xref: /linux/include/linux/sched.h (revision c537b994505099b7197e7d3125b942ecbcc51eb6)
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3 
4 #include <linux/auxvec.h>	/* For AT_VECTOR_SIZE */
5 
6 /*
7  * cloning flags:
8  */
9 #define CSIGNAL		0x000000ff	/* signal mask to be sent at exit */
10 #define CLONE_VM	0x00000100	/* set if VM shared between processes */
11 #define CLONE_FS	0x00000200	/* set if fs info shared between processes */
12 #define CLONE_FILES	0x00000400	/* set if open files shared between processes */
13 #define CLONE_SIGHAND	0x00000800	/* set if signal handlers and blocked signals shared */
14 #define CLONE_PTRACE	0x00002000	/* set if we want to let tracing continue on the child too */
15 #define CLONE_VFORK	0x00004000	/* set if the parent wants the child to wake it up on mm_release */
16 #define CLONE_PARENT	0x00008000	/* set if we want to have the same parent as the cloner */
17 #define CLONE_THREAD	0x00010000	/* Same thread group? */
18 #define CLONE_NEWNS	0x00020000	/* New namespace group? */
19 #define CLONE_SYSVSEM	0x00040000	/* share system V SEM_UNDO semantics */
20 #define CLONE_SETTLS	0x00080000	/* create a new TLS for the child */
21 #define CLONE_PARENT_SETTID	0x00100000	/* set the TID in the parent */
22 #define CLONE_CHILD_CLEARTID	0x00200000	/* clear the TID in the child */
23 #define CLONE_DETACHED		0x00400000	/* Unused, ignored */
24 #define CLONE_UNTRACED		0x00800000	/* set if the tracing process can't force CLONE_PTRACE on this clone */
25 #define CLONE_CHILD_SETTID	0x01000000	/* set the TID in the child */
26 #define CLONE_STOPPED		0x02000000	/* Start in stopped state */
27 #define CLONE_NEWUTS		0x04000000	/* New utsname group? */
28 #define CLONE_NEWIPC		0x08000000	/* New ipcs */
29 
30 /*
31  * Scheduling policies
32  */
33 #define SCHED_NORMAL		0
34 #define SCHED_FIFO		1
35 #define SCHED_RR		2
36 #define SCHED_BATCH		3
37 
38 #ifdef __KERNEL__
39 
40 struct sched_param {
41 	int sched_priority;
42 };
43 
44 #include <asm/param.h>	/* for HZ */
45 
46 #include <linux/capability.h>
47 #include <linux/threads.h>
48 #include <linux/kernel.h>
49 #include <linux/types.h>
50 #include <linux/timex.h>
51 #include <linux/jiffies.h>
52 #include <linux/rbtree.h>
53 #include <linux/thread_info.h>
54 #include <linux/cpumask.h>
55 #include <linux/errno.h>
56 #include <linux/nodemask.h>
57 
58 #include <asm/system.h>
59 #include <asm/semaphore.h>
60 #include <asm/page.h>
61 #include <asm/ptrace.h>
62 #include <asm/mmu.h>
63 #include <asm/cputime.h>
64 
65 #include <linux/smp.h>
66 #include <linux/sem.h>
67 #include <linux/signal.h>
68 #include <linux/securebits.h>
69 #include <linux/fs_struct.h>
70 #include <linux/compiler.h>
71 #include <linux/completion.h>
72 #include <linux/pid.h>
73 #include <linux/percpu.h>
74 #include <linux/topology.h>
75 #include <linux/seccomp.h>
76 #include <linux/rcupdate.h>
77 #include <linux/futex.h>
78 #include <linux/rtmutex.h>
79 
80 #include <linux/time.h>
81 #include <linux/param.h>
82 #include <linux/resource.h>
83 #include <linux/timer.h>
84 #include <linux/hrtimer.h>
85 #include <linux/task_io_accounting.h>
86 
87 #include <asm/processor.h>
88 
89 struct exec_domain;
90 struct futex_pi_state;
91 
92 /*
93  * List of flags we want to share for kernel threads,
94  * if only because they are not used by them anyway.
95  */
96 #define CLONE_KERNEL	(CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
97 
98 /*
99  * These are the constant used to fake the fixed-point load-average
100  * counting. Some notes:
101  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
102  *    a load-average precision of 10 bits integer + 11 bits fractional
103  *  - if you want to count load-averages more often, you need more
104  *    precision, or rounding will get you. With 2-second counting freq,
105  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
106  *    11 bit fractions.
107  */
108 extern unsigned long avenrun[];		/* Load averages */
109 
110 #define FSHIFT		11		/* nr of bits of precision */
111 #define FIXED_1		(1<<FSHIFT)	/* 1.0 as fixed-point */
112 #define LOAD_FREQ	(5*HZ)		/* 5 sec intervals */
113 #define EXP_1		1884		/* 1/exp(5sec/1min) as fixed-point */
114 #define EXP_5		2014		/* 1/exp(5sec/5min) */
115 #define EXP_15		2037		/* 1/exp(5sec/15min) */
116 
117 #define CALC_LOAD(load,exp,n) \
118 	load *= exp; \
119 	load += n*(FIXED_1-exp); \
120 	load >>= FSHIFT;
121 
122 extern unsigned long total_forks;
123 extern int nr_threads;
124 DECLARE_PER_CPU(unsigned long, process_counts);
125 extern int nr_processes(void);
126 extern unsigned long nr_running(void);
127 extern unsigned long nr_uninterruptible(void);
128 extern unsigned long nr_active(void);
129 extern unsigned long nr_iowait(void);
130 extern unsigned long weighted_cpuload(const int cpu);
131 
132 
133 /*
134  * Task state bitmask. NOTE! These bits are also
135  * encoded in fs/proc/array.c: get_task_state().
136  *
137  * We have two separate sets of flags: task->state
138  * is about runnability, while task->exit_state are
139  * about the task exiting. Confusing, but this way
140  * modifying one set can't modify the other one by
141  * mistake.
142  */
143 #define TASK_RUNNING		0
144 #define TASK_INTERRUPTIBLE	1
145 #define TASK_UNINTERRUPTIBLE	2
146 #define TASK_STOPPED		4
147 #define TASK_TRACED		8
148 /* in tsk->exit_state */
149 #define EXIT_ZOMBIE		16
150 #define EXIT_DEAD		32
151 /* in tsk->state again */
152 #define TASK_NONINTERACTIVE	64
153 #define TASK_DEAD		128
154 
155 #define __set_task_state(tsk, state_value)		\
156 	do { (tsk)->state = (state_value); } while (0)
157 #define set_task_state(tsk, state_value)		\
158 	set_mb((tsk)->state, (state_value))
159 
160 /*
161  * set_current_state() includes a barrier so that the write of current->state
162  * is correctly serialised wrt the caller's subsequent test of whether to
163  * actually sleep:
164  *
165  *	set_current_state(TASK_UNINTERRUPTIBLE);
166  *	if (do_i_need_to_sleep())
167  *		schedule();
168  *
169  * If the caller does not need such serialisation then use __set_current_state()
170  */
171 #define __set_current_state(state_value)			\
172 	do { current->state = (state_value); } while (0)
173 #define set_current_state(state_value)		\
174 	set_mb(current->state, (state_value))
175 
176 /* Task command name length */
177 #define TASK_COMM_LEN 16
178 
179 #include <linux/spinlock.h>
180 
181 /*
182  * This serializes "schedule()" and also protects
183  * the run-queue from deletions/modifications (but
184  * _adding_ to the beginning of the run-queue has
185  * a separate lock).
