xref: /linux/include/linux/sched.h (revision 25aee3debe0464f6c680173041fa3de30ec9ff54)
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 /* 0x02000000 was previously the unused CLONE_STOPPED (Start in stopped state)
25    and is now available for re-use. */
26 #define CLONE_NEWUTS		0x04000000	/* New utsname group? */
27 #define CLONE_NEWIPC		0x08000000	/* New ipcs */
28 #define CLONE_NEWUSER		0x10000000	/* New user namespace */
29 #define CLONE_NEWPID		0x20000000	/* New pid namespace */
30 #define CLONE_NEWNET		0x40000000	/* New network namespace */
31 #define CLONE_IO		0x80000000	/* Clone io context */
32 
33 /*
34  * Scheduling policies
35  */
36 #define SCHED_NORMAL		0
37 #define SCHED_FIFO		1
38 #define SCHED_RR		2
39 #define SCHED_BATCH		3
40 /* SCHED_ISO: reserved but not implemented yet */
41 #define SCHED_IDLE		5
42 /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
43 #define SCHED_RESET_ON_FORK     0x40000000
44 
45 #ifdef __KERNEL__
46 
47 struct sched_param {
48 	int sched_priority;
49 };
50 
51 #include <asm/param.h>	/* for HZ */
52 
53 #include <linux/capability.h>
54 #include <linux/threads.h>
55 #include <linux/kernel.h>
56 #include <linux/types.h>
57 #include <linux/timex.h>
58 #include <linux/jiffies.h>
59 #include <linux/rbtree.h>
60 #include <linux/thread_info.h>
61 #include <linux/cpumask.h>
62 #include <linux/errno.h>
63 #include <linux/nodemask.h>
64 #include <linux/mm_types.h>
65 
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/compiler.h>
74 #include <linux/completion.h>
75 #include <linux/pid.h>
76 #include <linux/percpu.h>
77 #include <linux/topology.h>
78 #include <linux/proportions.h>
79 #include <linux/seccomp.h>
80 #include <linux/rcupdate.h>
81 #include <linux/rculist.h>
82 #include <linux/rtmutex.h>
83 
84 #include <linux/time.h>
85 #include <linux/param.h>
86 #include <linux/resource.h>
87 #include <linux/timer.h>
88 #include <linux/hrtimer.h>
89 #include <linux/task_io_accounting.h>
90 #include <linux/latencytop.h>
91 #include <linux/cred.h>
92 #include <linux/llist.h>
93 #include <linux/uidgid.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_list;
101 struct fs_struct;
102 struct perf_event_context;
103 struct blk_plug;
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(int cpu);
144 extern unsigned long this_cpu_load(void);
145 
146 
147 extern void calc_global_load(unsigned long ticks);
148 extern void update_cpu_load_nohz(void);
149 
150 extern unsigned long get_parent_ip(unsigned long addr);
151 
152 struct seq_file;
153 struct cfs_rq;
154 struct task_group;
155 #ifdef CONFIG_SCHED_DEBUG
156 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
157 extern void proc_sched_set_task(struct task_struct *p);
158 extern void
159 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
160 #else
161 static inline void
162 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
163 {
164 }
165 static inline void proc_sched_set_task(struct task_struct *p)
166 {
167 }
168 static inline void
169 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
170 {
171 }
172 #endif
173 
174 /*
175  * Task state bitmask. NOTE! These bits are also
176  * encoded in fs/proc/array.c: get_task_state().
177  *
178  * We have two separate sets of flags: task->state
179  * is about runnability, while task->exit_state are
180  * about the task exiting. Confusing, but this way
181  * modifying one set can't modify the other one by
182  * mistake.
183  */
184 #define TASK_RUNNING		0
185 #define TASK_INTERRUPTIBLE	1
186 #define TASK_UNINTERRUPTIBLE	2
187 #define __TASK_STOPPED		4
188 #define __TASK_TRACED		8
189 /* in tsk->exit_state */
190 #define EXIT_ZOMBIE		16
191 #define EXIT_DEAD		32
192 /* in tsk->state again */
193 #define TASK_DEAD		64
194 #define TASK_WAKEKILL		128
195 #define TASK_WAKING		256
196 #define TASK_STATE_MAX		512
197 
198 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
199 
200 extern char ___assert_task_state[1 - 2*!!(
201 		sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
202 
203 /* Convenience macros for the sake of set_task_state */
204 #define TASK_KILLABLE		(TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
205 #define TASK_STOPPED		(TASK_WAKEKILL | __TASK_STOPPED)
206 #define TASK_TRACED		(TASK_WAKEKILL | __TASK_TRACED)
207 
208 /* Convenience macros for the sake of wake_up */
209 #define TASK_NORMAL		(TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
210 #define TASK_ALL		(TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
211 
212 /* get_task_state() */
213 #define TASK_REPORT		(TASK_RUNNING | TASK_INTERRUPTIBLE | \
214 				 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
215 				 __TASK_TRACED)
216 
217 #define task_is_traced(task)	((task->state & __TASK_TRACED) != 0)
218 #define task_is_stopped(task)	((task->state & __TASK_STOPPED) != 0)
219 #define task_is_dead(task)	((task)->exit_state != 0)
220 #define task_is_stopped_or_traced(task)	\
221 			((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
222 #define task_contributes_to_load(task)	\
223 				((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
224 				 (task->flags & PF_FROZEN) == 0)
225 
226 #define __set_task_state(tsk, state_value)		\
227 	do { (tsk)->state = (state_value); } while (0)
228 #define set_task_state(tsk, state_value)		\
229 	set_mb((tsk)->state, (state_value))
230 
231 /*
232  * set_current_state() includes a barrier so that the write of current->state
233  * is correctly serialised wrt the caller's subsequent test of whether to
234  * actually sleep:
235  *
236  *	set_current_state(TASK_UNINTERRUPTIBLE);
237  *	if (do_i_need_to_sleep())
238  *		schedule();
239  *
240  * If the caller does not need such serialisation then use __set_current_state()
241  */
242 #define __set_current_state(state_value)			\
243 	do { current->state = (state_value); } while (0)
244 #define set_current_state(state_value)		\
245 	set_mb(current->state, (state_value))
246 
247 /* Task command name length */
248 #define TASK_COMM_LEN 16
249 
250 #include <linux/spinlock.h>
251 
252 /*
253  * This serializes "schedule()" and also protects
254  * the run-queue from deletions/modifications (but
255  * _adding_ to the beginning of the run-queue has
256  * a separate lock).
257  */
258 extern rwlock_t tasklist_lock;
259 extern spinlock_t mmlist_lock;
260 
261 struct task_struct;
262 
263 #ifdef CONFIG_PROVE_RCU
264 extern int lockdep_tasklist_lock_is_held(void);
265 #endif /* #ifdef CONFIG_PROVE_RCU */
266 
267 extern void sched_init(void);
268 extern void sched_init_smp(void);
269 extern asmlinkage void schedule_tail(struct task_struct *prev);
270 extern void init_idle(struct task_struct *idle, int cpu);
271 extern void init_idle_bootup_task(struct task_struct *idle);
272 
273 extern int runqueue_is_locked(int cpu);
274 
275 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
276 extern void select_nohz_load_balancer(int stop_tick);
277 extern void set_cpu_sd_state_idle(void);
278 extern int get_nohz_timer_target(void);
279 #else
280 static inline void select_nohz_load_balancer(int stop_tick) { }
281 static inline void set_cpu_sd_state_idle(void) { }
282 #endif
283 
284 /*
285  * Only dump TASK_* tasks. (0 for all tasks)
286  */
287 extern void show_state_filter(unsigned long state_filter);
288 
289 static inline void show_state(void)
290 {
291 	show_state_filter(0);
292 }
293 
294 extern void show_regs(struct pt_regs *);
295 
296 /*
297  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
298  * task), SP is the stack pointer of the first frame that should be shown in the back
299  * trace (or NULL if the entire call-chain of the task should be shown).
300  */
301 extern void show_stack(struct task_struct *task, unsigned long *sp);
302 
303 void io_schedule(void);
304 long io_schedule_timeout(long timeout);
305 
306 extern void cpu_init (void);
307 extern void trap_init(void);
308 extern void update_process_times(int user);
309 extern void scheduler_tick(void);
310 
311 extern void sched_show_task(struct task_struct *p);
312 
313 #ifdef CONFIG_LOCKUP_DETECTOR
314 extern void touch_softlockup_watchdog(void);
315 extern void touch_softlockup_watchdog_sync(void);
316 extern void touch_all_softlockup_watchdogs(void);
317 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
318 				  void __user *buffer,
319 				  size_t *lenp, loff_t *ppos);
320 extern unsigned int  softlockup_panic;
321 void lockup_detector_init(void);
322 #else
323 static inline void touch_softlockup_watchdog(void)
324 {
325 }
326 static inline void touch_softlockup_watchdog_sync(void)
327 {
328 }
329 static inline void touch_all_softlockup_watchdogs(void)
330 {
331 }
332 static inline void lockup_detector_init(void)
333 {
334 }
335 #endif
336 
337 #if defined(CONFIG_LOCKUP_DETECTOR) && defined(CONFIG_SUSPEND)
338 void lockup_detector_bootcpu_resume(void);
339 #else
340 static inline void lockup_detector_bootcpu_resume(void)
341 {
342 }
343 #endif
344 
345 #ifdef CONFIG_DETECT_HUNG_TASK
346 extern unsigned int  sysctl_hung_task_panic;
347 extern unsigned long sysctl_hung_task_check_count;
348 extern unsigned long sysctl_hung_task_timeout_secs;
349 extern unsigned long sysctl_hung_task_warnings;
350 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
351 					 void __user *buffer,
352 					 size_t *lenp, loff_t *ppos);
353 #else
354 /* Avoid need for ifdefs elsewhere in the code */
355 enum { sysctl_hung_task_timeout_secs = 0 };
356 #endif
357 
358 /* Attach to any functions which should be ignored in wchan output. */
359 #define __sched		__attribute__((__section__(".sched.text")))
360 
361 /* Linker adds these: start and end of __sched functions */
362 extern char __sched_text_start[], __sched_text_end[];
363 
364 /* Is this address in the __sched functions? */
365 extern int in_sched_functions(unsigned long addr);
366 
367 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
368 extern signed long schedule_timeout(signed long timeout);
369 extern signed long schedule_timeout_interruptible(signed long timeout);
370 extern signed long schedule_timeout_killable(signed long timeout);
371 extern signed long schedule_timeout_uninterruptible(signed long timeout);
372 asmlinkage void schedule(void);
373 extern void schedule_preempt_disabled(void);
374 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
375 
376 struct nsproxy;
377 struct user_namespace;
378 
379 /*
380  * Default maximum number of active map areas, this limits the number of vmas
381  * per mm struct. Users can overwrite this number by sysctl but there is a
382  * problem.
383  *
384  * When a program's coredump is generated as ELF format, a section is created
385  * per a vma. In ELF, the number of sections is represented in unsigned short.
386  * This means the number of sections should be smaller than 65535 at coredump.
387  * Because the kernel adds some informative sections to a image of program at
388  * generating coredump, we need some margin. The number of extra sections is
389  * 1-3 now and depends on arch. We use "5" as safe margin, here.
