xref: /linux/include/linux/sched.h (revision 98b8788ae91694499d1995035625bea16a4db0c4)
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3 
4 /*
5  * cloning flags:
6  */
7 #define CSIGNAL		0x000000ff	/* signal mask to be sent at exit */
8 #define CLONE_VM	0x00000100	/* set if VM shared between processes */
9 #define CLONE_FS	0x00000200	/* set if fs info shared between processes */
10 #define CLONE_FILES	0x00000400	/* set if open files shared between processes */
11 #define CLONE_SIGHAND	0x00000800	/* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE	0x00002000	/* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK	0x00004000	/* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT	0x00008000	/* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD	0x00010000	/* Same thread group? */
16 #define CLONE_NEWNS	0x00020000	/* New namespace group? */
17 #define CLONE_SYSVSEM	0x00040000	/* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS	0x00080000	/* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID	0x00100000	/* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID	0x00200000	/* clear the TID in the child */
21 #define CLONE_DETACHED		0x00400000	/* Unused, ignored */
22 #define CLONE_UNTRACED		0x00800000	/* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID	0x01000000	/* set the TID in the child */
24 #define CLONE_STOPPED		0x02000000	/* Start in stopped state */
25 #define CLONE_NEWUTS		0x04000000	/* New utsname group? */
26 #define CLONE_NEWIPC		0x08000000	/* New ipcs */
27 #define CLONE_NEWUSER		0x10000000	/* New user namespace */
28 #define CLONE_NEWPID		0x20000000	/* New pid namespace */
29 #define CLONE_NEWNET		0x40000000	/* New network namespace */
30 #define CLONE_IO		0x80000000	/* Clone io context */
31 
32 /*
33  * Scheduling policies
34  */
35 #define SCHED_NORMAL		0
36 #define SCHED_FIFO		1
37 #define SCHED_RR		2
38 #define SCHED_BATCH		3
39 /* SCHED_ISO: reserved but not implemented yet */
40 #define SCHED_IDLE		5
41 /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
42 #define SCHED_RESET_ON_FORK     0x40000000
43 
44 #ifdef __KERNEL__
45 
46 struct sched_param {
47 	int sched_priority;
48 };
49 
50 #include <asm/param.h>	/* for HZ */
51 
52 #include <linux/capability.h>
53 #include <linux/threads.h>
54 #include <linux/kernel.h>
55 #include <linux/types.h>
56 #include <linux/timex.h>
57 #include <linux/jiffies.h>
58 #include <linux/rbtree.h>
59 #include <linux/thread_info.h>
60 #include <linux/cpumask.h>
61 #include <linux/errno.h>
62 #include <linux/nodemask.h>
63 #include <linux/mm_types.h>
64 
65 #include <asm/system.h>
66 #include <asm/page.h>
67 #include <asm/ptrace.h>
68 #include <asm/cputime.h>
69 
70 #include <linux/smp.h>
71 #include <linux/sem.h>
72 #include <linux/signal.h>
73 #include <linux/path.h>
74 #include <linux/compiler.h>
75 #include <linux/completion.h>
76 #include <linux/pid.h>
77 #include <linux/percpu.h>
78 #include <linux/topology.h>
79 #include <linux/proportions.h>
80 #include <linux/seccomp.h>
81 #include <linux/rcupdate.h>
82 #include <linux/rculist.h>
83 #include <linux/rtmutex.h>
84 
85 #include <linux/time.h>
86 #include <linux/param.h>
87 #include <linux/resource.h>
88 #include <linux/timer.h>
89 #include <linux/hrtimer.h>
90 #include <linux/task_io_accounting.h>
91 #include <linux/kobject.h>
92 #include <linux/latencytop.h>
93 #include <linux/cred.h>
94 
95 #include <asm/processor.h>
96 
97 struct exec_domain;
98 struct futex_pi_state;
99 struct robust_list_head;
100 struct bio;
101 struct fs_struct;
102 struct bts_context;
103 struct perf_event_context;
104 
105 /*
106  * List of flags we want to share for kernel threads,
107  * if only because they are not used by them anyway.
108  */
109 #define CLONE_KERNEL	(CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
110 
111 /*
112  * These are the constant used to fake the fixed-point load-average
113  * counting. Some notes:
114  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
115  *    a load-average precision of 10 bits integer + 11 bits fractional
116  *  - if you want to count load-averages more often, you need more
117  *    precision, or rounding will get you. With 2-second counting freq,
118  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
119  *    11 bit fractions.
120  */
121 extern unsigned long avenrun[];		/* Load averages */
122 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
123 
124 #define FSHIFT		11		/* nr of bits of precision */
125 #define FIXED_1		(1<<FSHIFT)	/* 1.0 as fixed-point */
126 #define LOAD_FREQ	(5*HZ+1)	/* 5 sec intervals */
127 #define EXP_1		1884		/* 1/exp(5sec/1min) as fixed-point */
128 #define EXP_5		2014		/* 1/exp(5sec/5min) */
129 #define EXP_15		2037		/* 1/exp(5sec/15min) */
130 
131 #define CALC_LOAD(load,exp,n) \
132 	load *= exp; \
133 	load += n*(FIXED_1-exp); \
134 	load >>= FSHIFT;
135 
136 extern unsigned long total_forks;
137 extern int nr_threads;
138 DECLARE_PER_CPU(unsigned long, process_counts);
139 extern int nr_processes(void);
140 extern unsigned long nr_running(void);
141 extern unsigned long nr_uninterruptible(void);
142 extern unsigned long nr_iowait(void);
143 extern unsigned long nr_iowait_cpu(void);
144 extern unsigned long this_cpu_load(void);
145 
146 
147 extern void calc_global_load(void);
148 
149 extern unsigned long get_parent_ip(unsigned long addr);
150 
151 struct seq_file;
152 struct cfs_rq;
153 struct task_group;
154 #ifdef CONFIG_SCHED_DEBUG
155 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
156 extern void proc_sched_set_task(struct task_struct *p);
157 extern void
158 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
159 #else
160 static inline void
161 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
162 {
163 }
164 static inline void proc_sched_set_task(struct task_struct *p)
165 {
166 }
167 static inline void
168 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
169 {
170 }
171 #endif
172 
173 /*
174  * Task state bitmask. NOTE! These bits are also
175  * encoded in fs/proc/array.c: get_task_state().
176  *
177  * We have two separate sets of flags: task->state
178  * is about runnability, while task->exit_state are
179  * about the task exiting. Confusing, but this way
180  * modifying one set can't modify the other one by
181  * mistake.
182  */
183 #define TASK_RUNNING		0
184 #define TASK_INTERRUPTIBLE	1
185 #define TASK_UNINTERRUPTIBLE	2
186 #define __TASK_STOPPED		4
187 #define __TASK_TRACED		8
188 /* in tsk->exit_state */
189 #define EXIT_ZOMBIE		16
190 #define EXIT_DEAD		32
191 /* in tsk->state again */
192 #define TASK_DEAD		64
193 #define TASK_WAKEKILL		128
194 #define TASK_WAKING		256
195 
196 /* Convenience macros for the sake of set_task_state */
197 #define TASK_KILLABLE		(TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
198 #define TASK_STOPPED		(TASK_WAKEKILL | __TASK_STOPPED)
199 #define TASK_TRACED		(TASK_WAKEKILL | __TASK_TRACED)
200 
201 /* Convenience macros for the sake of wake_up */
202 #define TASK_NORMAL		(TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
203 #define TASK_ALL		(TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
204 
205 /* get_task_state() */
206 #define TASK_REPORT		(TASK_RUNNING | TASK_INTERRUPTIBLE | \
207 				 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
208 				 __TASK_TRACED)
209 
210 #define task_is_traced(task)	((task->state & __TASK_TRACED) != 0)
211 #define task_is_stopped(task)	((task->state & __TASK_STOPPED) != 0)
212 #define task_is_stopped_or_traced(task)	\
213 			((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
214 #define task_contributes_to_load(task)	\
215 				((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
216 				 (task->flags & PF_FREEZING) == 0)
217 
218 #define __set_task_state(tsk, state_value)		\
219 	do { (tsk)->state = (state_value); } while (0)
220 #define set_task_state(tsk, state_value)		\
221 	set_mb((tsk)->state, (state_value))
222 
223 /*
224  * set_current_state() includes a barrier so that the write of current->state
225  * is correctly serialised wrt the caller's subsequent test of whether to
226  * actually sleep:
227  *
228  *	set_current_state(TASK_UNINTERRUPTIBLE);
229  *	if (do_i_need_to_sleep())
230  *		schedule();
231  *
232  * If the caller does not need such serialisation then use __set_current_state()
233  */
234 #define __set_current_state(state_value)			\
235 	do { current->state = (state_value); } while (0)
236 #define set_current_state(state_value)		\
237 	set_mb(current->state, (state_value))
238 
239 /* Task command name length */
240 #define TASK_COMM_LEN 16
241 
242 #include <linux/spinlock.h>
243 
244 /*
245  * This serializes "schedule()" and also protects
246  * the run-queue from deletions/modifications (but
247  * _adding_ to the beginning of the run-queue has
248  * a separate lock).
249  */
250 extern rwlock_t tasklist_lock;
251 extern spinlock_t mmlist_lock;
252 
253 struct task_struct;
254 
255 extern void sched_init(void);
256 extern void sched_init_smp(void);
257 extern asmlinkage void schedule_tail(struct task_struct *prev);
258 extern void init_idle(struct task_struct *idle, int cpu);
259 extern void init_idle_bootup_task(struct task_struct *idle);
260 
261 extern int runqueue_is_locked(int cpu);
262 extern void task_rq_unlock_wait(struct task_struct *p);
263 
264 extern cpumask_var_t nohz_cpu_mask;
265 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
266 extern int select_nohz_load_balancer(int cpu);
267 extern int get_nohz_load_balancer(void);
268 #else
269 static inline int select_nohz_load_balancer(int cpu)
270 {
271 	return 0;
272 }
273 #endif
274 
275 /*
276  * Only dump TASK_* tasks. (0 for all tasks)
277  */
278 extern void show_state_filter(unsigned long state_filter);
279 
280 static inline void show_state(void)
281 {
282 	show_state_filter(0);
283 }
284 
285 extern void show_regs(struct pt_regs *);
286 
287 /*
288  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
289  * task), SP is the stack pointer of the first frame that should be shown in the back
290  * trace (or NULL if the entire call-chain of the task should be shown).