186  */
187 extern rwlock_t tasklist_lock;
188 extern spinlock_t mmlist_lock;
189 
190 struct task_struct;
191 
192 extern void sched_init(void);
193 extern void sched_init_smp(void);
194 extern void init_idle(struct task_struct *idle, int cpu);
195 
196 extern cpumask_t nohz_cpu_mask;
197 
198 /*
199  * Only dump TASK_* tasks. (-1 for all tasks)
200  */
201 extern void show_state_filter(unsigned long state_filter);
202 
203 static inline void show_state(void)
204 {
205 	show_state_filter(-1);
206 }
207 
208 extern void show_regs(struct pt_regs *);
209 
210 /*
211  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
212  * task), SP is the stack pointer of the first frame that should be shown in the back
213  * trace (or NULL if the entire call-chain of the task should be shown).
214  */
215 extern void show_stack(struct task_struct *task, unsigned long *sp);
216 
217 void io_schedule(void);
218 long io_schedule_timeout(long timeout);
219 
220 extern void cpu_init (void);
221 extern void trap_init(void);
222 extern void update_process_times(int user);
223 extern void scheduler_tick(void);
224 
225 #ifdef CONFIG_DETECT_SOFTLOCKUP
226 extern void softlockup_tick(void);
227 extern void spawn_softlockup_task(void);
228 extern void touch_softlockup_watchdog(void);
229 #else
230 static inline void softlockup_tick(void)
231 {
232 }
233 static inline void spawn_softlockup_task(void)
234 {
235 }
236 static inline void touch_softlockup_watchdog(void)
237 {
238 }
239 #endif
240 
241 
242 /* Attach to any functions which should be ignored in wchan output. */
243 #define __sched		__attribute__((__section__(".sched.text")))
244 /* Is this address in the __sched functions? */
245 extern int in_sched_functions(unsigned long addr);
246 
247 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
248 extern signed long FASTCALL(schedule_timeout(signed long timeout));
249 extern signed long schedule_timeout_interruptible(signed long timeout);
250 extern signed long schedule_timeout_uninterruptible(signed long timeout);
251 asmlinkage void schedule(void);
252 
253 struct nsproxy;
254 
255 /* Maximum number of active map areas.. This is a random (large) number */
256 #define DEFAULT_MAX_MAP_COUNT	65536
257 
258 extern int sysctl_max_map_count;
259 
260 #include <linux/aio.h>
261 
262 extern unsigned long
263 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
264 		       unsigned long, unsigned long);
265 extern unsigned long
266 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
267 			  unsigned long len, unsigned long pgoff,
268 			  unsigned long flags);
269 extern void arch_unmap_area(struct mm_struct *, unsigned long);
270 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
271 
272 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
273 /*
274  * The mm counters are not protected by its page_table_lock,
275  * so must be incremented atomically.
276  */
277 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
278 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
279 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
280 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
281 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
282 typedef atomic_long_t mm_counter_t;
283 
284 #else  /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
285 /*
286  * The mm counters are protected by its page_table_lock,
287  * so can be incremented directly.
288  */
289 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
290 #define get_mm_counter(mm, member) ((mm)->_##member)
291 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
292 #define inc_mm_counter(mm, member) (mm)->_##member++
293 #define dec_mm_counter(mm, member) (mm)->_##member--
294 typedef unsigned long mm_counter_t;
295 
296 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
297 
298 #define get_mm_rss(mm)					\
299 	(get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
300 #define update_hiwater_rss(mm)	do {			\
301 	unsigned long _rss = get_mm_rss(mm);		\
302 	if ((mm)->hiwater_rss < _rss)			\
303 		(mm)->hiwater_rss = _rss;		\
304 } while (0)
305 #define update_hiwater_vm(mm)	do {			\
306 	if ((mm)->hiwater_vm < (mm)->total_vm)		\
307 		(mm)->hiwater_vm = (mm)->total_vm;	\
308 } while (0)
309 
310 struct mm_struct {
311 	struct vm_area_struct * mmap;		/* list of VMAs */
312 	struct rb_root mm_rb;
313 	struct vm_area_struct * mmap_cache;	/* last find_vma result */
314 	unsigned long (*get_unmapped_area) (struct file *filp,
315 				unsigned long addr, unsigned long len,
316 				unsigned long pgoff, unsigned long flags);
317 	void (*unmap_area) (struct mm_struct *mm, unsigned long addr);
318 	unsigned long mmap_base;		/* base of mmap area */
319 	unsigned long task_size;		/* size of task vm space */
320 	unsigned long cached_hole_size;         /* if non-zero, the largest hole below free_area_cache */
321 	unsigned long free_area_cache;		/* first hole of size cached_hole_size or larger */
322 	pgd_t * pgd;
323 	atomic_t mm_users;			/* How many users with user space? */
324 	atomic_t mm_count;			/* How many references to "struct mm_struct" (users count as 1) */
325 	int map_count;				/* number of VMAs */
326 	struct rw_semaphore mmap_sem;
327 	spinlock_t page_table_lock;		/* Protects page tables and some counters */
328 
329 	struct list_head mmlist;		/* List of maybe swapped mm's.  These are globally strung
330 						 * together off init_mm.mmlist, and are protected
331 						 * by mmlist_lock
332 						 */
333 
334 	/* Special counters, in some configurations protected by the
335 	 * page_table_lock, in other configurations by being atomic.
336 	 */
337 	mm_counter_t _file_rss;
338 	mm_counter_t _anon_rss;
339 
340 	unsigned long hiwater_rss;	/* High-watermark of RSS usage */
341 	unsigned long hiwater_vm;	/* High-water virtual memory usage */
342 
343 	unsigned long total_vm, locked_vm, shared_vm, exec_vm;
344 	unsigned long stack_vm, reserved_vm, def_flags, nr_ptes;
345 	unsigned long start_code, end_code, start_data, end_data;
346 	unsigned long start_brk, brk, start_stack;
347 	unsigned long arg_start, arg_end, env_start, env_end;
348 
349 	unsigned long saved_auxv[AT_VECTOR_SIZE]; /* for /proc/PID/auxv */
350 
351 	cpumask_t cpu_vm_mask;
352 
353 	/* Architecture-specific MM context */
354 	mm_context_t context;
355 
356 	/* Swap token stuff */
357 	/*
358 	 * Last value of global fault stamp as seen by this process.
359 	 * In other words, this value gives an indication of how long
360 	 * it has been since this task got the token.
361 	 * Look at mm/thrash.c
362 	 */
363 	unsigned int faultstamp;
364 	unsigned int token_priority;
365 	unsigned int last_interval;
366 
367 	unsigned char dumpable:2;
368 
369 	/* coredumping support */
370 	int core_waiters;
371 	struct completion *core_startup_done, core_done;
372 
373 	/* aio bits */
374 	rwlock_t		ioctx_list_lock;
375 	struct kioctx		*ioctx_list;
376 };
377 
378 struct sighand_struct {
379 	atomic_t		count;
380 	struct k_sigaction	action[_NSIG];
381 	spinlock_t		siglock;
382 };
383 
384 struct pacct_struct {
385 	int			ac_flag;
386 	long			ac_exitcode;
387 	unsigned long		ac_mem;
388 	cputime_t		ac_utime, ac_stime;
389 	unsigned long		ac_minflt, ac_majflt;
390 };
391 
392 /*
393  * NOTE! "signal_struct" does not have it's own
394  * locking, because a shared signal_struct always
395  * implies a shared sighand_struct, so locking
396  * sighand_struct is always a proper superset of
397  * the locking of signal_struct.
398  */
399 struct signal_struct {
400 	atomic_t		count;
401 	atomic_t		live;
402 
403 	wait_queue_head_t	wait_chldexit;	/* for wait4() */
404 
405 	/* current thread group signal load-balancing target: */
406 	struct task_struct	*curr_target;
407 
408 	/* shared signal handling: */
409 	struct sigpending	shared_pending;
410 
411 	/* thread group exit support */
412 	int			group_exit_code;
413 	/* overloaded:
414 	 * - notify group_exit_task when ->count is equal to notify_count
415 	 * - everyone except group_exit_task is stopped during signal delivery
416 	 *   of fatal signals, group_exit_task processes the signal.