390  */
391 #define MAPCOUNT_ELF_CORE_MARGIN	(5)
392 #define DEFAULT_MAX_MAP_COUNT	(USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
393 
394 extern int sysctl_max_map_count;
395 
396 #include <linux/aio.h>
397 
398 #ifdef CONFIG_MMU
399 extern void arch_pick_mmap_layout(struct mm_struct *mm);
400 extern unsigned long
401 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
402 		       unsigned long, unsigned long);
403 extern unsigned long
404 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
405 			  unsigned long len, unsigned long pgoff,
406 			  unsigned long flags);
407 extern void arch_unmap_area(struct mm_struct *, unsigned long);
408 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
409 #else
410 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
411 #endif
412 
413 
414 extern void set_dumpable(struct mm_struct *mm, int value);
415 extern int get_dumpable(struct mm_struct *mm);
416 
417 /* get/set_dumpable() values */
418 #define SUID_DUMPABLE_DISABLED	0
419 #define SUID_DUMPABLE_ENABLED	1
420 #define SUID_DUMPABLE_SAFE	2
421 
422 /* mm flags */
423 /* dumpable bits */
424 #define MMF_DUMPABLE      0  /* core dump is permitted */
425 #define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
426 
427 #define MMF_DUMPABLE_BITS 2
428 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
429 
430 /* coredump filter bits */
431 #define MMF_DUMP_ANON_PRIVATE	2
432 #define MMF_DUMP_ANON_SHARED	3
433 #define MMF_DUMP_MAPPED_PRIVATE	4
434 #define MMF_DUMP_MAPPED_SHARED	5
435 #define MMF_DUMP_ELF_HEADERS	6
436 #define MMF_DUMP_HUGETLB_PRIVATE 7
437 #define MMF_DUMP_HUGETLB_SHARED  8
438 
439 #define MMF_DUMP_FILTER_SHIFT	MMF_DUMPABLE_BITS
440 #define MMF_DUMP_FILTER_BITS	7
441 #define MMF_DUMP_FILTER_MASK \
442 	(((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
443 #define MMF_DUMP_FILTER_DEFAULT \
444 	((1 << MMF_DUMP_ANON_PRIVATE) |	(1 << MMF_DUMP_ANON_SHARED) |\
445 	 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
446 
447 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
448 # define MMF_DUMP_MASK_DEFAULT_ELF	(1 << MMF_DUMP_ELF_HEADERS)
449 #else
450 # define MMF_DUMP_MASK_DEFAULT_ELF	0
451 #endif
452 					/* leave room for more dump flags */
453 #define MMF_VM_MERGEABLE	16	/* KSM may merge identical pages */
454 #define MMF_VM_HUGEPAGE		17	/* set when VM_HUGEPAGE is set on vma */
455 #define MMF_EXE_FILE_CHANGED	18	/* see prctl_set_mm_exe_file() */
456 
457 #define MMF_INIT_MASK		(MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
458 
459 struct sighand_struct {
460 	atomic_t		count;
461 	struct k_sigaction	action[_NSIG];
462 	spinlock_t		siglock;
463 	wait_queue_head_t	signalfd_wqh;
464 };
465 
466 struct pacct_struct {
467 	int			ac_flag;
468 	long			ac_exitcode;
469 	unsigned long		ac_mem;
470 	cputime_t		ac_utime, ac_stime;
471 	unsigned long		ac_minflt, ac_majflt;
472 };
473 
474 struct cpu_itimer {
475 	cputime_t expires;
476 	cputime_t incr;
477 	u32 error;
478 	u32 incr_error;
479 };
480 
481 /**
482  * struct task_cputime - collected CPU time counts
483  * @utime:		time spent in user mode, in &cputime_t units
484  * @stime:		time spent in kernel mode, in &cputime_t units
485  * @sum_exec_runtime:	total time spent on the CPU, in nanoseconds
486  *
487  * This structure groups together three kinds of CPU time that are
488  * tracked for threads and thread groups.  Most things considering
489  * CPU time want to group these counts together and treat all three
490  * of them in parallel.
491  */
492 struct task_cputime {
493 	cputime_t utime;
494 	cputime_t stime;
495 	unsigned long long sum_exec_runtime;
496 };
497 /* Alternate field names when used to cache expirations. */
498 #define prof_exp	stime
499 #define virt_exp	utime
500 #define sched_exp	sum_exec_runtime
501 
502 #define INIT_CPUTIME	\
503 	(struct task_cputime) {					\
504 		.utime = 0,					\
505 		.stime = 0,					\
506 		.sum_exec_runtime = 0,				\
507 	}
508 
509 /*
510  * Disable preemption until the scheduler is running.
511  * Reset by start_kernel()->sched_init()->init_idle().
512  *
513  * We include PREEMPT_ACTIVE to avoid cond_resched() from working
514  * before the scheduler is active -- see should_resched().
515  */
516 #define INIT_PREEMPT_COUNT	(1 + PREEMPT_ACTIVE)
517 
518 /**
519  * struct thread_group_cputimer - thread group interval timer counts
520  * @cputime:		thread group interval timers.
521  * @running:		non-zero when there are timers running and
522  * 			@cputime receives updates.
523  * @lock:		lock for fields in this struct.
524  *
525  * This structure contains the version of task_cputime, above, that is
526  * used for thread group CPU timer calculations.
527  */
528 struct thread_group_cputimer {
529 	struct task_cputime cputime;
530 	int running;
531 	raw_spinlock_t lock;
532 };
533 
534 #include <linux/rwsem.h>
535 struct autogroup;
536 
537 /*
538  * NOTE! "signal_struct" does not have its own
539  * locking, because a shared signal_struct always
540  * implies a shared sighand_struct, so locking
541  * sighand_struct is always a proper superset of
542  * the locking of signal_struct.
543  */
544 struct signal_struct {
545 	atomic_t		sigcnt;
546 	atomic_t		live;
547 	int			nr_threads;
548 
549 	wait_queue_head_t	wait_chldexit;	/* for wait4() */
550 
551 	/* current thread group signal load-balancing target: */
552 	struct task_struct	*curr_target;
553 
554 	/* shared signal handling: */
555 	struct sigpending	shared_pending;
556 
557 	/* thread group exit support */
558 	int			group_exit_code;
559 	/* overloaded:
560 	 * - notify group_exit_task when ->count is equal to notify_count
561 	 * - everyone except group_exit_task is stopped during signal delivery
562 	 *   of fatal signals, group_exit_task processes the signal.
563 	 */
564 	int			notify_count;
565 	struct task_struct	*group_exit_task;
566 
567 	/* thread group stop support, overloads group_exit_code too */
568 	int			group_stop_count;
569 	unsigned int		flags; /* see SIGNAL_* flags below */
570 
571 	/*
572 	 * PR_SET_CHILD_SUBREAPER marks a process, like a service
573 	 * manager, to re-parent orphan (double-forking) child processes
574 	 * to this process instead of 'init'. The service manager is
575 	 * able to receive SIGCHLD signals and is able to investigate
576 	 * the process until it calls wait(). All children of this
577 	 * process will inherit a flag if they should look for a
578 	 * child_subreaper process at exit.
579 	 */
580 	unsigned int		is_child_subreaper:1;
581 	unsigned int		has_child_subreaper:1;
582 
583 	/* POSIX.1b Interval Timers */
584 	struct list_head posix_timers;
585 
586 	/* ITIMER_REAL timer for the process */
587 	struct hrtimer real_timer;
588 	struct pid *leader_pid;
589 	ktime_t it_real_incr;
590 
591 	/*
592 	 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
593 	 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
594 	 * values are defined to 0 and 1 respectively
595 	 */
596 	struct cpu_itimer it[2];
597 
598 	/*
599 	 * Thread group totals for process CPU timers.
600 	 * See thread_group_cputimer(), et al, for details.
601 	 */
602 	struct thread_group_cputimer cputimer;
603 
604 	/* Earliest-expiration cache. */
605 	struct task_cputime cputime_expires;
606 
607 	struct list_head cpu_timers[3];
608 
609 	struct pid *tty_old_pgrp;
610 
611 	/* boolean value for session group leader */
612 	int leader;
613 
614 	struct tty_struct *tty; /* NULL if no tty */
615 
616 #ifdef CONFIG_SCHED_AUTOGROUP
617 	struct autogroup *autogroup;
618 #endif
619 	/*
620 	 * Cumulative resource counters for dead threads in the group,
621 	 * and for reaped dead child processes forked by this group.
622 	 * Live threads maintain their own counters and add to these
623 	 * in __exit_signal, except for the group leader.
624 	 */
625 	cputime_t utime, stime, cutime, cstime;
626 	cputime_t gtime;
627 	cputime_t cgtime;
628 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
629 	cputime_t prev_utime, prev_stime;
630 #endif
631 	unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
632 	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
633 	unsigned long inblock, oublock, cinblock, coublock;
634 	unsigned long maxrss, cmaxrss;
635 	struct task_io_accounting ioac;
636 
637 	/*
638 	 * Cumulative ns of schedule CPU time fo dead threads in the
639 	 * group, not including a zombie group leader, (This only differs
640 	 * from jiffies_to_ns(utime + stime) if sched_clock uses something
641 	 * other than jiffies.)
642 	 */
643 	unsigned long long sum_sched_runtime;
644 
645 	/*
646 	 * We don't bother to synchronize most readers of this at all,
647 	 * because there is no reader checking a limit that actually needs
648 	 * to get both rlim_cur and rlim_max atomically, and either one
649 	 * alone is a single word that can safely be read normally.
650 	 * getrlimit/setrlimit use task_lock(current->group_leader) to
651 	 * protect this instead of the siglock, because they really
652 	 * have no need to disable irqs.
653 	 */
654 	struct rlimit rlim[RLIM_NLIMITS];
655 
656 #ifdef CONFIG_BSD_PROCESS_ACCT
657 	struct pacct_struct pacct;	/* per-process accounting information */
658 #endif
659 #ifdef CONFIG_TASKSTATS
660 	struct taskstats *stats;
661 #endif
662 #ifdef CONFIG_AUDIT
663 	unsigned audit_tty;
664 	struct tty_audit_buf *tty_audit_buf;
665 #endif
666 #ifdef CONFIG_CGROUPS
667 	/*
668 	 * group_rwsem prevents new tasks from entering the threadgroup and
669 	 * member tasks from exiting,a more specifically, setting of
670 	 * PF_EXITING.  fork and exit paths are protected with this rwsem
671 	 * using threadgroup_change_begin/end().  Users which require
672 	 * threadgroup to remain stable should use threadgroup_[un]lock()
673 	 * which also takes care of exec path.  Currently, cgroup is the
674 	 * only user.
675 	 */
676 	struct rw_semaphore group_rwsem;
677 #endif
678 
679 	int oom_adj;		/* OOM kill score adjustment (bit shift) */
680 	int oom_score_adj;	/* OOM kill score adjustment */
681 	int oom_score_adj_min;	/* OOM kill score adjustment minimum value.
682 				 * Only settable by CAP_SYS_RESOURCE. */
683 
684 	struct mutex cred_guard_mutex;	/* guard against foreign influences on
685 					 * credential calculations
686 					 * (notably. ptrace) */
687 };
688 
689 /* Context switch must be unlocked if interrupts are to be enabled */
690 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
691 # define __ARCH_WANT_UNLOCKED_CTXSW
692 #endif
693 
694 /*
695  * Bits in flags field of signal_struct.
696  */
697 #define SIGNAL_STOP_STOPPED	0x00000001 /* job control stop in effect */
698 #define SIGNAL_STOP_CONTINUED	0x00000002 /* SIGCONT since WCONTINUED reap */
699 #define SIGNAL_GROUP_EXIT	0x00000004 /* group exit in progress */
700 /*
701  * Pending notifications to parent.
702  */
703 #define SIGNAL_CLD_STOPPED	0x00000010
704 #define SIGNAL_CLD_CONTINUED	0x00000020
705 #define SIGNAL_CLD_MASK		(SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
706 
707 #define SIGNAL_UNKILLABLE	0x00000040 /* for init: ignore fatal signals */
708 
709 /* If true, all threads except ->group_exit_task have pending SIGKILL */
710 static inline int signal_group_exit(const struct signal_struct *sig)
711 {
712 	return	(sig->flags & SIGNAL_GROUP_EXIT) ||
713 		(sig->group_exit_task != NULL);
714 }
715 
716 /*
717  * Some day this will be a full-fledged user tracking system..
