xref: /linux/include/linux/sched.h (revision 5148fa52a12fa1b97c730b2fe321f2aad7ea041c)
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 
94 #include <asm/processor.h>
95 
96 struct exec_domain;
97 struct futex_pi_state;
98 struct robust_list_head;
99 struct bio_list;
100 struct fs_struct;
101 struct perf_event_context;
102 struct blk_plug;
103 
104 /*
105  * List of flags we want to share for kernel threads,
106  * if only because they are not used by them anyway.
107  */
108 #define CLONE_KERNEL	(CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
109 
110 /*
111  * These are the constant used to fake the fixed-point load-average
112  * counting. Some notes:
113  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
114  *    a load-average precision of 10 bits integer + 11 bits fractional
115  *  - if you want to count load-averages more often, you need more
116  *    precision, or rounding will get you. With 2-second counting freq,
117  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
118  *    11 bit fractions.
119  */
120 extern unsigned long avenrun[];		/* Load averages */
121 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
122 
123 #define FSHIFT		11		/* nr of bits of precision */
124 #define FIXED_1		(1<<FSHIFT)	/* 1.0 as fixed-point */
125 #define LOAD_FREQ	(5*HZ+1)	/* 5 sec intervals */
126 #define EXP_1		1884		/* 1/exp(5sec/1min) as fixed-point */
127 #define EXP_5		2014		/* 1/exp(5sec/5min) */
128 #define EXP_15		2037		/* 1/exp(5sec/15min) */
129 
130 #define CALC_LOAD(load,exp,n) \
131 	load *= exp; \
132 	load += n*(FIXED_1-exp); \
133 	load >>= FSHIFT;
134 
135 extern unsigned long total_forks;
136 extern int nr_threads;
137 DECLARE_PER_CPU(unsigned long, process_counts);
138 extern int nr_processes(void);
139 extern unsigned long nr_running(void);
140 extern unsigned long nr_uninterruptible(void);
141 extern unsigned long nr_iowait(void);
142 extern unsigned long nr_iowait_cpu(int cpu);
143 extern unsigned long this_cpu_load(void);
144 
145 
146 extern void calc_global_load(unsigned long ticks);
147 
148 extern unsigned long get_parent_ip(unsigned long addr);
149 
150 struct seq_file;
151 struct cfs_rq;
152 struct task_group;
153 #ifdef CONFIG_SCHED_DEBUG
154 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
155 extern void proc_sched_set_task(struct task_struct *p);
156 extern void
157 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
158 #else
159 static inline void
160 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
161 {
162 }
163 static inline void proc_sched_set_task(struct task_struct *p)
164 {
165 }
166 static inline void
167 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
168 {
169 }
170 #endif
171 
172 /*
173  * Task state bitmask. NOTE! These bits are also
174  * encoded in fs/proc/array.c: get_task_state().
175  *
176  * We have two separate sets of flags: task->state
177  * is about runnability, while task->exit_state are
178  * about the task exiting. Confusing, but this way
179  * modifying one set can't modify the other one by
180  * mistake.
181  */
182 #define TASK_RUNNING		0
183 #define TASK_INTERRUPTIBLE	1
184 #define TASK_UNINTERRUPTIBLE	2
185 #define __TASK_STOPPED		4
186 #define __TASK_TRACED		8
187 /* in tsk->exit_state */
188 #define EXIT_ZOMBIE		16
189 #define EXIT_DEAD		32
190 /* in tsk->state again */
191 #define TASK_DEAD		64
192 #define TASK_WAKEKILL		128
193 #define TASK_WAKING		256
194 #define TASK_STATE_MAX		512
195 
196 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
197 
198 extern char ___assert_task_state[1 - 2*!!(
199 		sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
200 
201 /* Convenience macros for the sake of set_task_state */
202 #define TASK_KILLABLE		(TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
203 #define TASK_STOPPED		(TASK_WAKEKILL | __TASK_STOPPED)
204 #define TASK_TRACED		(TASK_WAKEKILL | __TASK_TRACED)
205 
206 /* Convenience macros for the sake of wake_up */
207 #define TASK_NORMAL		(TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
208 #define TASK_ALL		(TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
209 
210 /* get_task_state() */
211 #define TASK_REPORT		(TASK_RUNNING | TASK_INTERRUPTIBLE | \
212 				 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
213 				 __TASK_TRACED)
214 
215 #define task_is_traced(task)	((task->state & __TASK_TRACED) != 0)
216 #define task_is_stopped(task)	((task->state & __TASK_STOPPED) != 0)
217 #define task_is_dead(task)	((task)->exit_state != 0)
218 #define task_is_stopped_or_traced(task)	\
219 			((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
220 #define task_contributes_to_load(task)	\
221 				((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
222 				 (task->flags & PF_FROZEN) == 0)
223 
224 #define __set_task_state(tsk, state_value)		\
225 	do { (tsk)->state = (state_value); } while (0)
226 #define set_task_state(tsk, state_value)		\
227 	set_mb((tsk)->state, (state_value))
228 
229 /*
230  * set_current_state() includes a barrier so that the write of current->state
231  * is correctly serialised wrt the caller's subsequent test of whether to
232  * actually sleep:
233  *
234  *	set_current_state(TASK_UNINTERRUPTIBLE);
235  *	if (do_i_need_to_sleep())
236  *		schedule();
237  *
238  * If the caller does not need such serialisation then use __set_current_state()
239  */
240 #define __set_current_state(state_value)			\
241 	do { current->state = (state_value); } while (0)
242 #define set_current_state(state_value)		\
243 	set_mb(current->state, (state_value))
244 
245 /* Task command name length */
246 #define TASK_COMM_LEN 16
247 
248 #include <linux/spinlock.h>
249 
250 /*
251  * This serializes "schedule()" and also protects
252  * the run-queue from deletions/modifications (but
253  * _adding_ to the beginning of the run-queue has
254  * a separate lock).
255  */
256 extern rwlock_t tasklist_lock;
257 extern spinlock_t mmlist_lock;
258 
259 struct task_struct;
260 
261 #ifdef CONFIG_PROVE_RCU
262 extern int lockdep_tasklist_lock_is_held(void);
263 #endif /* #ifdef CONFIG_PROVE_RCU */
264 
265 extern void sched_init(void);
266 extern void sched_init_smp(void);
267 extern asmlinkage void schedule_tail(struct task_struct *prev);
268 extern void init_idle(struct task_struct *idle, int cpu);
269 extern void init_idle_bootup_task(struct task_struct *idle);
270 
271 extern int runqueue_is_locked(int cpu);
272 
273 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
274 extern void select_nohz_load_balancer(int stop_tick);
275 extern void set_cpu_sd_state_idle(void);
276 extern int get_nohz_timer_target(void);
277 #else
278 static inline void select_nohz_load_balancer(int stop_tick) { }
279 static inline void set_cpu_sd_state_idle(void) { }
280 #endif
281 
282 /*
283  * Only dump TASK_* tasks. (0 for all tasks)
284  */
285 extern void show_state_filter(unsigned long state_filter);
286 
287 static inline void show_state(void)
288 {
289 	show_state_filter(0);
290 }
291 
292 extern void show_regs(struct pt_regs *);
293 
294 /*
295  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
296  * task), SP is the stack pointer of the first frame that should be shown in the back
297  * trace (or NULL if the entire call-chain of the task should be shown).
298  */
299 extern void show_stack(struct task_struct *task, unsigned long *sp);
300 
301 void io_schedule(void);
302 long io_schedule_timeout(long timeout);
303 
304 extern void cpu_init (void);
305 extern void trap_init(void);
306 extern void update_process_times(int user);
307 extern void scheduler_tick(void);
308 
309 extern void sched_show_task(struct task_struct *p);
310 
311 #ifdef CONFIG_LOCKUP_DETECTOR
312 extern void touch_softlockup_watchdog(void);
313 extern void touch_softlockup_watchdog_sync(void);
314 extern void touch_all_softlockup_watchdogs(void);
315 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
316 				  void __user *buffer,
317 				  size_t *lenp, loff_t *ppos);
318 extern unsigned int  softlockup_panic;
319 void lockup_detector_init(void);
320 #else
321 static inline void touch_softlockup_watchdog(void)
322 {
323 }
324 static inline void touch_softlockup_watchdog_sync(void)
325 {
326 }
327 static inline void touch_all_softlockup_watchdogs(void)
328 {
329 }
330 static inline void lockup_detector_init(void)
331 {
332 }
333 #endif
334 
335 #ifdef CONFIG_DETECT_HUNG_TASK
336 extern unsigned int  sysctl_hung_task_panic;
337 extern unsigned long sysctl_hung_task_check_count;
338 extern unsigned long sysctl_hung_task_timeout_secs;
339 extern unsigned long sysctl_hung_task_warnings;
340 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
341 					 void __user *buffer,
342 					 size_t *lenp, loff_t *ppos);
343 #else
344 /* Avoid need for ifdefs elsewhere in the code */
345 enum { sysctl_hung_task_timeout_secs = 0 };
346 #endif
347 
348 /* Attach to any functions which should be ignored in wchan output. */
349 #define __sched		__attribute__((__section__(".sched.text")))
350 
351 /* Linker adds these: start and end of __sched functions */
352 extern char __sched_text_start[], __sched_text_end[];
353 
354 /* Is this address in the __sched functions? */
355 extern int in_sched_functions(unsigned long addr);
356 
357 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
358 extern signed long schedule_timeout(signed long timeout);
359 extern signed long schedule_timeout_interruptible(signed long timeout);
360 extern signed long schedule_timeout_killable(signed long timeout);
361 extern signed long schedule_timeout_uninterruptible(signed long timeout);
362 asmlinkage void schedule(void);
363 extern void schedule_preempt_disabled(void);
364 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
365 
366 struct nsproxy;
367 struct user_namespace;
368 
369 /*
370  * Default maximum number of active map areas, this limits the number of vmas
371  * per mm struct. Users can overwrite this number by sysctl but there is a
372  * problem.
373  *
374  * When a program's coredump is generated as ELF format, a section is created
375  * per a vma. In ELF, the number of sections is represented in unsigned short.
376  * This means the number of sections should be smaller than 65535 at coredump.
377  * Because the kernel adds some informative sections to a image of program at
378  * generating coredump, we need some margin. The number of extra sections is
379  * 1-3 now and depends on arch. We use "5" as safe margin, here.