417 	 */
418 	struct task_struct	*group_exit_task;
419 	int			notify_count;
420 
421 	/* thread group stop support, overloads group_exit_code too */
422 	int			group_stop_count;
423 	unsigned int		flags; /* see SIGNAL_* flags below */
424 
425 	/* POSIX.1b Interval Timers */
426 	struct list_head posix_timers;
427 
428 	/* ITIMER_REAL timer for the process */
429 	struct hrtimer real_timer;
430 	struct task_struct *tsk;
431 	ktime_t it_real_incr;
432 
433 	/* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
434 	cputime_t it_prof_expires, it_virt_expires;
435 	cputime_t it_prof_incr, it_virt_incr;
436 
437 	/* job control IDs */
438 	pid_t pgrp;
439 	struct pid *tty_old_pgrp;
440 
441 	union {
442 		pid_t session __deprecated;
443 		pid_t __session;
444 	};
445 
446 	/* boolean value for session group leader */
447 	int leader;
448 
449 	struct tty_struct *tty; /* NULL if no tty */
450 
451 	/*
452 	 * Cumulative resource counters for dead threads in the group,
453 	 * and for reaped dead child processes forked by this group.
454 	 * Live threads maintain their own counters and add to these
455 	 * in __exit_signal, except for the group leader.
456 	 */
457 	cputime_t utime, stime, cutime, cstime;
458 	unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
459 	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
460 
461 	/*
462 	 * Cumulative ns of scheduled CPU time for dead threads in the
463 	 * group, not including a zombie group leader.  (This only differs
464 	 * from jiffies_to_ns(utime + stime) if sched_clock uses something
465 	 * other than jiffies.)
466 	 */
467 	unsigned long long sched_time;
468 
469 	/*
470 	 * We don't bother to synchronize most readers of this at all,
471 	 * because there is no reader checking a limit that actually needs
472 	 * to get both rlim_cur and rlim_max atomically, and either one
473 	 * alone is a single word that can safely be read normally.
474 	 * getrlimit/setrlimit use task_lock(current->group_leader) to
475 	 * protect this instead of the siglock, because they really
476 	 * have no need to disable irqs.
477 	 */
478 	struct rlimit rlim[RLIM_NLIMITS];
479 
480 	struct list_head cpu_timers[3];
481 
482 	/* keep the process-shared keyrings here so that they do the right
483 	 * thing in threads created with CLONE_THREAD */
484 #ifdef CONFIG_KEYS
485 	struct key *session_keyring;	/* keyring inherited over fork */
486 	struct key *process_keyring;	/* keyring private to this process */
487 #endif
488 #ifdef CONFIG_BSD_PROCESS_ACCT
489 	struct pacct_struct pacct;	/* per-process accounting information */
490 #endif
491 #ifdef CONFIG_TASKSTATS
492 	struct taskstats *stats;
493 #endif
494 };
495 
496 /* Context switch must be unlocked if interrupts are to be enabled */
497 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
498 # define __ARCH_WANT_UNLOCKED_CTXSW
499 #endif
500 
501 /*
502  * Bits in flags field of signal_struct.
503  */
504 #define SIGNAL_STOP_STOPPED	0x00000001 /* job control stop in effect */
505 #define SIGNAL_STOP_DEQUEUED	0x00000002 /* stop signal dequeued */
506 #define SIGNAL_STOP_CONTINUED	0x00000004 /* SIGCONT since WCONTINUED reap */
507 #define SIGNAL_GROUP_EXIT	0x00000008 /* group exit in progress */
508 
509 
510 /*
511  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
512  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
513  * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
514  * values are inverted: lower p->prio value means higher priority.
515  *
516  * The MAX_USER_RT_PRIO value allows the actual maximum
517  * RT priority to be separate from the value exported to
518  * user-space.  This allows kernel threads to set their
519  * priority to a value higher than any user task. Note:
520  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
521  */
522 
523 #define MAX_USER_RT_PRIO	100
524 #define MAX_RT_PRIO		MAX_USER_RT_PRIO
525 
526 #define MAX_PRIO		(MAX_RT_PRIO + 40)
527 
528 #define rt_prio(prio)		unlikely((prio) < MAX_RT_PRIO)
529 #define rt_task(p)		rt_prio((p)->prio)
530 #define batch_task(p)		(unlikely((p)->policy == SCHED_BATCH))
531 #define is_rt_policy(p)		((p) != SCHED_NORMAL && (p) != SCHED_BATCH)
532 #define has_rt_policy(p)	unlikely(is_rt_policy((p)->policy))
533 
534 /*
535  * Some day this will be a full-fledged user tracking system..
536  */
537 struct user_struct {
538 	atomic_t __count;	/* reference count */
539 	atomic_t processes;	/* How many processes does this user have? */
540 	atomic_t files;		/* How many open files does this user have? */
541 	atomic_t sigpending;	/* How many pending signals does this user have? */
542 #ifdef CONFIG_INOTIFY_USER
543 	atomic_t inotify_watches; /* How many inotify watches does this user have? */
544 	atomic_t inotify_devs;	/* How many inotify devs does this user have opened? */
545 #endif
546 	/* protected by mq_lock	*/
547 	unsigned long mq_bytes;	/* How many bytes can be allocated to mqueue? */
548 	unsigned long locked_shm; /* How many pages of mlocked shm ? */
549 
550 #ifdef CONFIG_KEYS
551 	struct key *uid_keyring;	/* UID specific keyring */
552 	struct key *session_keyring;	/* UID's default session keyring */
553 #endif
554 
555 	/* Hash table maintenance information */
556 	struct list_head uidhash_list;
557 	uid_t uid;
558 };
559 
560 extern struct user_struct *find_user(uid_t);
561 
562 extern struct user_struct root_user;
563 #define INIT_USER (&root_user)
564 
565 struct backing_dev_info;
566 struct reclaim_state;
567 
568 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
569 struct sched_info {
570 	/* cumulative counters */
571 	unsigned long	cpu_time,	/* time spent on the cpu */
572 			run_delay,	/* time spent waiting on a runqueue */
573 			pcnt;		/* # of timeslices run on this cpu */
574 
575 	/* timestamps */
576 	unsigned long	last_arrival,	/* when we last ran on a cpu */
577 			last_queued;	/* when we were last queued to run */
578 };
579 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
580 
581 #ifdef CONFIG_SCHEDSTATS
582 extern const struct file_operations proc_schedstat_operations;
583 #endif /* CONFIG_SCHEDSTATS */
584 
585 #ifdef CONFIG_TASK_DELAY_ACCT
586 struct task_delay_info {
587 	spinlock_t	lock;
588 	unsigned int	flags;	/* Private per-task flags */
589 
590 	/* For each stat XXX, add following, aligned appropriately
591 	 *
592 	 * struct timespec XXX_start, XXX_end;
593 	 * u64 XXX_delay;
594 	 * u32 XXX_count;
595 	 *
596 	 * Atomicity of updates to XXX_delay, XXX_count protected by
597 	 * single lock above (split into XXX_lock if contention is an issue).
598 	 */
599 
600 	/*
601 	 * XXX_count is incremented on every XXX operation, the delay
602 	 * associated with the operation is added to XXX_delay.
603 	 * XXX_delay contains the accumulated delay time in nanoseconds.