718  */
719 struct user_struct {
720 	atomic_t __count;	/* reference count */
721 	atomic_t processes;	/* How many processes does this user have? */
722 	atomic_t files;		/* How many open files does this user have? */
723 	atomic_t sigpending;	/* How many pending signals does this user have? */
724 #ifdef CONFIG_INOTIFY_USER
725 	atomic_t inotify_watches; /* How many inotify watches does this user have? */
726 	atomic_t inotify_devs;	/* How many inotify devs does this user have opened? */
727 #endif
728 #ifdef CONFIG_FANOTIFY
729 	atomic_t fanotify_listeners;
730 #endif
731 #ifdef CONFIG_EPOLL
732 	atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
733 #endif
734 #ifdef CONFIG_POSIX_MQUEUE
735 	/* protected by mq_lock	*/
736 	unsigned long mq_bytes;	/* How many bytes can be allocated to mqueue? */
737 #endif
738 	unsigned long locked_shm; /* How many pages of mlocked shm ? */
739 
740 #ifdef CONFIG_KEYS
741 	struct key *uid_keyring;	/* UID specific keyring */
742 	struct key *session_keyring;	/* UID's default session keyring */
743 #endif
744 
745 	/* Hash table maintenance information */
746 	struct hlist_node uidhash_node;
747 	kuid_t uid;
748 
749 #ifdef CONFIG_PERF_EVENTS
750 	atomic_long_t locked_vm;
751 #endif
752 };
753 
754 extern int uids_sysfs_init(void);
755 
756 extern struct user_struct *find_user(kuid_t);
757 
758 extern struct user_struct root_user;
759 #define INIT_USER (&root_user)
760 
761 
762 struct backing_dev_info;
763 struct reclaim_state;
764 
765 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
766 struct sched_info {
767 	/* cumulative counters */
768 	unsigned long pcount;	      /* # of times run on this cpu */
769 	unsigned long long run_delay; /* time spent waiting on a runqueue */
770 
771 	/* timestamps */
772 	unsigned long long last_arrival,/* when we last ran on a cpu */
773 			   last_queued;	/* when we were last queued to run */
774 };
775 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
776 
777 #ifdef CONFIG_TASK_DELAY_ACCT
778 struct task_delay_info {
779 	spinlock_t	lock;
780 	unsigned int	flags;	/* Private per-task flags */
781 
782 	/* For each stat XXX, add following, aligned appropriately
783 	 *
784 	 * struct timespec XXX_start, XXX_end;
785 	 * u64 XXX_delay;
786 	 * u32 XXX_count;
787 	 *
788 	 * Atomicity of updates to XXX_delay, XXX_count protected by
789 	 * single lock above (split into XXX_lock if contention is an issue).
790 	 */
791 
792 	/*
793 	 * XXX_count is incremented on every XXX operation, the delay
794 	 * associated with the operation is added to XXX_delay.
795 	 * XXX_delay contains the accumulated delay time in nanoseconds.
796 	 */
797 	struct timespec blkio_start, blkio_end;	/* Shared by blkio, swapin */
798 	u64 blkio_delay;	/* wait for sync block io completion */
799 	u64 swapin_delay;	/* wait for swapin block io completion */
800 	u32 blkio_count;	/* total count of the number of sync block */
801 				/* io operations performed */
802 	u32 swapin_count;	/* total count of the number of swapin block */
803 				/* io operations performed */
804 
805 	struct timespec freepages_start, freepages_end;
806 	u64 freepages_delay;	/* wait for memory reclaim */
807 	u32 freepages_count;	/* total count of memory reclaim */
808 };
809 #endif	/* CONFIG_TASK_DELAY_ACCT */
810 
811 static inline int sched_info_on(void)
812 {
813 #ifdef CONFIG_SCHEDSTATS
814 	return 1;
815 #elif defined(CONFIG_TASK_DELAY_ACCT)
816 	extern int delayacct_on;
817 	return delayacct_on;
818 #else
819 	return 0;
820 #endif
821 }
822 
823 enum cpu_idle_type {
824 	CPU_IDLE,
825 	CPU_NOT_IDLE,
826 	CPU_NEWLY_IDLE,
827 	CPU_MAX_IDLE_TYPES
828 };
829 
830 /*
831  * Increase resolution of nice-level calculations for 64-bit architectures.
832  * The extra resolution improves shares distribution and load balancing of
833  * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
834  * hierarchies, especially on larger systems. This is not a user-visible change
835  * and does not change the user-interface for setting shares/weights.
836  *
837  * We increase resolution only if we have enough bits to allow this increased
838  * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
839  * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
840  * increased costs.
841  */
842 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load  */
843 # define SCHED_LOAD_RESOLUTION	10
844 # define scale_load(w)		((w) << SCHED_LOAD_RESOLUTION)
845 # define scale_load_down(w)	((w) >> SCHED_LOAD_RESOLUTION)
846 #else
847 # define SCHED_LOAD_RESOLUTION	0
848 # define scale_load(w)		(w)
849 # define scale_load_down(w)	(w)
850 #endif
851 
852 #define SCHED_LOAD_SHIFT	(10 + SCHED_LOAD_RESOLUTION)
853 #define SCHED_LOAD_SCALE	(1L << SCHED_LOAD_SHIFT)
854 
855 /*
856  * Increase resolution of cpu_power calculations
857  */
858 #define SCHED_POWER_SHIFT	10
859 #define SCHED_POWER_SCALE	(1L << SCHED_POWER_SHIFT)
860 
861 /*
862  * sched-domains (multiprocessor balancing) declarations:
863  */
864 #ifdef CONFIG_SMP
865 #define SD_LOAD_BALANCE		0x0001	/* Do load balancing on this domain. */
866 #define SD_BALANCE_NEWIDLE	0x0002	/* Balance when about to become idle */
867 #define SD_BALANCE_EXEC		0x0004	/* Balance on exec */
868 #define SD_BALANCE_FORK		0x0008	/* Balance on fork, clone */
869 #define SD_BALANCE_WAKE		0x0010  /* Balance on wakeup */
870 #define SD_WAKE_AFFINE		0x0020	/* Wake task to waking CPU */
871 #define SD_PREFER_LOCAL		0x0040  /* Prefer to keep tasks local to this domain */
872 #define SD_SHARE_CPUPOWER	0x0080	/* Domain members share cpu power */
873 #define SD_SHARE_PKG_RESOURCES	0x0200	/* Domain members share cpu pkg resources */
874 #define SD_SERIALIZE		0x0400	/* Only a single load balancing instance */
875 #define SD_ASYM_PACKING		0x0800  /* Place busy groups earlier in the domain */
876 #define SD_PREFER_SIBLING	0x1000	/* Prefer to place tasks in a sibling domain */
877 #define SD_OVERLAP		0x2000	/* sched_domains of this level overlap */
878 
879 extern int __weak arch_sd_sibiling_asym_packing(void);
880 
881 struct sched_group_power {
882 	atomic_t ref;
883 	/*
884 	 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
885 	 * single CPU.
886 	 */
887 	unsigned int power, power_orig;
888 	unsigned long next_update;
889 	/*
890 	 * Number of busy cpus in this group.
891 	 */
892 	atomic_t nr_busy_cpus;
893 
894 	unsigned long cpumask[0]; /* iteration mask */
895 };
896 
897 struct sched_group {
898 	struct sched_group *next;	/* Must be a circular list */
899 	atomic_t ref;
900 
901 	unsigned int group_weight;
902 	struct sched_group_power *sgp;
903 
904 	/*
905 	 * The CPUs this group covers.
906 	 *
907 	 * NOTE: this field is variable length. (Allocated dynamically
908 	 * by attaching extra space to the end of the structure,
909 	 * depending on how many CPUs the kernel has booted up with)
910 	 */
911 	unsigned long cpumask[0];
912 };
913 
914 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
915 {
916 	return to_cpumask(sg->cpumask);
917 }
918 
919 /*
920  * cpumask masking which cpus in the group are allowed to iterate up the domain
921  * tree.
922  */
923 static inline struct cpumask *sched_group_mask(struct sched_group *sg)
924 {
925 	return to_cpumask(sg->sgp->cpumask);
926 }
927 
928 /**
929  * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
930  * @group: The group whose first cpu is to be returned.
931  */
932 static inline unsigned int group_first_cpu(struct sched_group *group)
933 {
934 	return cpumask_first(sched_group_cpus(group));
935 }
936 
937 struct sched_domain_attr {
938 	int relax_domain_level;
939 };
940 
941 #define SD_ATTR_INIT	(struct sched_domain_attr) {	\
942 	.relax_domain_level = -1,			\
943 }
944 
945 extern int sched_domain_level_max;
946 
947 struct sched_domain {
948 	/* These fields must be setup */
949 	struct sched_domain *parent;	/* top domain must be null terminated */
950 	struct sched_domain *child;	/* bottom domain must be null terminated */
951 	struct sched_group *groups;	/* the balancing groups of the domain */
952 	unsigned long min_interval;	/* Minimum balance interval ms */
953 	unsigned long max_interval;	/* Maximum balance interval ms */
954 	unsigned int busy_factor;	/* less balancing by factor if busy */
955 	unsigned int imbalance_pct;	/* No balance until over watermark */
956 	unsigned int cache_nice_tries;	/* Leave cache hot tasks for # tries */
957 	unsigned int busy_idx;
958 	unsigned int idle_idx;
959 	unsigned int newidle_idx;
960 	unsigned int wake_idx;
961 	unsigned int forkexec_idx;
962 	unsigned int smt_gain;
963 	int flags;			/* See SD_* */
964 	int level;
965 	int idle_buddy;			/* cpu assigned to select_idle_sibling() */
966 
967 	/* Runtime fields. */
968 	unsigned long last_balance;	/* init to jiffies. units in jiffies */
969 	unsigned int balance_interval;	/* initialise to 1. units in ms. */
970 	unsigned int nr_balance_failed; /* initialise to 0 */
971 
972 	u64 last_update;
973 
974 #ifdef CONFIG_SCHEDSTATS
975 	/* load_balance() stats */
976 	unsigned int lb_count[CPU_MAX_IDLE_TYPES];
977 	unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
978 	unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
979 	unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
980 	unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
981 	unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
982 	unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
983 	unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
984 
985 	/* Active load balancing */
986 	unsigned int alb_count;
987 	unsigned int alb_failed;
988 	unsigned int alb_pushed;
989 
990 	/* SD_BALANCE_EXEC stats */
991 	unsigned int sbe_count;
992 	unsigned int sbe_balanced;
993 	unsigned int sbe_pushed;
994 
995 	/* SD_BALANCE_FORK stats */
996 	unsigned int sbf_count;
997 	unsigned int sbf_balanced;
998 	unsigned int sbf_pushed;
999 
1000 	/* try_to_wake_up() stats */
1001 	unsigned int ttwu_wake_remote;
1002 	unsigned int ttwu_move_affine;
1003 	unsigned int ttwu_move_balance;
1004 #endif
1005 #ifdef CONFIG_SCHED_DEBUG
1006 	char *name;
1007 #endif
1008 	union {
1009 		void *private;		/* used during construction */
1010 		struct rcu_head rcu;	/* used during destruction */
1011 	};
1012 
1013 	unsigned int span_weight;
1014 	/*
1015 	 * Span of all CPUs in this domain.
