since last update */
862 #endif
863 #ifdef CONFIG_NUMA
864   	struct mempolicy *mempolicy;
865 	short il_next;
866 #endif
867 #ifdef CONFIG_CPUSETS
868 	struct cpuset *cpuset;
869 	nodemask_t mems_allowed;
870 	int cpuset_mems_generation;
871 #endif
872 	atomic_t fs_excl;	/* holding fs exclusive resources */
873 	struct rcu_head rcu;
874 };
875 
876 static inline pid_t process_group(struct task_struct *tsk)
877 {
878 	return tsk->signal->pgrp;
879 }
880 
881 /**
882  * pid_alive - check that a task structure is not stale
883  * @p: Task structure to be checked.
884  *
885  * Test if a process is not yet dead (at most zombie state)
886  * If pid_alive fails, then pointers within the task structure
887  * can be stale and must not be dereferenced.
888  */
889 static inline int pid_alive(struct task_struct *p)
890 {
891 	return p->pids[PIDTYPE_PID].nr != 0;
892 }
893 
894 extern void free_task(struct task_struct *tsk);
895 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
896 
897 extern void __put_task_struct_cb(struct rcu_head *rhp);
898 
899 static inline void put_task_struct(struct task_struct *t)
900 {
901 	if (atomic_dec_and_test(&t->usage))
902 		call_rcu(&t->rcu, __put_task_struct_cb);
903 }
904 
905 /*
906  * Per process flags
907  */
908 #define PF_ALIGNWARN	0x00000001	/* Print alignment warning msgs */
909 					/* Not implemented yet, only for 486*/
910 #define PF_STARTING	0x00000002	/* being created */
911 #define PF_EXITING	0x00000004	/* getting shut down */
912 #define PF_DEAD		0x00000008	/* Dead */
913 #define PF_FORKNOEXEC	0x00000040	/* forked but didn't exec */
914 #define PF_SUPERPRIV	0x00000100	/* used super-user privileges */
915 #define PF_DUMPCORE	0x00000200	/* dumped core */
916 #define PF_SIGNALED	0x00000400	/* killed by a signal */
917 #define PF_MEMALLOC	0x00000800	/* Allocating memory */
918 #define PF_FLUSHER	0x00001000	/* responsible for disk writeback */
919 #define PF_USED_MATH	0x00002000	/* if unset the fpu must be initialized before use */
920 #define PF_FREEZE	0x00004000	/* this task is being frozen for suspend now */
921 #define PF_NOFREEZE	0x00008000	/* this thread should not be frozen */
922 #define PF_FROZEN	0x00010000	/* frozen for system suspend */
923 #define PF_FSTRANS	0x00020000	/* inside a filesystem transaction */
924 #define PF_KSWAPD	0x00040000	/* I am kswapd */
925 #define PF_SWAPOFF	0x00080000	/* I am in swapoff */
926 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
927 #define PF_SYNCWRITE	0x00200000	/* I am doing a sync write */
928 #define PF_BORROWED_MM	0x00400000	/* I am a kthread doing use_mm */
929 #define PF_RANDOMIZE	0x00800000	/* randomize virtual address space */
930 #define PF_SWAPWRITE	0x01000000	/* Allowed to write to swap */
931 
932 /*
933  * Only the _current_ task can read/write to tsk->flags, but other
934  * tasks can access tsk->flags in readonly mode for example
935  * with tsk_used_math (like during threaded core dumping).
936  * There is however an exception to this rule during ptrace
937  * or during fork: the ptracer task is allowed to write to the
938  * child->flags of its traced child (same goes for fork, the parent
939  * can write to the child->flags), because we're guaranteed the
940  * child is not running and in turn not changing child->flags
941  * at the same time the parent does it.