604 	 */
605 	struct timespec blkio_start, blkio_end;	/* Shared by blkio, swapin */
606 	u64 blkio_delay;	/* wait for sync block io completion */
607 	u64 swapin_delay;	/* wait for swapin block io completion */
608 	u32 blkio_count;	/* total count of the number of sync block */
609 				/* io operations performed */
610 	u32 swapin_count;	/* total count of the number of swapin block */
611 				/* io operations performed */
612 };
613 #endif	/* CONFIG_TASK_DELAY_ACCT */
614 
615 static inline int sched_info_on(void)
616 {
617 #ifdef CONFIG_SCHEDSTATS
618 	return 1;
619 #elif defined(CONFIG_TASK_DELAY_ACCT)
620 	extern int delayacct_on;
621 	return delayacct_on;
622 #else
623 	return 0;
624 #endif
625 }
626 
627 enum idle_type
628 {
629 	SCHED_IDLE,
630 	NOT_IDLE,
631 	NEWLY_IDLE,
632 	MAX_IDLE_TYPES
633 };
634 
635 /*
636  * sched-domains (multiprocessor balancing) declarations:
637  */
638 #define SCHED_LOAD_SCALE	128UL	/* increase resolution of load */
639 
640 #ifdef CONFIG_SMP
641 #define SD_LOAD_BALANCE		1	/* Do load balancing on this domain. */
642 #define SD_BALANCE_NEWIDLE	2	/* Balance when about to become idle */
643 #define SD_BALANCE_EXEC		4	/* Balance on exec */
644 #define SD_BALANCE_FORK		8	/* Balance on fork, clone */
645 #define SD_WAKE_IDLE		16	/* Wake to idle CPU on task wakeup */
646 #define SD_WAKE_AFFINE		32	/* Wake task to waking CPU */
647 #define SD_WAKE_BALANCE		64	/* Perform balancing at task wakeup */
648 #define SD_SHARE_CPUPOWER	128	/* Domain members share cpu power */
649 #define SD_POWERSAVINGS_BALANCE	256	/* Balance for power savings */
650 #define SD_SHARE_PKG_RESOURCES	512	/* Domain members share cpu pkg resources */
651 #define SD_SERIALIZE		1024	/* Only a single load balancing instance */
652 
653 #define BALANCE_FOR_MC_POWER	\
654 	(sched_smt_power_savings ? SD_POWERSAVINGS_BALANCE : 0)
655 
656 #define BALANCE_FOR_PKG_POWER	\
657 	((sched_mc_power_savings || sched_smt_power_savings) ?	\
658 	 SD_POWERSAVINGS_BALANCE : 0)
659 
660 #define test_sd_parent(sd, flag)	((sd->parent &&		\
661 					 (sd->parent->flags & flag)) ? 1 : 0)
662 
663 
664 struct sched_group {
665 	struct sched_group *next;	/* Must be a circular list */
666 	cpumask_t cpumask;
667 
668 	/*
669 	 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
670 	 * single CPU. This is read only (except for setup, hotplug CPU).
671 	 */
672 	unsigned long cpu_power;
673 };
674 
675 struct sched_domain {
676 	/* These fields must be setup */
677 	struct sched_domain *parent;	/* top domain must be null terminated */
678 	struct sched_domain *child;	/* bottom domain must be null terminated */
679 	struct sched_group *groups;	/* the balancing groups of the domain */
680 	cpumask_t span;			/* span of all CPUs in this domain */
681 	unsigned long min_interval;	/* Minimum balance interval ms */
682 	unsigned long max_interval;	/* Maximum balance interval ms */
683 	unsigned int busy_factor;	/* less balancing by factor if busy */
684 	unsigned int imbalance_pct;	/* No balance until over watermark */
685 	unsigned long long cache_hot_time; /* Task considered cache hot (ns) */
686 	unsigned int cache_nice_tries;	/* Leave cache hot tasks for # tries */
687 	unsigned int busy_idx;
688 	unsigned int idle_idx;
689 	unsigned int newidle_idx;
690 	unsigned int wake_idx;
691 	unsigned int forkexec_idx;
692 	int flags;			/* See SD_* */
693 
694 	/* Runtime fields. */
695 	unsigned long last_balance;	/* init to jiffies. units in jiffies */
696 	unsigned int balance_interval;	/* initialise to 1. units in ms. */
697 	unsigned int nr_balance_failed; /* initialise to 0 */
698 
699 #ifdef CONFIG_SCHEDSTATS
700 	/* load_balance() stats */
701 	unsigned long lb_cnt[MAX_IDLE_TYPES];
702 	unsigned long lb_failed[MAX_IDLE_TYPES];
703 	unsigned long lb_balanced[MAX_IDLE_TYPES];
704 	unsigned long lb_imbalance[MAX_IDLE_TYPES];
705 	unsigned long lb_gained[MAX_IDLE_TYPES];
706 	unsigned long lb_hot_gained[MAX_IDLE_TYPES];
707 	unsigned long lb_nobusyg[MAX_IDLE_TYPES];
708 	unsigned long lb_nobusyq[MAX_IDLE_TYPES];
709 
710 	/* Active load balancing */
711 	unsigned long alb_cnt;
712 	unsigned long alb_failed;
713 	unsigned long alb_pushed;
714 
715 	/* SD_BALANCE_EXEC stats */
716 	unsigned long sbe_cnt;
717 	unsigned long sbe_balanced;
718 	unsigned long sbe_pushed;
719 
720 	/* SD_BALANCE_FORK stats */
721 	unsigned long sbf_cnt;
722 	unsigned long sbf_balanced;
723 	unsigned long sbf_pushed;
724 
725 	/* try_to_wake_up() stats */
726 	unsigned long ttwu_wake_remote;
727 	unsigned long ttwu_move_affine;
728 	unsigned long ttwu_move_balance;
729 #endif
730 };
731 
732 extern int partition_sched_domains(cpumask_t *partition1,
733 				    cpumask_t *partition2);
734 
735 /*
736  * Maximum cache size the migration-costs auto-tuning code will
737  * search from:
738  */
739 extern unsigned int max_cache_size;
740 
741 #endif	/* CONFIG_SMP */
742 
743 
744 struct io_context;			/* See blkdev.h */
745 struct cpuset;
746 
747 #define NGROUPS_SMALL		32
748 #define NGROUPS_PER_BLOCK	((int)(PAGE_SIZE / sizeof(gid_t)))
749 struct group_info {
750 	int ngroups;
751 	atomic_t usage;
752 	gid_t small_block[NGROUPS_SMALL];
753 	int nblocks;
754 	gid_t *blocks[0];
755 };
756 
757 /*
758  * get_group_info() must be called with the owning task locked (via task_lock())
759  * when task != current.  The reason being that the vast majority of callers are
760  * looking at current->group_info, which can not be changed except by the
761  * current task.  Changing current->group_info requires the task lock, too.
762  */
763 #define get_group_info(group_info) do { \
764 	atomic_inc(&(group_info)->usage); \
765 } while (0)
766 
767 #define put_group_info(group_info) do { \
768 	if (atomic_dec_and_test(&(group_info)->usage)) \
769 		groups_free(group_info); \
770 } while (0)
771 
772 extern struct group_info *groups_alloc(int gidsetsize);
773 extern void groups_free(struct group_info *group_info);
774 extern int set_current_groups(struct group_info *group_info);
775 extern int groups_search(struct group_info *group_info, gid_t grp);
776 /* access the groups "array" with this macro */
777 #define GROUP_AT(gi, i) \
778     ((gi)->blocks[(i)/NGROUPS_PER_BLOCK][(i)%NGROUPS_PER_BLOCK])
779 
780 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
781 extern void prefetch_stack(struct task_struct *t);
782 #else
783 static inline void prefetch_stack(struct task_struct *t) { }
784 #endif
785 
786 struct audit_context;		/* See audit.c */
787 struct mempolicy;
788 struct pipe_inode_info;
789 struct uts_namespace;
790 
791 enum sleep_type {
792 	SLEEP_NORMAL,
793 	SLEEP_NONINTERACTIVE,
794 	SLEEP_INTERACTIVE,
795 	SLEEP_INTERRUPTED,
796 };
797 
798 struct prio_array;
799 
800 struct task_struct {
801 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
802 	struct thread_info *thread_info;
803 	atomic_t usage;
804 	unsigned long flags;	/* per process flags, defined below */
805 	unsigned long ptrace;
806 
807 	int lock_depth;		/* BKL lock depth */
808 
809 #ifdef CONFIG_SMP
810 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
811 	int oncpu;
812 #endif
813 #endif
814 	int load_weight;	/* for niceness load balancing purposes */
815 	int prio, static_prio, normal_prio;
816 	struct list_head run_list;
817 	struct prio_array *array;
818 
819 	unsigned short ioprio;
820 #ifdef CONFIG_BLK_DEV_IO_TRACE
821 	unsigned int btrace_seq;
822 #endif
823 	unsigned long sleep_avg;
824 	unsigned long long timestamp, last_ran;
825 	unsigned long long sched_time; /* sched_clock time spent running */
826 	enum sleep_type sleep_type;
827 
828 	unsigned long policy;
829 	cpumask_t cpus_allowed;
830 	unsigned int time_slice, first_time_slice;
831 
832 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
833 	struct sched_info sched_info;
834 #endif
835 
836 	struct list_head tasks;
837 	/*
838 	 * ptrace_list/ptrace_children forms the list of my children
839 	 * that were stolen by a ptracer.