1016 	 *
1017 	 * NOTE: this field is variable length. (Allocated dynamically
1018 	 * by attaching extra space to the end of the structure,
1019 	 * depending on how many CPUs the kernel has booted up with)
1020 	 */
1021 	unsigned long span[0];
1022 };
1023 
1024 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1025 {
1026 	return to_cpumask(sd->span);
1027 }
1028 
1029 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1030 				    struct sched_domain_attr *dattr_new);
1031 
1032 /* Allocate an array of sched domains, for partition_sched_domains(). */
1033 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1034 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1035 
1036 /* Test a flag in parent sched domain */
1037 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1038 {
1039 	if (sd->parent && (sd->parent->flags & flag))
1040 		return 1;
1041 
1042 	return 0;
1043 }
1044 
1045 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1046 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1047 
1048 bool cpus_share_cache(int this_cpu, int that_cpu);
1049 
1050 #else /* CONFIG_SMP */
1051 
1052 struct sched_domain_attr;
1053 
1054 static inline void
1055 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1056 			struct sched_domain_attr *dattr_new)
1057 {
1058 }
1059 
1060 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1061 {
1062 	return true;
1063 }
1064 
1065 #endif	/* !CONFIG_SMP */
1066 
1067 
1068 struct io_context;			/* See blkdev.h */
1069 
1070 
1071 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1072 extern void prefetch_stack(struct task_struct *t);
1073 #else
1074 static inline void prefetch_stack(struct task_struct *t) { }
1075 #endif
1076 
1077 struct audit_context;		/* See audit.c */
1078 struct mempolicy;
1079 struct pipe_inode_info;
1080 struct uts_namespace;
1081 
1082 struct rq;
1083 struct sched_domain;
1084 
1085 /*
1086  * wake flags
1087  */
1088 #define WF_SYNC		0x01		/* waker goes to sleep after wakup */
1089 #define WF_FORK		0x02		/* child wakeup after fork */
1090 #define WF_MIGRATED	0x04		/* internal use, task got migrated */
1091 
1092 #define ENQUEUE_WAKEUP		1
1093 #define ENQUEUE_HEAD		2
1094 #ifdef CONFIG_SMP
1095 #define ENQUEUE_WAKING		4	/* sched_class::task_waking was called */
1096 #else
1097 #define ENQUEUE_WAKING		0
1098 #endif
1099 
1100 #define DEQUEUE_SLEEP		1
1101 
1102 struct sched_class {
1103 	const struct sched_class *next;
1104 
1105 	void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1106 	void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1107 	void (*yield_task) (struct rq *rq);
1108 	bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1109 
1110 	void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1111 
1112 	struct task_struct * (*pick_next_task) (struct rq *rq);
1113 	void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1114 
1115 #ifdef CONFIG_SMP
1116 	int  (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1117 
1118 	void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1119 	void (*post_schedule) (struct rq *this_rq);
1120 	void (*task_waking) (struct task_struct *task);
1121 	void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1122 
1123 	void (*set_cpus_allowed)(struct task_struct *p,
1124 				 const struct cpumask *newmask);
1125 
1126 	void (*rq_online)(struct rq *rq);
1127 	void (*rq_offline)(struct rq *rq);
1128 #endif
1129 
1130 	void (*set_curr_task) (struct rq *rq);
1131 	void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1132 	void (*task_fork) (struct task_struct *p);
1133 
1134 	void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1135 	void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1136 	void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1137 			     int oldprio);
1138 
1139 	unsigned int (*get_rr_interval) (struct rq *rq,
1140 					 struct task_struct *task);
1141 
1142 #ifdef CONFIG_FAIR_GROUP_SCHED
1143 	void (*task_move_group) (struct task_struct *p, int on_rq);
1144 #endif
1145 };
1146 
1147 struct load_weight {
1148 	unsigned long weight, inv_weight;
1149 };
1150 
1151 #ifdef CONFIG_SCHEDSTATS
1152 struct sched_statistics {
1153 	u64			wait_start;
1154 	u64			wait_max;
1155 	u64			wait_count;
1156 	u64			wait_sum;
1157 	u64			iowait_count;
1158 	u64			iowait_sum;
1159 
1160 	u64			sleep_start;
1161 	u64			sleep_max;
1162 	s64			sum_sleep_runtime;
1163 
1164 	u64			block_start;
1165 	u64			block_max;
1166 	u64			exec_max;
1167 	u64			slice_max;
1168 
1169 	u64			nr_migrations_cold;
1170 	u64			nr_failed_migrations_affine;
1171 	u64			nr_failed_migrations_running;
1172 	u64			nr_failed_migrations_hot;
1173 	u64			nr_forced_migrations;
1174 
1175 	u64			nr_wakeups;
1176 	u64			nr_wakeups_sync;
1177 	u64			nr_wakeups_migrate;
1178 	u64			nr_wakeups_local;
1179 	u64			nr_wakeups_remote;
1180 	u64			nr_wakeups_affine;
1181 	u64			nr_wakeups_affine_attempts;
1182 	u64			nr_wakeups_passive;
1183 	u64			nr_wakeups_idle;
1184 };
1185 #endif
1186 
1187 struct sched_entity {
1188 	struct load_weight	load;		/* for load-balancing */
1189 	struct rb_node		run_node;
1190 	struct list_head	group_node;
1191 	unsigned int		on_rq;
1192 
1193 	u64			exec_start;
1194 	u64			sum_exec_runtime;
1195 	u64			vruntime;
1196 	u64			prev_sum_exec_runtime;
1197 
1198 	u64			nr_migrations;
1199 
1200 #ifdef CONFIG_SCHEDSTATS
1201 	struct sched_statistics statistics;
1202 #endif
1203 
1204 #ifdef CONFIG_FAIR_GROUP_SCHED
1205 	struct sched_entity	*parent;
1206 	/* rq on which this entity is (to be) queued: */
1207 	struct cfs_rq		*cfs_rq;
1208 	/* rq "owned" by this entity/group: */
1209 	struct cfs_rq		*my_q;
1210 #endif
1211 };
1212 
1213 struct sched_rt_entity {
1214 	struct list_head run_list;
1215 	unsigned long timeout;
1216 	unsigned int time_slice;
1217 
1218 	struct sched_rt_entity *back;
1219 #ifdef CONFIG_RT_GROUP_SCHED
1220 	struct sched_rt_entity	*parent;
1221 	/* rq on which this entity is (to be) queued: */
1222 	struct rt_rq		*rt_rq;
1223 	/* rq "owned" by this entity/group: */
1224 	struct rt_rq		*my_q;
1225 #endif
1226 };
1227 
1228 /*
1229  * default timeslice is 100 msecs (used only for SCHED_RR tasks).
1230  * Timeslices get refilled after they expire.
1231  */
1232 #define RR_TIMESLICE		(100 * HZ / 1000)
1233 
1234 struct rcu_node;
1235 
1236 enum perf_event_task_context {
1237 	perf_invalid_context = -1,
1238 	perf_hw_context = 0,
1239 	perf_sw_context,
1240 	perf_nr_task_contexts,
1241 };
1242 
1243 struct task_struct {
1244 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
1245 	void *stack;
1246 	atomic_t usage;
1247 	unsigned int flags;	/* per process flags, defined below */
1248 	unsigned int ptrace;
1249 
1250 #ifdef CONFIG_SMP
1251 	struct llist_node wake_entry;
1252 	int on_cpu;
1253 #endif
1254 	int on_rq;
1255 
1256 	int prio, static_prio, normal_prio;
1257 	unsigned int rt_priority;
1258 	const struct sched_class *sched_class;
1259 	struct sched_entity se;
1260 	struct sched_rt_entity rt;
1261 #ifdef CONFIG_CGROUP_SCHED
1262 	struct task_group *sched_task_group;
1263 #endif
1264 
1265 #ifdef CONFIG_PREEMPT_NOTIFIERS
1266 	/* list of struct preempt_notifier: */
1267 	struct hlist_head preempt_notifiers;
1268 #endif
1269 
1270 	/*
1271 	 * fpu_counter contains the number of consecutive context switches
1272 	 * that the FPU is used. If this is over a threshold, the lazy fpu
1273 	 * saving becomes unlazy to save the trap. This is an unsigned char
1274 	 * so that after 256 times the counter wraps and the behavior turns
1275 	 * lazy again; this to deal with bursty apps that only use FPU for
1276 	 * a short time
1277 	 */
1278 	unsigned char fpu_counter;
1279 #ifdef CONFIG_BLK_DEV_IO_TRACE
1280 	unsigned int btrace_seq;
1281 #endif
1282 
1283 	unsigned int policy;
1284 	int nr_cpus_allowed;
1285 	cpumask_t cpus_allowed;
1286 
1287 #ifdef CONFIG_PREEMPT_RCU
1288 	int rcu_read_lock_nesting;
1289 	char rcu_read_unlock_special;
1290 	struct list_head rcu_node_entry;
1291 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1292 #ifdef CONFIG_TREE_PREEMPT_RCU
1293 	struct rcu_node *rcu_blocked_node;
1294 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1295 #ifdef CONFIG_RCU_BOOST
1296 	struct rt_mutex *rcu_boost_mutex;
1297 #endif /* #ifdef CONFIG_RCU_BOOST */
1298 
1299 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1300 	struct sched_info sched_info;
1301 #endif
1302 
1303 	struct list_head tasks;
1304 #ifdef CONFIG_SMP
1305 	struct plist_node pushable_tasks;
1306 #endif
1307 
1308 	struct mm_struct *mm, *active_mm;
1309 #ifdef CONFIG_COMPAT_BRK
1310 	unsigned brk_randomized:1;
1311 #endif
1312 #if defined(SPLIT_RSS_COUNTING)
1313 	struct task_rss_stat	rss_stat;
1314 #endif
1315 /* task state */
1316 	int exit_state;
1317 	int exit_code, exit_signal;
1318 	int pdeath_signal;  /*  The signal sent when the parent dies  */
1319 	unsigned int jobctl;	/* JOBCTL_*, siglock protected */
1320 	/* ??? */
1321 	unsigned int personality;
1322 	unsigned did_exec:1;
1323 	unsigned in_execve:1;	/* Tell the LSMs that the process is doing an
1324 				 * execve */
1325 	unsigned in_iowait:1;
1326 
1327 	/* task may not gain privileges */
1328 	unsigned no_new_privs:1;
1329 
1330 	/* Revert to default priority/policy when forking */
1331 	unsigned sched_reset_on_fork:1;
1332 	unsigned sched_contributes_to_load:1;
1333 
1334 	pid_t pid;
1335 	pid_t tgid;
1336 
1337 #ifdef CONFIG_CC_STACKPROTECTOR
1338 	/* Canary value for the -fstack-protector gcc feature */
1339 	unsigned long stack_canary;
1340 #endif
1341 	/*
1342 	 * pointers to (original) parent process, youngest child, younger sibling,
1343 	 * older sibling, respectively.  (p->father can be replaced with
1344 	 * p->real_parent->pid)
1345 	 */
1346 	struct task_struct __rcu *real_parent; /* real parent process */
1347 	struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1348 	/*
1349 	 * children/sibling forms the list of my natural children
1350 	 */
1351 	struct list_head children;	/* list of my children */
1352 	struct list_head sibling;	/* linkage in my parent's children list */
1353 	struct task_struct *group_leader;	/* threadgroup leader */
1354 
1355 	/*
1356 	 * ptraced is the list of tasks this task is using ptrace on.
1357 	 * This includes both natural children and PTRACE_ATTACH targets.