291  */
292 extern void show_stack(struct task_struct *task, unsigned long *sp);
293 
294 void io_schedule(void);
295 long io_schedule_timeout(long timeout);
296 
297 extern void cpu_init (void);
298 extern void trap_init(void);
299 extern void update_process_times(int user);
300 extern void scheduler_tick(void);
301 
302 extern void sched_show_task(struct task_struct *p);
303 
304 #ifdef CONFIG_DETECT_SOFTLOCKUP
305 extern void softlockup_tick(void);
306 extern void touch_softlockup_watchdog(void);
307 extern void touch_all_softlockup_watchdogs(void);
308 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
309 				    void __user *buffer,
310 				    size_t *lenp, loff_t *ppos);
311 extern unsigned int  softlockup_panic;
312 extern int softlockup_thresh;
313 #else
314 static inline void softlockup_tick(void)
315 {
316 }
317 static inline void touch_softlockup_watchdog(void)
318 {
319 }
320 static inline void touch_all_softlockup_watchdogs(void)
321 {
322 }
323 #endif
324 
325 #ifdef CONFIG_DETECT_HUNG_TASK
326 extern unsigned int  sysctl_hung_task_panic;
327 extern unsigned long sysctl_hung_task_check_count;
328 extern unsigned long sysctl_hung_task_timeout_secs;
329 extern unsigned long sysctl_hung_task_warnings;
330 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
331 					 void __user *buffer,
332 					 size_t *lenp, loff_t *ppos);
333 #endif
334 
335 /* Attach to any functions which should be ignored in wchan output. */
336 #define __sched		__attribute__((__section__(".sched.text")))
337 
338 /* Linker adds these: start and end of __sched functions */
339 extern char __sched_text_start[], __sched_text_end[];
340 
341 /* Is this address in the __sched functions? */
342 extern int in_sched_functions(unsigned long addr);
343 
344 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
345 extern signed long schedule_timeout(signed long timeout);
346 extern signed long schedule_timeout_interruptible(signed long timeout);
347 extern signed long schedule_timeout_killable(signed long timeout);
348 extern signed long schedule_timeout_uninterruptible(signed long timeout);
349 asmlinkage void schedule(void);
350 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
351 
352 struct nsproxy;
353 struct user_namespace;
354 
355 /*
356  * Default maximum number of active map areas, this limits the number of vmas
357  * per mm struct. Users can overwrite this number by sysctl but there is a
358  * problem.
359  *
360  * When a program's coredump is generated as ELF format, a section is created
361  * per a vma. In ELF, the number of sections is represented in unsigned short.
362  * This means the number of sections should be smaller than 65535 at coredump.
363  * Because the kernel adds some informative sections to a image of program at
364  * generating coredump, we need some margin. The number of extra sections is
365  * 1-3 now and depends on arch. We use "5" as safe margin, here.
366  */
367 #define MAPCOUNT_ELF_CORE_MARGIN	(5)
368 #define DEFAULT_MAX_MAP_COUNT	(USHORT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
369 
370 extern int sysctl_max_map_count;
371 
372 #include <linux/aio.h>
373 
374 extern unsigned long
375 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
376 		       unsigned long, unsigned long);
377 extern unsigned long
378 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
379 			  unsigned long len, unsigned long pgoff,
380 			  unsigned long flags);
381 extern void arch_unmap_area(struct mm_struct *, unsigned long);
382 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
383 
384 #if USE_SPLIT_PTLOCKS
385 /*
386  * The mm counters are not protected by its page_table_lock,
387  * so must be incremented atomically.
388  */
389 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
390 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
391 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
392 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
393 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
394 
395 #else  /* !USE_SPLIT_PTLOCKS */
396 /*
397  * The mm counters are protected by its page_table_lock,
398  * so can be incremented directly.
399  */
400 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
401 #define get_mm_counter(mm, member) ((mm)->_##member)
402 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
403 #define inc_mm_counter(mm, member) (mm)->_##member++
404 #define dec_mm_counter(mm, member) (mm)->_##member--
405 
406 #endif /* !USE_SPLIT_PTLOCKS */
407 
408 #define get_mm_rss(mm)					\
409 	(get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
410 #define update_hiwater_rss(mm)	do {			\
411 	unsigned long _rss = get_mm_rss(mm);		\
412 	if ((mm)->hiwater_rss < _rss)			\
413 		(mm)->hiwater_rss = _rss;		\
414 } while (0)
415 #define update_hiwater_vm(mm)	do {			\
416 	if ((mm)->hiwater_vm < (mm)->total_vm)		\
417 		(mm)->hiwater_vm = (mm)->total_vm;	\
418 } while (0)
419 
420 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
421 {
422 	return max(mm->hiwater_rss, get_mm_rss(mm));
423 }
424 
425 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
426 					 struct mm_struct *mm)
427 {
428 	unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
429 
430 	if (*maxrss < hiwater_rss)
431 		*maxrss = hiwater_rss;
432 }
433 
434 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
435 {
436 	return max(mm->hiwater_vm, mm->total_vm);
437 }
438 
439 extern void set_dumpable(struct mm_struct *mm, int value);
440 extern int get_dumpable(struct mm_struct *mm);
441 
442 /* mm flags */
443 /* dumpable bits */
444 #define MMF_DUMPABLE      0  /* core dump is permitted */
445 #define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
446 
447 #define MMF_DUMPABLE_BITS 2
448 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
449 
450 /* coredump filter bits */
451 #define MMF_DUMP_ANON_PRIVATE	2
452 #define MMF_DUMP_ANON_SHARED	3
453 #define MMF_DUMP_MAPPED_PRIVATE	4
454 #define MMF_DUMP_MAPPED_SHARED	5
455 #define MMF_DUMP_ELF_HEADERS	6
456 #define MMF_DUMP_HUGETLB_PRIVATE 7
457 #define MMF_DUMP_HUGETLB_SHARED  8
458 
459 #define MMF_DUMP_FILTER_SHIFT	MMF_DUMPABLE_BITS
460 #define MMF_DUMP_FILTER_BITS	7
461 #define MMF_DUMP_FILTER_MASK \
462 	(((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
463 #define MMF_DUMP_FILTER_DEFAULT \
464 	((1 << MMF_DUMP_ANON_PRIVATE) |	(1 << MMF_DUMP_ANON_SHARED) |\
465 	 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
466 
467 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
468 # define MMF_DUMP_MASK_DEFAULT_ELF	(1 << MMF_DUMP_ELF_HEADERS)
469 #else
470 # define MMF_DUMP_MASK_DEFAULT_ELF	0
471 #endif
472 					/* leave room for more dump flags */
473 #define MMF_VM_MERGEABLE	16	/* KSM may merge identical pages */
474 
475 #define MMF_INIT_MASK		(MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
476 
477 struct sighand_struct {
478 	atomic_t		count;
479 	struct k_sigaction	action[_NSIG];
480 	spinlock_t		siglock;
481 	wait_queue_head_t	signalfd_wqh;
482 };
483 
484 struct pacct_struct {
485 	int			ac_flag;
486 	long			ac_exitcode;
487 	unsigned long		ac_mem;
488 	cputime_t		ac_utime, ac_stime;
489 	unsigned long		ac_minflt, ac_majflt;
490 };
491 
492 struct cpu_itimer {
493 	cputime_t expires;
494 	cputime_t incr;
495 	u32 error;
496 	u32 incr_error;
497 };
498 
499 /**
500  * struct task_cputime - collected CPU time counts
501  * @utime:		time spent in user mode, in &cputime_t units
502  * @stime:		time spent in kernel mode, in &cputime_t units
503  * @sum_exec_runtime:	total time spent on the CPU, in nanoseconds
504  *
505  * This structure groups together three kinds of CPU time that are
506  * tracked for threads and thread groups.  Most things considering
507  * CPU time want to group these counts together and treat all three
508  * of them in parallel.
509  */
510 struct task_cputime {
511 	cputime_t utime;
512 	cputime_t stime;
513 	unsigned long long sum_exec_runtime;
514 };
515 /* Alternate field names when used to cache expirations. */
516 #define prof_exp	stime
517 #define virt_exp	utime
518 #define sched_exp	sum_exec_runtime
519 
520 #define INIT_CPUTIME	\
521 	(struct task_cputime) {					\
522 		.utime = cputime_zero,				\
523 		.stime = cputime_zero,				\
524 		.sum_exec_runtime = 0,				\
525 	}
526 
527 /*
528  * Disable preemption until the scheduler is running.
529  * Reset by start_kernel()->sched_init()->init_idle().
530  *
531  * We include PREEMPT_ACTIVE to avoid cond_resched() from working
532  * before the scheduler is active -- see should_resched().
533  */
534 #define INIT_PREEMPT_COUNT	(1 + PREEMPT_ACTIVE)
535 
536 /**
537  * struct thread_group_cputimer - thread group interval timer counts
538  * @cputime:		thread group interval timers.
539  * @running:		non-zero when there are timers running and
540  * 			@cputime receives updates.
541  * @lock:		lock for fields in this struct.
542  *
543  * This structure contains the version of task_cputime, above, that is
544  * used for thread group CPU timer calculations.
545  */
546 struct thread_group_cputimer {
547 	struct task_cputime cputime;
548 	int running;
549 	spinlock_t lock;
550 };
551 
552 /*
553  * NOTE! "signal_struct" does not have it's own
554  * locking, because a shared signal_struct always
555  * implies a shared sighand_struct, so locking
556  * sighand_struct is always a proper superset of
557  * the locking of signal_struct.
558  */
559 struct signal_struct {
560 	atomic_t		count;
561 	atomic_t		live;
562 
563 	wait_queue_head_t	wait_chldexit;	/* for wait4() */
564 
565 	/* current thread group signal load-balancing target: */
566 	struct task_struct	*curr_target;
567 
568 	/* shared signal handling: */
569 	struct sigpending	shared_pending;
570 
571 	/* thread group exit support */
572 	int			group_exit_code;
573 	/* overloaded:
574 	 * - notify group_exit_task when ->count is equal to notify_count
575 	 * - everyone except group_exit_task is stopped during signal delivery
576 	 *   of fatal signals, group_exit_task processes the signal.
577 	 */
578 	int			notify_count;
579 	struct task_struct	*group_exit_task;
580 
581 	/* thread group stop support, overloads group_exit_code too */
582 	int			group_stop_count;
583 	unsigned int		flags; /* see SIGNAL_* flags below */
584 
585 	/* POSIX.1b Interval Timers */
586 	struct list_head posix_timers;
587 
588 	/* ITIMER_REAL timer for the process */
589 	struct hrtimer real_timer;
590 	struct pid *leader_pid;
591 	ktime_t it_real_incr;
592 
593 	/*
594 	 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
595 	 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
596 	 * values are defined to 0 and 1 respectively
597 	 */
598 	struct cpu_itimer it[2];
599 
600 	/*
601 	 * Thread group totals for process CPU timers.
602 	 * See thread_group_cputimer(), et al, for details.
603 	 */
604 	struct thread_group_cputimer cputimer;
605 
606 	/* Earliest-expiration cache. */
607 	struct task_cputime cputime_expires;
608 
609 	struct list_head cpu_timers[3];
610 
611 	struct pid *tty_old_pgrp;
612 
613 	/* boolean value for session group leader */
614 	int leader;
615 
616 	struct tty_struct *tty; /* NULL if no tty */
617 
618 	/*
619 	 * Cumulative resource counters for dead threads in the group,
620 	 * and for reaped dead child processes forked by this group.
621 	 * Live threads maintain their own counters and add to these
622 	 * in __exit_signal, except for the group leader.
623 	 */
624 	cputime_t utime, stime, cutime, cstime;
625 	cputime_t gtime;
626 	cputime_t cgtime;
627 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
628 	cputime_t prev_utime, prev_stime;
629 #endif
630 	unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
631 	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
632 	unsigned long inblock, oublock, cinblock, coublock;
633 	unsigned long maxrss, cmaxrss;
634 	struct task_io_accounting ioac;
635 
636 	/*
637 	 * Cumulative ns of schedule CPU time fo dead threads in the
638 	 * group, not including a zombie group leader, (This only differs
639 	 * from jiffies_to_ns(utime + stime) if sched_clock uses something
640 	 * other than jiffies.)