840 	 */
841 	struct list_head ptrace_children;
842 	struct list_head ptrace_list;
843 
844 	struct mm_struct *mm, *active_mm;
845 
846 /* task state */
847 	struct linux_binfmt *binfmt;
848 	long exit_state;
849 	int exit_code, exit_signal;
850 	int pdeath_signal;  /*  The signal sent when the parent dies  */
851 	/* ??? */
852 	unsigned long personality;
853 	unsigned did_exec:1;
854 	pid_t pid;
855 	pid_t tgid;
856 
857 #ifdef CONFIG_CC_STACKPROTECTOR
858 	/* Canary value for the -fstack-protector gcc feature */
859 	unsigned long stack_canary;
860 #endif
861 	/*
862 	 * pointers to (original) parent process, youngest child, younger sibling,
863 	 * older sibling, respectively.  (p->father can be replaced with
864 	 * p->parent->pid)
865 	 */
866 	struct task_struct *real_parent; /* real parent process (when being debugged) */
867 	struct task_struct *parent;	/* parent process */
868 	/*
869 	 * children/sibling forms the list of my children plus the
870 	 * tasks I'm ptracing.
871 	 */
872 	struct list_head children;	/* list of my children */
873 	struct list_head sibling;	/* linkage in my parent's children list */
874 	struct task_struct *group_leader;	/* threadgroup leader */
875 
876 	/* PID/PID hash table linkage. */
877 	struct pid_link pids[PIDTYPE_MAX];
878 	struct list_head thread_group;
879 
880 	struct completion *vfork_done;		/* for vfork() */
881 	int __user *set_child_tid;		/* CLONE_CHILD_SETTID */
882 	int __user *clear_child_tid;		/* CLONE_CHILD_CLEARTID */
883 
884 	unsigned long rt_priority;
885 	cputime_t utime, stime;
886 	unsigned long nvcsw, nivcsw; /* context switch counts */
887 	struct timespec start_time;
888 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
889 	unsigned long min_flt, maj_flt;
890 
891   	cputime_t it_prof_expires, it_virt_expires;
892 	unsigned long long it_sched_expires;
893 	struct list_head cpu_timers[3];
894 
895 /* process credentials */
896 	uid_t uid,euid,suid,fsuid;
897 	gid_t gid,egid,sgid,fsgid;
898 	struct group_info *group_info;
899 	kernel_cap_t   cap_effective, cap_inheritable, cap_permitted;
900 	unsigned keep_capabilities:1;
901 	struct user_struct *user;
902 #ifdef CONFIG_KEYS
903 	struct key *request_key_auth;	/* assumed request_key authority */
904 	struct key *thread_keyring;	/* keyring private to this thread */
905 	unsigned char jit_keyring;	/* default keyring to attach requested keys to */
906 #endif
907 	/*
908 	 * fpu_counter contains the number of consecutive context switches
909 	 * that the FPU is used. If this is over a threshold, the lazy fpu
910 	 * saving becomes unlazy to save the trap. This is an unsigned char
911 	 * so that after 256 times the counter wraps and the behavior turns
912 	 * lazy again; this to deal with bursty apps that only use FPU for
913 	 * a short time
914 	 */
915 	unsigned char fpu_counter;
916 	int oomkilladj; /* OOM kill score adjustment (bit shift). */
917 	char comm[TASK_COMM_LEN]; /* executable name excluding path
918 				     - access with [gs]et_task_comm (which lock
919 				       it with task_lock())
920 				     - initialized normally by flush_old_exec */
921 /* file system info */
922 	int link_count, total_link_count;
923 #ifdef CONFIG_SYSVIPC
924 /* ipc stuff */
925 	struct sysv_sem sysvsem;
926 #endif
927 /* CPU-specific state of this task */
928 	struct thread_struct thread;
929 /* filesystem information */
930 	struct fs_struct *fs;
931 /* open file information */
932 	struct files_struct *files;
933 /* namespaces */
934 	struct nsproxy *nsproxy;
935 /* signal handlers */
936 	struct signal_struct *signal;
937 	struct sighand_struct *sighand;
938 
939 	sigset_t blocked, real_blocked;
940 	sigset_t saved_sigmask;		/* To be restored with TIF_RESTORE_SIGMASK */
941 	struct sigpending pending;
942 
943 	unsigned long sas_ss_sp;
944 	size_t sas_ss_size;
945 	int (*notifier)(void *priv);
946 	void *notifier_data;
947 	sigset_t *notifier_mask;
948 
949 	void *security;
950 	struct audit_context *audit_context;
951 	seccomp_t seccomp;
952 
953 /* Thread group tracking */
954    	u32 parent_exec_id;
955    	u32 self_exec_id;
956 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
957 	spinlock_t alloc_lock;
958 
959 	/* Protection of the PI data structures: */
960 	spinlock_t pi_lock;
961 
962 #ifdef CONFIG_RT_MUTEXES
963 	/* PI waiters blocked on a rt_mutex held by this task */
964 	struct plist_head pi_waiters;
965 	/* Deadlock detection and priority inheritance handling */
966 	struct rt_mutex_waiter *pi_blocked_on;
967 #endif
968 
969 #ifdef CONFIG_DEBUG_MUTEXES
970 	/* mutex deadlock detection */
971 	struct mutex_waiter *blocked_on;
972 #endif
973 #ifdef CONFIG_TRACE_IRQFLAGS
974 	unsigned int irq_events;
975 	int hardirqs_enabled;
976 	unsigned long hardirq_enable_ip;
977 	unsigned int hardirq_enable_event;
978 	unsigned long hardirq_disable_ip;
979 	unsigned int hardirq_disable_event;
980 	int softirqs_enabled;
981 	unsigned long softirq_disable_ip;
982 	unsigned int softirq_disable_event;
983 	unsigned long softirq_enable_ip;
984 	unsigned int softirq_enable_event;
985 	int hardirq_context;
986 	int softirq_context;
987 #endif
988 #ifdef CONFIG_LOCKDEP
989 # define MAX_LOCK_DEPTH 30UL
990 	u64 curr_chain_key;
991 	int lockdep_depth;
992 	struct held_lock held_locks[MAX_LOCK_DEPTH];
993 	unsigned int lockdep_recursion;
994 #endif
995 
996 /* journalling filesystem info */
997 	void *journal_info;
998 
999 /* VM state */
1000 	struct reclaim_state *reclaim_state;
1001 
1002 	struct backing_dev_info *backing_dev_info;
1003 
1004 	struct io_context *io_context;
1005 
1006 	unsigned long ptrace_message;
1007 	siginfo_t *last_siginfo; /* For ptrace use.  */
1008 /*
1009  * current io wait handle: wait queue entry to use for io waits
1010  * If this thread is processing aio, this points at the waitqueue
1011  * inside the currently handled kiocb. It may be NULL (i.e. default
1012  * to a stack based synchronous wait) if its doing sync IO.