1358 	 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1359 	 */
1360 	struct list_head ptraced;
1361 	struct list_head ptrace_entry;
1362 
1363 	/* PID/PID hash table linkage. */
1364 	struct pid_link pids[PIDTYPE_MAX];
1365 	struct list_head thread_group;
1366 
1367 	struct completion *vfork_done;		/* for vfork() */
1368 	int __user *set_child_tid;		/* CLONE_CHILD_SETTID */
1369 	int __user *clear_child_tid;		/* CLONE_CHILD_CLEARTID */
1370 
1371 	cputime_t utime, stime, utimescaled, stimescaled;
1372 	cputime_t gtime;
1373 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1374 	cputime_t prev_utime, prev_stime;
1375 #endif
1376 	unsigned long nvcsw, nivcsw; /* context switch counts */
1377 	struct timespec start_time; 		/* monotonic time */
1378 	struct timespec real_start_time;	/* boot based time */
1379 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1380 	unsigned long min_flt, maj_flt;
1381 
1382 	struct task_cputime cputime_expires;
1383 	struct list_head cpu_timers[3];
1384 
1385 /* process credentials */
1386 	const struct cred __rcu *real_cred; /* objective and real subjective task
1387 					 * credentials (COW) */
1388 	const struct cred __rcu *cred;	/* effective (overridable) subjective task
1389 					 * credentials (COW) */
1390 	char comm[TASK_COMM_LEN]; /* executable name excluding path
1391 				     - access with [gs]et_task_comm (which lock
1392 				       it with task_lock())
1393 				     - initialized normally by setup_new_exec */
1394 /* file system info */
1395 	int link_count, total_link_count;
1396 #ifdef CONFIG_SYSVIPC
1397 /* ipc stuff */
1398 	struct sysv_sem sysvsem;
1399 #endif
1400 #ifdef CONFIG_DETECT_HUNG_TASK
1401 /* hung task detection */
1402 	unsigned long last_switch_count;
1403 #endif
1404 /* CPU-specific state of this task */
1405 	struct thread_struct thread;
1406 /* filesystem information */
1407 	struct fs_struct *fs;
1408 /* open file information */
1409 	struct files_struct *files;
1410 /* namespaces */
1411 	struct nsproxy *nsproxy;
1412 /* signal handlers */
1413 	struct signal_struct *signal;
1414 	struct sighand_struct *sighand;
1415 
1416 	sigset_t blocked, real_blocked;
1417 	sigset_t saved_sigmask;	/* restored if set_restore_sigmask() was used */
1418 	struct sigpending pending;
1419 
1420 	unsigned long sas_ss_sp;
1421 	size_t sas_ss_size;
1422 	int (*notifier)(void *priv);
1423 	void *notifier_data;
1424 	sigset_t *notifier_mask;
1425 	struct callback_head *task_works;
1426 
1427 	struct audit_context *audit_context;
1428 #ifdef CONFIG_AUDITSYSCALL
1429 	uid_t loginuid;
1430 	unsigned int sessionid;
1431 #endif
1432 	struct seccomp seccomp;
1433 
1434 /* Thread group tracking */
1435    	u32 parent_exec_id;
1436    	u32 self_exec_id;
1437 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1438  * mempolicy */
1439 	spinlock_t alloc_lock;
1440 
1441 	/* Protection of the PI data structures: */
1442 	raw_spinlock_t pi_lock;
1443 
1444 #ifdef CONFIG_RT_MUTEXES
1445 	/* PI waiters blocked on a rt_mutex held by this task */
1446 	struct plist_head pi_waiters;
1447 	/* Deadlock detection and priority inheritance handling */
1448 	struct rt_mutex_waiter *pi_blocked_on;
1449 #endif
1450 
1451 #ifdef CONFIG_DEBUG_MUTEXES
1452 	/* mutex deadlock detection */
1453 	struct mutex_waiter *blocked_on;
1454 #endif
1455 #ifdef CONFIG_TRACE_IRQFLAGS
1456 	unsigned int irq_events;
1457 	unsigned long hardirq_enable_ip;
1458 	unsigned long hardirq_disable_ip;
1459 	unsigned int hardirq_enable_event;
1460 	unsigned int hardirq_disable_event;
1461 	int hardirqs_enabled;
1462 	int hardirq_context;
1463 	unsigned long softirq_disable_ip;
1464 	unsigned long softirq_enable_ip;
1465 	unsigned int softirq_disable_event;
1466 	unsigned int softirq_enable_event;
1467 	int softirqs_enabled;
1468 	int softirq_context;
1469 #endif
1470 #ifdef CONFIG_LOCKDEP
1471 # define MAX_LOCK_DEPTH 48UL
1472 	u64 curr_chain_key;
1473 	int lockdep_depth;
1474 	unsigned int lockdep_recursion;
1475 	struct held_lock held_locks[MAX_LOCK_DEPTH];
1476 	gfp_t lockdep_reclaim_gfp;
1477 #endif
1478 
1479 /* journalling filesystem info */
1480 	void *journal_info;
1481 
1482 /* stacked block device info */
1483 	struct bio_list *bio_list;
1484 
1485 #ifdef CONFIG_BLOCK
1486 /* stack plugging */
1487 	struct blk_plug *plug;
1488 #endif
1489 
1490 /* VM state */
1491 	struct reclaim_state *reclaim_state;
1492 
1493 	struct backing_dev_info *backing_dev_info;
1494 
1495 	struct io_context *io_context;
1496 
1497 	unsigned long ptrace_message;
1498 	siginfo_t *last_siginfo; /* For ptrace use.  */
1499 	struct task_io_accounting ioac;
1500 #if defined(CONFIG_TASK_XACCT)
1501 	u64 acct_rss_mem1;	/* accumulated rss usage */
1502 	u64 acct_vm_mem1;	/* accumulated virtual memory usage */
1503 	cputime_t acct_timexpd;	/* stime + utime since last update */
1504 #endif
1505 #ifdef CONFIG_CPUSETS
1506 	nodemask_t mems_allowed;	/* Protected by alloc_lock */
1507 	seqcount_t mems_allowed_seq;	/* Seqence no to catch updates */
1508 	int cpuset_mem_spread_rotor;
1509 	int cpuset_slab_spread_rotor;
1510 #endif
1511 #ifdef CONFIG_CGROUPS
1512 	/* Control Group info protected by css_set_lock */
1513 	struct css_set __rcu *cgroups;
1514 	/* cg_list protected by css_set_lock and tsk->alloc_lock */
1515 	struct list_head cg_list;
1516 #endif
1517 #ifdef CONFIG_FUTEX
1518 	struct robust_list_head __user *robust_list;
1519 #ifdef CONFIG_COMPAT
1520 	struct compat_robust_list_head __user *compat_robust_list;
1521 #endif
1522 	struct list_head pi_state_list;
1523 	struct futex_pi_state *pi_state_cache;
1524 #endif
1525 #ifdef CONFIG_PERF_EVENTS
1526 	struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1527 	struct mutex perf_event_mutex;
1528 	struct list_head perf_event_list;
1529 #endif
1530 #ifdef CONFIG_NUMA
1531 	struct mempolicy *mempolicy;	/* Protected by alloc_lock */
1532 	short il_next;
1533 	short pref_node_fork;
1534 #endif
1535 	struct rcu_head rcu;
1536 
1537 	/*
1538 	 * cache last used pipe for splice
1539 	 */
1540 	struct pipe_inode_info *splice_pipe;
1541 #ifdef	CONFIG_TASK_DELAY_ACCT
1542 	struct task_delay_info *delays;
1543 #endif
1544 #ifdef CONFIG_FAULT_INJECTION
1545 	int make_it_fail;
1546 #endif
1547 	/*
1548 	 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1549 	 * balance_dirty_pages() for some dirty throttling pause
1550 	 */
1551 	int nr_dirtied;
1552 	int nr_dirtied_pause;
1553 	unsigned long dirty_paused_when; /* start of a write-and-pause period */
1554 
1555 #ifdef CONFIG_LATENCYTOP
1556 	int latency_record_count;
1557 	struct latency_record latency_record[LT_SAVECOUNT];
1558 #endif
1559 	/*
1560 	 * time slack values; these are used to round up poll() and
1561 	 * select() etc timeout values. These are in nanoseconds.
1562 	 */
1563 	unsigned long timer_slack_ns;
1564 	unsigned long default_timer_slack_ns;
1565 
1566 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1567 	/* Index of current stored address in ret_stack */
1568 	int curr_ret_stack;
1569 	/* Stack of return addresses for return function tracing */
1570 	struct ftrace_ret_stack	*ret_stack;
1571 	/* time stamp for last schedule */
1572 	unsigned long long ftrace_timestamp;
1573 	/*
1574 	 * Number of functions that haven't been traced
1575 	 * because of depth overrun.
1576 	 */
1577 	atomic_t trace_overrun;
1578 	/* Pause for the tracing */
1579 	atomic_t tracing_graph_pause;
1580 #endif
1581 #ifdef CONFIG_TRACING
1582 	/* state flags for use by tracers */
1583 	unsigned long trace;
1584 	/* bitmask and counter of trace recursion */
1585 	unsigned long trace_recursion;
1586 #endif /* CONFIG_TRACING */
1587 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1588 	struct memcg_batch_info {
1589 		int do_batch;	/* incremented when batch uncharge started */
1590 		struct mem_cgroup *memcg; /* target memcg of uncharge */
1591 		unsigned long nr_pages;	/* uncharged usage */
1592 		unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1593 	} memcg_batch;
1594 #endif
1595 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1596 	atomic_t ptrace_bp_refcnt;
1597 #endif
1598 #ifdef CONFIG_UPROBES
1599 	struct uprobe_task *utask;
1600 #endif
1601 };
1602 
1603 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1604 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1605 
1606 /*
1607  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1608  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1609  * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1610  * values are inverted: lower p->prio value means higher priority.
1611  *
1612  * The MAX_USER_RT_PRIO value allows the actual maximum
1613  * RT priority to be separate from the value exported to
1614  * user-space.  This allows kernel threads to set their
1615  * priority to a value higher than any user task. Note:
1616  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1617  */
1618 
1619 #define MAX_USER_RT_PRIO	100
1620 #define MAX_RT_PRIO		MAX_USER_RT_PRIO
1621 
1622 #define MAX_PRIO		(MAX_RT_PRIO + 40)
1623 #define DEFAULT_PRIO		(MAX_RT_PRIO + 20)
1624 
1625 static inline int rt_prio(int prio)
1626 {
1627 	if (unlikely(prio < MAX_RT_PRIO))
1628 		return 1;
1629 	return 0;
1630 }
1631 
1632 static inline int rt_task(struct task_struct *p)
1633 {
1634 	return rt_prio(p->prio);
1635 }
1636 
1637 static inline struct pid *task_pid(struct task_struct *task)
1638 {
1639 	return task->pids[PIDTYPE_PID].pid;
1640 }
1641 
1642 static inline struct pid *task_tgid(struct task_struct *task)
1643 {
1644 	return task->group_leader->pids[PIDTYPE_PID].pid;
1645 }
1646 
1647 /*
1648  * Without tasklist or rcu lock it is not safe to dereference
1649  * the result of task_pgrp/task_session even if task == current,
1650  * we can race with another thread doing sys_setsid/sys_setpgid.
1651  */
1652 static inline struct pid *task_pgrp(struct task_struct *task)
1653 {
1654 	return task->group_leader->pids[PIDTYPE_PGID].pid;
1655 }
1656 
1657 static inline struct pid *task_session(struct task_struct *task)
1658 {
1659 	return task->group_leader->pids[PIDTYPE_SID].pid;
1660 }
1661 
1662 struct pid_namespace;
1663 
1664 /*
1665  * the helpers to get the task's different pids as they are seen
1666  * from various namespaces
1667  *
1668  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1669  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1670  *                     current.
1671  * task_xid_nr_ns()  : id seen from the ns specified;
1672  *
1673  * set_task_vxid()   : assigns a virtual id to a task;
1674  *
1675  * see also pid_nr() etc in include/linux/pid.h
1676  */
1677 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1678 			struct pid_namespace *ns);
1679 
1680 static inline pid_t task_pid_nr(struct task_struct *tsk)
1681 {
1682 	return tsk->pid;
1683 }
1684 
1685 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1686 					struct pid_namespace *ns)
1687 {
1688 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1689 }
1690 
1691 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1692 {
1693 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1694 }
1695 
1696 
1697 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1698 {
1699 	return tsk->tgid;
1700 }
1701 
1702 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1703 
1704 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1705 {
1706 	return pid_vnr(task_tgid(tsk));
1707 }
1708 
1709 
1710 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1711 					struct pid_namespace *ns)
1712 {
1713 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1714 }
1715 
1716 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1717 {
1718 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1719 }
1720 
1721 
1722 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1723 					struct pid_namespace *ns)
1724 {
1725 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1726 }
1727 
1728 static inline pid_t task_session_vnr(struct task_struct *tsk)
1729 {
1730 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1731 }
1732 
1733 /* obsolete, do not use */
1734 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1735 {
1736 	return task_pgrp_nr_ns(tsk, &init_pid_ns);
1737 }
1738 
1739 /**
1740  * pid_alive - check that a task structure is not stale
1741  * @p: Task structure to be checked.
1742  *
1743  * Test if a process is not yet dead (at most zombie state)
1744  * If pid_alive fails, then pointers within the task structure
1745  * can be stale and must not be dereferenced.
1746  */
1747 static inline int pid_alive(struct task_struct *p)
1748 {
1749 	return p->pids[PIDTYPE_PID].pid != NULL;
1750 }
1751 
1752 /**
1753  * is_global_init - check if a task structure is init
1754  * @tsk: Task structure to be checked.
1755  *
1756  * Check if a task structure is the first user space task the kernel created.
1757  */
1758 static inline int is_global_init(struct task_struct *tsk)
1759 {
1760 	return tsk->pid == 1;
1761 }
1762 
1763 /*
1764  * is_container_init:
1765  * check whether in the task is init in its own pid namespace.