942  */
943 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
944 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
945 #define clear_used_math() clear_stopped_child_used_math(current)
946 #define set_used_math() set_stopped_child_used_math(current)
947 #define conditional_stopped_child_used_math(condition, child) \
948 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
949 #define conditional_used_math(condition) \
950 	conditional_stopped_child_used_math(condition, current)
951 #define copy_to_stopped_child_used_math(child) \
952 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
953 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
954 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
955 #define used_math() tsk_used_math(current)
956 
957 #ifdef CONFIG_SMP
958 extern int set_cpus_allowed(task_t *p, cpumask_t new_mask);
959 #else
960 static inline int set_cpus_allowed(task_t *p, cpumask_t new_mask)
961 {
962 	if (!cpu_isset(0, new_mask))
963 		return -EINVAL;
964 	return 0;
965 }
966 #endif
967 
968 extern unsigned long long sched_clock(void);
969 extern unsigned long long current_sched_time(const task_t *current_task);
970 
971 /* sched_exec is called by processes performing an exec */
972 #ifdef CONFIG_SMP
973 extern void sched_exec(void);
974 #else
975 #define sched_exec()   {}
976 #endif
977 
978 #ifdef CONFIG_HOTPLUG_CPU
979 extern void idle_task_exit(void);
980 #else
981 static inline void idle_task_exit(void) {}
982 #endif
983 
984 extern void sched_idle_next(void);
985 extern void set_user_nice(task_t *p, long nice);
986 extern int task_prio(const task_t *p);
987 extern int task_nice(const task_t *p);
988 extern int can_nice(const task_t *p, const int nice);
989 extern int task_curr(const task_t *p);
990 extern int idle_cpu(int cpu);
991 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
992 extern task_t *idle_task(int cpu);
993 extern task_t *curr_task(int cpu);
994 extern void set_curr_task(int cpu, task_t *p);
995 
996 void yield(void);
997 
998 /*
999  * The default (Linux) execution domain.
1000  */
1001 extern struct exec_domain	default_exec_domain;
1002 
1003 union thread_union {
1004 	struct thread_info thread_info;
1005 	unsigned long stack[THREAD_SIZE/sizeof(long)];
1006 };
1007 
1008 #ifndef __HAVE_ARCH_KSTACK_END
1009 static inline int kstack_end(void *addr)
1010 {
1011 	/* Reliable end of stack detection:
1012 	 * Some APM bios versions misalign the stack
1013 	 */
1014 	return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1015 }
1016 #endif
1017 
1018 extern union thread_union init_thread_union;
1019 extern struct task_struct init_task;
1020 
1021 extern struct   mm_struct init_mm;
1022 
1023 #define find_task_by_pid(nr)	find_task_by_pid_type(PIDTYPE_PID, nr)
1024 extern struct task_struct *find_task_by_pid_type(int type, int pid);
1025 extern void set_special_pids(pid_t session, pid_t pgrp);
1026 extern void __set_special_pids(pid_t session, pid_t pgrp);
1027 
1028 /* per-UID process charging. */
1029 extern struct user_struct * alloc_uid(uid_t);
1030 static inline struct user_struct *get_uid(struct user_struct *u)
1031 {
1032 	atomic_inc(&u->__count);
1033 	return u;
1034 }
1035 extern void free_uid(struct user_struct *);
1036 extern void switch_uid(struct user_struct *);
1037 
1038 #include <asm/current.h>
1039 
1040 extern void do_timer(struct pt_regs *);
1041 
1042 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1043 extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1044 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1045 						unsigned long clone_flags));
1046 #ifdef CONFIG_SMP
1047  extern void kick_process(struct task_struct *tsk);
1048 #else
1049  static inline void kick_process(struct task_struct *tsk) { }
1050 #endif
1051 extern void FASTCALL(sched_fork(task_t * p, int clone_flags));
1052 extern void FASTCALL(sched_exit(task_t * p));
1053 
1054 extern int in_group_p(gid_t);
1055 extern int in_egroup_p(gid_t);
1056 
1057 extern void proc_caches_init(void);
1058 extern void flush_signals(struct task_struct *);
1059 extern void flush_signal_handlers(struct task_struct *, int force_default);
1060 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1061 
1062 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1063 {
1064 	unsigned long flags;
1065 	int ret;
1066 
1067 	spin_lock_irqsave(&tsk->sighand->siglock, flags);
1068 	ret = dequeue_signal(tsk, mask, info);
1069 	spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1070 
1071 	return ret;
1072 }
1073 
1074 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1075 			      sigset_t *mask);
1076 extern void unblock_all_signals(void);
1077 extern void release_task(struct task_struct * p);