380  */
381 #define MAPCOUNT_ELF_CORE_MARGIN	(5)
382 #define DEFAULT_MAX_MAP_COUNT	(USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
383 
384 extern int sysctl_max_map_count;
385 
386 #include <linux/aio.h>
387 
388 #ifdef CONFIG_MMU
389 extern void arch_pick_mmap_layout(struct mm_struct *mm);
390 extern unsigned long
391 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
392 		       unsigned long, unsigned long);
393 extern unsigned long
394 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
395 			  unsigned long len, unsigned long pgoff,
396 			  unsigned long flags);
397 extern void arch_unmap_area(struct mm_struct *, unsigned long);
398 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
399 #else
400 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
401 #endif
402 
403 
404 extern void set_dumpable(struct mm_struct *mm, int value);
405 extern int get_dumpable(struct mm_struct *mm);
406 
407 /* mm flags */
408 /* dumpable bits */
409 #define MMF_DUMPABLE      0  /* core dump is permitted */
410 #define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
411 
412 #define MMF_DUMPABLE_BITS 2
413 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
414 
415 /* coredump filter bits */
416 #define MMF_DUMP_ANON_PRIVATE	2
417 #define MMF_DUMP_ANON_SHARED	3
418 #define MMF_DUMP_MAPPED_PRIVATE	4
419 #define MMF_DUMP_MAPPED_SHARED	5
420 #define MMF_DUMP_ELF_HEADERS	6
421 #define MMF_DUMP_HUGETLB_PRIVATE 7
422 #define MMF_DUMP_HUGETLB_SHARED  8
423 
424 #define MMF_DUMP_FILTER_SHIFT	MMF_DUMPABLE_BITS
425 #define MMF_DUMP_FILTER_BITS	7
426 #define MMF_DUMP_FILTER_MASK \
427 	(((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
428 #define MMF_DUMP_FILTER_DEFAULT \
429 	((1 << MMF_DUMP_ANON_PRIVATE) |	(1 << MMF_DUMP_ANON_SHARED) |\
430 	 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
431 
432 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
433 # define MMF_DUMP_MASK_DEFAULT_ELF	(1 << MMF_DUMP_ELF_HEADERS)
434 #else
435 # define MMF_DUMP_MASK_DEFAULT_ELF	0
436 #endif
437 					/* leave room for more dump flags */
438 #define MMF_VM_MERGEABLE	16	/* KSM may merge identical pages */
439 #define MMF_VM_HUGEPAGE		17	/* set when VM_HUGEPAGE is set on vma */
440 
441 #define MMF_INIT_MASK		(MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
442 
443 struct sighand_struct {
444 	atomic_t		count;
445 	struct k_sigaction	action[_NSIG];
446 	spinlock_t		siglock;
447 	wait_queue_head_t	signalfd_wqh;
448 };
449 
450 struct pacct_struct {
451 	int			ac_flag;
452 	long			ac_exitcode;
453 	unsigned long		ac_mem;
454 	cputime_t		ac_utime, ac_stime;
455 	unsigned long		ac_minflt, ac_majflt;
456 };
457 
458 struct cpu_itimer {
459 	cputime_t expires;
460 	cputime_t incr;
461 	u32 error;
462 	u32 incr_error;
463 };
464 
465 /**
466  * struct task_cputime - collected CPU time counts
467  * @utime:		time spent in user mode, in &cputime_t units
468  * @stime:		time spent in kernel mode, in &cputime_t units
469  * @sum_exec_runtime:	total time spent on the CPU, in nanoseconds
470  *
471  * This structure groups together three kinds of CPU time that are
472  * tracked for threads and thread groups.  Most things considering
473  * CPU time want to group these counts together and treat all three
474  * of them in parallel.
475  */
476 struct task_cputime {
477 	cputime_t utime;
478 	cputime_t stime;
479 	unsigned long long sum_exec_runtime;
480 };
481 /* Alternate field names when used to cache expirations. */
482 #define prof_exp	stime
483 #define virt_exp	utime
484 #define sched_exp	sum_exec_runtime
485 
486 #define INIT_CPUTIME	\
487 	(struct task_cputime) {					\
488 		.utime = 0,					\
489 		.stime = 0,					\
490 		.sum_exec_runtime = 0,				\
491 	}
492 
493 /*
494  * Disable preemption until the scheduler is running.
495  * Reset by start_kernel()->sched_init()->init_idle().
496  *
497  * We include PREEMPT_ACTIVE to avoid cond_resched() from working
498  * before the scheduler is active -- see should_resched().
499  */
500 #define INIT_PREEMPT_COUNT	(1 + PREEMPT_ACTIVE)
501 
502 /**
503  * struct thread_group_cputimer - thread group interval timer counts
504  * @cputime:		thread group interval timers.
505  * @running:		non-zero when there are timers running and
506  * 			@cputime receives updates.
507  * @lock:		lock for fields in this struct.
508  *
509  * This structure contains the version of task_cputime, above, that is
510  * used for thread group CPU timer calculations.
511  */
512 struct thread_group_cputimer {
513 	struct task_cputime cputime;
514 	int running;
515 	raw_spinlock_t lock;
516 };
517 
518 #include <linux/rwsem.h>
519 struct autogroup;
520 
521 /*
522  * NOTE! "signal_struct" does not have its own
523  * locking, because a shared signal_struct always
524  * implies a shared sighand_struct, so locking
525  * sighand_struct is always a proper superset of
526  * the locking of signal_struct.
527  */
528 struct signal_struct {
529 	atomic_t		sigcnt;
530 	atomic_t		live;
531 	int			nr_threads;
532 
533 	wait_queue_head_t	wait_chldexit;	/* for wait4() */
534 
535 	/* current thread group signal load-balancing target: */
536 	struct task_struct	*curr_target;
537 
538 	/* shared signal handling: */
539 	struct sigpending	shared_pending;
540 
541 	/* thread group exit support */
542 	int			group_exit_code;
543 	/* overloaded:
544 	 * - notify group_exit_task when ->count is equal to notify_count
545 	 * - everyone except group_exit_task is stopped during signal delivery
546 	 *   of fatal signals, group_exit_task processes the signal.
547 	 */
548 	int			notify_count;
549 	struct task_struct	*group_exit_task;
550 
551 	/* thread group stop support, overloads group_exit_code too */
552 	int			group_stop_count;
553 	unsigned int		flags; /* see SIGNAL_* flags below */
554 
555 	/*
556 	 * PR_SET_CHILD_SUBREAPER marks a process, like a service
557 	 * manager, to re-parent orphan (double-forking) child processes
558 	 * to this process instead of 'init'. The service manager is
559 	 * able to receive SIGCHLD signals and is able to investigate
560 	 * the process until it calls wait(). All children of this
561 	 * process will inherit a flag if they should look for a
562 	 * child_subreaper process at exit.
563 	 */
564 	unsigned int		is_child_subreaper:1;
565 	unsigned int		has_child_subreaper:1;
566 
567 	/* POSIX.1b Interval Timers */
568 	struct list_head posix_timers;
569 
570 	/* ITIMER_REAL timer for the process */
571 	struct hrtimer real_timer;
572 	struct pid *leader_pid;
573 	ktime_t it_real_incr;
574 
575 	/*
576 	 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
577 	 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
578 	 * values are defined to 0 and 1 respectively
579 	 */
580 	struct cpu_itimer it[2];
581 
582 	/*
583 	 * Thread group totals for process CPU timers.
584 	 * See thread_group_cputimer(), et al, for details.
585 	 */
586 	struct thread_group_cputimer cputimer;
587 
588 	/* Earliest-expiration cache. */
589 	struct task_cputime cputime_expires;
590 
591 	struct list_head cpu_timers[3];
592 
593 	struct pid *tty_old_pgrp;
594 
595 	/* boolean value for session group leader */
596 	int leader;
597 
598 	struct tty_struct *tty; /* NULL if no tty */
599 
600 #ifdef CONFIG_SCHED_AUTOGROUP
601 	struct autogroup *autogroup;
602 #endif
603 	/*
604 	 * Cumulative resource counters for dead threads in the group,
605 	 * and for reaped dead child processes forked by this group.
606 	 * Live threads maintain their own counters and add to these
607 	 * in __exit_signal, except for the group leader.
608 	 */
609 	cputime_t utime, stime, cutime, cstime;
610 	cputime_t gtime;
611 	cputime_t cgtime;
612 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
613 	cputime_t prev_utime, prev_stime;
614 #endif
615 	unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
616 	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
617 	unsigned long inblock, oublock, cinblock, coublock;
618 	unsigned long maxrss, cmaxrss;
619 	struct task_io_accounting ioac;
620 
621 	/*
622 	 * Cumulative ns of schedule CPU time fo dead threads in the
623 	 * group, not including a zombie group leader, (This only differs
624 	 * from jiffies_to_ns(utime + stime) if sched_clock uses something
625 	 * other than jiffies.)
626 	 */
627 	unsigned long long sum_sched_runtime;
628 
629 	/*
630 	 * We don't bother to synchronize most readers of this at all,
631 	 * because there is no reader checking a limit that actually needs
632 	 * to get both rlim_cur and rlim_max atomically, and either one
633 	 * alone is a single word that can safely be read normally.
634 	 * getrlimit/setrlimit use task_lock(current->group_leader) to
635 	 * protect this instead of the siglock, because they really
636 	 * have no need to disable irqs.
637 	 */
638 	struct rlimit rlim[RLIM_NLIMITS];
639 
640 #ifdef CONFIG_BSD_PROCESS_ACCT
641 	struct pacct_struct pacct;	/* per-process accounting information */
642 #endif
643 #ifdef CONFIG_TASKSTATS
644 	struct taskstats *stats;
645 #endif
646 #ifdef CONFIG_AUDIT
647 	unsigned audit_tty;
648 	struct tty_audit_buf *tty_audit_buf;
649 #endif
650 #ifdef CONFIG_CGROUPS
651 	/*
652 	 * group_rwsem prevents new tasks from entering the threadgroup and
653 	 * member tasks from exiting,a more specifically, setting of
654 	 * PF_EXITING.  fork and exit paths are protected with this rwsem
655 	 * using threadgroup_change_begin/end().  Users which require
656 	 * threadgroup to remain stable should use threadgroup_[un]lock()
657 	 * which also takes care of exec path.  Currently, cgroup is the
658 	 * only user.
659 	 */
660 	struct rw_semaphore group_rwsem;
661 #endif
662 
663 	int oom_adj;		/* OOM kill score adjustment (bit shift) */
664 	int oom_score_adj;	/* OOM kill score adjustment */
665 	int oom_score_adj_min;	/* OOM kill score adjustment minimum value.
666 				 * Only settable by CAP_SYS_RESOURCE. */
667 
668 	struct mutex cred_guard_mutex;	/* guard against foreign influences on
669 					 * credential calculations
670 					 * (notably. ptrace) */
671 };
672 
673 /* Context switch must be unlocked if interrupts are to be enabled */
674 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
675 # define __ARCH_WANT_UNLOCKED_CTXSW
676 #endif
677 
678 /*
679  * Bits in flags field of signal_struct.