641 	 */
642 	unsigned long long sum_sched_runtime;
643 
644 	/*
645 	 * We don't bother to synchronize most readers of this at all,
646 	 * because there is no reader checking a limit that actually needs
647 	 * to get both rlim_cur and rlim_max atomically, and either one
648 	 * alone is a single word that can safely be read normally.
649 	 * getrlimit/setrlimit use task_lock(current->group_leader) to
650 	 * protect this instead of the siglock, because they really
651 	 * have no need to disable irqs.
652 	 */
653 	struct rlimit rlim[RLIM_NLIMITS];
654 
655 #ifdef CONFIG_BSD_PROCESS_ACCT
656 	struct pacct_struct pacct;	/* per-process accounting information */
657 #endif
658 #ifdef CONFIG_TASKSTATS
659 	struct taskstats *stats;
660 #endif
661 #ifdef CONFIG_AUDIT
662 	unsigned audit_tty;
663 	struct tty_audit_buf *tty_audit_buf;
664 #endif
665 
666 	int oom_adj;	/* OOM kill score adjustment (bit shift) */
667 };
668 
669 /* Context switch must be unlocked if interrupts are to be enabled */
670 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
671 # define __ARCH_WANT_UNLOCKED_CTXSW
672 #endif
673 
674 /*
675  * Bits in flags field of signal_struct.
676  */
677 #define SIGNAL_STOP_STOPPED	0x00000001 /* job control stop in effect */
678 #define SIGNAL_STOP_DEQUEUED	0x00000002 /* stop signal dequeued */
679 #define SIGNAL_STOP_CONTINUED	0x00000004 /* SIGCONT since WCONTINUED reap */
680 #define SIGNAL_GROUP_EXIT	0x00000008 /* group exit in progress */
681 /*
682  * Pending notifications to parent.
683  */
684 #define SIGNAL_CLD_STOPPED	0x00000010
685 #define SIGNAL_CLD_CONTINUED	0x00000020
686 #define SIGNAL_CLD_MASK		(SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
687 
688 #define SIGNAL_UNKILLABLE	0x00000040 /* for init: ignore fatal signals */
689 
690 /* If true, all threads except ->group_exit_task have pending SIGKILL */
691 static inline int signal_group_exit(const struct signal_struct *sig)
692 {
693 	return	(sig->flags & SIGNAL_GROUP_EXIT) ||
694 		(sig->group_exit_task != NULL);
695 }
696 
697 /*
698  * Some day this will be a full-fledged user tracking system..
699  */
700 struct user_struct {
701 	atomic_t __count;	/* reference count */
702 	atomic_t processes;	/* How many processes does this user have? */
703 	atomic_t files;		/* How many open files does this user have? */
704 	atomic_t sigpending;	/* How many pending signals does this user have? */
705 #ifdef CONFIG_INOTIFY_USER
706 	atomic_t inotify_watches; /* How many inotify watches does this user have? */
707 	atomic_t inotify_devs;	/* How many inotify devs does this user have opened? */
708 #endif
709 #ifdef CONFIG_EPOLL
710 	atomic_t epoll_watches;	/* The number of file descriptors currently watched */
711 #endif
712 #ifdef CONFIG_POSIX_MQUEUE
713 	/* protected by mq_lock	*/
714 	unsigned long mq_bytes;	/* How many bytes can be allocated to mqueue? */
715 #endif
716 	unsigned long locked_shm; /* How many pages of mlocked shm ? */
717 
718 #ifdef CONFIG_KEYS
719 	struct key *uid_keyring;	/* UID specific keyring */
720 	struct key *session_keyring;	/* UID's default session keyring */
721 #endif
722 
723 	/* Hash table maintenance information */
724 	struct hlist_node uidhash_node;
725 	uid_t uid;
726 	struct user_namespace *user_ns;
727 
728 #ifdef CONFIG_USER_SCHED
729 	struct task_group *tg;
730 #ifdef CONFIG_SYSFS
731 	struct kobject kobj;
732 	struct delayed_work work;
733 #endif
734 #endif
735 
736 #ifdef CONFIG_PERF_EVENTS
737 	atomic_long_t locked_vm;
738 #endif
739 };
740 
741 extern int uids_sysfs_init(void);
742 
743 extern struct user_struct *find_user(uid_t);
744 
745 extern struct user_struct root_user;
746 #define INIT_USER (&root_user)
747 
748 
749 struct backing_dev_info;
750 struct reclaim_state;
751 
752 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
753 struct sched_info {
754 	/* cumulative counters */
755 	unsigned long pcount;	      /* # of times run on this cpu */
756 	unsigned long long run_delay; /* time spent waiting on a runqueue */
757 
758 	/* timestamps */
759 	unsigned long long last_arrival,/* when we last ran on a cpu */
760 			   last_queued;	/* when we were last queued to run */
761 #ifdef CONFIG_SCHEDSTATS
762 	/* BKL stats */
763 	unsigned int bkl_count;
764 #endif
765 };
766 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
767 
768 #ifdef CONFIG_TASK_DELAY_ACCT
769 struct task_delay_info {
770 	spinlock_t	lock;
771 	unsigned int	flags;	/* Private per-task flags */
772 
773 	/* For each stat XXX, add following, aligned appropriately
774 	 *
775 	 * struct timespec XXX_start, XXX_end;
776 	 * u64 XXX_delay;
777 	 * u32 XXX_count;
778 	 *
779 	 * Atomicity of updates to XXX_delay, XXX_count protected by
780 	 * single lock above (split into XXX_lock if contention is an issue).
781 	 */
782 
783 	/*
784 	 * XXX_count is incremented on every XXX operation, the delay
785 	 * associated with the operation is added to XXX_delay.
786 	 * XXX_delay contains the accumulated delay time in nanoseconds.
787 	 */
788 	struct timespec blkio_start, blkio_end;	/* Shared by blkio, swapin */
789 	u64 blkio_delay;	/* wait for sync block io completion */
790 	u64 swapin_delay;	/* wait for swapin block io completion */
791 	u32 blkio_count;	/* total count of the number of sync block */
792 				/* io operations performed */
793 	u32 swapin_count;	/* total count of the number of swapin block */
794 				/* io operations performed */
795 
796 	struct timespec freepages_start, freepages_end;
797 	u64 freepages_delay;	/* wait for memory reclaim */
798 	u32 freepages_count;	/* total count of memory reclaim */
799 };
800 #endif	/* CONFIG_TASK_DELAY_ACCT */
801 
802 static inline int sched_info_on(void)
803 {
804 #ifdef CONFIG_SCHEDSTATS
805 	return 1;
806 #elif defined(CONFIG_TASK_DELAY_ACCT)
807 	extern int delayacct_on;
808 	return delayacct_on;
809 #else
810 	return 0;
811 #endif
812 }
813 
814 enum cpu_idle_type {
815 	CPU_IDLE,
816 	CPU_NOT_IDLE,
817 	CPU_NEWLY_IDLE,
818 	CPU_MAX_IDLE_TYPES
819 };
820 
821 /*
822  * sched-domains (multiprocessor balancing) declarations:
823  */
824 
825 /*
826  * Increase resolution of nice-level calculations:
827  */
828 #define SCHED_LOAD_SHIFT	10
829 #define SCHED_LOAD_SCALE	(1L << SCHED_LOAD_SHIFT)
830 
831 #define SCHED_LOAD_SCALE_FUZZ	SCHED_LOAD_SCALE
832 
833 #ifdef CONFIG_SMP
834 #define SD_LOAD_BALANCE		0x0001	/* Do load balancing on this domain. */
835 #define SD_BALANCE_NEWIDLE	0x0002	/* Balance when about to become idle */
836 #define SD_BALANCE_EXEC		0x0004	/* Balance on exec */
837 #define SD_BALANCE_FORK		0x0008	/* Balance on fork, clone */
838 #define SD_BALANCE_WAKE		0x0010  /* Balance on wakeup */
839 #define SD_WAKE_AFFINE		0x0020	/* Wake task to waking CPU */
840 #define SD_PREFER_LOCAL		0x0040  /* Prefer to keep tasks local to this domain */
841 #define SD_SHARE_CPUPOWER	0x0080	/* Domain members share cpu power */
842 #define SD_POWERSAVINGS_BALANCE	0x0100	/* Balance for power savings */
843 #define SD_SHARE_PKG_RESOURCES	0x0200	/* Domain members share cpu pkg resources */
844 #define SD_SERIALIZE		0x0400	/* Only a single load balancing instance */
845 
846 #define SD_PREFER_SIBLING	0x1000	/* Prefer to place tasks in a sibling domain */
847 
848 enum powersavings_balance_level {
849 	POWERSAVINGS_BALANCE_NONE = 0,  /* No power saving load balance */
850 	POWERSAVINGS_BALANCE_BASIC,	/* Fill one thread/core/package
851 					 * first for long running threads
852 					 */
853 	POWERSAVINGS_BALANCE_WAKEUP,	/* Also bias task wakeups to semi-idle
854 					 * cpu package for power savings
855 					 */
856 	MAX_POWERSAVINGS_BALANCE_LEVELS
857 };
858 
859 extern int sched_mc_power_savings, sched_smt_power_savings;
860 
861 static inline int sd_balance_for_mc_power(void)
862 {
863 	if (sched_smt_power_savings)
864 		return SD_POWERSAVINGS_BALANCE;
865 
866 	return SD_PREFER_SIBLING;
867 }
868 
869 static inline int sd_balance_for_package_power(void)
870 {
871 	if (sched_mc_power_savings | sched_smt_power_savings)
872 		return SD_POWERSAVINGS_BALANCE;
873 
874 	return SD_PREFER_SIBLING;
875 }
876 
877 /*
878  * Optimise SD flags for power savings:
879  * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
880  * Keep default SD flags if sched_{smt,mc}_power_saving=0
881  */
882 
883 static inline int sd_power_saving_flags(void)
884 {
885 	if (sched_mc_power_savings | sched_smt_power_savings)
886 		return SD_BALANCE_NEWIDLE;
887 
888 	return 0;
889 }
890 
891 struct sched_group {
892 	struct sched_group *next;	/* Must be a circular list */
893 
894 	/*
895 	 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
896 	 * single CPU.
897 	 */
898 	unsigned int cpu_power;
899 
900 	/*
901 	 * The CPUs this group covers.