1078 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1079 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1080 extern int force_sigsegv(int, struct task_struct *);
1081 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1082 extern int __kill_pg_info(int sig, struct siginfo *info, pid_t pgrp);
1083 extern int kill_pg_info(int, struct siginfo *, pid_t);
1084 extern int kill_proc_info(int, struct siginfo *, pid_t);
1085 extern int kill_proc_info_as_uid(int, struct siginfo *, pid_t, uid_t, uid_t);
1086 extern void do_notify_parent(struct task_struct *, int);
1087 extern void force_sig(int, struct task_struct *);
1088 extern void force_sig_specific(int, struct task_struct *);
1089 extern int send_sig(int, struct task_struct *, int);
1090 extern void zap_other_threads(struct task_struct *p);
1091 extern int kill_pg(pid_t, int, int);
1092 extern int kill_sl(pid_t, int, int);
1093 extern int kill_proc(pid_t, int, int);
1094 extern struct sigqueue *sigqueue_alloc(void);
1095 extern void sigqueue_free(struct sigqueue *);
1096 extern int send_sigqueue(int, struct sigqueue *,  struct task_struct *);
1097 extern int send_group_sigqueue(int, struct sigqueue *,  struct task_struct *);
1098 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1099 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1100 
1101 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
1102 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1103 #define SEND_SIG_PRIV	((struct siginfo *) 1)
1104 #define SEND_SIG_FORCED	((struct siginfo *) 2)
1105 
1106 static inline int is_si_special(const struct siginfo *info)
1107 {
1108 	return info <= SEND_SIG_FORCED;
1109 }
1110 
1111 /* True if we are on the alternate signal stack.  */
1112 
1113 static inline int on_sig_stack(unsigned long sp)
1114 {
1115 	return (sp - current->sas_ss_sp < current->sas_ss_size);
1116 }
1117 
1118 static inline int sas_ss_flags(unsigned long sp)
1119 {
1120 	return (current->sas_ss_size == 0 ? SS_DISABLE
1121 		: on_sig_stack(sp) ? SS_ONSTACK : 0);
1122 }
1123 
1124 /*
1125  * Routines for handling mm_structs
1126  */
1127 extern struct mm_struct * mm_alloc(void);
1128 
1129 /* mmdrop drops the mm and the page tables */
1130 extern void FASTCALL(__mmdrop(struct mm_struct *));
1131 static inline void mmdrop(struct mm_struct * mm)
1132 {
1133 	if (atomic_dec_and_test(&mm->mm_count))
1134 		__mmdrop(mm);
1135 }
1136 
1137 /* mmput gets rid of the mappings and all user-space */
1138 extern void mmput(struct mm_struct *);
1139 /* Grab a reference to a task's mm, if it is not already going away */
1140 extern struct mm_struct *get_task_mm(struct task_struct *task);
1141 /* Remove the current tasks stale references to the old mm_struct */
1142 extern void mm_release(struct task_struct *, struct mm_struct *);
1143 
1144 extern int  copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1145 extern void flush_thread(void);
1146 extern void exit_thread(void);
1147 
1148 extern void exit_files(struct task_struct *);
1149 extern void exit_signal(struct task_struct *);
1150 extern void __exit_signal(struct task_struct *);
1151 extern void exit_sighand(struct task_struct *);
1152 extern void __exit_sighand(struct task_struct *);
1153 extern void exit_itimers(struct signal_struct *);
1154 
1155 extern NORET_TYPE void do_group_exit(int);
1156 
1157 extern void daemonize(const char *, ...);
1158 extern int allow_signal(int);
1159 extern int disallow_signal(int);
1160 extern task_t *child_reaper;
1161 
1162 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1163 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1164 task_t *fork_idle(int);
1165 
1166 extern void set_task_comm(struct task_struct *tsk, char *from);
1167 extern void get_task_comm(char *to, struct task_struct *tsk);
1168 
1169 #ifdef CONFIG_SMP
1170 extern void wait_task_inactive(task_t * p);
1171 #else
1172 #define wait_task_inactive(p)	do { } while (0)
1173 #endif
1174 
1175 #define remove_parent(p)	list_del_init(&(p)->sibling)
1176 #define add_parent(p, parent)	list_add_tail(&(p)->sibling,&(parent)->children)
1177 
1178 #define REMOVE_LINKS(p) do {					\
1179 	if (thread_group_leader(p))				\
1180 		list_del_init(&(p)->tasks);			\
1181 	remove_parent(p);					\
1182 	} while (0)
1183 
1184 #define SET_LINKS(p) do {					\
1185 	if (thread_group_leader(p))				\
1186 		list_add_tail(&(p)->tasks,&init_task.tasks);	\
1187 	add_parent(p, (p)->parent);				\
1188 	} while (0)
1189 
1190 #define next_task(p)	list_entry((p)->tasks.next, struct task_struct, tasks)
1191 #define prev_task(p)	list_entry((p)->tasks.prev, struct task_struct, tasks)
1192 
1193 #define for_each_process(p) \
1194 	for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1195 
1196 /*
1197  * Careful: do_each_thread/while_each_thread is a double loop so
1198  *          'break' will not work as expected - use goto instead.