680  */
681 #define SIGNAL_STOP_STOPPED	0x00000001 /* job control stop in effect */
682 #define SIGNAL_STOP_CONTINUED	0x00000002 /* SIGCONT since WCONTINUED reap */
683 #define SIGNAL_GROUP_EXIT	0x00000004 /* group exit in progress */
684 /*
685  * Pending notifications to parent.
686  */
687 #define SIGNAL_CLD_STOPPED	0x00000010
688 #define SIGNAL_CLD_CONTINUED	0x00000020
689 #define SIGNAL_CLD_MASK		(SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
690 
691 #define SIGNAL_UNKILLABLE	0x00000040 /* for init: ignore fatal signals */
692 
693 /* If true, all threads except ->group_exit_task have pending SIGKILL */
694 static inline int signal_group_exit(const struct signal_struct *sig)
695 {
696 	return	(sig->flags & SIGNAL_GROUP_EXIT) ||
697 		(sig->group_exit_task != NULL);
698 }
699 
700 /*
701  * Some day this will be a full-fledged user tracking system..
702  */
703 struct user_struct {
704 	atomic_t __count;	/* reference count */
705 	atomic_t processes;	/* How many processes does this user have? */
706 	atomic_t files;		/* How many open files does this user have? */
707 	atomic_t sigpending;	/* How many pending signals does this user have? */
708 #ifdef CONFIG_INOTIFY_USER
709 	atomic_t inotify_watches; /* How many inotify watches does this user have? */
710 	atomic_t inotify_devs;	/* How many inotify devs does this user have opened? */
711 #endif
712 #ifdef CONFIG_FANOTIFY
713 	atomic_t fanotify_listeners;
714 #endif
715 #ifdef CONFIG_EPOLL
716 	atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
717 #endif
718 #ifdef CONFIG_POSIX_MQUEUE
719 	/* protected by mq_lock	*/
720 	unsigned long mq_bytes;	/* How many bytes can be allocated to mqueue? */
721 #endif
722 	unsigned long locked_shm; /* How many pages of mlocked shm ? */
723 
724 #ifdef CONFIG_KEYS
725 	struct key *uid_keyring;	/* UID specific keyring */
726 	struct key *session_keyring;	/* UID's default session keyring */
727 #endif
728 
729 	/* Hash table maintenance information */
730 	struct hlist_node uidhash_node;
731 	uid_t uid;
732 	struct user_namespace *user_ns;
733 
734 #ifdef CONFIG_PERF_EVENTS
735 	atomic_long_t locked_vm;
736 #endif
737 };
738 
739 extern int uids_sysfs_init(void);
740 
741 extern struct user_struct *find_user(uid_t);
742 
743 extern struct user_struct root_user;
744 #define INIT_USER (&root_user)
745 
746 
747 struct backing_dev_info;
748 struct reclaim_state;
749 
750 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
751 struct sched_info {
752 	/* cumulative counters */
753 	unsigned long pcount;	      /* # of times run on this cpu */
754 	unsigned long long run_delay; /* time spent waiting on a runqueue */
755 
756 	/* timestamps */
757 	unsigned long long last_arrival,/* when we last ran on a cpu */
758 			   last_queued;	/* when we were last queued to run */
759 };
760 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
761 
762 #ifdef CONFIG_TASK_DELAY_ACCT
763 struct task_delay_info {
764 	spinlock_t	lock;
765 	unsigned int	flags;	/* Private per-task flags */
766 
767 	/* For each stat XXX, add following, aligned appropriately
768 	 *
769 	 * struct timespec XXX_start, XXX_end;
770 	 * u64 XXX_delay;
771 	 * u32 XXX_count;
772 	 *
773 	 * Atomicity of updates to XXX_delay, XXX_count protected by
774 	 * single lock above (split into XXX_lock if contention is an issue).
775 	 */
776 
777 	/*
778 	 * XXX_count is incremented on every XXX operation, the delay
779 	 * associated with the operation is added to XXX_delay.
780 	 * XXX_delay contains the accumulated delay time in nanoseconds.
781 	 */
782 	struct timespec blkio_start, blkio_end;	/* Shared by blkio, swapin */
783 	u64 blkio_delay;	/* wait for sync block io completion */
784 	u64 swapin_delay;	/* wait for swapin block io completion */
785 	u32 blkio_count;	/* total count of the number of sync block */
786 				/* io operations performed */
787 	u32 swapin_count;	/* total count of the number of swapin block */
788 				/* io operations performed */
789 
790 	struct timespec freepages_start, freepages_end;
791 	u64 freepages_delay;	/* wait for memory reclaim */
792 	u32 freepages_count;	/* total count of memory reclaim */
793 };
794 #endif	/* CONFIG_TASK_DELAY_ACCT */
795 
796 static inline int sched_info_on(void)
797 {
798 #ifdef CONFIG_SCHEDSTATS
799 	return 1;
800 #elif defined(CONFIG_TASK_DELAY_ACCT)
801 	extern int delayacct_on;
802 	return delayacct_on;
803 #else
804 	return 0;
805 #endif
806 }
807 
808 enum cpu_idle_type {
809 	CPU_IDLE,
810 	CPU_NOT_IDLE,
811 	CPU_NEWLY_IDLE,
812 	CPU_MAX_IDLE_TYPES
813 };
814 
815 /*
816  * Increase resolution of nice-level calculations for 64-bit architectures.
817  * The extra resolution improves shares distribution and load balancing of
818  * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
819  * hierarchies, especially on larger systems. This is not a user-visible change
820  * and does not change the user-interface for setting shares/weights.
821  *
822  * We increase resolution only if we have enough bits to allow this increased
823  * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
824  * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
825  * increased costs.
826  */
827 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load  */
828 # define SCHED_LOAD_RESOLUTION	10
829 # define scale_load(w)		((w) << SCHED_LOAD_RESOLUTION)
830 # define scale_load_down(w)	((w) >> SCHED_LOAD_RESOLUTION)
831 #else
832 # define SCHED_LOAD_RESOLUTION	0
833 # define scale_load(w)		(w)
834 # define scale_load_down(w)	(w)
835 #endif
836 
837 #define SCHED_LOAD_SHIFT	(10 + SCHED_LOAD_RESOLUTION)
838 #define SCHED_LOAD_SCALE	(1L << SCHED_LOAD_SHIFT)
839 
840 /*
841  * Increase resolution of cpu_power calculations
842  */
843 #define SCHED_POWER_SHIFT	10
844 #define SCHED_POWER_SCALE	(1L << SCHED_POWER_SHIFT)
845 
846 /*
847  * sched-domains (multiprocessor balancing) declarations:
848  */
849 #ifdef CONFIG_SMP
850 #define SD_LOAD_BALANCE		0x0001	/* Do load balancing on this domain. */
851 #define SD_BALANCE_NEWIDLE	0x0002	/* Balance when about to become idle */
852 #define SD_BALANCE_EXEC		0x0004	/* Balance on exec */
853 #define SD_BALANCE_FORK		0x0008	/* Balance on fork, clone */
854 #define SD_BALANCE_WAKE		0x0010  /* Balance on wakeup */
855 #define SD_WAKE_AFFINE		0x0020	/* Wake task to waking CPU */
856 #define SD_PREFER_LOCAL		0x0040  /* Prefer to keep tasks local to this domain */
857 #define SD_SHARE_CPUPOWER	0x0080	/* Domain members share cpu power */
858 #define SD_SHARE_PKG_RESOURCES	0x0200	/* Domain members share cpu pkg resources */
859 #define SD_SERIALIZE		0x0400	/* Only a single load balancing instance */
860 #define SD_ASYM_PACKING		0x0800  /* Place busy groups earlier in the domain */
861 #define SD_PREFER_SIBLING	0x1000	/* Prefer to place tasks in a sibling domain */
862 #define SD_OVERLAP		0x2000	/* sched_domains of this level overlap */
863 
864 extern int __weak arch_sd_sibiling_asym_packing(void);
865 
866 struct sched_group_power {
867 	atomic_t ref;
868 	/*
869 	 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
870 	 * single CPU.
871 	 */
872 	unsigned int power, power_orig;
873 	unsigned long next_update;
874 	/*
875 	 * Number of busy cpus in this group.
876 	 */
877 	atomic_t nr_busy_cpus;
878 };
879 
880 struct sched_group {
881 	struct sched_group *next;	/* Must be a circular list */
882 	atomic_t ref;
883 
884 	unsigned int group_weight;
885 	struct sched_group_power *sgp;
886 
887 	/*
888 	 * The CPUs this group covers.
889 	 *
890 	 * NOTE: this field is variable length. (Allocated dynamically
891 	 * by attaching extra space to the end of the structure,
892 	 * depending on how many CPUs the kernel has booted up with)
893 	 */
894 	unsigned long cpumask[0];
895 };
896 
897 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
898 {
899 	return to_cpumask(sg->cpumask);
900 }
901 
902 /**
903  * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
904  * @group: The group whose first cpu is to be returned.
905  */
906 static inline unsigned int group_first_cpu(struct sched_group *group)
907 {
908 	return cpumask_first(sched_group_cpus(group));
909 }
910 
911 struct sched_domain_attr {
912 	int relax_domain_level;
913 };
914 
915 #define SD_ATTR_INIT	(struct sched_domain_attr) {	\
916 	.relax_domain_level = -1,			\
917 }
918 
919 extern int sched_domain_level_max;
920 
921 struct sched_domain {
922 	/* These fields must be setup */
923 	struct sched_domain *parent;	/* top domain must be null terminated */
924 	struct sched_domain *child;	/* bottom domain must be null terminated */
925 	struct sched_group *groups;	/* the balancing groups of the domain */
926 	unsigned long min_interval;	/* Minimum balance interval ms */
927 	unsigned long max_interval;	/* Maximum balance interval ms */
928 	unsigned int busy_factor;	/* less balancing by factor if busy */
929 	unsigned int imbalance_pct;	/* No balance until over watermark */
930 	unsigned int cache_nice_tries;	/* Leave cache hot tasks for # tries */
931 	unsigned int busy_idx;
932 	unsigned int idle_idx;
933 	unsigned int newidle_idx;
934 	unsigned int wake_idx;
935 	unsigned int forkexec_idx;
936 	unsigned int smt_gain;
937 	int flags;			/* See SD_* */
938 	int level;
939 
940 	/* Runtime fields. */
941 	unsigned long last_balance;	/* init to jiffies. units in jiffies */
942 	unsigned int balance_interval;	/* initialise to 1. units in ms. */
943 	unsigned int nr_balance_failed; /* initialise to 0 */
944 
945 	u64 last_update;
946 
947 #ifdef CONFIG_SCHEDSTATS
948 	/* load_balance() stats */
949 	unsigned int lb_count[CPU_MAX_IDLE_TYPES];
950 	unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
951 	unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
952 	unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
953 	unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
954 	unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
955 	unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
956 	unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
957 
958 	/* Active load balancing */
959 	unsigned int alb_count;
960 	unsigned int alb_failed;
961 	unsigned int alb_pushed;
962 
963 	/* SD_BALANCE_EXEC stats */
964 	unsigned int sbe_count;
965 	unsigned int sbe_balanced;
966 	unsigned int sbe_pushed;
967 
968 	/* SD_BALANCE_FORK stats */
969 	unsigned int sbf_count;
970 	unsigned int sbf_balanced;
971 	unsigned int sbf_pushed;
972 
973 	/* try_to_wake_up() stats */
974 	unsigned int ttwu_wake_remote;
975 	unsigned int ttwu_move_affine;
976 	unsigned int ttwu_move_balance;
977 #endif
978 #ifdef CONFIG_SCHED_DEBUG
979 	char *name;
980 #endif
981 	union {
982 		void *private;		/* used during construction */
983 		struct rcu_head rcu;	/* used during destruction */
984 	};
985 
986 	unsigned int span_weight;
987 	/*
988 	 * Span of all CPUs in this domain.