1766  */
1767 extern int is_container_init(struct task_struct *tsk);
1768 
1769 extern struct pid *cad_pid;
1770 
1771 extern void free_task(struct task_struct *tsk);
1772 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1773 
1774 extern void __put_task_struct(struct task_struct *t);
1775 
1776 static inline void put_task_struct(struct task_struct *t)
1777 {
1778 	if (atomic_dec_and_test(&t->usage))
1779 		__put_task_struct(t);
1780 }
1781 
1782 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1783 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1784 
1785 /*
1786  * Per process flags
1787  */
1788 #define PF_EXITING	0x00000004	/* getting shut down */
1789 #define PF_EXITPIDONE	0x00000008	/* pi exit done on shut down */
1790 #define PF_VCPU		0x00000010	/* I'm a virtual CPU */
1791 #define PF_WQ_WORKER	0x00000020	/* I'm a workqueue worker */
1792 #define PF_FORKNOEXEC	0x00000040	/* forked but didn't exec */
1793 #define PF_MCE_PROCESS  0x00000080      /* process policy on mce errors */
1794 #define PF_SUPERPRIV	0x00000100	/* used super-user privileges */
1795 #define PF_DUMPCORE	0x00000200	/* dumped core */
1796 #define PF_SIGNALED	0x00000400	/* killed by a signal */
1797 #define PF_MEMALLOC	0x00000800	/* Allocating memory */
1798 #define PF_NPROC_EXCEEDED 0x00001000	/* set_user noticed that RLIMIT_NPROC was exceeded */
1799 #define PF_USED_MATH	0x00002000	/* if unset the fpu must be initialized before use */
1800 #define PF_NOFREEZE	0x00008000	/* this thread should not be frozen */
1801 #define PF_FROZEN	0x00010000	/* frozen for system suspend */
1802 #define PF_FSTRANS	0x00020000	/* inside a filesystem transaction */
1803 #define PF_KSWAPD	0x00040000	/* I am kswapd */
1804 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
1805 #define PF_KTHREAD	0x00200000	/* I am a kernel thread */
1806 #define PF_RANDOMIZE	0x00400000	/* randomize virtual address space */
1807 #define PF_SWAPWRITE	0x00800000	/* Allowed to write to swap */
1808 #define PF_SPREAD_PAGE	0x01000000	/* Spread page cache over cpuset */
1809 #define PF_SPREAD_SLAB	0x02000000	/* Spread some slab caches over cpuset */
1810 #define PF_THREAD_BOUND	0x04000000	/* Thread bound to specific cpu */
1811 #define PF_MCE_EARLY    0x08000000      /* Early kill for mce process policy */
1812 #define PF_MEMPOLICY	0x10000000	/* Non-default NUMA mempolicy */
1813 #define PF_MUTEX_TESTER	0x20000000	/* Thread belongs to the rt mutex tester */
1814 #define PF_FREEZER_SKIP	0x40000000	/* Freezer should not count it as freezable */
1815 
1816 /*
1817  * Only the _current_ task can read/write to tsk->flags, but other
1818  * tasks can access tsk->flags in readonly mode for example
1819  * with tsk_used_math (like during threaded core dumping).
1820  * There is however an exception to this rule during ptrace
1821  * or during fork: the ptracer task is allowed to write to the
1822  * child->flags of its traced child (same goes for fork, the parent
1823  * can write to the child->flags), because we're guaranteed the
1824  * child is not running and in turn not changing child->flags
1825  * at the same time the parent does it.
1826  */
1827 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1828 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1829 #define clear_used_math() clear_stopped_child_used_math(current)
1830 #define set_used_math() set_stopped_child_used_math(current)
1831 #define conditional_stopped_child_used_math(condition, child) \
1832 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1833 #define conditional_used_math(condition) \
1834 	conditional_stopped_child_used_math(condition, current)
1835 #define copy_to_stopped_child_used_math(child) \
1836 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1837 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1838 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1839 #define used_math() tsk_used_math(current)
1840 
1841 /*
1842  * task->jobctl flags
1843  */
1844 #define JOBCTL_STOP_SIGMASK	0xffff	/* signr of the last group stop */
1845 
1846 #define JOBCTL_STOP_DEQUEUED_BIT 16	/* stop signal dequeued */
1847 #define JOBCTL_STOP_PENDING_BIT	17	/* task should stop for group stop */
1848 #define JOBCTL_STOP_CONSUME_BIT	18	/* consume group stop count */
1849 #define JOBCTL_TRAP_STOP_BIT	19	/* trap for STOP */
1850 #define JOBCTL_TRAP_NOTIFY_BIT	20	/* trap for NOTIFY */
1851 #define JOBCTL_TRAPPING_BIT	21	/* switching to TRACED */
1852 #define JOBCTL_LISTENING_BIT	22	/* ptracer is listening for events */
1853 
1854 #define JOBCTL_STOP_DEQUEUED	(1 << JOBCTL_STOP_DEQUEUED_BIT)
1855 #define JOBCTL_STOP_PENDING	(1 << JOBCTL_STOP_PENDING_BIT)
1856 #define JOBCTL_STOP_CONSUME	(1 << JOBCTL_STOP_CONSUME_BIT)
1857 #define JOBCTL_TRAP_STOP	(1 << JOBCTL_TRAP_STOP_BIT)
1858 #define JOBCTL_TRAP_NOTIFY	(1 << JOBCTL_TRAP_NOTIFY_BIT)
1859 #define JOBCTL_TRAPPING		(1 << JOBCTL_TRAPPING_BIT)
1860 #define JOBCTL_LISTENING	(1 << JOBCTL_LISTENING_BIT)
1861 
1862 #define JOBCTL_TRAP_MASK	(JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1863 #define JOBCTL_PENDING_MASK	(JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1864 
1865 extern bool task_set_jobctl_pending(struct task_struct *task,
1866 				    unsigned int mask);
1867 extern void task_clear_jobctl_trapping(struct task_struct *task);
1868 extern void task_clear_jobctl_pending(struct task_struct *task,
1869 				      unsigned int mask);
1870 
1871 #ifdef CONFIG_PREEMPT_RCU
1872 
1873 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1874 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1875 
1876 static inline void rcu_copy_process(struct task_struct *p)
1877 {
1878 	p->rcu_read_lock_nesting = 0;
1879 	p->rcu_read_unlock_special = 0;
1880 #ifdef CONFIG_TREE_PREEMPT_RCU
1881 	p->rcu_blocked_node = NULL;
1882 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1883 #ifdef CONFIG_RCU_BOOST
1884 	p->rcu_boost_mutex = NULL;
1885 #endif /* #ifdef CONFIG_RCU_BOOST */
1886 	INIT_LIST_HEAD(&p->rcu_node_entry);
1887 }
1888 
1889 #else
1890 
1891 static inline void rcu_copy_process(struct task_struct *p)
1892 {
1893 }
1894 
1895 #endif
1896 
1897 static inline void tsk_restore_flags(struct task_struct *task,
1898 				unsigned long orig_flags, unsigned long flags)
1899 {
1900 	task->flags &= ~flags;
1901 	task->flags |= orig_flags & flags;
1902 }
1903 
1904 #ifdef CONFIG_SMP
1905 extern void do_set_cpus_allowed(struct task_struct *p,
1906 			       const struct cpumask *new_mask);
1907 
1908 extern int set_cpus_allowed_ptr(struct task_struct *p,
1909 				const struct cpumask *new_mask);
1910 #else
1911 static inline void do_set_cpus_allowed(struct task_struct *p,
1912 				      const struct cpumask *new_mask)
1913 {
1914 }
1915 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1916 				       const struct cpumask *new_mask)
1917 {
1918 	if (!cpumask_test_cpu(0, new_mask))
1919 		return -EINVAL;
1920 	return 0;
1921 }
1922 #endif
1923 
1924 #ifdef CONFIG_NO_HZ
1925 void calc_load_enter_idle(void);
1926 void calc_load_exit_idle(void);
1927 #else
1928 static inline void calc_load_enter_idle(void) { }
1929 static inline void calc_load_exit_idle(void) { }
1930 #endif /* CONFIG_NO_HZ */
1931 
1932 #ifndef CONFIG_CPUMASK_OFFSTACK
1933 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1934 {
1935 	return set_cpus_allowed_ptr(p, &new_mask);
1936 }
1937 #endif
1938 
1939 /*
1940  * Do not use outside of architecture code which knows its limitations.
1941  *
1942  * sched_clock() has no promise of monotonicity or bounded drift between
1943  * CPUs, use (which you should not) requires disabling IRQs.
1944  *
1945  * Please use one of the three interfaces below.
1946  */
1947 extern unsigned long long notrace sched_clock(void);
1948 /*
1949  * See the comment in kernel/sched/clock.c
1950  */
1951 extern u64 cpu_clock(int cpu);
1952 extern u64 local_clock(void);
1953 extern u64 sched_clock_cpu(int cpu);
1954 
1955 
1956 extern void sched_clock_init(void);
1957 
1958 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1959 static inline void sched_clock_tick(void)
1960 {
1961 }
1962 
1963 static inline void sched_clock_idle_sleep_event(void)
1964 {
1965 }
1966 
1967 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1968 {
1969 }
1970 #else
1971 /*
1972  * Architectures can set this to 1 if they have specified
1973  * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1974  * but then during bootup it turns out that sched_clock()
1975  * is reliable after all:
1976  */
1977 extern int sched_clock_stable;
1978 
1979 extern void sched_clock_tick(void);
1980 extern void sched_clock_idle_sleep_event(void);
1981 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1982 #endif
1983 
1984 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1985 /*
1986  * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1987  * The reason for this explicit opt-in is not to have perf penalty with
1988  * slow sched_clocks.
1989  */
1990 extern void enable_sched_clock_irqtime(void);
1991 extern void disable_sched_clock_irqtime(void);
1992 #else
1993 static inline void enable_sched_clock_irqtime(void) {}
1994 static inline void disable_sched_clock_irqtime(void) {}
1995 #endif
1996 
1997 extern unsigned long long
1998 task_sched_runtime(struct task_struct *task);
1999 
2000 /* sched_exec is called by processes performing an exec */
2001 #ifdef CONFIG_SMP
2002 extern void sched_exec(void);
2003 #else
2004 #define sched_exec()   {}
2005 #endif
2006 
2007 extern void sched_clock_idle_sleep_event(void);
2008 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2009 
2010 #ifdef CONFIG_HOTPLUG_CPU
2011 extern void idle_task_exit(void);
2012 #else
2013 static inline void idle_task_exit(void) {}
2014 #endif
2015 
2016 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
2017 extern void wake_up_idle_cpu(int cpu);
2018 #else
2019 static inline void wake_up_idle_cpu(int cpu) { }
2020 #endif
2021 
2022 extern unsigned int sysctl_sched_latency;
2023 extern unsigned int sysctl_sched_min_granularity;
2024 extern unsigned int sysctl_sched_wakeup_granularity;
2025 extern unsigned int sysctl_sched_child_runs_first;
2026 
2027 enum sched_tunable_scaling {
2028 	SCHED_TUNABLESCALING_NONE,
2029 	SCHED_TUNABLESCALING_LOG,
2030 	SCHED_TUNABLESCALING_LINEAR,
2031 	SCHED_TUNABLESCALING_END,
2032 };
2033 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
2034 
2035 #ifdef CONFIG_SCHED_DEBUG
2036 extern unsigned int sysctl_sched_migration_cost;
2037 extern unsigned int sysctl_sched_nr_migrate;
2038 extern unsigned int sysctl_sched_time_avg;
2039 extern unsigned int sysctl_timer_migration;
2040 extern unsigned int sysctl_sched_shares_window;
2041 
2042 int sched_proc_update_handler(struct ctl_table *table, int write,
2043 		void __user *buffer, size_t *length,
2044 		loff_t *ppos);
2045 #endif
2046 #ifdef CONFIG_SCHED_DEBUG
2047 static inline unsigned int get_sysctl_timer_migration(void)
2048 {
2049 	return sysctl_timer_migration;
2050 }
2051 #else
2052 static inline unsigned int get_sysctl_timer_migration(void)
2053 {
2054 	return 1;
2055 }
2056 #endif
2057 extern unsigned int sysctl_sched_rt_period;
2058 extern int sysctl_sched_rt_runtime;
2059 
2060 int sched_rt_handler(struct ctl_table *table, int write,
2061 		void __user *buffer, size_t *lenp,
2062 		loff_t *ppos);
2063 
2064 #ifdef CONFIG_SCHED_AUTOGROUP
2065 extern unsigned int sysctl_sched_autogroup_enabled;
2066 
2067 extern void sched_autogroup_create_attach(struct task_struct *p);
2068 extern void sched_autogroup_detach(struct task_struct *p);
2069 extern void sched_autogroup_fork(struct signal_struct *sig);
2070 extern void sched_autogroup_exit(struct signal_struct *sig);
2071 #ifdef CONFIG_PROC_FS
2072 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2073 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2074 #endif
2075 #else
2076 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2077 static inline void sched_autogroup_detach(struct task_struct *p) { }
2078 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2079 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2080 #endif
2081 
2082 #ifdef CONFIG_CFS_BANDWIDTH
2083 extern unsigned int sysctl_sched_cfs_bandwidth_slice;
2084 #endif
2085 
2086 #ifdef CONFIG_RT_MUTEXES
2087 extern int rt_mutex_getprio(struct task_struct *p);
2088 extern void rt_mutex_setprio(struct task_struct *p, int prio);
2089 extern void rt_mutex_adjust_pi(struct task_struct *p);
2090 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2091 {
2092 	return tsk->pi_blocked_on != NULL;
2093 }
2094 #else
2095 static inline int rt_mutex_getprio(struct task_struct *p)
2096 {
2097 	return p->normal_prio;
2098 }
2099 # define rt_mutex_adjust_pi(p)		do { } while (0)
2100 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2101 {
2102 	return false;
2103 }
2104 #endif
2105 
2106 extern bool yield_to(struct task_struct *p, bool preempt);
2107 extern void set_user_nice(struct task_struct *p, long nice);
2108 extern int task_prio(const struct task_struct *p);
2109 extern int task_nice(const struct task_struct *p);
2110 extern int can_nice(const struct task_struct *p, const int nice);
2111 extern int task_curr(const struct task_struct *p);
2112 extern int idle_cpu(int cpu);
2113 extern int sched_setscheduler(struct task_struct *, int,
2114 			      const struct sched_param *);
2115 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2116 				      const struct sched_param *);
2117 extern struct task_struct *idle_task(int cpu);
2118 /**
2119  * is_idle_task - is the specified task an idle task?