902 	 *
903 	 * NOTE: this field is variable length. (Allocated dynamically
904 	 * by attaching extra space to the end of the structure,
905 	 * depending on how many CPUs the kernel has booted up with)
906 	 *
907 	 * It is also be embedded into static data structures at build
908 	 * time. (See 'struct static_sched_group' in kernel/sched.c)
909 	 */
910 	unsigned long cpumask[0];
911 };
912 
913 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
914 {
915 	return to_cpumask(sg->cpumask);
916 }
917 
918 enum sched_domain_level {
919 	SD_LV_NONE = 0,
920 	SD_LV_SIBLING,
921 	SD_LV_MC,
922 	SD_LV_CPU,
923 	SD_LV_NODE,
924 	SD_LV_ALLNODES,
925 	SD_LV_MAX
926 };
927 
928 struct sched_domain_attr {
929 	int relax_domain_level;
930 };
931 
932 #define SD_ATTR_INIT	(struct sched_domain_attr) {	\
933 	.relax_domain_level = -1,			\
934 }
935 
936 struct sched_domain {
937 	/* These fields must be setup */
938 	struct sched_domain *parent;	/* top domain must be null terminated */
939 	struct sched_domain *child;	/* bottom domain must be null terminated */
940 	struct sched_group *groups;	/* the balancing groups of the domain */
941 	unsigned long min_interval;	/* Minimum balance interval ms */
942 	unsigned long max_interval;	/* Maximum balance interval ms */
943 	unsigned int busy_factor;	/* less balancing by factor if busy */
944 	unsigned int imbalance_pct;	/* No balance until over watermark */
945 	unsigned int cache_nice_tries;	/* Leave cache hot tasks for # tries */
946 	unsigned int busy_idx;
947 	unsigned int idle_idx;
948 	unsigned int newidle_idx;
949 	unsigned int wake_idx;
950 	unsigned int forkexec_idx;
951 	unsigned int smt_gain;
952 	int flags;			/* See SD_* */
953 	enum sched_domain_level level;
954 
955 	/* Runtime fields. */
956 	unsigned long last_balance;	/* init to jiffies. units in jiffies */
957 	unsigned int balance_interval;	/* initialise to 1. units in ms. */
958 	unsigned int nr_balance_failed; /* initialise to 0 */
959 
960 	u64 last_update;
961 
962 #ifdef CONFIG_SCHEDSTATS
963 	/* load_balance() stats */
964 	unsigned int lb_count[CPU_MAX_IDLE_TYPES];
965 	unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
966 	unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
967 	unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
968 	unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
969 	unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
970 	unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
971 	unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
972 
973 	/* Active load balancing */
974 	unsigned int alb_count;
975 	unsigned int alb_failed;
976 	unsigned int alb_pushed;
977 
978 	/* SD_BALANCE_EXEC stats */
979 	unsigned int sbe_count;
980 	unsigned int sbe_balanced;
981 	unsigned int sbe_pushed;
982 
983 	/* SD_BALANCE_FORK stats */
984 	unsigned int sbf_count;
985 	unsigned int sbf_balanced;
986 	unsigned int sbf_pushed;
987 
988 	/* try_to_wake_up() stats */
989 	unsigned int ttwu_wake_remote;
990 	unsigned int ttwu_move_affine;
991 	unsigned int ttwu_move_balance;
992 #endif
993 #ifdef CONFIG_SCHED_DEBUG
994 	char *name;
995 #endif
996 
997 	/*
998 	 * Span of all CPUs in this domain.
999 	 *
1000 	 * NOTE: this field is variable length. (Allocated dynamically
1001 	 * by attaching extra space to the end of the structure,
1002 	 * depending on how many CPUs the kernel has booted up with)
1003 	 *
1004 	 * It is also be embedded into static data structures at build
1005 	 * time. (See 'struct static_sched_domain' in kernel/sched.c)
1006 	 */
1007 	unsigned long span[0];
1008 };
1009 
1010 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1011 {
1012 	return to_cpumask(sd->span);
1013 }
1014 
1015 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1016 				    struct sched_domain_attr *dattr_new);
1017 
1018 /* Allocate an array of sched domains, for partition_sched_domains(). */
1019 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1020 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1021 
1022 /* Test a flag in parent sched domain */
1023 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1024 {
1025 	if (sd->parent && (sd->parent->flags & flag))
1026 		return 1;
1027 
1028 	return 0;
1029 }
1030 
1031 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1032 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1033 
1034 #else /* CONFIG_SMP */
1035 
1036 struct sched_domain_attr;
1037 
1038 static inline void
1039 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1040 			struct sched_domain_attr *dattr_new)
1041 {
1042 }
1043 #endif	/* !CONFIG_SMP */
1044 
1045 
1046 struct io_context;			/* See blkdev.h */
1047 
1048 
1049 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1050 extern void prefetch_stack(struct task_struct *t);
1051 #else
1052 static inline void prefetch_stack(struct task_struct *t) { }
1053 #endif
1054 
1055 struct audit_context;		/* See audit.c */
1056 struct mempolicy;
1057 struct pipe_inode_info;
1058 struct uts_namespace;
1059 
1060 struct rq;
1061 struct sched_domain;
1062 
1063 /*
1064  * wake flags
1065  */
1066 #define WF_SYNC		0x01		/* waker goes to sleep after wakup */
1067 #define WF_FORK		0x02		/* child wakeup after fork */
1068 
1069 struct sched_class {
1070 	const struct sched_class *next;
1071 
1072 	void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
1073 	void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
1074 	void (*yield_task) (struct rq *rq);
1075 
1076 	void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1077 
1078 	struct task_struct * (*pick_next_task) (struct rq *rq);
1079 	void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1080 
1081 #ifdef CONFIG_SMP
1082 	int  (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1083 
1084 	unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
1085 			struct rq *busiest, unsigned long max_load_move,
1086 			struct sched_domain *sd, enum cpu_idle_type idle,
1087 			int *all_pinned, int *this_best_prio);
1088 
1089 	int (*move_one_task) (struct rq *this_rq, int this_cpu,
1090 			      struct rq *busiest, struct sched_domain *sd,
1091 			      enum cpu_idle_type idle);
1092 	void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1093 	void (*post_schedule) (struct rq *this_rq);
1094 	void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
1095 
1096 	void (*set_cpus_allowed)(struct task_struct *p,
1097 				 const struct cpumask *newmask);
1098 
1099 	void (*rq_online)(struct rq *rq);
1100 	void (*rq_offline)(struct rq *rq);
1101 #endif
1102 
1103 	void (*set_curr_task) (struct rq *rq);
1104 	void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1105 	void (*task_new) (struct rq *rq, struct task_struct *p);
1106 
1107 	void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1108 			       int running);
1109 	void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1110 			     int running);
1111 	void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1112 			     int oldprio, int running);
1113 
1114 	unsigned int (*get_rr_interval) (struct task_struct *task);
1115 
1116 #ifdef CONFIG_FAIR_GROUP_SCHED
1117 	void (*moved_group) (struct task_struct *p);
1118 #endif
1119 };
1120 
1121 struct load_weight {
1122 	unsigned long weight, inv_weight;
1123 };
1124 
1125 /*
1126  * CFS stats for a schedulable entity (task, task-group etc)
1127  *
1128  * Current field usage histogram:
1129  *
1130  *     4 se->block_start
1131  *     4 se->run_node
1132  *     4 se->sleep_start
1133  *     6 se->load.weight
1134  */
1135 struct sched_entity {
1136 	struct load_weight	load;		/* for load-balancing */
1137 	struct rb_node		run_node;
1138 	struct list_head	group_node;
1139 	unsigned int		on_rq;
1140 
1141 	u64			exec_start;
1142 	u64			sum_exec_runtime;
1143 	u64			vruntime;
1144 	u64			prev_sum_exec_runtime;
1145 
1146 	u64			last_wakeup;
1147 	u64			avg_overlap;
1148 
1149 	u64			nr_migrations;
1150 
1151 	u64			start_runtime;
1152 	u64			avg_wakeup;
1153 
1154 	u64			avg_running;
1155 
1156 #ifdef CONFIG_SCHEDSTATS
1157 	u64			wait_start;
1158 	u64			wait_max;
1159 	u64			wait_count;
1160 	u64			wait_sum;
1161 	u64			iowait_count;
1162 	u64			iowait_sum;
1163 
1164 	u64			sleep_start;
1165 	u64			sleep_max;
1166 	s64			sum_sleep_runtime;
1167 
1168 	u64			block_start;
1169 	u64			block_max;
1170 	u64			exec_max;
1171 	u64			slice_max;
1172 
1173 	u64			nr_migrations_cold;
1174 	u64			nr_failed_migrations_affine;
1175 	u64			nr_failed_migrations_running;
1176 	u64			nr_failed_migrations_hot;
1177 	u64			nr_forced_migrations;
1178 	u64			nr_forced2_migrations;
1179 
1180 	u64			nr_wakeups;
1181 	u64			nr_wakeups_sync;
1182 	u64			nr_wakeups_migrate;
1183 	u64			nr_wakeups_local;
1184 	u64			nr_wakeups_remote;
1185 	u64			nr_wakeups_affine;
1186 	u64			nr_wakeups_affine_attempts;
1187 	u64			nr_wakeups_passive;
1188 	u64			nr_wakeups_idle;
1189 #endif
1190 
1191 #ifdef CONFIG_FAIR_GROUP_SCHED
1192 	struct sched_entity	*parent;
1193 	/* rq on which this entity is (to be) queued: */
1194 	struct cfs_rq		*cfs_rq;
1195 	/* rq "owned" by this entity/group: */
1196 	struct cfs_rq		*my_q;
1197 #endif
1198 };
1199 
1200 struct sched_rt_entity {
1201 	struct list_head run_list;
1202 	unsigned long timeout;
1203 	unsigned int time_slice;
1204 	int nr_cpus_allowed;
1205 
1206 	struct sched_rt_entity *back;
1207 #ifdef CONFIG_RT_GROUP_SCHED
1208 	struct sched_rt_entity	*parent;
1209 	/* rq on which this entity is (to be) queued: */
1210 	struct rt_rq		*rt_rq;
1211 	/* rq "owned" by this entity/group: */
1212 	struct rt_rq		*my_q;
1213 #endif
1214 };
1215 
1216 struct rcu_node;
1217 
1218 struct task_struct {
1219 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
1220 	void *stack;
1221 	atomic_t usage;
1222 	unsigned int flags;	/* per process flags, defined below */
1223 	unsigned int ptrace;
1224 
1225 	int lock_depth;		/* BKL lock depth */
1226 
1227 #ifdef CONFIG_SMP
1228 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1229 	int oncpu;
1230 #endif
1231 #endif
1232 
1233 	int prio, static_prio, normal_prio;
1234 	unsigned int rt_priority;
1235 	const struct sched_class *sched_class;
1236 	struct sched_entity se;
1237 	struct sched_rt_entity rt;
1238 
1239 #ifdef CONFIG_PREEMPT_NOTIFIERS
1240 	/* list of struct preempt_notifier: */
1241 	struct hlist_head preempt_notifiers;
1242 #endif
1243 
1244 	/*
1245 	 * fpu_counter contains the number of consecutive context switches
1246 	 * that the FPU is used. If this is over a threshold, the lazy fpu
1247 	 * saving becomes unlazy to save the trap. This is an unsigned char
1248 	 * so that after 256 times the counter wraps and the behavior turns
1249 	 * lazy again; this to deal with bursty apps that only use FPU for
1250 	 * a short time
1251 	 */
1252 	unsigned char fpu_counter;
1253 #ifdef CONFIG_BLK_DEV_IO_TRACE
1254 	unsigned int btrace_seq;
1255 #endif
1256 
1257 	unsigned int policy;
1258 	cpumask_t cpus_allowed;
1259 
1260 #ifdef CONFIG_TREE_PREEMPT_RCU
1261 	int rcu_read_lock_nesting;
1262 	char rcu_read_unlock_special;
1263 	struct rcu_node *rcu_blocked_node;
1264 	struct list_head rcu_node_entry;
1265 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1266 
1267 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1268 	struct sched_info sched_info;
1269 #endif
1270 
1271 	struct list_head tasks;
1272 	struct plist_node pushable_tasks;
1273 
1274 	struct mm_struct *mm, *active_mm;
1275 
1276 /* task state */
1277 	int exit_state;
1278 	int exit_code, exit_signal;
1279 	int pdeath_signal;  /*  The signal sent when the parent dies  */
1280 	/* ??? */
1281 	unsigned int personality;
1282 	unsigned did_exec:1;
1283 	unsigned in_execve:1;	/* Tell the LSMs that the process is doing an
1284 				 * execve */
1285 	unsigned in_iowait:1;
1286 
1287 
1288 	/* Revert to default priority/policy when forking */
1289 	unsigned sched_reset_on_fork:1;
1290 
1291 	pid_t pid;
1292 	pid_t tgid;
1293 
1294 #ifdef CONFIG_CC_STACKPROTECTOR
1295 	/* Canary value for the -fstack-protector gcc feature */
1296 	unsigned long stack_canary;
1297 #endif
1298 
1299 	/*
1300 	 * pointers to (original) parent process, youngest child, younger sibling,
1301 	 * older sibling, respectively.  (p->father can be replaced with
1302 	 * p->real_parent->pid)
1303 	 */
1304 	struct task_struct *real_parent; /* real parent process */
1305 	struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1306 	/*
1307 	 * children/sibling forms the list of my natural children
1308 	 */
1309 	struct list_head children;	/* list of my children */
1310 	struct list_head sibling;	/* linkage in my parent's children list */
1311 	struct task_struct *group_leader;	/* threadgroup leader */
1312 
1313 	/*
1314 	 * ptraced is the list of tasks this task is using ptrace on.