989 	 *
990 	 * NOTE: this field is variable length. (Allocated dynamically
991 	 * by attaching extra space to the end of the structure,
992 	 * depending on how many CPUs the kernel has booted up with)
993 	 */
994 	unsigned long span[0];
995 };
996 
997 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
998 {
999 	return to_cpumask(sd->span);
1000 }
1001 
1002 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1003 				    struct sched_domain_attr *dattr_new);
1004 
1005 /* Allocate an array of sched domains, for partition_sched_domains(). */
1006 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1007 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1008 
1009 /* Test a flag in parent sched domain */
1010 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1011 {
1012 	if (sd->parent && (sd->parent->flags & flag))
1013 		return 1;
1014 
1015 	return 0;
1016 }
1017 
1018 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1019 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1020 
1021 bool cpus_share_cache(int this_cpu, int that_cpu);
1022 
1023 #else /* CONFIG_SMP */
1024 
1025 struct sched_domain_attr;
1026 
1027 static inline void
1028 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1029 			struct sched_domain_attr *dattr_new)
1030 {
1031 }
1032 
1033 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1034 {
1035 	return true;
1036 }
1037 
1038 #endif	/* !CONFIG_SMP */
1039 
1040 
1041 struct io_context;			/* See blkdev.h */
1042 
1043 
1044 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1045 extern void prefetch_stack(struct task_struct *t);
1046 #else
1047 static inline void prefetch_stack(struct task_struct *t) { }
1048 #endif
1049 
1050 struct audit_context;		/* See audit.c */
1051 struct mempolicy;
1052 struct pipe_inode_info;
1053 struct uts_namespace;
1054 
1055 struct rq;
1056 struct sched_domain;
1057 
1058 /*
1059  * wake flags
1060  */
1061 #define WF_SYNC		0x01		/* waker goes to sleep after wakup */
1062 #define WF_FORK		0x02		/* child wakeup after fork */
1063 #define WF_MIGRATED	0x04		/* internal use, task got migrated */
1064 
1065 #define ENQUEUE_WAKEUP		1
1066 #define ENQUEUE_HEAD		2
1067 #ifdef CONFIG_SMP
1068 #define ENQUEUE_WAKING		4	/* sched_class::task_waking was called */
1069 #else
1070 #define ENQUEUE_WAKING		0
1071 #endif
1072 
1073 #define DEQUEUE_SLEEP		1
1074 
1075 struct sched_class {
1076 	const struct sched_class *next;
1077 
1078 	void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1079 	void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1080 	void (*yield_task) (struct rq *rq);
1081 	bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1082 
1083 	void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1084 
1085 	struct task_struct * (*pick_next_task) (struct rq *rq);
1086 	void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1087 
1088 #ifdef CONFIG_SMP
1089 	int  (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1090 
1091 	void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1092 	void (*post_schedule) (struct rq *this_rq);
1093 	void (*task_waking) (struct task_struct *task);
1094 	void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1095 
1096 	void (*set_cpus_allowed)(struct task_struct *p,
1097 				 const struct cpumask *newmask);
1098 
1099 	void (*rq_online)(struct rq *rq);
1100 	void (*rq_offline)(struct rq *rq);
1101 #endif
1102 
1103 	void (*set_curr_task) (struct rq *rq);
1104 	void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1105 	void (*task_fork) (struct task_struct *p);
1106 
1107 	void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1108 	void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1109 	void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1110 			     int oldprio);
1111 
1112 	unsigned int (*get_rr_interval) (struct rq *rq,
1113 					 struct task_struct *task);
1114 
1115 #ifdef CONFIG_FAIR_GROUP_SCHED
1116 	void (*task_move_group) (struct task_struct *p, int on_rq);
1117 #endif
1118 };
1119 
1120 struct load_weight {
1121 	unsigned long weight, inv_weight;
1122 };
1123 
1124 #ifdef CONFIG_SCHEDSTATS
1125 struct sched_statistics {
1126 	u64			wait_start;
1127 	u64			wait_max;
1128 	u64			wait_count;
1129 	u64			wait_sum;
1130 	u64			iowait_count;
1131 	u64			iowait_sum;
1132 
1133 	u64			sleep_start;
1134 	u64			sleep_max;
1135 	s64			sum_sleep_runtime;
1136 
1137 	u64			block_start;
1138 	u64			block_max;
1139 	u64			exec_max;
1140 	u64			slice_max;
1141 
1142 	u64			nr_migrations_cold;
1143 	u64			nr_failed_migrations_affine;
1144 	u64			nr_failed_migrations_running;
1145 	u64			nr_failed_migrations_hot;
1146 	u64			nr_forced_migrations;
1147 
1148 	u64			nr_wakeups;
1149 	u64			nr_wakeups_sync;
1150 	u64			nr_wakeups_migrate;
1151 	u64			nr_wakeups_local;
1152 	u64			nr_wakeups_remote;
1153 	u64			nr_wakeups_affine;
1154 	u64			nr_wakeups_affine_attempts;
1155 	u64			nr_wakeups_passive;
1156 	u64			nr_wakeups_idle;
1157 };
1158 #endif
1159 
1160 struct sched_entity {
1161 	struct load_weight	load;		/* for load-balancing */
1162 	struct rb_node		run_node;
1163 	struct list_head	group_node;
1164 	unsigned int		on_rq;
1165 
1166 	u64			exec_start;
1167 	u64			sum_exec_runtime;
1168 	u64			vruntime;
1169 	u64			prev_sum_exec_runtime;
1170 
1171 	u64			nr_migrations;
1172 
1173 #ifdef CONFIG_SCHEDSTATS
1174 	struct sched_statistics statistics;
1175 #endif
1176 
1177 #ifdef CONFIG_FAIR_GROUP_SCHED
1178 	struct sched_entity	*parent;
1179 	/* rq on which this entity is (to be) queued: */
1180 	struct cfs_rq		*cfs_rq;
1181 	/* rq "owned" by this entity/group: */
1182 	struct cfs_rq		*my_q;
1183 #endif
1184 };
1185 
1186 struct sched_rt_entity {
1187 	struct list_head run_list;
1188 	unsigned long timeout;
1189 	unsigned int time_slice;
1190 	int nr_cpus_allowed;
1191 
1192 	struct sched_rt_entity *back;
1193 #ifdef CONFIG_RT_GROUP_SCHED
1194 	struct sched_rt_entity	*parent;
1195 	/* rq on which this entity is (to be) queued: */
1196 	struct rt_rq		*rt_rq;
1197 	/* rq "owned" by this entity/group: */
1198 	struct rt_rq		*my_q;
1199 #endif
1200 };
1201 
1202 /*
1203  * default timeslice is 100 msecs (used only for SCHED_RR tasks).
1204  * Timeslices get refilled after they expire.
1205  */
1206 #define RR_TIMESLICE		(100 * HZ / 1000)
1207 
1208 struct rcu_node;
1209 
1210 enum perf_event_task_context {
1211 	perf_invalid_context = -1,
1212 	perf_hw_context = 0,
1213 	perf_sw_context,
1214 	perf_nr_task_contexts,
1215 };
1216 
1217 struct task_struct {
1218 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
1219 	void *stack;
1220 	atomic_t usage;
1221 	unsigned int flags;	/* per process flags, defined below */
1222 	unsigned int ptrace;
1223 
1224 #ifdef CONFIG_SMP
1225 	struct llist_node wake_entry;
1226 	int on_cpu;
1227 #endif
1228 	int on_rq;
1229 
1230 	int prio, static_prio, normal_prio;
1231 	unsigned int rt_priority;
1232 	const struct sched_class *sched_class;
1233 	struct sched_entity se;
1234 	struct sched_rt_entity rt;
1235 
1236 #ifdef CONFIG_PREEMPT_NOTIFIERS
1237 	/* list of struct preempt_notifier: */
1238 	struct hlist_head preempt_notifiers;
1239 #endif
1240 
1241 	/*
1242 	 * fpu_counter contains the number of consecutive context switches
1243 	 * that the FPU is used. If this is over a threshold, the lazy fpu
1244 	 * saving becomes unlazy to save the trap. This is an unsigned char
1245 	 * so that after 256 times the counter wraps and the behavior turns
1246 	 * lazy again; this to deal with bursty apps that only use FPU for
1247 	 * a short time
1248 	 */
1249 	unsigned char fpu_counter;
1250 #ifdef CONFIG_BLK_DEV_IO_TRACE
1251 	unsigned int btrace_seq;
1252 #endif
1253 
1254 	unsigned int policy;
1255 	cpumask_t cpus_allowed;
1256 
1257 #ifdef CONFIG_PREEMPT_RCU
1258 	int rcu_read_lock_nesting;
1259 	char rcu_read_unlock_special;
1260 	struct list_head rcu_node_entry;
1261 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1262 #ifdef CONFIG_TREE_PREEMPT_RCU
1263 	struct rcu_node *rcu_blocked_node;
1264 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1265 #ifdef CONFIG_RCU_BOOST
1266 	struct rt_mutex *rcu_boost_mutex;
1267 #endif /* #ifdef CONFIG_RCU_BOOST */
1268 
1269 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1270 	struct sched_info sched_info;
1271 #endif
1272 
1273 	struct list_head tasks;
1274 #ifdef CONFIG_SMP
1275 	struct plist_node pushable_tasks;
1276 #endif
1277 
1278 	struct mm_struct *mm, *active_mm;
1279 #ifdef CONFIG_COMPAT_BRK
1280 	unsigned brk_randomized:1;
1281 #endif
1282 #if defined(SPLIT_RSS_COUNTING)
1283 	struct task_rss_stat	rss_stat;
1284 #endif
1285 /* task state */
1286 	int exit_state;
1287 	int exit_code, exit_signal;
1288 	int pdeath_signal;  /*  The signal sent when the parent dies  */
1289 	unsigned int jobctl;	/* JOBCTL_*, siglock protected */
1290 	/* ??? */
1291 	unsigned int personality;
1292 	unsigned did_exec:1;
1293 	unsigned in_execve:1;	/* Tell the LSMs that the process is doing an
1294 				 * execve */
1295 	unsigned in_iowait:1;
1296 
1297 	/* task may not gain privileges */
1298 	unsigned no_new_privs:1;
1299 
1300 	/* Revert to default priority/policy when forking */
1301 	unsigned sched_reset_on_fork:1;
1302 	unsigned sched_contributes_to_load:1;
1303 
1304 #ifdef CONFIG_GENERIC_HARDIRQS
1305 	/* IRQ handler threads */
1306 	unsigned irq_thread:1;
1307 #endif
1308 
1309 	pid_t pid;
1310 	pid_t tgid;
1311 
1312 #ifdef CONFIG_CC_STACKPROTECTOR
1313 	/* Canary value for the -fstack-protector gcc feature */
1314 	unsigned long stack_canary;
1315 #endif
1316 
1317 	/*
1318 	 * pointers to (original) parent process, youngest child, younger sibling,
1319 	 * older sibling, respectively.  (p->father can be replaced with
1320 	 * p->real_parent->pid)
1321 	 */
1322 	struct task_struct __rcu *real_parent; /* real parent process */
1323 	struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1324 	/*
1325 	 * children/sibling forms the list of my natural children
1326 	 */
1327 	struct list_head children;	/* list of my children */
1328 	struct list_head sibling;	/* linkage in my parent's children list */
1329 	struct task_struct *group_leader;	/* threadgroup leader */
1330 
1331 	/*
1332 	 * ptraced is the list of tasks this task is using ptrace on.