1199  */
1200 #define do_each_thread(g, t) \
1201 	for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1202 
1203 #define while_each_thread(g, t) \
1204 	while ((t = next_thread(t)) != g)
1205 
1206 extern task_t * FASTCALL(next_thread(const task_t *p));
1207 
1208 #define thread_group_leader(p)	(p->pid == p->tgid)
1209 
1210 static inline int thread_group_empty(task_t *p)
1211 {
1212 	return list_empty(&p->pids[PIDTYPE_TGID].pid_list);
1213 }
1214 
1215 #define delay_group_leader(p) \
1216 		(thread_group_leader(p) && !thread_group_empty(p))
1217 
1218 extern void unhash_process(struct task_struct *p);
1219 
1220 /*
1221  * Protects ->fs, ->files, ->mm, ->ptrace, ->group_info, ->comm, keyring
1222  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
1223  * pins the final release of task.io_context.  Also protects ->cpuset.
1224  *
1225  * Nests both inside and outside of read_lock(&tasklist_lock).
1226  * It must not be nested with write_lock_irq(&tasklist_lock),
1227  * neither inside nor outside.
1228  */
1229 static inline void task_lock(struct task_struct *p)
1230 {
1231 	spin_lock(&p->alloc_lock);
1232 }
1233 
1234 static inline void task_unlock(struct task_struct *p)
1235 {
1236 	spin_unlock(&p->alloc_lock);
1237 }
1238 
1239 #ifndef __HAVE_THREAD_FUNCTIONS
1240 
1241 #define task_thread_info(task) (task)->thread_info
1242 #define task_stack_page(task) ((void*)((task)->thread_info))
1243 
1244 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1245 {
1246 	*task_thread_info(p) = *task_thread_info(org);
1247 	task_thread_info(p)->task = p;
1248 }
1249 
1250 static inline unsigned long *end_of_stack(struct task_struct *p)
1251 {
1252 	return (unsigned long *)(p->thread_info + 1);
1253 }
1254 
1255 #endif
1256 
1257 /* set thread flags in other task's structures
1258  * - see asm/thread_info.h for TIF_xxxx flags available
1259  */
1260 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1261 {
1262 	set_ti_thread_flag(task_thread_info(tsk), flag);
1263 }
1264 
1265 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1266 {
1267 	clear_ti_thread_flag(task_thread_info(tsk), flag);
1268 }
1269 
1270 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1271 {
1272 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1273 }
1274 
1275 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1276 {
1277 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1278 }
1279 
1280 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1281 {
1282 	return test_ti_thread_flag(task_thread_info(tsk), flag);
1283 }
1284 
1285 static inline void set_tsk_need_resched(struct task_struct *tsk)
1286 {
1287 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1288 }
1289 
1290 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1291 {
1292 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1293 }
1294 
1295 static inline int signal_pending(struct task_struct *p)
1296 {
1297 	return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1298 }
1299 
1300 static inline int need_resched(void)
1301 {
1302 	return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1303 }
1304 
1305 /*
1306  * cond_resched() and cond_resched_lock(): latency reduction via
1307  * explicit rescheduling in places that are safe. The return
1308  * value indicates whether a reschedule was done in fact.
1309  * cond_resched_lock() will drop the spinlock before scheduling,
1310  * cond_resched_softirq() will enable bhs before scheduling.