1315 	 * This includes both natural children and PTRACE_ATTACH targets.
1316 	 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1317 	 */
1318 	struct list_head ptraced;
1319 	struct list_head ptrace_entry;
1320 
1321 	/*
1322 	 * This is the tracer handle for the ptrace BTS extension.
1323 	 * This field actually belongs to the ptracer task.
1324 	 */
1325 	struct bts_context *bts;
1326 
1327 	/* PID/PID hash table linkage. */
1328 	struct pid_link pids[PIDTYPE_MAX];
1329 	struct list_head thread_group;
1330 
1331 	struct completion *vfork_done;		/* for vfork() */
1332 	int __user *set_child_tid;		/* CLONE_CHILD_SETTID */
1333 	int __user *clear_child_tid;		/* CLONE_CHILD_CLEARTID */
1334 
1335 	cputime_t utime, stime, utimescaled, stimescaled;
1336 	cputime_t gtime;
1337 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1338 	cputime_t prev_utime, prev_stime;
1339 #endif
1340 	unsigned long nvcsw, nivcsw; /* context switch counts */
1341 	struct timespec start_time; 		/* monotonic time */
1342 	struct timespec real_start_time;	/* boot based time */
1343 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1344 	unsigned long min_flt, maj_flt;
1345 
1346 	struct task_cputime cputime_expires;
1347 	struct list_head cpu_timers[3];
1348 
1349 /* process credentials */
1350 	const struct cred *real_cred;	/* objective and real subjective task
1351 					 * credentials (COW) */
1352 	const struct cred *cred;	/* effective (overridable) subjective task
1353 					 * credentials (COW) */
1354 	struct mutex cred_guard_mutex;	/* guard against foreign influences on
1355 					 * credential calculations
1356 					 * (notably. ptrace) */
1357 	struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1358 
1359 	char comm[TASK_COMM_LEN]; /* executable name excluding path
1360 				     - access with [gs]et_task_comm (which lock
1361 				       it with task_lock())
1362 				     - initialized normally by flush_old_exec */
1363 /* file system info */
1364 	int link_count, total_link_count;
1365 #ifdef CONFIG_SYSVIPC
1366 /* ipc stuff */
1367 	struct sysv_sem sysvsem;
1368 #endif
1369 #ifdef CONFIG_DETECT_HUNG_TASK
1370 /* hung task detection */
1371 	unsigned long last_switch_count;
1372 #endif
1373 /* CPU-specific state of this task */
1374 	struct thread_struct thread;
1375 /* filesystem information */
1376 	struct fs_struct *fs;
1377 /* open file information */
1378 	struct files_struct *files;
1379 /* namespaces */
1380 	struct nsproxy *nsproxy;
1381 /* signal handlers */
1382 	struct signal_struct *signal;
1383 	struct sighand_struct *sighand;
1384 
1385 	sigset_t blocked, real_blocked;
1386 	sigset_t saved_sigmask;	/* restored if set_restore_sigmask() was used */
1387 	struct sigpending pending;
1388 
1389 	unsigned long sas_ss_sp;
1390 	size_t sas_ss_size;
1391 	int (*notifier)(void *priv);
1392 	void *notifier_data;
1393 	sigset_t *notifier_mask;
1394 	struct audit_context *audit_context;
1395 #ifdef CONFIG_AUDITSYSCALL
1396 	uid_t loginuid;
1397 	unsigned int sessionid;
1398 #endif
1399 	seccomp_t seccomp;
1400 
1401 /* Thread group tracking */
1402    	u32 parent_exec_id;
1403    	u32 self_exec_id;
1404 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1405  * mempolicy */
1406 	spinlock_t alloc_lock;
1407 
1408 #ifdef CONFIG_GENERIC_HARDIRQS
1409 	/* IRQ handler threads */
1410 	struct irqaction *irqaction;
1411 #endif
1412 
1413 	/* Protection of the PI data structures: */
1414 	spinlock_t pi_lock;
1415 
1416 #ifdef CONFIG_RT_MUTEXES
1417 	/* PI waiters blocked on a rt_mutex held by this task */
1418 	struct plist_head pi_waiters;
1419 	/* Deadlock detection and priority inheritance handling */
1420 	struct rt_mutex_waiter *pi_blocked_on;
1421 #endif
1422 
1423 #ifdef CONFIG_DEBUG_MUTEXES
1424 	/* mutex deadlock detection */
1425 	struct mutex_waiter *blocked_on;
1426 #endif
1427 #ifdef CONFIG_TRACE_IRQFLAGS
1428 	unsigned int irq_events;
1429 	unsigned long hardirq_enable_ip;
1430 	unsigned long hardirq_disable_ip;
1431 	unsigned int hardirq_enable_event;
1432 	unsigned int hardirq_disable_event;
1433 	int hardirqs_enabled;
1434 	int hardirq_context;
1435 	unsigned long softirq_disable_ip;
1436 	unsigned long softirq_enable_ip;
1437 	unsigned int softirq_disable_event;
1438 	unsigned int softirq_enable_event;
1439 	int softirqs_enabled;
1440 	int softirq_context;
1441 #endif
1442 #ifdef CONFIG_LOCKDEP
1443 # define MAX_LOCK_DEPTH 48UL
1444 	u64 curr_chain_key;
1445 	int lockdep_depth;
1446 	unsigned int lockdep_recursion;
1447 	struct held_lock held_locks[MAX_LOCK_DEPTH];
1448 	gfp_t lockdep_reclaim_gfp;
1449 #endif
1450 
1451 /* journalling filesystem info */
1452 	void *journal_info;
1453 
1454 /* stacked block device info */
1455 	struct bio *bio_list, **bio_tail;
1456 
1457 /* VM state */
1458 	struct reclaim_state *reclaim_state;
1459 
1460 	struct backing_dev_info *backing_dev_info;
1461 
1462 	struct io_context *io_context;
1463 
1464 	unsigned long ptrace_message;
1465 	siginfo_t *last_siginfo; /* For ptrace use.  */
1466 	struct task_io_accounting ioac;
1467 #if defined(CONFIG_TASK_XACCT)
1468 	u64 acct_rss_mem1;	/* accumulated rss usage */
1469 	u64 acct_vm_mem1;	/* accumulated virtual memory usage */
1470 	cputime_t acct_timexpd;	/* stime + utime since last update */
1471 #endif
1472 #ifdef CONFIG_CPUSETS
1473 	nodemask_t mems_allowed;	/* Protected by alloc_lock */
1474 	int cpuset_mem_spread_rotor;
1475 #endif
1476 #ifdef CONFIG_CGROUPS
1477 	/* Control Group info protected by css_set_lock */
1478 	struct css_set *cgroups;
1479 	/* cg_list protected by css_set_lock and tsk->alloc_lock */
1480 	struct list_head cg_list;
1481 #endif
1482 #ifdef CONFIG_FUTEX
1483 	struct robust_list_head __user *robust_list;
1484 #ifdef CONFIG_COMPAT
1485 	struct compat_robust_list_head __user *compat_robust_list;
1486 #endif
1487 	struct list_head pi_state_list;
1488 	struct futex_pi_state *pi_state_cache;
1489 #endif
1490 #ifdef CONFIG_PERF_EVENTS
1491 	struct perf_event_context *perf_event_ctxp;
1492 	struct mutex perf_event_mutex;
1493 	struct list_head perf_event_list;
1494 #endif
1495 #ifdef CONFIG_NUMA
1496 	struct mempolicy *mempolicy;	/* Protected by alloc_lock */
1497 	short il_next;
1498 #endif
1499 	atomic_t fs_excl;	/* holding fs exclusive resources */
1500 	struct rcu_head rcu;
1501 
1502 	/*
1503 	 * cache last used pipe for splice
1504 	 */
1505 	struct pipe_inode_info *splice_pipe;
1506 #ifdef	CONFIG_TASK_DELAY_ACCT
1507 	struct task_delay_info *delays;
1508 #endif
1509 #ifdef CONFIG_FAULT_INJECTION
1510 	int make_it_fail;
1511 #endif
1512 	struct prop_local_single dirties;
1513 #ifdef CONFIG_LATENCYTOP
1514 	int latency_record_count;
1515 	struct latency_record latency_record[LT_SAVECOUNT];
1516 #endif
1517 	/*
1518 	 * time slack values; these are used to round up poll() and
1519 	 * select() etc timeout values. These are in nanoseconds.
1520 	 */
1521 	unsigned long timer_slack_ns;
1522 	unsigned long default_timer_slack_ns;
1523 
1524 	struct list_head	*scm_work_list;
1525 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1526 	/* Index of current stored adress in ret_stack */
1527 	int curr_ret_stack;
1528 	/* Stack of return addresses for return function tracing */
1529 	struct ftrace_ret_stack	*ret_stack;
1530 	/* time stamp for last schedule */
1531 	unsigned long long ftrace_timestamp;
1532 	/*
1533 	 * Number of functions that haven't been traced
1534 	 * because of depth overrun.
1535 	 */
1536 	atomic_t trace_overrun;
1537 	/* Pause for the tracing */
1538 	atomic_t tracing_graph_pause;
1539 #endif
1540 #ifdef CONFIG_TRACING
1541 	/* state flags for use by tracers */
1542 	unsigned long trace;
1543 	/* bitmask of trace recursion */
1544 	unsigned long trace_recursion;
1545 #endif /* CONFIG_TRACING */
1546 	unsigned long stack_start;
1547 };
1548 
1549 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1550 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1551 
1552 /*
1553  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1554  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1555  * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1556  * values are inverted: lower p->prio value means higher priority.