2120  * @p: the task in question.
2121  */
2122 static inline bool is_idle_task(const struct task_struct *p)
2123 {
2124 	return p->pid == 0;
2125 }
2126 extern struct task_struct *curr_task(int cpu);
2127 extern void set_curr_task(int cpu, struct task_struct *p);
2128 
2129 void yield(void);
2130 
2131 /*
2132  * The default (Linux) execution domain.
2133  */
2134 extern struct exec_domain	default_exec_domain;
2135 
2136 union thread_union {
2137 	struct thread_info thread_info;
2138 	unsigned long stack[THREAD_SIZE/sizeof(long)];
2139 };
2140 
2141 #ifndef __HAVE_ARCH_KSTACK_END
2142 static inline int kstack_end(void *addr)
2143 {
2144 	/* Reliable end of stack detection:
2145 	 * Some APM bios versions misalign the stack
2146 	 */
2147 	return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2148 }
2149 #endif
2150 
2151 extern union thread_union init_thread_union;
2152 extern struct task_struct init_task;
2153 
2154 extern struct   mm_struct init_mm;
2155 
2156 extern struct pid_namespace init_pid_ns;
2157 
2158 /*
2159  * find a task by one of its numerical ids
2160  *
2161  * find_task_by_pid_ns():
2162  *      finds a task by its pid in the specified namespace
2163  * find_task_by_vpid():
2164  *      finds a task by its virtual pid
2165  *
2166  * see also find_vpid() etc in include/linux/pid.h
2167  */
2168 
2169 extern struct task_struct *find_task_by_vpid(pid_t nr);
2170 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2171 		struct pid_namespace *ns);
2172 
2173 extern void __set_special_pids(struct pid *pid);
2174 
2175 /* per-UID process charging. */
2176 extern struct user_struct * alloc_uid(kuid_t);
2177 static inline struct user_struct *get_uid(struct user_struct *u)
2178 {
2179 	atomic_inc(&u->__count);
2180 	return u;
2181 }
2182 extern void free_uid(struct user_struct *);
2183 
2184 #include <asm/current.h>
2185 
2186 extern void xtime_update(unsigned long ticks);
2187 
2188 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2189 extern int wake_up_process(struct task_struct *tsk);
2190 extern void wake_up_new_task(struct task_struct *tsk);
2191 #ifdef CONFIG_SMP
2192  extern void kick_process(struct task_struct *tsk);
2193 #else
2194  static inline void kick_process(struct task_struct *tsk) { }
2195 #endif
2196 extern void sched_fork(struct task_struct *p);
2197 extern void sched_dead(struct task_struct *p);
2198 
2199 extern void proc_caches_init(void);
2200 extern void flush_signals(struct task_struct *);
2201 extern void __flush_signals(struct task_struct *);
2202 extern void ignore_signals(struct task_struct *);
2203 extern void flush_signal_handlers(struct task_struct *, int force_default);
2204 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2205 
2206 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2207 {
2208 	unsigned long flags;
2209 	int ret;
2210 
2211 	spin_lock_irqsave(&tsk->sighand->siglock, flags);
2212 	ret = dequeue_signal(tsk, mask, info);
2213 	spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2214 
2215 	return ret;
2216 }
2217 
2218 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2219 			      sigset_t *mask);
2220 extern void unblock_all_signals(void);
2221 extern void release_task(struct task_struct * p);
2222 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2223 extern int force_sigsegv(int, struct task_struct *);
2224 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2225 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2226 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2227 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2228 				const struct cred *, u32);
2229 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2230 extern int kill_pid(struct pid *pid, int sig, int priv);
2231 extern int kill_proc_info(int, struct siginfo *, pid_t);
2232 extern __must_check bool do_notify_parent(struct task_struct *, int);
2233 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2234 extern void force_sig(int, struct task_struct *);
2235 extern int send_sig(int, struct task_struct *, int);
2236 extern int zap_other_threads(struct task_struct *p);
2237 extern struct sigqueue *sigqueue_alloc(void);
2238 extern void sigqueue_free(struct sigqueue *);
2239 extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
2240 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2241 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2242 
2243 static inline void restore_saved_sigmask(void)
2244 {
2245 	if (test_and_clear_restore_sigmask())
2246 		__set_current_blocked(&current->saved_sigmask);
2247 }
2248 
2249 static inline sigset_t *sigmask_to_save(void)
2250 {
2251 	sigset_t *res = &current->blocked;
2252 	if (unlikely(test_restore_sigmask()))
2253 		res = &current->saved_sigmask;
2254 	return res;
2255 }
2256 
2257 static inline int kill_cad_pid(int sig, int priv)
2258 {
2259 	return kill_pid(cad_pid, sig, priv);
2260 }
2261 
2262 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
2263 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2264 #define SEND_SIG_PRIV	((struct siginfo *) 1)
2265 #define SEND_SIG_FORCED	((struct siginfo *) 2)
2266 
2267 /*
2268  * True if we are on the alternate signal stack.
2269  */
2270 static inline int on_sig_stack(unsigned long sp)
2271 {
2272 #ifdef CONFIG_STACK_GROWSUP
2273 	return sp >= current->sas_ss_sp &&
2274 		sp - current->sas_ss_sp < current->sas_ss_size;
2275 #else
2276 	return sp > current->sas_ss_sp &&
2277 		sp - current->sas_ss_sp <= current->sas_ss_size;
2278 #endif
2279 }
2280 
2281 static inline int sas_ss_flags(unsigned long sp)
2282 {
2283 	return (current->sas_ss_size == 0 ? SS_DISABLE
2284 		: on_sig_stack(sp) ? SS_ONSTACK : 0);
2285 }
2286 
2287 /*
2288  * Routines for handling mm_structs
2289  */
2290 extern struct mm_struct * mm_alloc(void);
2291 
2292 /* mmdrop drops the mm and the page tables */
2293 extern void __mmdrop(struct mm_struct *);
2294 static inline void mmdrop(struct mm_struct * mm)
2295 {
2296 	if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2297 		__mmdrop(mm);
2298 }
2299 
2300 /* mmput gets rid of the mappings and all user-space */
2301 extern void mmput(struct mm_struct *);
2302 /* Grab a reference to a task's mm, if it is not already going away */
2303 extern struct mm_struct *get_task_mm(struct task_struct *task);
2304 /*
2305  * Grab a reference to a task's mm, if it is not already going away
2306  * and ptrace_may_access with the mode parameter passed to it
2307  * succeeds.
2308  */
2309 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2310 /* Remove the current tasks stale references to the old mm_struct */
2311 extern void mm_release(struct task_struct *, struct mm_struct *);
2312 /* Allocate a new mm structure and copy contents from tsk->mm */
2313 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2314 
2315 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2316 			struct task_struct *, struct pt_regs *);
2317 extern void flush_thread(void);
2318 extern void exit_thread(void);
2319 
2320 extern void exit_files(struct task_struct *);
2321 extern void __cleanup_sighand(struct sighand_struct *);
2322 
2323 extern void exit_itimers(struct signal_struct *);
2324 extern void flush_itimer_signals(void);
2325 
2326 extern void do_group_exit(int);
2327 
2328 extern void daemonize(const char *, ...);
2329 extern int allow_signal(int);
2330 extern int disallow_signal(int);
2331 
2332 extern int do_execve(const char *,
2333 		     const char __user * const __user *,
2334 		     const char __user * const __user *, struct pt_regs *);
2335 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2336 struct task_struct *fork_idle(int);
2337 
2338 extern void set_task_comm(struct task_struct *tsk, char *from);
2339 extern char *get_task_comm(char *to, struct task_struct *tsk);
2340 
2341 #ifdef CONFIG_SMP
2342 void scheduler_ipi(void);
2343 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2344 #else
2345 static inline void scheduler_ipi(void) { }
2346 static inline unsigned long wait_task_inactive(struct task_struct *p,
2347 					       long match_state)
2348 {
2349 	return 1;
2350 }
2351 #endif
2352 
2353 #define next_task(p) \
2354 	list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2355 
2356 #define for_each_process(p) \
2357 	for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2358 
2359 extern bool current_is_single_threaded(void);
2360 
2361 /*
2362  * Careful: do_each_thread/while_each_thread is a double loop so
2363  *          'break' will not work as expected - use goto instead.
2364  */
2365 #define do_each_thread(g, t) \
2366 	for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2367 
2368 #define while_each_thread(g, t) \
2369 	while ((t = next_thread(t)) != g)
2370 
2371 static inline int get_nr_threads(struct task_struct *tsk)
2372 {
2373 	return tsk->signal->nr_threads;
2374 }
2375 
2376 static inline bool thread_group_leader(struct task_struct *p)
2377 {
2378 	return p->exit_signal >= 0;
2379 }
2380 
2381 /* Do to the insanities of de_thread it is possible for a process
2382  * to have the pid of the thread group leader without actually being
2383  * the thread group leader.  For iteration through the pids in proc
2384  * all we care about is that we have a task with the appropriate
2385  * pid, we don't actually care if we have the right task.
2386  */
2387 static inline int has_group_leader_pid(struct task_struct *p)
2388 {
2389 	return p->pid == p->tgid;
2390 }
2391 
2392 static inline
2393 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2394 {
2395 	return p1->tgid == p2->tgid;
2396 }
2397 
2398 static inline struct task_struct *next_thread(const struct task_struct *p)
2399 {
2400 	return list_entry_rcu(p->thread_group.next,
2401 			      struct task_struct, thread_group);
2402 }
2403 
2404 static inline int thread_group_empty(struct task_struct *p)
2405 {
2406 	return list_empty(&p->thread_group);
2407 }
2408 
2409 #define delay_group_leader(p) \
2410 		(thread_group_leader(p) && !thread_group_empty(p))
2411 
2412 /*
2413  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2414  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2415  * pins the final release of task.io_context.  Also protects ->cpuset and
2416  * ->cgroup.subsys[]. And ->vfork_done.
2417  *
2418  * Nests both inside and outside of read_lock(&tasklist_lock).