1013  */
1014 	wait_queue_t *io_wait;
1015 #ifdef CONFIG_TASK_XACCT
1016 /* i/o counters(bytes read/written, #syscalls */
1017 	u64 rchar, wchar, syscr, syscw;
1018 #endif
1019 	struct task_io_accounting ioac;
1020 #if defined(CONFIG_TASK_XACCT)
1021 	u64 acct_rss_mem1;	/* accumulated rss usage */
1022 	u64 acct_vm_mem1;	/* accumulated virtual memory usage */
1023 	cputime_t acct_stimexpd;/* stime since last update */
1024 #endif
1025 #ifdef CONFIG_NUMA
1026   	struct mempolicy *mempolicy;
1027 	short il_next;
1028 #endif
1029 #ifdef CONFIG_CPUSETS
1030 	struct cpuset *cpuset;
1031 	nodemask_t mems_allowed;
1032 	int cpuset_mems_generation;
1033 	int cpuset_mem_spread_rotor;
1034 #endif
1035 	struct robust_list_head __user *robust_list;
1036 #ifdef CONFIG_COMPAT
1037 	struct compat_robust_list_head __user *compat_robust_list;
1038 #endif
1039 	struct list_head pi_state_list;
1040 	struct futex_pi_state *pi_state_cache;
1041 
1042 	atomic_t fs_excl;	/* holding fs exclusive resources */
1043 	struct rcu_head rcu;
1044 
1045 	/*
1046 	 * cache last used pipe for splice
1047 	 */
1048 	struct pipe_inode_info *splice_pipe;
1049 #ifdef	CONFIG_TASK_DELAY_ACCT
1050 	struct task_delay_info *delays;
1051 #endif
1052 #ifdef CONFIG_FAULT_INJECTION
1053 	int make_it_fail;
1054 #endif
1055 };
1056 
1057 static inline pid_t process_group(struct task_struct *tsk)
1058 {
1059 	return tsk->signal->pgrp;
1060 }
1061 
1062 static inline pid_t signal_session(struct signal_struct *sig)
1063 {
1064 	return sig->__session;
1065 }
1066 
1067 static inline pid_t process_session(struct task_struct *tsk)
1068 {
1069 	return signal_session(tsk->signal);
1070 }
1071 
1072 static inline void set_signal_session(struct signal_struct *sig, pid_t session)
1073 {
1074 	sig->__session = session;
1075 }
1076 
1077 static inline struct pid *task_pid(struct task_struct *task)
1078 {
1079 	return task->pids[PIDTYPE_PID].pid;
1080 }
1081 
1082 static inline struct pid *task_tgid(struct task_struct *task)
1083 {
1084 	return task->group_leader->pids[PIDTYPE_PID].pid;
1085 }
1086 
1087 static inline struct pid *task_pgrp(struct task_struct *task)
1088 {
1089 	return task->group_leader->pids[PIDTYPE_PGID].pid;
1090 }
1091 
1092 static inline struct pid *task_session(struct task_struct *task)
1093 {
1094 	return task->group_leader->pids[PIDTYPE_SID].pid;
1095 }
1096 
1097 /**
1098  * pid_alive - check that a task structure is not stale
1099  * @p: Task structure to be checked.
1100  *
1101  * Test if a process is not yet dead (at most zombie state)
1102  * If pid_alive fails, then pointers within the task structure
1103  * can be stale and must not be dereferenced.
1104  */
1105 static inline int pid_alive(struct task_struct *p)
1106 {
1107 	return p->pids[PIDTYPE_PID].pid != NULL;
1108 }
1109 
1110 /**
1111  * is_init - check if a task structure is init
1112  * @tsk: Task structure to be checked.
1113  *
1114  * Check if a task structure is the first user space task the kernel created.
1115  */
1116 static inline int is_init(struct task_struct *tsk)
1117 {
1118 	return tsk->pid == 1;
1119 }
1120 
1121 extern struct pid *cad_pid;
1122 
1123 extern void free_task(struct task_struct *tsk);
1124 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1125 
1126 extern void __put_task_struct(struct task_struct *t);
1127 
1128 static inline void put_task_struct(struct task_struct *t)
1129 {
1130 	if (atomic_dec_and_test(&t->usage))
1131 		__put_task_struct(t);
1132 }
1133 
1134 /*
1135  * Per process flags
1136  */
1137 #define PF_ALIGNWARN	0x00000001	/* Print alignment warning msgs */
1138 					/* Not implemented yet, only for 486*/
1139 #define PF_STARTING	0x00000002	/* being created */
1140 #define PF_EXITING	0x00000004	/* getting shut down */
1141 #define PF_FORKNOEXEC	0x00000040	/* forked but didn't exec */
1142 #define PF_SUPERPRIV	0x00000100	/* used super-user privileges */
1143 #define PF_DUMPCORE	0x00000200	/* dumped core */
1144 #define PF_SIGNALED	0x00000400	/* killed by a signal */
1145 #define PF_MEMALLOC	0x00000800	/* Allocating memory */
1146 #define PF_FLUSHER	0x00001000	/* responsible for disk writeback */
1147 #define PF_USED_MATH	0x00002000	/* if unset the fpu must be initialized before use */
1148 #define PF_NOFREEZE	0x00008000	/* this thread should not be frozen */
1149 #define PF_FROZEN	0x00010000	/* frozen for system suspend */
1150 #define PF_FSTRANS	0x00020000	/* inside a filesystem transaction */
1151 #define PF_KSWAPD	0x00040000	/* I am kswapd */
1152 #define PF_SWAPOFF	0x00080000	/* I am in swapoff */
1153 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
1154 #define PF_BORROWED_MM	0x00200000	/* I am a kthread doing use_mm */
1155 #define PF_RANDOMIZE	0x00400000	/* randomize virtual address space */
1156 #define PF_SWAPWRITE	0x00800000	/* Allowed to write to swap */
1157 #define PF_SPREAD_PAGE	0x01000000	/* Spread page cache over cpuset */
1158 #define PF_SPREAD_SLAB	0x02000000	/* Spread some slab caches over cpuset */
1159 #define PF_MEMPOLICY	0x10000000	/* Non-default NUMA mempolicy */
1160 #define PF_MUTEX_TESTER	0x20000000	/* Thread belongs to the rt mutex tester */
1161 
1162 /*
1163  * Only the _current_ task can read/write to tsk->flags, but other
1164  * tasks can access tsk->flags in readonly mode for example
1165  * with tsk_used_math (like during threaded core dumping).
1166  * There is however an exception to this rule during ptrace
1167  * or during fork: the ptracer task is allowed to write to the
1168  * child->flags of its traced child (same goes for fork, the parent
1169  * can write to the child->flags), because we're guaranteed the
1170  * child is not running and in turn not changing child->flags
1171  * at the same time the parent does it.
1172  */
1173 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1174 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1175 #define clear_used_math() clear_stopped_child_used_math(current)
1176 #define set_used_math() set_stopped_child_used_math(current)
1177 #define conditional_stopped_child_used_math(condition, child) \
1178 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1179 #define conditional_used_math(condition) \
1180 	conditional_stopped_child_used_math(condition, current)
1181 #define copy_to_stopped_child_used_math(child) \
1182 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1183 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1184 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1185 #define used_math() tsk_used_math(current)
1186 
1187 #ifdef CONFIG_SMP
1188 extern int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask);
1189 #else
1190 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1191 {
1192 	if (!cpu_isset(0, new_mask))
1193 		return -EINVAL;
1194 	return 0;
1195 }
1196 #endif
1197 
1198 extern unsigned long long sched_clock(void);
1199 extern unsigned long long
1200 current_sched_time(const struct task_struct *current_task);
1201 
1202 /* sched_exec is called by processes performing an exec */
1203 #ifdef CONFIG_SMP
1204 extern void sched_exec(void);
1205 #else
1206 #define sched_exec()   {}
1207 #endif
1208 
1209 #ifdef CONFIG_HOTPLUG_CPU
1210 extern void idle_task_exit(void);
1211 #else
1212 static inline void idle_task_exit(void) {}
1213 #endif
1214 
1215 extern void sched_idle_next(void);
1216 
1217 #ifdef CONFIG_RT_MUTEXES
1218 extern int rt_mutex_getprio(struct task_struct *p);
1219 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1220 extern void rt_mutex_adjust_pi(struct task_struct *p);
1221 #else
1222 static inline int rt_mutex_getprio(struct task_struct *p)
1223 {
1224 	return p->normal_prio;
1225 }
1226 # define rt_mutex_adjust_pi(p)		do { } while (0)
1227 #endif
1228 
1229 extern void set_user_nice(struct task_struct *p, long nice);
1230 extern int task_prio(const struct task_struct *p);
1231 extern int task_nice(const struct task_struct *p);
1232 extern int can_nice(const struct task_struct *p, const int nice);
1233 extern int task_curr(const struct task_struct *p);
1234 extern int idle_cpu(int cpu);
1235 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1236 extern struct task_struct *idle_task(int cpu);
1237 extern struct task_struct *curr_task(int cpu);
1238 extern void set_curr_task(int cpu, struct task_struct *p);
1239 
1240 void yield(void);
1241 
1242 /*
1243  * The default (Linux) execution domain.