1311  */
1312 extern int cond_resched(void);
1313 extern int cond_resched_lock(spinlock_t * lock);
1314 extern int cond_resched_softirq(void);
1315 
1316 /*
1317  * Does a critical section need to be broken due to another
1318  * task waiting?:
1319  */
1320 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1321 # define need_lockbreak(lock) ((lock)->break_lock)
1322 #else
1323 # define need_lockbreak(lock) 0
1324 #endif
1325 
1326 /*
1327  * Does a critical section need to be broken due to another
1328  * task waiting or preemption being signalled:
1329  */
1330 static inline int lock_need_resched(spinlock_t *lock)
1331 {
1332 	if (need_lockbreak(lock) || need_resched())
1333 		return 1;
1334 	return 0;
1335 }
1336 
1337 /* Reevaluate whether the task has signals pending delivery.
1338    This is required every time the blocked sigset_t changes.
1339    callers must hold sighand->siglock.  */
1340 
1341 extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1342 extern void recalc_sigpending(void);
1343 
1344 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1345 
1346 /*
1347  * Wrappers for p->thread_info->cpu access. No-op on UP.
1348  */
1349 #ifdef CONFIG_SMP
1350 
1351 static inline unsigned int task_cpu(const struct task_struct *p)
1352 {
1353 	return task_thread_info(p)->cpu;
1354 }
1355 
1356 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1357 {
1358 	task_thread_info(p)->cpu = cpu;
1359 }
1360 
1361 #else
1362 
1363 static inline unsigned int task_cpu(const struct task_struct *p)
1364 {
1365 	return 0;
1366 }
1367 
1368 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1369 {
1370 }
1371 
1372 #endif /* CONFIG_SMP */
1373 
1374 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1375 extern void arch_pick_mmap_layout(struct mm_struct *mm);
1376 #else
1377 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1378 {
1379 	mm->mmap_base = TASK_UNMAPPED_BASE;
1380 	mm->get_unmapped_area = arch_get_unmapped_area;
1381 	mm->unmap_area = arch_unmap_area;
1382 }
1383 #endif
1384 
1385 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1386 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1387 
1388 extern void normalize_rt_tasks(void);
1389 
1390 #ifdef CONFIG_PM
1391 /*
1392  * Check if a process has been frozen
1393  */
1394 static inline int frozen(struct task_struct *p)
1395 {
1396 	return p->flags & PF_FROZEN;
1397 }
1398 
1399 /*
1400  * Check if there is a request to freeze a process
1401  */
1402 static inline int freezing(struct task_struct *p)
1403 {
1404 	return p->flags & PF_FREEZE;
1405 }
1406 
1407 /*
1408  * Request that a process be frozen
1409  * FIXME: SMP problem. We may not modify other process' flags!
1410  */
1411 static inline void freeze(struct task_struct *p)
1412 {
1413 	p->flags |= PF_FREEZE;
1414 }
1415 
1416 /*
1417  * Wake up a frozen process
1418  */
1419 static inline int thaw_process(struct task_struct *p)
1420 {
1421 	if (frozen(p)) {
1422 		p->flags &= ~PF_FROZEN;
1423 		wake_up_process(p);
1424 		return 1;
1425 	}
1426 	return 0;
1427 }
1428 
1429 /*
1430  * freezing is complete, mark process as frozen
1431  */
1432 static inline void frozen_process(struct task_struct *p)
1433 {
1434 	p->flags = (p->flags & ~PF_FREEZE) | PF_FROZEN;
1435 }
1436 
1437 extern void refrigerator(void);
1438 extern int freeze_processes(void);
1439 extern void thaw_processes(void);
1440 
1441 static inline int try_to_freeze(void)
1442 {
1443 	if (freezing(current)) {
1444 		refrigerator();
1445 		return 1;
1446 	} else
1447 		return 0;
1448 }
1449 #else
1450 static inline int frozen(struct task_struct *p) { return 0; }
1451 static inline int freezing(struct task_struct *p) { return 0; }
1452 static inline void freeze(struct task_struct *p) { BUG(); }
1453 static inline int thaw_process(struct task_struct *p) { return 1; }
1454 static inline void frozen_process(struct task_struct *p) { BUG(); }
1455 
1456 static inline void refrigerator(void) {}
1457 static inline int freeze_processes(void) { BUG(); return 0; }
1458 static inline void thaw_processes(void) {}
1459 
1460 static inline int try_to_freeze(void) { return 0; }
1461 
1462 #endif /* CONFIG_PM */
1463 #endif /* __KERNEL__ */
1464 
1465 #endif
1466