1557  *
1558  * The MAX_USER_RT_PRIO value allows the actual maximum
1559  * RT priority to be separate from the value exported to
1560  * user-space.  This allows kernel threads to set their
1561  * priority to a value higher than any user task. Note:
1562  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1563  */
1564 
1565 #define MAX_USER_RT_PRIO	100
1566 #define MAX_RT_PRIO		MAX_USER_RT_PRIO
1567 
1568 #define MAX_PRIO		(MAX_RT_PRIO + 40)
1569 #define DEFAULT_PRIO		(MAX_RT_PRIO + 20)
1570 
1571 static inline int rt_prio(int prio)
1572 {
1573 	if (unlikely(prio < MAX_RT_PRIO))
1574 		return 1;
1575 	return 0;
1576 }
1577 
1578 static inline int rt_task(struct task_struct *p)
1579 {
1580 	return rt_prio(p->prio);
1581 }
1582 
1583 static inline struct pid *task_pid(struct task_struct *task)
1584 {
1585 	return task->pids[PIDTYPE_PID].pid;
1586 }
1587 
1588 static inline struct pid *task_tgid(struct task_struct *task)
1589 {
1590 	return task->group_leader->pids[PIDTYPE_PID].pid;
1591 }
1592 
1593 /*
1594  * Without tasklist or rcu lock it is not safe to dereference
1595  * the result of task_pgrp/task_session even if task == current,
1596  * we can race with another thread doing sys_setsid/sys_setpgid.
1597  */
1598 static inline struct pid *task_pgrp(struct task_struct *task)
1599 {
1600 	return task->group_leader->pids[PIDTYPE_PGID].pid;
1601 }
1602 
1603 static inline struct pid *task_session(struct task_struct *task)
1604 {
1605 	return task->group_leader->pids[PIDTYPE_SID].pid;
1606 }
1607 
1608 struct pid_namespace;
1609 
1610 /*
1611  * the helpers to get the task's different pids as they are seen
1612  * from various namespaces
1613  *
1614  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1615  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1616  *                     current.
1617  * task_xid_nr_ns()  : id seen from the ns specified;
1618  *
1619  * set_task_vxid()   : assigns a virtual id to a task;
1620  *
1621  * see also pid_nr() etc in include/linux/pid.h
1622  */
1623 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1624 			struct pid_namespace *ns);
1625 
1626 static inline pid_t task_pid_nr(struct task_struct *tsk)
1627 {
1628 	return tsk->pid;
1629 }
1630 
1631 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1632 					struct pid_namespace *ns)
1633 {
1634 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1635 }
1636 
1637 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1638 {
1639 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1640 }
1641 
1642 
1643 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1644 {
1645 	return tsk->tgid;
1646 }
1647 
1648 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1649 
1650 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1651 {
1652 	return pid_vnr(task_tgid(tsk));
1653 }
1654 
1655 
1656 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1657 					struct pid_namespace *ns)
1658 {
1659 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1660 }
1661 
1662 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1663 {
1664 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1665 }
1666 
1667 
1668 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1669 					struct pid_namespace *ns)
1670 {
1671 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1672 }
1673 
1674 static inline pid_t task_session_vnr(struct task_struct *tsk)
1675 {
1676 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1677 }
1678 
1679 /* obsolete, do not use */
1680 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1681 {
1682 	return task_pgrp_nr_ns(tsk, &init_pid_ns);
1683 }
1684 
1685 /**
1686  * pid_alive - check that a task structure is not stale
1687  * @p: Task structure to be checked.
1688  *
1689  * Test if a process is not yet dead (at most zombie state)
1690  * If pid_alive fails, then pointers within the task structure
1691  * can be stale and must not be dereferenced.
1692  */
1693 static inline int pid_alive(struct task_struct *p)
1694 {
1695 	return p->pids[PIDTYPE_PID].pid != NULL;
1696 }
1697 
1698 /**
1699  * is_global_init - check if a task structure is init
1700  * @tsk: Task structure to be checked.
1701  *
1702  * Check if a task structure is the first user space task the kernel created.
1703  */
1704 static inline int is_global_init(struct task_struct *tsk)
1705 {
1706 	return tsk->pid == 1;
1707 }
1708 
1709 /*
1710  * is_container_init:
1711  * check whether in the task is init in its own pid namespace.
1712  */
1713 extern int is_container_init(struct task_struct *tsk);
1714 
1715 extern struct pid *cad_pid;
1716 
1717 extern void free_task(struct task_struct *tsk);
1718 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1719 
1720 extern void __put_task_struct(struct task_struct *t);
1721 
1722 static inline void put_task_struct(struct task_struct *t)
1723 {
1724 	if (atomic_dec_and_test(&t->usage))
1725 		__put_task_struct(t);
1726 }
1727 
1728 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1729 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1730 
1731 /*
1732  * Per process flags
1733  */
1734 #define PF_ALIGNWARN	0x00000001	/* Print alignment warning msgs */
1735 					/* Not implemented yet, only for 486*/
1736 #define PF_STARTING	0x00000002	/* being created */
1737 #define PF_EXITING	0x00000004	/* getting shut down */
1738 #define PF_EXITPIDONE	0x00000008	/* pi exit done on shut down */
1739 #define PF_VCPU		0x00000010	/* I'm a virtual CPU */
1740 #define PF_FORKNOEXEC	0x00000040	/* forked but didn't exec */
1741 #define PF_MCE_PROCESS  0x00000080      /* process policy on mce errors */
1742 #define PF_SUPERPRIV	0x00000100	/* used super-user privileges */
1743 #define PF_DUMPCORE	0x00000200	/* dumped core */
1744 #define PF_SIGNALED	0x00000400	/* killed by a signal */
1745 #define PF_MEMALLOC	0x00000800	/* Allocating memory */
1746 #define PF_FLUSHER	0x00001000	/* responsible for disk writeback */
1747 #define PF_USED_MATH	0x00002000	/* if unset the fpu must be initialized before use */
1748 #define PF_FREEZING	0x00004000	/* freeze in progress. do not account to load */
1749 #define PF_NOFREEZE	0x00008000	/* this thread should not be frozen */
1750 #define PF_FROZEN	0x00010000	/* frozen for system suspend */
1751 #define PF_FSTRANS	0x00020000	/* inside a filesystem transaction */
1752 #define PF_KSWAPD	0x00040000	/* I am kswapd */
1753 #define PF_OOM_ORIGIN	0x00080000	/* Allocating much memory to others */
1754 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
1755 #define PF_KTHREAD	0x00200000	/* I am a kernel thread */
1756 #define PF_RANDOMIZE	0x00400000	/* randomize virtual address space */
1757 #define PF_SWAPWRITE	0x00800000	/* Allowed to write to swap */
1758 #define PF_SPREAD_PAGE	0x01000000	/* Spread page cache over cpuset */
1759 #define PF_SPREAD_SLAB	0x02000000	/* Spread some slab caches over cpuset */
1760 #define PF_THREAD_BOUND	0x04000000	/* Thread bound to specific cpu */
1761 #define PF_MCE_EARLY    0x08000000      /* Early kill for mce process policy */
1762 #define PF_MEMPOLICY	0x10000000	/* Non-default NUMA mempolicy */
1763 #define PF_MUTEX_TESTER	0x20000000	/* Thread belongs to the rt mutex tester */
1764 #define PF_FREEZER_SKIP	0x40000000	/* Freezer should not count it as freezeable */
1765 #define PF_FREEZER_NOSIG 0x80000000	/* Freezer won't send signals to it */
1766 
1767 /*
1768  * Only the _current_ task can read/write to tsk->flags, but other
1769  * tasks can access tsk->flags in readonly mode for example
1770  * with tsk_used_math (like during threaded core dumping).
1771  * There is however an exception to this rule during ptrace
1772  * or during fork: the ptracer task is allowed to write to the
1773  * child->flags of its traced child (same goes for fork, the parent
1774  * can write to the child->flags), because we're guaranteed the
1775  * child is not running and in turn not changing child->flags
1776  * at the same time the parent does it.