1244  */
1245 extern struct exec_domain	default_exec_domain;
1246 
1247 union thread_union {
1248 	struct thread_info thread_info;
1249 	unsigned long stack[THREAD_SIZE/sizeof(long)];
1250 };
1251 
1252 #ifndef __HAVE_ARCH_KSTACK_END
1253 static inline int kstack_end(void *addr)
1254 {
1255 	/* Reliable end of stack detection:
1256 	 * Some APM bios versions misalign the stack
1257 	 */
1258 	return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1259 }
1260 #endif
1261 
1262 extern union thread_union init_thread_union;
1263 extern struct task_struct init_task;
1264 
1265 extern struct   mm_struct init_mm;
1266 
1267 #define find_task_by_pid(nr)	find_task_by_pid_type(PIDTYPE_PID, nr)
1268 extern struct task_struct *find_task_by_pid_type(int type, int pid);
1269 extern void __set_special_pids(pid_t session, pid_t pgrp);
1270 
1271 /* per-UID process charging. */
1272 extern struct user_struct * alloc_uid(uid_t);
1273 static inline struct user_struct *get_uid(struct user_struct *u)
1274 {
1275 	atomic_inc(&u->__count);
1276 	return u;
1277 }
1278 extern void free_uid(struct user_struct *);
1279 extern void switch_uid(struct user_struct *);
1280 
1281 #include <asm/current.h>
1282 
1283 extern void do_timer(unsigned long ticks);
1284 
1285 extern int FASTCALL(wake_up_state(struct task_struct * tsk, unsigned int state));
1286 extern int FASTCALL(wake_up_process(struct task_struct * tsk));
1287 extern void FASTCALL(wake_up_new_task(struct task_struct * tsk,
1288 						unsigned long clone_flags));
1289 #ifdef CONFIG_SMP
1290  extern void kick_process(struct task_struct *tsk);
1291 #else
1292  static inline void kick_process(struct task_struct *tsk) { }
1293 #endif
1294 extern void FASTCALL(sched_fork(struct task_struct * p, int clone_flags));
1295 extern void FASTCALL(sched_exit(struct task_struct * p));
1296 
1297 extern int in_group_p(gid_t);
1298 extern int in_egroup_p(gid_t);
1299 
1300 extern void proc_caches_init(void);
1301 extern void flush_signals(struct task_struct *);
1302 extern void flush_signal_handlers(struct task_struct *, int force_default);
1303 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
1304 
1305 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
1306 {
1307 	unsigned long flags;
1308 	int ret;
1309 
1310 	spin_lock_irqsave(&tsk->sighand->siglock, flags);
1311 	ret = dequeue_signal(tsk, mask, info);
1312 	spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
1313 
1314 	return ret;
1315 }
1316 
1317 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
1318 			      sigset_t *mask);
1319 extern void unblock_all_signals(void);
1320 extern void release_task(struct task_struct * p);
1321 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
1322 extern int send_group_sig_info(int, struct siginfo *, struct task_struct *);
1323 extern int force_sigsegv(int, struct task_struct *);
1324 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
1325 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1326 extern int kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
1327 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
1328 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
1329 extern int kill_pgrp(struct pid *pid, int sig, int priv);
1330 extern int kill_pid(struct pid *pid, int sig, int priv);
1331 extern int kill_proc_info(int, struct siginfo *, pid_t);
1332 extern void do_notify_parent(struct task_struct *, int);
1333 extern void force_sig(int, struct task_struct *);
1334 extern void force_sig_specific(int, struct task_struct *);
1335 extern int send_sig(int, struct task_struct *, int);
1336 extern void zap_other_threads(struct task_struct *p);
1337 extern int kill_proc(pid_t, int, int);
1338 extern struct sigqueue *sigqueue_alloc(void);
1339 extern void sigqueue_free(struct sigqueue *);
1340 extern int send_sigqueue(int, struct sigqueue *,  struct task_struct *);
1341 extern int send_group_sigqueue(int, struct sigqueue *,  struct task_struct *);
1342 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
1343 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
1344 
1345 static inline int kill_cad_pid(int sig, int priv)
1346 {
1347 	return kill_pid(cad_pid, sig, priv);
1348 }
1349 
1350 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
1351 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
1352 #define SEND_SIG_PRIV	((struct siginfo *) 1)
1353 #define SEND_SIG_FORCED	((struct siginfo *) 2)
1354 
1355 static inline int is_si_special(const struct siginfo *info)
1356 {
1357 	return info <= SEND_SIG_FORCED;
1358 }
1359 
1360 /* True if we are on the alternate signal stack.  */
1361 
1362 static inline int on_sig_stack(unsigned long sp)
1363 {
1364 	return (sp - current->sas_ss_sp < current->sas_ss_size);
1365 }
1366 
1367 static inline int sas_ss_flags(unsigned long sp)
1368 {
1369 	return (current->sas_ss_size == 0 ? SS_DISABLE
1370 		: on_sig_stack(sp) ? SS_ONSTACK : 0);
1371 }
1372 
1373 /*
1374  * Routines for handling mm_structs
1375  */
1376 extern struct mm_struct * mm_alloc(void);
1377 
1378 /* mmdrop drops the mm and the page tables */
1379 extern void FASTCALL(__mmdrop(struct mm_struct *));
1380 static inline void mmdrop(struct mm_struct * mm)
1381 {
1382 	if (atomic_dec_and_test(&mm->mm_count))
1383 		__mmdrop(mm);
1384 }
1385 
1386 /* mmput gets rid of the mappings and all user-space */
1387 extern void mmput(struct mm_struct *);
1388 /* Grab a reference to a task's mm, if it is not already going away */
1389 extern struct mm_struct *get_task_mm(struct task_struct *task);
1390 /* Remove the current tasks stale references to the old mm_struct */
1391 extern void mm_release(struct task_struct *, struct mm_struct *);
1392 
1393 extern int  copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
1394 extern void flush_thread(void);
1395 extern void exit_thread(void);
1396 
1397 extern void exit_files(struct task_struct *);
1398 extern void __cleanup_signal(struct signal_struct *);
1399 extern void __cleanup_sighand(struct sighand_struct *);
1400 extern void exit_itimers(struct signal_struct *);
1401 
1402 extern NORET_TYPE void do_group_exit(int);
1403 
1404 extern void daemonize(const char *, ...);
1405 extern int allow_signal(int);
1406 extern int disallow_signal(int);
1407 
1408 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
1409 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
1410 struct task_struct *fork_idle(int);
1411 
1412 extern void set_task_comm(struct task_struct *tsk, char *from);
1413 extern void get_task_comm(char *to, struct task_struct *tsk);
1414 
1415 #ifdef CONFIG_SMP
1416 extern void wait_task_inactive(struct task_struct * p);
1417 #else
1418 #define wait_task_inactive(p)	do { } while (0)
1419 #endif
1420 
1421 #define remove_parent(p)	list_del_init(&(p)->sibling)
1422 #define add_parent(p)		list_add_tail(&(p)->sibling,&(p)->parent->children)
1423 
1424 #define next_task(p)	list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
1425 
1426 #define for_each_process(p) \
1427 	for (p = &init_task ; (p = next_task(p)) != &init_task ; )
1428 
1429 /*
1430  * Careful: do_each_thread/while_each_thread is a double loop so
1431  *          'break' will not work as expected - use goto instead.