1333 	 * This includes both natural children and PTRACE_ATTACH targets.
1334 	 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1335 	 */
1336 	struct list_head ptraced;
1337 	struct list_head ptrace_entry;
1338 
1339 	/* PID/PID hash table linkage. */
1340 	struct pid_link pids[PIDTYPE_MAX];
1341 	struct list_head thread_group;
1342 
1343 	struct completion *vfork_done;		/* for vfork() */
1344 	int __user *set_child_tid;		/* CLONE_CHILD_SETTID */
1345 	int __user *clear_child_tid;		/* CLONE_CHILD_CLEARTID */
1346 
1347 	cputime_t utime, stime, utimescaled, stimescaled;
1348 	cputime_t gtime;
1349 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1350 	cputime_t prev_utime, prev_stime;
1351 #endif
1352 	unsigned long nvcsw, nivcsw; /* context switch counts */
1353 	struct timespec start_time; 		/* monotonic time */
1354 	struct timespec real_start_time;	/* boot based time */
1355 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1356 	unsigned long min_flt, maj_flt;
1357 
1358 	struct task_cputime cputime_expires;
1359 	struct list_head cpu_timers[3];
1360 
1361 /* process credentials */
1362 	const struct cred __rcu *real_cred; /* objective and real subjective task
1363 					 * credentials (COW) */
1364 	const struct cred __rcu *cred;	/* effective (overridable) subjective task
1365 					 * credentials (COW) */
1366 	struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1367 
1368 	char comm[TASK_COMM_LEN]; /* executable name excluding path
1369 				     - access with [gs]et_task_comm (which lock
1370 				       it with task_lock())
1371 				     - initialized normally by setup_new_exec */
1372 /* file system info */
1373 	int link_count, total_link_count;
1374 #ifdef CONFIG_SYSVIPC
1375 /* ipc stuff */
1376 	struct sysv_sem sysvsem;
1377 #endif
1378 #ifdef CONFIG_DETECT_HUNG_TASK
1379 /* hung task detection */
1380 	unsigned long last_switch_count;
1381 #endif
1382 /* CPU-specific state of this task */
1383 	struct thread_struct thread;
1384 /* filesystem information */
1385 	struct fs_struct *fs;
1386 /* open file information */
1387 	struct files_struct *files;
1388 /* namespaces */
1389 	struct nsproxy *nsproxy;
1390 /* signal handlers */
1391 	struct signal_struct *signal;
1392 	struct sighand_struct *sighand;
1393 
1394 	sigset_t blocked, real_blocked;
1395 	sigset_t saved_sigmask;	/* restored if set_restore_sigmask() was used */
1396 	struct sigpending pending;
1397 
1398 	unsigned long sas_ss_sp;
1399 	size_t sas_ss_size;
1400 	int (*notifier)(void *priv);
1401 	void *notifier_data;
1402 	sigset_t *notifier_mask;
1403 	struct audit_context *audit_context;
1404 #ifdef CONFIG_AUDITSYSCALL
1405 	uid_t loginuid;
1406 	unsigned int sessionid;
1407 #endif
1408 	struct seccomp seccomp;
1409 
1410 /* Thread group tracking */
1411    	u32 parent_exec_id;
1412    	u32 self_exec_id;
1413 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1414  * mempolicy */
1415 	spinlock_t alloc_lock;
1416 
1417 	/* Protection of the PI data structures: */
1418 	raw_spinlock_t pi_lock;
1419 
1420 #ifdef CONFIG_RT_MUTEXES
1421 	/* PI waiters blocked on a rt_mutex held by this task */
1422 	struct plist_head pi_waiters;
1423 	/* Deadlock detection and priority inheritance handling */
1424 	struct rt_mutex_waiter *pi_blocked_on;
1425 #endif
1426 
1427 #ifdef CONFIG_DEBUG_MUTEXES
1428 	/* mutex deadlock detection */
1429 	struct mutex_waiter *blocked_on;
1430 #endif
1431 #ifdef CONFIG_TRACE_IRQFLAGS
1432 	unsigned int irq_events;
1433 	unsigned long hardirq_enable_ip;
1434 	unsigned long hardirq_disable_ip;
1435 	unsigned int hardirq_enable_event;
1436 	unsigned int hardirq_disable_event;
1437 	int hardirqs_enabled;
1438 	int hardirq_context;
1439 	unsigned long softirq_disable_ip;
1440 	unsigned long softirq_enable_ip;
1441 	unsigned int softirq_disable_event;
1442 	unsigned int softirq_enable_event;
1443 	int softirqs_enabled;
1444 	int softirq_context;
1445 #endif
1446 #ifdef CONFIG_LOCKDEP
1447 # define MAX_LOCK_DEPTH 48UL
1448 	u64 curr_chain_key;
1449 	int lockdep_depth;
1450 	unsigned int lockdep_recursion;
1451 	struct held_lock held_locks[MAX_LOCK_DEPTH];
1452 	gfp_t lockdep_reclaim_gfp;
1453 #endif
1454 
1455 /* journalling filesystem info */
1456 	void *journal_info;
1457 
1458 /* stacked block device info */
1459 	struct bio_list *bio_list;
1460 
1461 #ifdef CONFIG_BLOCK
1462 /* stack plugging */
1463 	struct blk_plug *plug;
1464 #endif
1465 
1466 /* VM state */
1467 	struct reclaim_state *reclaim_state;
1468 
1469 	struct backing_dev_info *backing_dev_info;
1470 
1471 	struct io_context *io_context;
1472 
1473 	unsigned long ptrace_message;
1474 	siginfo_t *last_siginfo; /* For ptrace use.  */
1475 	struct task_io_accounting ioac;
1476 #if defined(CONFIG_TASK_XACCT)
1477 	u64 acct_rss_mem1;	/* accumulated rss usage */
1478 	u64 acct_vm_mem1;	/* accumulated virtual memory usage */
1479 	cputime_t acct_timexpd;	/* stime + utime since last update */
1480 #endif
1481 #ifdef CONFIG_CPUSETS
1482 	nodemask_t mems_allowed;	/* Protected by alloc_lock */
1483 	seqcount_t mems_allowed_seq;	/* Seqence no to catch updates */
1484 	int cpuset_mem_spread_rotor;
1485 	int cpuset_slab_spread_rotor;
1486 #endif
1487 #ifdef CONFIG_CGROUPS
1488 	/* Control Group info protected by css_set_lock */
1489 	struct css_set __rcu *cgroups;
1490 	/* cg_list protected by css_set_lock and tsk->alloc_lock */
1491 	struct list_head cg_list;
1492 #endif
1493 #ifdef CONFIG_FUTEX
1494 	struct robust_list_head __user *robust_list;
1495 #ifdef CONFIG_COMPAT
1496 	struct compat_robust_list_head __user *compat_robust_list;
1497 #endif
1498 	struct list_head pi_state_list;
1499 	struct futex_pi_state *pi_state_cache;
1500 #endif
1501 #ifdef CONFIG_PERF_EVENTS
1502 	struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1503 	struct mutex perf_event_mutex;
1504 	struct list_head perf_event_list;
1505 #endif
1506 #ifdef CONFIG_NUMA
1507 	struct mempolicy *mempolicy;	/* Protected by alloc_lock */
1508 	short il_next;
1509 	short pref_node_fork;
1510 #endif
1511 	struct rcu_head rcu;
1512 
1513 	/*
1514 	 * cache last used pipe for splice
1515 	 */
1516 	struct pipe_inode_info *splice_pipe;
1517 #ifdef	CONFIG_TASK_DELAY_ACCT
1518 	struct task_delay_info *delays;
1519 #endif
1520 #ifdef CONFIG_FAULT_INJECTION
1521 	int make_it_fail;
1522 #endif
1523 	/*
1524 	 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1525 	 * balance_dirty_pages() for some dirty throttling pause
1526 	 */
1527 	int nr_dirtied;
1528 	int nr_dirtied_pause;
1529 	unsigned long dirty_paused_when; /* start of a write-and-pause period */
1530 
1531 #ifdef CONFIG_LATENCYTOP
1532 	int latency_record_count;
1533 	struct latency_record latency_record[LT_SAVECOUNT];
1534 #endif
1535 	/*
1536 	 * time slack values; these are used to round up poll() and
1537 	 * select() etc timeout values. These are in nanoseconds.
1538 	 */
1539 	unsigned long timer_slack_ns;
1540 	unsigned long default_timer_slack_ns;
1541 
1542 	struct list_head	*scm_work_list;
1543 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1544 	/* Index of current stored address in ret_stack */
1545 	int curr_ret_stack;
1546 	/* Stack of return addresses for return function tracing */
1547 	struct ftrace_ret_stack	*ret_stack;
1548 	/* time stamp for last schedule */
1549 	unsigned long long ftrace_timestamp;
1550 	/*
1551 	 * Number of functions that haven't been traced
1552 	 * because of depth overrun.