2419  * It must not be nested with write_lock_irq(&tasklist_lock),
2420  * neither inside nor outside.
2421  */
2422 static inline void task_lock(struct task_struct *p)
2423 {
2424 	spin_lock(&p->alloc_lock);
2425 }
2426 
2427 static inline void task_unlock(struct task_struct *p)
2428 {
2429 	spin_unlock(&p->alloc_lock);
2430 }
2431 
2432 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2433 							unsigned long *flags);
2434 
2435 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2436 						       unsigned long *flags)
2437 {
2438 	struct sighand_struct *ret;
2439 
2440 	ret = __lock_task_sighand(tsk, flags);
2441 	(void)__cond_lock(&tsk->sighand->siglock, ret);
2442 	return ret;
2443 }
2444 
2445 static inline void unlock_task_sighand(struct task_struct *tsk,
2446 						unsigned long *flags)
2447 {
2448 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2449 }
2450 
2451 #ifdef CONFIG_CGROUPS
2452 static inline void threadgroup_change_begin(struct task_struct *tsk)
2453 {
2454 	down_read(&tsk->signal->group_rwsem);
2455 }
2456 static inline void threadgroup_change_end(struct task_struct *tsk)
2457 {
2458 	up_read(&tsk->signal->group_rwsem);
2459 }
2460 
2461 /**
2462  * threadgroup_lock - lock threadgroup
2463  * @tsk: member task of the threadgroup to lock
2464  *
2465  * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
2466  * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2467  * perform exec.  This is useful for cases where the threadgroup needs to
2468  * stay stable across blockable operations.
2469  *
2470  * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2471  * synchronization.  While held, no new task will be added to threadgroup
2472  * and no existing live task will have its PF_EXITING set.
2473  *
2474  * During exec, a task goes and puts its thread group through unusual
2475  * changes.  After de-threading, exclusive access is assumed to resources
2476  * which are usually shared by tasks in the same group - e.g. sighand may
2477  * be replaced with a new one.  Also, the exec'ing task takes over group
2478  * leader role including its pid.  Exclude these changes while locked by
2479  * grabbing cred_guard_mutex which is used to synchronize exec path.
2480  */
2481 static inline void threadgroup_lock(struct task_struct *tsk)
2482 {
2483 	/*
2484 	 * exec uses exit for de-threading nesting group_rwsem inside
2485 	 * cred_guard_mutex. Grab cred_guard_mutex first.
2486 	 */
2487 	mutex_lock(&tsk->signal->cred_guard_mutex);
2488 	down_write(&tsk->signal->group_rwsem);
2489 }
2490 
2491 /**
2492  * threadgroup_unlock - unlock threadgroup
2493  * @tsk: member task of the threadgroup to unlock
2494  *
2495  * Reverse threadgroup_lock().
2496  */
2497 static inline void threadgroup_unlock(struct task_struct *tsk)
2498 {
2499 	up_write(&tsk->signal->group_rwsem);
2500 	mutex_unlock(&tsk->signal->cred_guard_mutex);
2501 }
2502 #else
2503 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2504 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2505 static inline void threadgroup_lock(struct task_struct *tsk) {}
2506 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2507 #endif
2508 
2509 #ifndef __HAVE_THREAD_FUNCTIONS
2510 
2511 #define task_thread_info(task)	((struct thread_info *)(task)->stack)
2512 #define task_stack_page(task)	((task)->stack)
2513 
2514 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2515 {
2516 	*task_thread_info(p) = *task_thread_info(org);
2517 	task_thread_info(p)->task = p;
2518 }
2519 
2520 static inline unsigned long *end_of_stack(struct task_struct *p)
2521 {
2522 	return (unsigned long *)(task_thread_info(p) + 1);
2523 }
2524 
2525 #endif
2526 
2527 static inline int object_is_on_stack(void *obj)
2528 {
2529 	void *stack = task_stack_page(current);
2530 
2531 	return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2532 }
2533 
2534 extern void thread_info_cache_init(void);
2535 
2536 #ifdef CONFIG_DEBUG_STACK_USAGE
2537 static inline unsigned long stack_not_used(struct task_struct *p)
2538 {
2539 	unsigned long *n = end_of_stack(p);
2540 
2541 	do { 	/* Skip over canary */
2542 		n++;
2543 	} while (!*n);
2544 
2545 	return (unsigned long)n - (unsigned long)end_of_stack(p);
2546 }
2547 #endif
2548 
2549 /* set thread flags in other task's structures
2550  * - see asm/thread_info.h for TIF_xxxx flags available
2551  */
2552 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2553 {
2554 	set_ti_thread_flag(task_thread_info(tsk), flag);
2555 }
2556 
2557 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2558 {
2559 	clear_ti_thread_flag(task_thread_info(tsk), flag);
2560 }
2561 
2562 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2563 {
2564 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2565 }
2566 
2567 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2568 {
2569 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2570 }
2571 
2572 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2573 {
2574 	return test_ti_thread_flag(task_thread_info(tsk), flag);
2575 }
2576 
2577 static inline void set_tsk_need_resched(struct task_struct *tsk)
2578 {
2579 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2580 }
2581 
2582 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2583 {
2584 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2585 }
2586 
2587 static inline int test_tsk_need_resched(struct task_struct *tsk)
2588 {
2589 	return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2590 }
2591 
2592 static inline int restart_syscall(void)
2593 {
2594 	set_tsk_thread_flag(current, TIF_SIGPENDING);
2595 	return -ERESTARTNOINTR;
2596 }
2597 
2598 static inline int signal_pending(struct task_struct *p)
2599 {
2600 	return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2601 }
2602 
2603 static inline int __fatal_signal_pending(struct task_struct *p)
2604 {
2605 	return unlikely(sigismember(&p->pending.signal, SIGKILL));
2606 }
2607 
2608 static inline int fatal_signal_pending(struct task_struct *p)
2609 {
2610 	return signal_pending(p) && __fatal_signal_pending(p);
2611 }
2612 
2613 static inline int signal_pending_state(long state, struct task_struct *p)
2614 {
2615 	if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2616 		return 0;
2617 	if (!signal_pending(p))
2618 		return 0;
2619 
2620 	return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2621 }
2622 
2623 static inline int need_resched(void)
2624 {
2625 	return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2626 }
2627 
2628 /*
2629  * cond_resched() and cond_resched_lock(): latency reduction via
2630  * explicit rescheduling in places that are safe. The return
2631  * value indicates whether a reschedule was done in fact.
2632  * cond_resched_lock() will drop the spinlock before scheduling,
2633  * cond_resched_softirq() will enable bhs before scheduling.
2634  */
2635 extern int _cond_resched(void);
2636 
2637 #define cond_resched() ({			\
2638 	__might_sleep(__FILE__, __LINE__, 0);	\
2639 	_cond_resched();			\
2640 })
2641 
2642 extern int __cond_resched_lock(spinlock_t *lock);
2643 
2644 #ifdef CONFIG_PREEMPT_COUNT
2645 #define PREEMPT_LOCK_OFFSET	PREEMPT_OFFSET
2646 #else
2647 #define PREEMPT_LOCK_OFFSET	0
2648 #endif
2649 
2650 #define cond_resched_lock(lock) ({				\
2651 	__might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);	\
2652 	__cond_resched_lock(lock);				\
2653 })
2654 
2655 extern int __cond_resched_softirq(void);
2656 
2657 #define cond_resched_softirq() ({					\
2658 	__might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET);	\
2659 	__cond_resched_softirq();					\
2660 })
2661 
2662 /*
2663  * Does a critical section need to be broken due to another
2664  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2665  * but a general need for low latency)
2666  */
2667 static inline int spin_needbreak(spinlock_t *lock)
2668 {
2669 #ifdef CONFIG_PREEMPT
2670 	return spin_is_contended(lock);
2671 #else
2672 	return 0;
2673 #endif
2674 }
2675 
2676 /*
2677  * Thread group CPU time accounting.
2678  */
2679 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2680 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2681 
2682 static inline void thread_group_cputime_init(struct signal_struct *sig)
2683 {
2684 	raw_spin_lock_init(&sig->cputimer.lock);
2685 }
2686 
2687 /*
2688  * Reevaluate whether the task has signals pending delivery.
2689  * Wake the task if so.
2690  * This is required every time the blocked sigset_t changes.
2691  * callers must hold sighand->siglock.
2692  */
2693 extern void recalc_sigpending_and_wake(struct task_struct *t);
2694 extern void recalc_sigpending(void);
2695 
2696 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2697 
2698 /*
2699  * Wrappers for p->thread_info->cpu access. No-op on UP.
2700  */
2701 #ifdef CONFIG_SMP
2702 
2703 static inline unsigned int task_cpu(const struct task_struct *p)
2704 {
2705 	return task_thread_info(p)->cpu;
2706 }
2707 
2708 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2709 
2710 #else
2711 
2712 static inline unsigned int task_cpu(const struct task_struct *p)
2713 {
2714 	return 0;
2715 }
2716 
2717 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2718 {
2719 }
2720 
2721 #endif /* CONFIG_SMP */
2722 
2723 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2724 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2725 
2726 extern void normalize_rt_tasks(void);
2727 
2728 #ifdef CONFIG_CGROUP_SCHED
2729 
2730 extern struct task_group root_task_group;
2731 
2732 extern struct task_group *sched_create_group(struct task_group *parent);
2733 extern void sched_destroy_group(struct task_group *tg);
2734 extern void sched_move_task(struct task_struct *tsk);
2735 #ifdef CONFIG_FAIR_GROUP_SCHED
2736 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2737 extern unsigned long sched_group_shares(struct task_group *tg);
2738 #endif
2739 #ifdef CONFIG_RT_GROUP_SCHED
2740 extern int sched_group_set_rt_runtime(struct task_group *tg,
2741 				      long rt_runtime_us);
2742 extern long sched_group_rt_runtime(struct task_group *tg);
2743 extern int sched_group_set_rt_period(struct task_group *tg,
2744 				      long rt_period_us);
2745 extern long sched_group_rt_period(struct task_group *tg);
2746 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2747 #endif
2748 #endif /* CONFIG_CGROUP_SCHED */
2749 
2750 extern int task_can_switch_user(struct user_struct *up,
2751 					struct task_struct *tsk);
2752 
2753 #ifdef CONFIG_TASK_XACCT
2754 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2755 {
2756 	tsk->ioac.rchar += amt;
2757 }
2758 
2759 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2760 {
2761 	tsk->ioac.wchar += amt;
2762 }
2763 
2764 static inline void inc_syscr(struct task_struct *tsk)
2765 {
2766 	tsk->ioac.syscr++;
2767 }
2768 
2769 static inline void inc_syscw(struct task_struct *tsk)
2770 {
2771 	tsk->ioac.syscw++;
2772 }
2773 #else
2774 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2775 {
2776 }
2777 
2778 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2779 {
2780 }
2781 
2782 static inline void inc_syscr(struct task_struct *tsk)
2783 {
2784 }
2785 
2786 static inline void inc_syscw(struct task_struct *tsk)
2787 {
2788 }
2789 #endif
2790 
2791 #ifndef TASK_SIZE_OF
2792 #define TASK_SIZE_OF(tsk)	TASK_SIZE
2793 #endif
2794 
2795 #ifdef CONFIG_MM_OWNER
2796 extern void mm_update_next_owner(struct mm_struct *mm);
2797 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2798 #else
2799 static inline void mm_update_next_owner(struct mm_struct *mm)
2800 {
2801 }
2802 
2803 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2804 {
2805 }
2806 #endif /* CONFIG_MM_OWNER */
2807 
2808 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2809 		unsigned int limit)
2810 {
2811 	return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2812 }
2813 
2814 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2815 		unsigned int limit)
2816 {
2817 	return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2818 }
2819 
2820 static inline unsigned long rlimit(unsigned int limit)
2821 {
2822 	return task_rlimit(current, limit);
2823 }
2824 
2825 static inline unsigned long rlimit_max(unsigned int limit)
2826 {
2827 	return task_rlimit_max(current, limit);
2828 }
2829 
2830 #endif /* __KERNEL__ */
2831 
2832 #endif
2833