1432  */
1433 #define do_each_thread(g, t) \
1434 	for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
1435 
1436 #define while_each_thread(g, t) \
1437 	while ((t = next_thread(t)) != g)
1438 
1439 /* de_thread depends on thread_group_leader not being a pid based check */
1440 #define thread_group_leader(p)	(p == p->group_leader)
1441 
1442 /* Do to the insanities of de_thread it is possible for a process
1443  * to have the pid of the thread group leader without actually being
1444  * the thread group leader.  For iteration through the pids in proc
1445  * all we care about is that we have a task with the appropriate
1446  * pid, we don't actually care if we have the right task.
1447  */
1448 static inline int has_group_leader_pid(struct task_struct *p)
1449 {
1450 	return p->pid == p->tgid;
1451 }
1452 
1453 static inline struct task_struct *next_thread(const struct task_struct *p)
1454 {
1455 	return list_entry(rcu_dereference(p->thread_group.next),
1456 			  struct task_struct, thread_group);
1457 }
1458 
1459 static inline int thread_group_empty(struct task_struct *p)
1460 {
1461 	return list_empty(&p->thread_group);
1462 }
1463 
1464 #define delay_group_leader(p) \
1465 		(thread_group_leader(p) && !thread_group_empty(p))
1466 
1467 /*
1468  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1469  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
1470  * pins the final release of task.io_context.  Also protects ->cpuset.
1471  *
1472  * Nests both inside and outside of read_lock(&tasklist_lock).
1473  * It must not be nested with write_lock_irq(&tasklist_lock),
1474  * neither inside nor outside.
1475  */
1476 static inline void task_lock(struct task_struct *p)
1477 {
1478 	spin_lock(&p->alloc_lock);
1479 }
1480 
1481 static inline void task_unlock(struct task_struct *p)
1482 {
1483 	spin_unlock(&p->alloc_lock);
1484 }
1485 
1486 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
1487 							unsigned long *flags);
1488 
1489 static inline void unlock_task_sighand(struct task_struct *tsk,
1490 						unsigned long *flags)
1491 {
1492 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
1493 }
1494 
1495 #ifndef __HAVE_THREAD_FUNCTIONS
1496 
1497 #define task_thread_info(task) (task)->thread_info
1498 #define task_stack_page(task) ((void*)((task)->thread_info))
1499 
1500 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1501 {
1502 	*task_thread_info(p) = *task_thread_info(org);
1503 	task_thread_info(p)->task = p;
1504 }
1505 
1506 static inline unsigned long *end_of_stack(struct task_struct *p)
1507 {
1508 	return (unsigned long *)(p->thread_info + 1);
1509 }
1510 
1511 #endif
1512 
1513 /* set thread flags in other task's structures
1514  * - see asm/thread_info.h for TIF_xxxx flags available
1515  */
1516 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1517 {
1518 	set_ti_thread_flag(task_thread_info(tsk), flag);
1519 }
1520 
1521 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1522 {
1523 	clear_ti_thread_flag(task_thread_info(tsk), flag);
1524 }
1525 
1526 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1527 {
1528 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1529 }
1530 
1531 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1532 {
1533 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1534 }
1535 
1536 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1537 {
1538 	return test_ti_thread_flag(task_thread_info(tsk), flag);
1539 }
1540 
1541 static inline void set_tsk_need_resched(struct task_struct *tsk)
1542 {
1543 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1544 }
1545 
1546 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1547 {
1548 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1549 }
1550 
1551 static inline int signal_pending(struct task_struct *p)
1552 {
1553 	return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
1554 }
1555 
1556 static inline int need_resched(void)
1557 {
1558 	return unlikely(test_thread_flag(TIF_NEED_RESCHED));
1559 }
1560 
1561 /*
1562  * cond_resched() and cond_resched_lock(): latency reduction via
1563  * explicit rescheduling in places that are safe. The return
1564  * value indicates whether a reschedule was done in fact.
1565  * cond_resched_lock() will drop the spinlock before scheduling,
1566  * cond_resched_softirq() will enable bhs before scheduling.
1567  */
1568 extern int cond_resched(void);
1569 extern int cond_resched_lock(spinlock_t * lock);
1570 extern int cond_resched_softirq(void);
1571 
1572 /*
1573  * Does a critical section need to be broken due to another
1574  * task waiting?:
1575  */
1576 #if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP)
1577 # define need_lockbreak(lock) ((lock)->break_lock)
1578 #else
1579 # define need_lockbreak(lock) 0
1580 #endif
1581 
1582 /*
1583  * Does a critical section need to be broken due to another
1584  * task waiting or preemption being signalled:
1585  */
1586 static inline int lock_need_resched(spinlock_t *lock)
1587 {
1588 	if (need_lockbreak(lock) || need_resched())
1589 		return 1;
1590 	return 0;
1591 }
1592 
1593 /* Reevaluate whether the task has signals pending delivery.
1594    This is required every time the blocked sigset_t changes.
1595    callers must hold sighand->siglock.  */
1596 
1597 extern FASTCALL(void recalc_sigpending_tsk(struct task_struct *t));
1598 extern void recalc_sigpending(void);
1599 
1600 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
1601 
1602 /*
1603  * Wrappers for p->thread_info->cpu access. No-op on UP.
1604  */
1605 #ifdef CONFIG_SMP
1606 
1607 static inline unsigned int task_cpu(const struct task_struct *p)
1608 {
1609 	return task_thread_info(p)->cpu;
1610 }
1611 
1612 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1613 {
1614 	task_thread_info(p)->cpu = cpu;
1615 }
1616 
1617 #else
1618 
1619 static inline unsigned int task_cpu(const struct task_struct *p)
1620 {
1621 	return 0;
1622 }
1623 
1624 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1625 {
1626 }
1627 
1628 #endif /* CONFIG_SMP */
1629 
1630 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1631 extern void arch_pick_mmap_layout(struct mm_struct *mm);
1632 #else
1633 static inline void arch_pick_mmap_layout(struct mm_struct *mm)
1634 {
1635 	mm->mmap_base = TASK_UNMAPPED_BASE;
1636 	mm->get_unmapped_area = arch_get_unmapped_area;
1637 	mm->unmap_area = arch_unmap_area;
1638 }
1639 #endif
1640 
1641 extern long sched_setaffinity(pid_t pid, cpumask_t new_mask);
1642 extern long sched_getaffinity(pid_t pid, cpumask_t *mask);
1643 
1644 #include <linux/sysdev.h>
1645 extern int sched_mc_power_savings, sched_smt_power_savings;
1646 extern struct sysdev_attribute attr_sched_mc_power_savings, attr_sched_smt_power_savings;
1647 extern int sched_create_sysfs_power_savings_entries(struct sysdev_class *cls);
1648 
1649 extern void normalize_rt_tasks(void);
1650 
1651 #ifdef CONFIG_TASK_XACCT
1652 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
1653 {
1654 	tsk->rchar += amt;
1655 }
1656 
1657 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
1658 {
1659 	tsk->wchar += amt;
1660 }
1661 
1662 static inline void inc_syscr(struct task_struct *tsk)
1663 {
1664 	tsk->syscr++;
1665 }
1666 
1667 static inline void inc_syscw(struct task_struct *tsk)
1668 {
1669 	tsk->syscw++;
1670 }
1671 #else
1672 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
1673 {
1674 }
1675 
1676 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
1677 {
1678 }
1679 
1680 static inline void inc_syscr(struct task_struct *tsk)
1681 {
1682 }
1683 
1684 static inline void inc_syscw(struct task_struct *tsk)
1685 {
1686 }
1687 #endif
1688 
1689 #endif /* __KERNEL__ */
1690 
1691 #endif
1692