1553 	 */
1554 	atomic_t trace_overrun;
1555 	/* Pause for the tracing */
1556 	atomic_t tracing_graph_pause;
1557 #endif
1558 #ifdef CONFIG_TRACING
1559 	/* state flags for use by tracers */
1560 	unsigned long trace;
1561 	/* bitmask and counter of trace recursion */
1562 	unsigned long trace_recursion;
1563 #endif /* CONFIG_TRACING */
1564 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1565 	struct memcg_batch_info {
1566 		int do_batch;	/* incremented when batch uncharge started */
1567 		struct mem_cgroup *memcg; /* target memcg of uncharge */
1568 		unsigned long nr_pages;	/* uncharged usage */
1569 		unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1570 	} memcg_batch;
1571 #endif
1572 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1573 	atomic_t ptrace_bp_refcnt;
1574 #endif
1575 };
1576 
1577 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1578 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1579 
1580 /*
1581  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1582  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1583  * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1584  * values are inverted: lower p->prio value means higher priority.
1585  *
1586  * The MAX_USER_RT_PRIO value allows the actual maximum
1587  * RT priority to be separate from the value exported to
1588  * user-space.  This allows kernel threads to set their
1589  * priority to a value higher than any user task. Note:
1590  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1591  */
1592 
1593 #define MAX_USER_RT_PRIO	100
1594 #define MAX_RT_PRIO		MAX_USER_RT_PRIO
1595 
1596 #define MAX_PRIO		(MAX_RT_PRIO + 40)
1597 #define DEFAULT_PRIO		(MAX_RT_PRIO + 20)
1598 
1599 static inline int rt_prio(int prio)
1600 {
1601 	if (unlikely(prio < MAX_RT_PRIO))
1602 		return 1;
1603 	return 0;
1604 }
1605 
1606 static inline int rt_task(struct task_struct *p)
1607 {
1608 	return rt_prio(p->prio);
1609 }
1610 
1611 static inline struct pid *task_pid(struct task_struct *task)
1612 {
1613 	return task->pids[PIDTYPE_PID].pid;
1614 }
1615 
1616 static inline struct pid *task_tgid(struct task_struct *task)
1617 {
1618 	return task->group_leader->pids[PIDTYPE_PID].pid;
1619 }
1620 
1621 /*
1622  * Without tasklist or rcu lock it is not safe to dereference
1623  * the result of task_pgrp/task_session even if task == current,
1624  * we can race with another thread doing sys_setsid/sys_setpgid.
1625  */
1626 static inline struct pid *task_pgrp(struct task_struct *task)
1627 {
1628 	return task->group_leader->pids[PIDTYPE_PGID].pid;
1629 }
1630 
1631 static inline struct pid *task_session(struct task_struct *task)
1632 {
1633 	return task->group_leader->pids[PIDTYPE_SID].pid;
1634 }
1635 
1636 struct pid_namespace;
1637 
1638 /*
1639  * the helpers to get the task's different pids as they are seen
1640  * from various namespaces
1641  *
1642  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1643  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1644  *                     current.
1645  * task_xid_nr_ns()  : id seen from the ns specified;
1646  *
1647  * set_task_vxid()   : assigns a virtual id to a task;
1648  *
1649  * see also pid_nr() etc in include/linux/pid.h
1650  */
1651 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1652 			struct pid_namespace *ns);
1653 
1654 static inline pid_t task_pid_nr(struct task_struct *tsk)
1655 {
1656 	return tsk->pid;
1657 }
1658 
1659 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1660 					struct pid_namespace *ns)
1661 {
1662 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1663 }
1664 
1665 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1666 {
1667 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1668 }
1669 
1670 
1671 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1672 {
1673 	return tsk->tgid;
1674 }
1675 
1676 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1677 
1678 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1679 {
1680 	return pid_vnr(task_tgid(tsk));
1681 }
1682 
1683 
1684 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1685 					struct pid_namespace *ns)
1686 {
1687 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1688 }
1689 
1690 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1691 {
1692 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1693 }
1694 
1695 
1696 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1697 					struct pid_namespace *ns)
1698 {
1699 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1700 }
1701 
1702 static inline pid_t task_session_vnr(struct task_struct *tsk)
1703 {
1704 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1705 }
1706 
1707 /* obsolete, do not use */
1708 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1709 {
1710 	return task_pgrp_nr_ns(tsk, &init_pid_ns);
1711 }
1712 
1713 /**
1714  * pid_alive - check that a task structure is not stale
1715  * @p: Task structure to be checked.
1716  *
1717  * Test if a process is not yet dead (at most zombie state)
1718  * If pid_alive fails, then pointers within the task structure
1719  * can be stale and must not be dereferenced.
1720  */
1721 static inline int pid_alive(struct task_struct *p)
1722 {
1723 	return p->pids[PIDTYPE_PID].pid != NULL;
1724 }
1725 
1726 /**
1727  * is_global_init - check if a task structure is init
1728  * @tsk: Task structure to be checked.
1729  *
1730  * Check if a task structure is the first user space task the kernel created.
1731  */
1732 static inline int is_global_init(struct task_struct *tsk)
1733 {
1734 	return tsk->pid == 1;
1735 }
1736 
1737 /*
1738  * is_container_init:
1739  * check whether in the task is init in its own pid namespace.
1740  */
1741 extern int is_container_init(struct task_struct *tsk);
1742 
1743 extern struct pid *cad_pid;
1744 
1745 extern void free_task(struct task_struct *tsk);
1746 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1747 
1748 extern void __put_task_struct(struct task_struct *t);
1749 
1750 static inline void put_task_struct(struct task_struct *t)
1751 {
1752 	if (atomic_dec_and_test(&t->usage))
1753 		__put_task_struct(t);
1754 }
1755 
1756 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1757 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1758 
1759 /*
1760  * Per process flags
1761  */
1762 #define PF_EXITING	0x00000004	/* getting shut down */
1763 #define PF_EXITPIDONE	0x00000008	/* pi exit done on shut down */
1764 #define PF_VCPU		0x00000010	/* I'm a virtual CPU */
1765 #define PF_WQ_WORKER	0x00000020	/* I'm a workqueue worker */
1766 #define PF_FORKNOEXEC	0x00000040	/* forked but didn't exec */
1767 #define PF_MCE_PROCESS  0x00000080      /* process policy on mce errors */
1768 #define PF_SUPERPRIV	0x00000100	/* used super-user privileges */
1769 #define PF_DUMPCORE	0x00000200	/* dumped core */
1770 #define PF_SIGNALED	0x00000400	/* killed by a signal */
1771 #define PF_MEMALLOC	0x00000800	/* Allocating memory */
1772 #define PF_NPROC_EXCEEDED 0x00001000	/* set_user noticed that RLIMIT_NPROC was exceeded */
1773 #define PF_USED_MATH	0x00002000	/* if unset the fpu must be initialized before use */
1774 #define PF_NOFREEZE	0x00008000	/* this thread should not be frozen */
1775 #define PF_FROZEN	0x00010000	/* frozen for system suspend */
1776 #define PF_FSTRANS	0x00020000	/* inside a filesystem transaction */
1777 #define PF_KSWAPD	0x00040000	/* I am kswapd */
1778 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
1779 #define PF_KTHREAD	0x00200000	/* I am a kernel thread */
1780 #define PF_RANDOMIZE	0x00400000	/* randomize virtual address space */
1781 #define PF_SWAPWRITE	0x00800000	/* Allowed to write to swap */
1782 #define PF_SPREAD_PAGE	0x01000000	/* Spread page cache over cpuset */
1783 #define PF_SPREAD_SLAB	0x02000000	/* Spread some slab caches over cpuset */
1784 #define PF_THREAD_BOUND	0x04000000	/* Thread bound to specific cpu */
1785 #define PF_MCE_EARLY    0x08000000      /* Early kill for mce process policy */
1786 #define PF_MEMPOLICY	0x10000000	/* Non-default NUMA mempolicy */
1787 #define PF_MUTEX_TESTER	0x20000000	/* Thread belongs to the rt mutex tester */
1788 #define PF_FREEZER_SKIP	0x40000000	/* Freezer should not count it as freezable */
1789 
1790 /*
1791  * Only the _current_ task can read/write to tsk->flags, but other
1792  * tasks can access tsk->flags in readonly mode for example
1793  * with tsk_used_math (like during threaded core dumping).
1794  * There is however an exception to this rule during ptrace
1795  * or during fork: the ptracer task is allowed to write to the
1796  * child->flags of its traced child (same goes for fork, the parent
1797  * can write to the child->flags), because we're guaranteed the
1798  * child is not running and in turn not changing child->flags
1799  * at the same time the parent does it.