1777  */
1778 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1779 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1780 #define clear_used_math() clear_stopped_child_used_math(current)
1781 #define set_used_math() set_stopped_child_used_math(current)
1782 #define conditional_stopped_child_used_math(condition, child) \
1783 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1784 #define conditional_used_math(condition) \
1785 	conditional_stopped_child_used_math(condition, current)
1786 #define copy_to_stopped_child_used_math(child) \
1787 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1788 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1789 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1790 #define used_math() tsk_used_math(current)
1791 
1792 #ifdef CONFIG_TREE_PREEMPT_RCU
1793 
1794 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1795 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1796 
1797 static inline void rcu_copy_process(struct task_struct *p)
1798 {
1799 	p->rcu_read_lock_nesting = 0;
1800 	p->rcu_read_unlock_special = 0;
1801 	p->rcu_blocked_node = NULL;
1802 	INIT_LIST_HEAD(&p->rcu_node_entry);
1803 }
1804 
1805 #else
1806 
1807 static inline void rcu_copy_process(struct task_struct *p)
1808 {
1809 }
1810 
1811 #endif
1812 
1813 #ifdef CONFIG_SMP
1814 extern int set_cpus_allowed_ptr(struct task_struct *p,
1815 				const struct cpumask *new_mask);
1816 #else
1817 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1818 				       const struct cpumask *new_mask)
1819 {
1820 	if (!cpumask_test_cpu(0, new_mask))
1821 		return -EINVAL;
1822 	return 0;
1823 }
1824 #endif
1825 
1826 #ifndef CONFIG_CPUMASK_OFFSTACK
1827 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1828 {
1829 	return set_cpus_allowed_ptr(p, &new_mask);
1830 }
1831 #endif
1832 
1833 /*
1834  * Architectures can set this to 1 if they have specified
1835  * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1836  * but then during bootup it turns out that sched_clock()
1837  * is reliable after all:
1838  */
1839 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1840 extern int sched_clock_stable;
1841 #endif
1842 
1843 /* ftrace calls sched_clock() directly */
1844 extern unsigned long long notrace sched_clock(void);
1845 
1846 extern void sched_clock_init(void);
1847 extern u64 sched_clock_cpu(int cpu);
1848 
1849 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1850 static inline void sched_clock_tick(void)
1851 {
1852 }
1853 
1854 static inline void sched_clock_idle_sleep_event(void)
1855 {
1856 }
1857 
1858 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1859 {
1860 }
1861 #else
1862 extern void sched_clock_tick(void);
1863 extern void sched_clock_idle_sleep_event(void);
1864 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1865 #endif
1866 
1867 /*
1868  * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1869  * clock constructed from sched_clock():
1870  */
1871 extern unsigned long long cpu_clock(int cpu);
1872 
1873 extern unsigned long long
1874 task_sched_runtime(struct task_struct *task);
1875 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1876 
1877 /* sched_exec is called by processes performing an exec */
1878 #ifdef CONFIG_SMP
1879 extern void sched_exec(void);
1880 #else
1881 #define sched_exec()   {}
1882 #endif
1883 
1884 extern void sched_clock_idle_sleep_event(void);
1885 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1886 
1887 #ifdef CONFIG_HOTPLUG_CPU
1888 extern void idle_task_exit(void);
1889 #else
1890 static inline void idle_task_exit(void) {}
1891 #endif
1892 
1893 extern void sched_idle_next(void);
1894 
1895 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1896 extern void wake_up_idle_cpu(int cpu);
1897 #else
1898 static inline void wake_up_idle_cpu(int cpu) { }
1899 #endif
1900 
1901 extern unsigned int sysctl_sched_latency;
1902 extern unsigned int sysctl_sched_min_granularity;
1903 extern unsigned int sysctl_sched_wakeup_granularity;
1904 extern unsigned int sysctl_sched_shares_ratelimit;
1905 extern unsigned int sysctl_sched_shares_thresh;
1906 extern unsigned int sysctl_sched_child_runs_first;
1907 #ifdef CONFIG_SCHED_DEBUG
1908 extern unsigned int sysctl_sched_features;
1909 extern unsigned int sysctl_sched_migration_cost;
1910 extern unsigned int sysctl_sched_nr_migrate;
1911 extern unsigned int sysctl_sched_time_avg;
1912 extern unsigned int sysctl_timer_migration;
1913 
1914 int sched_nr_latency_handler(struct ctl_table *table, int write,
1915 		void __user *buffer, size_t *length,
1916 		loff_t *ppos);
1917 #endif
1918 #ifdef CONFIG_SCHED_DEBUG
1919 static inline unsigned int get_sysctl_timer_migration(void)
1920 {
1921 	return sysctl_timer_migration;
1922 }
1923 #else
1924 static inline unsigned int get_sysctl_timer_migration(void)
1925 {
1926 	return 1;
1927 }
1928 #endif
1929 extern unsigned int sysctl_sched_rt_period;
1930 extern int sysctl_sched_rt_runtime;
1931 
1932 int sched_rt_handler(struct ctl_table *table, int write,
1933 		void __user *buffer, size_t *lenp,
1934 		loff_t *ppos);
1935 
1936 extern unsigned int sysctl_sched_compat_yield;
1937 
1938 #ifdef CONFIG_RT_MUTEXES
1939 extern int rt_mutex_getprio(struct task_struct *p);
1940 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1941 extern void rt_mutex_adjust_pi(struct task_struct *p);
1942 #else
1943 static inline int rt_mutex_getprio(struct task_struct *p)
1944 {
1945 	return p->normal_prio;
1946 }
1947 # define rt_mutex_adjust_pi(p)		do { } while (0)
1948 #endif
1949 
1950 extern void set_user_nice(struct task_struct *p, long nice);
1951 extern int task_prio(const struct task_struct *p);
1952 extern int task_nice(const struct task_struct *p);
1953 extern int can_nice(const struct task_struct *p, const int nice);
1954 extern int task_curr(const struct task_struct *p);
1955 extern int idle_cpu(int cpu);
1956 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1957 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1958 				      struct sched_param *);
1959 extern struct task_struct *idle_task(int cpu);
1960 extern struct task_struct *curr_task(int cpu);
1961 extern void set_curr_task(int cpu, struct task_struct *p);
1962 
1963 void yield(void);
1964 
1965 /*
1966  * The default (Linux) execution domain.
1967  */
1968 extern struct exec_domain	default_exec_domain;
1969 
1970 union thread_union {
1971 	struct thread_info thread_info;
1972 	unsigned long stack[THREAD_SIZE/sizeof(long)];
1973 };
1974 
1975 #ifndef __HAVE_ARCH_KSTACK_END
1976 static inline int kstack_end(void *addr)
1977 {
1978 	/* Reliable end of stack detection:
1979 	 * Some APM bios versions misalign the stack
1980 	 */
1981 	return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1982 }
1983 #endif
1984 
1985 extern union thread_union init_thread_union;
1986 extern struct task_struct init_task;
1987 
1988 extern struct   mm_struct init_mm;
1989 
1990 extern struct pid_namespace init_pid_ns;
1991 
1992 /*
1993  * find a task by one of its numerical ids
1994  *
1995  * find_task_by_pid_ns():
1996  *      finds a task by its pid in the specified namespace
1997  * find_task_by_vpid():
1998  *      finds a task by its virtual pid
1999  *
2000  * see also find_vpid() etc in include/linux/pid.h
2001  */
2002 
2003 extern struct task_struct *find_task_by_vpid(pid_t nr);
2004 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2005 		struct pid_namespace *ns);
2006 
2007 extern void __set_special_pids(struct pid *pid);
2008 
2009 /* per-UID process charging. */
2010 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2011 static inline struct user_struct *get_uid(struct user_struct *u)
2012 {
2013 	atomic_inc(&u->__count);
2014 	return u;
2015 }
2016 extern void free_uid(struct user_struct *);
2017 extern void release_uids(struct user_namespace *ns);
2018 
2019 #include <asm/current.h>
2020 
2021 extern void do_timer(unsigned long ticks);
2022 
2023 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2024 extern int wake_up_process(struct task_struct *tsk);
2025 extern void wake_up_new_task(struct task_struct *tsk,
2026 				unsigned long clone_flags);
2027 #ifdef CONFIG_SMP
2028  extern void kick_process(struct task_struct *tsk);
2029 #else
2030  static inline void kick_process(struct task_struct *tsk) { }
2031 #endif
2032 extern void sched_fork(struct task_struct *p, int clone_flags);
2033 extern void sched_dead(struct task_struct *p);
2034 
2035 extern void proc_caches_init(void);
2036 extern void flush_signals(struct task_struct *);
2037 extern void __flush_signals(struct task_struct *);
2038 extern void ignore_signals(struct task_struct *);
2039 extern void flush_signal_handlers(struct task_struct *, int force_default);
2040 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2041 
2042 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2043 {
2044 	unsigned long flags;
2045 	int ret;
2046 
2047 	spin_lock_irqsave(&tsk->sighand->siglock, flags);
2048 	ret = dequeue_signal(tsk, mask, info);
2049 	spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2050 
2051 	return ret;
2052 }
2053 
2054 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2055 			      sigset_t *mask);
2056 extern void unblock_all_signals(void);
2057 extern void release_task(struct task_struct * p);
2058 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2059 extern int force_sigsegv(int, struct task_struct *);
2060 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2061 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2062 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2063 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2064 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2065 extern int kill_pid(struct pid *pid, int sig, int priv);
2066 extern int kill_proc_info(int, struct siginfo *, pid_t);
2067 extern int do_notify_parent(struct task_struct *, int);
2068 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2069 extern void force_sig(int, struct task_struct *);
2070 extern void force_sig_specific(int, struct task_struct *);
2071 extern int send_sig(int, struct task_struct *, int);
2072 extern void zap_other_threads(struct task_struct *p);
2073 extern struct sigqueue *sigqueue_alloc(void);
2074 extern void sigqueue_free(struct sigqueue *);
2075 extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
2076 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2077 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2078 
2079 static inline int kill_cad_pid(int sig, int priv)
2080 {
2081 	return kill_pid(cad_pid, sig, priv);
2082 }
2083 
2084 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
2085 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2086 #define SEND_SIG_PRIV	((struct siginfo *) 1)
2087 #define SEND_SIG_FORCED	((struct siginfo *) 2)
2088 
2089 static inline int is_si_special(const struct siginfo *info)
2090 {
2091 	return info <= SEND_SIG_FORCED;
2092 }
2093 
2094 /*
2095  * True if we are on the alternate signal stack.
2096  */
2097 static inline int on_sig_stack(unsigned long sp)
2098 {
2099 #ifdef CONFIG_STACK_GROWSUP
2100 	return sp >= current->sas_ss_sp &&
2101 		sp - current->sas_ss_sp < current->sas_ss_size;
2102 #else
2103 	return sp > current->sas_ss_sp &&
2104 		sp - current->sas_ss_sp <= current->sas_ss_size;
2105 #endif
2106 }
2107 
2108 static inline int sas_ss_flags(unsigned long sp)
2109 {
2110 	return (current->sas_ss_size == 0 ? SS_DISABLE
2111 		: on_sig_stack(sp) ? SS_ONSTACK : 0);
2112 }
2113 
2114 /*
2115  * Routines for handling mm_structs
2116  */
2117 extern struct mm_struct * mm_alloc(void);
2118 
2119 /* mmdrop drops the mm and the page tables */
2120 extern void __mmdrop(struct mm_struct *);
2121 static inline void mmdrop(struct mm_struct * mm)
2122 {
2123 	if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2124 		__mmdrop(mm);
2125 }
2126 
2127 /* mmput gets rid of the mappings and all user-space */
2128 extern void mmput(struct mm_struct *);
2129 /* Grab a reference to a task's mm, if it is not already going away */
2130 extern struct mm_struct *get_task_mm(struct task_struct *task);
2131 /* Remove the current tasks stale references to the old mm_struct */
2132 extern void mm_release(struct task_struct *, struct mm_struct *);
2133 /* Allocate a new mm structure and copy contents from tsk->mm */
2134 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2135 
2136 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2137 			struct task_struct *, struct pt_regs *);
2138 extern void flush_thread(void);
2139 extern void exit_thread(void);
2140 
2141 extern void exit_files(struct task_struct *);
2142 extern void __cleanup_signal(struct signal_struct *);
2143 extern void __cleanup_sighand(struct sighand_struct *);
2144 
2145 extern void exit_itimers(struct signal_struct *);
2146 extern void flush_itimer_signals(void);
2147 
2148 extern NORET_TYPE void do_group_exit(int);
2149 
2150 extern void daemonize(const char *, ...);
2151 extern int allow_signal(int);
2152 extern int disallow_signal(int);
2153 
2154 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2155 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2156 struct task_struct *fork_idle(int);
2157 
2158 extern void set_task_comm(struct task_struct *tsk, char *from);
2159 extern char *get_task_comm(char *to, struct task_struct *tsk);
2160 
2161 #ifdef CONFIG_SMP
2162 extern void wait_task_context_switch(struct task_struct *p);
2163 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2164 #else
2165 static inline void wait_task_context_switch(struct task_struct *p) {}
2166 static inline unsigned long wait_task_inactive(struct task_struct *p,
2167 					       long match_state)
2168 {
2169 	return 1;
2170 }
2171 #endif
2172 
2173 #define next_task(p) \
2174 	list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2175 
2176 #define for_each_process(p) \
2177 	for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2178 
2179 extern bool current_is_single_threaded(void);
2180 
2181 /*
2182  * Careful: do_each_thread/while_each_thread is a double loop so
2183  *          'break' will not work as expected - use goto instead.
2184  */
2185 #define do_each_thread(g, t) \
2186 	for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2187 
2188 #define while_each_thread(g, t) \
2189 	while ((t = next_thread(t)) != g)
2190 
2191 /* de_thread depends on thread_group_leader not being a pid based check */
2192 #define thread_group_leader(p)	(p == p->group_leader)
2193 
2194 /* Do to the insanities of de_thread it is possible for a process
2195  * to have the pid of the thread group leader without actually being
2196  * the thread group leader.  For iteration through the pids in proc
2197  * all we care about is that we have a task with the appropriate
2198  * pid, we don't actually care if we have the right task.