1800  */
1801 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1802 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1803 #define clear_used_math() clear_stopped_child_used_math(current)
1804 #define set_used_math() set_stopped_child_used_math(current)
1805 #define conditional_stopped_child_used_math(condition, child) \
1806 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1807 #define conditional_used_math(condition) \
1808 	conditional_stopped_child_used_math(condition, current)
1809 #define copy_to_stopped_child_used_math(child) \
1810 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1811 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1812 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1813 #define used_math() tsk_used_math(current)
1814 
1815 /*
1816  * task->jobctl flags
1817  */
1818 #define JOBCTL_STOP_SIGMASK	0xffff	/* signr of the last group stop */
1819 
1820 #define JOBCTL_STOP_DEQUEUED_BIT 16	/* stop signal dequeued */
1821 #define JOBCTL_STOP_PENDING_BIT	17	/* task should stop for group stop */
1822 #define JOBCTL_STOP_CONSUME_BIT	18	/* consume group stop count */
1823 #define JOBCTL_TRAP_STOP_BIT	19	/* trap for STOP */
1824 #define JOBCTL_TRAP_NOTIFY_BIT	20	/* trap for NOTIFY */
1825 #define JOBCTL_TRAPPING_BIT	21	/* switching to TRACED */
1826 #define JOBCTL_LISTENING_BIT	22	/* ptracer is listening for events */
1827 
1828 #define JOBCTL_STOP_DEQUEUED	(1 << JOBCTL_STOP_DEQUEUED_BIT)
1829 #define JOBCTL_STOP_PENDING	(1 << JOBCTL_STOP_PENDING_BIT)
1830 #define JOBCTL_STOP_CONSUME	(1 << JOBCTL_STOP_CONSUME_BIT)
1831 #define JOBCTL_TRAP_STOP	(1 << JOBCTL_TRAP_STOP_BIT)
1832 #define JOBCTL_TRAP_NOTIFY	(1 << JOBCTL_TRAP_NOTIFY_BIT)
1833 #define JOBCTL_TRAPPING		(1 << JOBCTL_TRAPPING_BIT)
1834 #define JOBCTL_LISTENING	(1 << JOBCTL_LISTENING_BIT)
1835 
1836 #define JOBCTL_TRAP_MASK	(JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1837 #define JOBCTL_PENDING_MASK	(JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1838 
1839 extern bool task_set_jobctl_pending(struct task_struct *task,
1840 				    unsigned int mask);
1841 extern void task_clear_jobctl_trapping(struct task_struct *task);
1842 extern void task_clear_jobctl_pending(struct task_struct *task,
1843 				      unsigned int mask);
1844 
1845 #ifdef CONFIG_PREEMPT_RCU
1846 
1847 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1848 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1849 
1850 static inline void rcu_copy_process(struct task_struct *p)
1851 {
1852 	p->rcu_read_lock_nesting = 0;
1853 	p->rcu_read_unlock_special = 0;
1854 #ifdef CONFIG_TREE_PREEMPT_RCU
1855 	p->rcu_blocked_node = NULL;
1856 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1857 #ifdef CONFIG_RCU_BOOST
1858 	p->rcu_boost_mutex = NULL;
1859 #endif /* #ifdef CONFIG_RCU_BOOST */
1860 	INIT_LIST_HEAD(&p->rcu_node_entry);
1861 }
1862 
1863 static inline void rcu_switch_from(struct task_struct *prev)
1864 {
1865 	if (prev->rcu_read_lock_nesting != 0)
1866 		rcu_preempt_note_context_switch();
1867 }
1868 
1869 #else
1870 
1871 static inline void rcu_copy_process(struct task_struct *p)
1872 {
1873 }
1874 
1875 static inline void rcu_switch_from(struct task_struct *prev)
1876 {
1877 }
1878 
1879 #endif
1880 
1881 #ifdef CONFIG_SMP
1882 extern void do_set_cpus_allowed(struct task_struct *p,
1883 			       const struct cpumask *new_mask);
1884 
1885 extern int set_cpus_allowed_ptr(struct task_struct *p,
1886 				const struct cpumask *new_mask);
1887 #else
1888 static inline void do_set_cpus_allowed(struct task_struct *p,
1889 				      const struct cpumask *new_mask)
1890 {
1891 }
1892 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1893 				       const struct cpumask *new_mask)
1894 {
1895 	if (!cpumask_test_cpu(0, new_mask))
1896 		return -EINVAL;
1897 	return 0;
1898 }
1899 #endif
1900 
1901 #ifndef CONFIG_CPUMASK_OFFSTACK
1902 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1903 {
1904 	return set_cpus_allowed_ptr(p, &new_mask);
1905 }
1906 #endif
1907 
1908 /*
1909  * Do not use outside of architecture code which knows its limitations.
1910  *
1911  * sched_clock() has no promise of monotonicity or bounded drift between
1912  * CPUs, use (which you should not) requires disabling IRQs.
1913  *
1914  * Please use one of the three interfaces below.
1915  */
1916 extern unsigned long long notrace sched_clock(void);
1917 /*
1918  * See the comment in kernel/sched/clock.c
1919  */
1920 extern u64 cpu_clock(int cpu);
1921 extern u64 local_clock(void);
1922 extern u64 sched_clock_cpu(int cpu);
1923 
1924 
1925 extern void sched_clock_init(void);
1926 
1927 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1928 static inline void sched_clock_tick(void)
1929 {
1930 }
1931 
1932 static inline void sched_clock_idle_sleep_event(void)
1933 {
1934 }
1935 
1936 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1937 {
1938 }
1939 #else
1940 /*
1941  * Architectures can set this to 1 if they have specified
1942  * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1943  * but then during bootup it turns out that sched_clock()
1944  * is reliable after all:
1945  */
1946 extern int sched_clock_stable;
1947 
1948 extern void sched_clock_tick(void);
1949 extern void sched_clock_idle_sleep_event(void);
1950 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1951 #endif
1952 
1953 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1954 /*
1955  * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1956  * The reason for this explicit opt-in is not to have perf penalty with
1957  * slow sched_clocks.
1958  */
1959 extern void enable_sched_clock_irqtime(void);
1960 extern void disable_sched_clock_irqtime(void);
1961 #else
1962 static inline void enable_sched_clock_irqtime(void) {}
1963 static inline void disable_sched_clock_irqtime(void) {}
1964 #endif
1965 
1966 extern unsigned long long
1967 task_sched_runtime(struct task_struct *task);
1968 
1969 /* sched_exec is called by processes performing an exec */
1970 #ifdef CONFIG_SMP
1971 extern void sched_exec(void);
1972 #else
1973 #define sched_exec()   {}
1974 #endif
1975 
1976 extern void sched_clock_idle_sleep_event(void);
1977 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1978 
1979 #ifdef CONFIG_HOTPLUG_CPU
1980 extern void idle_task_exit(void);
1981 #else
1982 static inline void idle_task_exit(void) {}
1983 #endif
1984 
1985 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1986 extern void wake_up_idle_cpu(int cpu);
1987 #else
1988 static inline void wake_up_idle_cpu(int cpu) { }
1989 #endif
1990 
1991 extern unsigned int sysctl_sched_latency;
1992 extern unsigned int sysctl_sched_min_granularity;
1993 extern unsigned int sysctl_sched_wakeup_granularity;
1994 extern unsigned int sysctl_sched_child_runs_first;
1995 
1996 enum sched_tunable_scaling {
1997 	SCHED_TUNABLESCALING_NONE,
1998 	SCHED_TUNABLESCALING_LOG,
1999 	SCHED_TUNABLESCALING_LINEAR,
2000 	SCHED_TUNABLESCALING_END,
2001 };
2002 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
2003 
2004 #ifdef CONFIG_SCHED_DEBUG
2005 extern unsigned int sysctl_sched_migration_cost;
2006 extern unsigned int sysctl_sched_nr_migrate;
2007 extern unsigned int sysctl_sched_time_avg;
2008 extern unsigned int sysctl_timer_migration;
2009 extern unsigned int sysctl_sched_shares_window;
2010 
2011 int sched_proc_update_handler(struct ctl_table *table, int write,
2012 		void __user *buffer, size_t *length,
2013 		loff_t *ppos);
2014 #endif
2015 #ifdef CONFIG_SCHED_DEBUG
2016 static inline unsigned int get_sysctl_timer_migration(void)
2017 {
2018 	return sysctl_timer_migration;
2019 }
2020 #else
2021 static inline unsigned int get_sysctl_timer_migration(void)
2022 {
2023 	return 1;
2024 }
2025 #endif
2026 extern unsigned int sysctl_sched_rt_period;
2027 extern int sysctl_sched_rt_runtime;
2028 
2029 int sched_rt_handler(struct ctl_table *table, int write,
2030 		void __user *buffer, size_t *lenp,
2031 		loff_t *ppos);
2032 
2033 #ifdef CONFIG_SCHED_AUTOGROUP
2034 extern unsigned int sysctl_sched_autogroup_enabled;
2035 
2036 extern void sched_autogroup_create_attach(struct task_struct *p);
2037 extern void sched_autogroup_detach(struct task_struct *p);
2038 extern void sched_autogroup_fork(struct signal_struct *sig);
2039 extern void sched_autogroup_exit(struct signal_struct *sig);
2040 #ifdef CONFIG_PROC_FS
2041 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2042 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2043 #endif
2044 #else
2045 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2046 static inline void sched_autogroup_detach(struct task_struct *p) { }
2047 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2048 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2049 #endif
2050 
2051 #ifdef CONFIG_CFS_BANDWIDTH
2052 extern unsigned int sysctl_sched_cfs_bandwidth_slice;
2053 #endif
2054 
2055 #ifdef CONFIG_RT_MUTEXES
2056 extern int rt_mutex_getprio(struct task_struct *p);
2057 extern void rt_mutex_setprio(struct task_struct *p, int prio);
2058 extern void rt_mutex_adjust_pi(struct task_struct *p);
2059 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2060 {
2061 	return tsk->pi_blocked_on != NULL;
2062 }
2063 #else
2064 static inline int rt_mutex_getprio(struct task_struct *p)
2065 {
2066 	return p->normal_prio;
2067 }
2068 # define rt_mutex_adjust_pi(p)		do { } while (0)
2069 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2070 {
2071 	return false;
2072 }
2073 #endif
2074 
2075 extern bool yield_to(struct task_struct *p, bool preempt);
2076 extern void set_user_nice(struct task_struct *p, long nice);
2077 extern int task_prio(const struct task_struct *p);
2078 extern int task_nice(const struct task_struct *p);
2079 extern int can_nice(const struct task_struct *p, const int nice);
2080 extern int task_curr(const struct task_struct *p);
2081 extern int idle_cpu(int cpu);
2082 extern int sched_setscheduler(struct task_struct *, int,
2083 			      const struct sched_param *);
2084 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2085 				      const struct sched_param *);
2086 extern struct task_struct *idle_task(int cpu);
2087 /**
2088  * is_idle_task - is the specified task an idle task?
2089  * @p: the task in question.
2090  */
2091 static inline bool is_idle_task(const struct task_struct *p)
2092 {
2093 	return p->pid == 0;
2094 }
2095 extern struct task_struct *curr_task(int cpu);
2096 extern void set_curr_task(int cpu, struct task_struct *p);
2097 
2098 void yield(void);
2099 
2100 /*
2101  * The default (Linux) execution domain.