2199  */
2200 static inline int has_group_leader_pid(struct task_struct *p)
2201 {
2202 	return p->pid == p->tgid;
2203 }
2204 
2205 static inline
2206 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2207 {
2208 	return p1->tgid == p2->tgid;
2209 }
2210 
2211 static inline struct task_struct *next_thread(const struct task_struct *p)
2212 {
2213 	return list_entry_rcu(p->thread_group.next,
2214 			      struct task_struct, thread_group);
2215 }
2216 
2217 static inline int thread_group_empty(struct task_struct *p)
2218 {
2219 	return list_empty(&p->thread_group);
2220 }
2221 
2222 #define delay_group_leader(p) \
2223 		(thread_group_leader(p) && !thread_group_empty(p))
2224 
2225 static inline int task_detached(struct task_struct *p)
2226 {
2227 	return p->exit_signal == -1;
2228 }
2229 
2230 /*
2231  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2232  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2233  * pins the final release of task.io_context.  Also protects ->cpuset and
2234  * ->cgroup.subsys[].
2235  *
2236  * Nests both inside and outside of read_lock(&tasklist_lock).
2237  * It must not be nested with write_lock_irq(&tasklist_lock),
2238  * neither inside nor outside.
2239  */
2240 static inline void task_lock(struct task_struct *p)
2241 {
2242 	spin_lock(&p->alloc_lock);
2243 }
2244 
2245 static inline void task_unlock(struct task_struct *p)
2246 {
2247 	spin_unlock(&p->alloc_lock);
2248 }
2249 
2250 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2251 							unsigned long *flags);
2252 
2253 static inline void unlock_task_sighand(struct task_struct *tsk,
2254 						unsigned long *flags)
2255 {
2256 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2257 }
2258 
2259 #ifndef __HAVE_THREAD_FUNCTIONS
2260 
2261 #define task_thread_info(task)	((struct thread_info *)(task)->stack)
2262 #define task_stack_page(task)	((task)->stack)
2263 
2264 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2265 {
2266 	*task_thread_info(p) = *task_thread_info(org);
2267 	task_thread_info(p)->task = p;
2268 }
2269 
2270 static inline unsigned long *end_of_stack(struct task_struct *p)
2271 {
2272 	return (unsigned long *)(task_thread_info(p) + 1);
2273 }
2274 
2275 #endif
2276 
2277 static inline int object_is_on_stack(void *obj)
2278 {
2279 	void *stack = task_stack_page(current);
2280 
2281 	return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2282 }
2283 
2284 extern void thread_info_cache_init(void);
2285 
2286 #ifdef CONFIG_DEBUG_STACK_USAGE
2287 static inline unsigned long stack_not_used(struct task_struct *p)
2288 {
2289 	unsigned long *n = end_of_stack(p);
2290 
2291 	do { 	/* Skip over canary */
2292 		n++;
2293 	} while (!*n);
2294 
2295 	return (unsigned long)n - (unsigned long)end_of_stack(p);
2296 }
2297 #endif
2298 
2299 /* set thread flags in other task's structures
2300  * - see asm/thread_info.h for TIF_xxxx flags available
2301  */
2302 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2303 {
2304 	set_ti_thread_flag(task_thread_info(tsk), flag);
2305 }
2306 
2307 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2308 {
2309 	clear_ti_thread_flag(task_thread_info(tsk), flag);
2310 }
2311 
2312 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2313 {
2314 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2315 }
2316 
2317 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2318 {
2319 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2320 }
2321 
2322 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2323 {
2324 	return test_ti_thread_flag(task_thread_info(tsk), flag);
2325 }
2326 
2327 static inline void set_tsk_need_resched(struct task_struct *tsk)
2328 {
2329 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2330 }
2331 
2332 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2333 {
2334 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2335 }
2336 
2337 static inline int test_tsk_need_resched(struct task_struct *tsk)
2338 {
2339 	return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2340 }
2341 
2342 static inline int restart_syscall(void)
2343 {
2344 	set_tsk_thread_flag(current, TIF_SIGPENDING);
2345 	return -ERESTARTNOINTR;
2346 }
2347 
2348 static inline int signal_pending(struct task_struct *p)
2349 {
2350 	return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2351 }
2352 
2353 static inline int __fatal_signal_pending(struct task_struct *p)
2354 {
2355 	return unlikely(sigismember(&p->pending.signal, SIGKILL));
2356 }
2357 
2358 static inline int fatal_signal_pending(struct task_struct *p)
2359 {
2360 	return signal_pending(p) && __fatal_signal_pending(p);
2361 }
2362 
2363 static inline int signal_pending_state(long state, struct task_struct *p)
2364 {
2365 	if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2366 		return 0;
2367 	if (!signal_pending(p))
2368 		return 0;
2369 
2370 	return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2371 }
2372 
2373 static inline int need_resched(void)
2374 {
2375 	return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2376 }
2377 
2378 /*
2379  * cond_resched() and cond_resched_lock(): latency reduction via
2380  * explicit rescheduling in places that are safe. The return
2381  * value indicates whether a reschedule was done in fact.
2382  * cond_resched_lock() will drop the spinlock before scheduling,
2383  * cond_resched_softirq() will enable bhs before scheduling.
2384  */
2385 extern int _cond_resched(void);
2386 
2387 #define cond_resched() ({			\
2388 	__might_sleep(__FILE__, __LINE__, 0);	\
2389 	_cond_resched();			\
2390 })
2391 
2392 extern int __cond_resched_lock(spinlock_t *lock);
2393 
2394 #ifdef CONFIG_PREEMPT
2395 #define PREEMPT_LOCK_OFFSET	PREEMPT_OFFSET
2396 #else
2397 #define PREEMPT_LOCK_OFFSET	0
2398 #endif
2399 
2400 #define cond_resched_lock(lock) ({				\
2401 	__might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);	\
2402 	__cond_resched_lock(lock);				\
2403 })
2404 
2405 extern int __cond_resched_softirq(void);
2406 
2407 #define cond_resched_softirq() ({				\
2408 	__might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET);	\
2409 	__cond_resched_softirq();				\
2410 })
2411 
2412 /*
2413  * Does a critical section need to be broken due to another
2414  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2415  * but a general need for low latency)
2416  */
2417 static inline int spin_needbreak(spinlock_t *lock)
2418 {
2419 #ifdef CONFIG_PREEMPT
2420 	return spin_is_contended(lock);
2421 #else
2422 	return 0;
2423 #endif
2424 }
2425 
2426 /*
2427  * Thread group CPU time accounting.
2428  */
2429 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2430 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2431 
2432 static inline void thread_group_cputime_init(struct signal_struct *sig)
2433 {
2434 	sig->cputimer.cputime = INIT_CPUTIME;
2435 	spin_lock_init(&sig->cputimer.lock);
2436 	sig->cputimer.running = 0;
2437 }
2438 
2439 static inline void thread_group_cputime_free(struct signal_struct *sig)
2440 {
2441 }
2442 
2443 /*
2444  * Reevaluate whether the task has signals pending delivery.
2445  * Wake the task if so.
2446  * This is required every time the blocked sigset_t changes.
2447  * callers must hold sighand->siglock.
2448  */
2449 extern void recalc_sigpending_and_wake(struct task_struct *t);
2450 extern void recalc_sigpending(void);
2451 
2452 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2453 
2454 /*
2455  * Wrappers for p->thread_info->cpu access. No-op on UP.
2456  */
2457 #ifdef CONFIG_SMP
2458 
2459 static inline unsigned int task_cpu(const struct task_struct *p)
2460 {
2461 	return task_thread_info(p)->cpu;
2462 }
2463 
2464 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2465 
2466 #else
2467 
2468 static inline unsigned int task_cpu(const struct task_struct *p)
2469 {
2470 	return 0;
2471 }
2472 
2473 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2474 {
2475 }
2476 
2477 #endif /* CONFIG_SMP */
2478 
2479 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2480 
2481 #ifdef CONFIG_TRACING
2482 extern void
2483 __trace_special(void *__tr, void *__data,
2484 		unsigned long arg1, unsigned long arg2, unsigned long arg3);
2485 #else
2486 static inline void
2487 __trace_special(void *__tr, void *__data,
2488 		unsigned long arg1, unsigned long arg2, unsigned long arg3)
2489 {
2490 }
2491 #endif
2492 
2493 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2494 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2495 
2496 extern void normalize_rt_tasks(void);
2497 
2498 #ifdef CONFIG_GROUP_SCHED
2499 
2500 extern struct task_group init_task_group;
2501 #ifdef CONFIG_USER_SCHED
2502 extern struct task_group root_task_group;
2503 extern void set_tg_uid(struct user_struct *user);
2504 #endif
2505 
2506 extern struct task_group *sched_create_group(struct task_group *parent);
2507 extern void sched_destroy_group(struct task_group *tg);
2508 extern void sched_move_task(struct task_struct *tsk);
2509 #ifdef CONFIG_FAIR_GROUP_SCHED
2510 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2511 extern unsigned long sched_group_shares(struct task_group *tg);
2512 #endif
2513 #ifdef CONFIG_RT_GROUP_SCHED
2514 extern int sched_group_set_rt_runtime(struct task_group *tg,
2515 				      long rt_runtime_us);
2516 extern long sched_group_rt_runtime(struct task_group *tg);
2517 extern int sched_group_set_rt_period(struct task_group *tg,
2518 				      long rt_period_us);
2519 extern long sched_group_rt_period(struct task_group *tg);
2520 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2521 #endif
2522 #endif
2523 
2524 extern int task_can_switch_user(struct user_struct *up,
2525 					struct task_struct *tsk);
2526 
2527 #ifdef CONFIG_TASK_XACCT
2528 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2529 {
2530 	tsk->ioac.rchar += amt;
2531 }
2532 
2533 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2534 {
2535 	tsk->ioac.wchar += amt;
2536 }
2537 
2538 static inline void inc_syscr(struct task_struct *tsk)
2539 {
2540 	tsk->ioac.syscr++;
2541 }
2542 
2543 static inline void inc_syscw(struct task_struct *tsk)
2544 {
2545 	tsk->ioac.syscw++;
2546 }
2547 #else
2548 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2549 {
2550 }
2551 
2552 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2553 {
2554 }
2555 
2556 static inline void inc_syscr(struct task_struct *tsk)
2557 {
2558 }
2559 
2560 static inline void inc_syscw(struct task_struct *tsk)
2561 {
2562 }
2563 #endif
2564 
2565 #ifndef TASK_SIZE_OF
2566 #define TASK_SIZE_OF(tsk)	TASK_SIZE
2567 #endif
2568 
2569 /*
2570  * Call the function if the target task is executing on a CPU right now:
2571  */
2572 extern void task_oncpu_function_call(struct task_struct *p,
2573 				     void (*func) (void *info), void *info);
2574 
2575 
2576 #ifdef CONFIG_MM_OWNER
2577 extern void mm_update_next_owner(struct mm_struct *mm);
2578 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2579 #else
2580 static inline void mm_update_next_owner(struct mm_struct *mm)
2581 {
2582 }
2583 
2584 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2585 {
2586 }
2587 #endif /* CONFIG_MM_OWNER */
2588 
2589 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2590 
2591 #endif /* __KERNEL__ */
2592 
2593 #endif
2594