2102  */
2103 extern struct exec_domain	default_exec_domain;
2104 
2105 union thread_union {
2106 	struct thread_info thread_info;
2107 	unsigned long stack[THREAD_SIZE/sizeof(long)];
2108 };
2109 
2110 #ifndef __HAVE_ARCH_KSTACK_END
2111 static inline int kstack_end(void *addr)
2112 {
2113 	/* Reliable end of stack detection:
2114 	 * Some APM bios versions misalign the stack
2115 	 */
2116 	return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2117 }
2118 #endif
2119 
2120 extern union thread_union init_thread_union;
2121 extern struct task_struct init_task;
2122 
2123 extern struct   mm_struct init_mm;
2124 
2125 extern struct pid_namespace init_pid_ns;
2126 
2127 /*
2128  * find a task by one of its numerical ids
2129  *
2130  * find_task_by_pid_ns():
2131  *      finds a task by its pid in the specified namespace
2132  * find_task_by_vpid():
2133  *      finds a task by its virtual pid
2134  *
2135  * see also find_vpid() etc in include/linux/pid.h
2136  */
2137 
2138 extern struct task_struct *find_task_by_vpid(pid_t nr);
2139 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2140 		struct pid_namespace *ns);
2141 
2142 extern void __set_special_pids(struct pid *pid);
2143 
2144 /* per-UID process charging. */
2145 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2146 static inline struct user_struct *get_uid(struct user_struct *u)
2147 {
2148 	atomic_inc(&u->__count);
2149 	return u;
2150 }
2151 extern void free_uid(struct user_struct *);
2152 extern void release_uids(struct user_namespace *ns);
2153 
2154 #include <asm/current.h>
2155 
2156 extern void xtime_update(unsigned long ticks);
2157 
2158 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2159 extern int wake_up_process(struct task_struct *tsk);
2160 extern void wake_up_new_task(struct task_struct *tsk);
2161 #ifdef CONFIG_SMP
2162  extern void kick_process(struct task_struct *tsk);
2163 #else
2164  static inline void kick_process(struct task_struct *tsk) { }
2165 #endif
2166 extern void sched_fork(struct task_struct *p);
2167 extern void sched_dead(struct task_struct *p);
2168 
2169 extern void proc_caches_init(void);
2170 extern void flush_signals(struct task_struct *);
2171 extern void __flush_signals(struct task_struct *);
2172 extern void ignore_signals(struct task_struct *);
2173 extern void flush_signal_handlers(struct task_struct *, int force_default);
2174 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2175 
2176 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2177 {
2178 	unsigned long flags;
2179 	int ret;
2180 
2181 	spin_lock_irqsave(&tsk->sighand->siglock, flags);
2182 	ret = dequeue_signal(tsk, mask, info);
2183 	spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2184 
2185 	return ret;
2186 }
2187 
2188 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2189 			      sigset_t *mask);
2190 extern void unblock_all_signals(void);
2191 extern void release_task(struct task_struct * p);
2192 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2193 extern int force_sigsegv(int, struct task_struct *);
2194 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2195 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2196 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2197 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2198 				const struct cred *, u32);
2199 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2200 extern int kill_pid(struct pid *pid, int sig, int priv);
2201 extern int kill_proc_info(int, struct siginfo *, pid_t);
2202 extern __must_check bool do_notify_parent(struct task_struct *, int);
2203 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2204 extern void force_sig(int, struct task_struct *);
2205 extern int send_sig(int, struct task_struct *, int);
2206 extern int zap_other_threads(struct task_struct *p);
2207 extern struct sigqueue *sigqueue_alloc(void);
2208 extern void sigqueue_free(struct sigqueue *);
2209 extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
2210 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2211 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2212 
2213 static inline int kill_cad_pid(int sig, int priv)
2214 {
2215 	return kill_pid(cad_pid, sig, priv);
2216 }
2217 
2218 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
2219 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2220 #define SEND_SIG_PRIV	((struct siginfo *) 1)
2221 #define SEND_SIG_FORCED	((struct siginfo *) 2)
2222 
2223 /*
2224  * True if we are on the alternate signal stack.
2225  */
2226 static inline int on_sig_stack(unsigned long sp)
2227 {
2228 #ifdef CONFIG_STACK_GROWSUP
2229 	return sp >= current->sas_ss_sp &&
2230 		sp - current->sas_ss_sp < current->sas_ss_size;
2231 #else
2232 	return sp > current->sas_ss_sp &&
2233 		sp - current->sas_ss_sp <= current->sas_ss_size;
2234 #endif
2235 }
2236 
2237 static inline int sas_ss_flags(unsigned long sp)
2238 {
2239 	return (current->sas_ss_size == 0 ? SS_DISABLE
2240 		: on_sig_stack(sp) ? SS_ONSTACK : 0);
2241 }
2242 
2243 /*
2244  * Routines for handling mm_structs
2245  */
2246 extern struct mm_struct * mm_alloc(void);
2247 
2248 /* mmdrop drops the mm and the page tables */
2249 extern void __mmdrop(struct mm_struct *);
2250 static inline void mmdrop(struct mm_struct * mm)
2251 {
2252 	if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2253 		__mmdrop(mm);
2254 }
2255 
2256 /* mmput gets rid of the mappings and all user-space */
2257 extern void mmput(struct mm_struct *);
2258 /* Grab a reference to a task's mm, if it is not already going away */
2259 extern struct mm_struct *get_task_mm(struct task_struct *task);
2260 /*
2261  * Grab a reference to a task's mm, if it is not already going away
2262  * and ptrace_may_access with the mode parameter passed to it
2263  * succeeds.
2264  */
2265 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2266 /* Remove the current tasks stale references to the old mm_struct */
2267 extern void mm_release(struct task_struct *, struct mm_struct *);
2268 /* Allocate a new mm structure and copy contents from tsk->mm */
2269 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2270 
2271 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2272 			struct task_struct *, struct pt_regs *);
2273 extern void flush_thread(void);
2274 extern void exit_thread(void);
2275 
2276 extern void exit_files(struct task_struct *);
2277 extern void __cleanup_sighand(struct sighand_struct *);
2278 
2279 extern void exit_itimers(struct signal_struct *);
2280 extern void flush_itimer_signals(void);
2281 
2282 extern void do_group_exit(int);
2283 
2284 extern void daemonize(const char *, ...);
2285 extern int allow_signal(int);
2286 extern int disallow_signal(int);
2287 
2288 extern int do_execve(const char *,
2289 		     const char __user * const __user *,
2290 		     const char __user * const __user *, struct pt_regs *);
2291 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2292 struct task_struct *fork_idle(int);
2293 
2294 extern void set_task_comm(struct task_struct *tsk, char *from);
2295 extern char *get_task_comm(char *to, struct task_struct *tsk);
2296 
2297 #ifdef CONFIG_SMP
2298 void scheduler_ipi(void);
2299 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2300 #else
2301 static inline void scheduler_ipi(void) { }
2302 static inline unsigned long wait_task_inactive(struct task_struct *p,
2303 					       long match_state)
2304 {
2305 	return 1;
2306 }
2307 #endif
2308 
2309 #define next_task(p) \
2310 	list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2311 
2312 #define for_each_process(p) \
2313 	for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2314 
2315 extern bool current_is_single_threaded(void);
2316 
2317 /*
2318  * Careful: do_each_thread/while_each_thread is a double loop so
2319  *          'break' will not work as expected - use goto instead.
2320  */
2321 #define do_each_thread(g, t) \
2322 	for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2323 
2324 #define while_each_thread(g, t) \
2325 	while ((t = next_thread(t)) != g)
2326 
2327 static inline int get_nr_threads(struct task_struct *tsk)
2328 {
2329 	return tsk->signal->nr_threads;
2330 }
2331 
2332 static inline bool thread_group_leader(struct task_struct *p)
2333 {
2334 	return p->exit_signal >= 0;
2335 }
2336 
2337 /* Do to the insanities of de_thread it is possible for a process
2338  * to have the pid of the thread group leader without actually being
2339  * the thread group leader.  For iteration through the pids in proc
2340  * all we care about is that we have a task with the appropriate
2341  * pid, we don't actually care if we have the right task.
2342  */
2343 static inline int has_group_leader_pid(struct task_struct *p)
2344 {
2345 	return p->pid == p->tgid;
2346 }
2347 
2348 static inline
2349 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2350 {
2351 	return p1->tgid == p2->tgid;
2352 }
2353 
2354 static inline struct task_struct *next_thread(const struct task_struct *p)
2355 {
2356 	return list_entry_rcu(p->thread_group.next,
2357 			      struct task_struct, thread_group);
2358 }
2359 
2360 static inline int thread_group_empty(struct task_struct *p)
2361 {
2362 	return list_empty(&p->thread_group);
2363 }
2364 
2365 #define delay_group_leader(p) \
2366 		(thread_group_leader(p) && !thread_group_empty(p))
2367 
2368 /*
2369  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2370  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2371  * pins the final release of task.io_context.  Also protects ->cpuset and
2372  * ->cgroup.subsys[]. And ->vfork_done.
2373  *
2374  * Nests both inside and outside of read_lock(&tasklist_lock).
2375  * It must not be nested with write_lock_irq(&tasklist_lock),
2376  * neither inside nor outside.
2377  */
2378 static inline void task_lock(struct task_struct *p)
2379 {
2380 	spin_lock(&p->alloc_lock);
2381 }
2382 
2383 static inline void task_unlock(struct task_struct *p)
2384 {
2385 	spin_unlock(&p->alloc_lock);
2386 }
2387 
2388 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2389 							unsigned long *flags);
2390 
2391 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2392 						       unsigned long *flags)
2393 {
2394 	struct sighand_struct *ret;
2395 
2396 	ret = __lock_task_sighand(tsk, flags);
2397 	(void)__cond_lock(&tsk->sighand->siglock, ret);
2398 	return ret;
2399 }
2400 
2401 static inline void unlock_task_sighand(struct task_struct *tsk,
2402 						unsigned long *flags)
2403 {
2404 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2405 }
2406 
2407 #ifdef CONFIG_CGROUPS
2408 static inline void threadgroup_change_begin(struct task_struct *tsk)
2409 {
2410 	down_read(&tsk->signal->group_rwsem);
2411 }
2412 static inline void threadgroup_change_end(struct task_struct *tsk)
2413 {
2414 	up_read(&tsk->signal->group_rwsem);
2415 }
2416 
2417 /**
2418  * threadgroup_lock - lock threadgroup
2419  * @tsk: member task of the threadgroup to lock
2420  *
2421  * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
2422  * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2423  * perform exec.  This is useful for cases where the threadgroup needs to
2424  * stay stable across blockable operations.
2425  *
2426  * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2427  * synchronization.  While held, no new task will be added to threadgroup
2428  * and no existing live task will have its PF_EXITING set.
2429  *
2430  * During exec, a task goes and puts its thread group through unusual
2431  * changes.  After de-threading, exclusive access is assumed to resources
2432  * which are usually shared by tasks in the same group - e.g. sighand may
2433  * be replaced with a new one.  Also, the exec'ing task takes over group
2434  * leader role including its pid.  Exclude these changes while locked by
2435  * grabbing cred_guard_mutex which is used to synchronize exec path.
2436  */
2437 static inline void threadgroup_lock(struct task_struct *tsk)
2438 {
2439 	/*
2440 	 * exec uses exit for de-threading nesting group_rwsem inside
2441 	 * cred_guard_mutex. Grab cred_guard_mutex first.
2442 	 */
2443 	mutex_lock(&tsk->signal->cred_guard_mutex);
2444 	down_write(&tsk->signal->group_rwsem);
2445 }
2446 
2447 /**
2448  * threadgroup_unlock - unlock threadgroup
2449  * @tsk: member task of the threadgroup to unlock
2450  *
2451  * Reverse threadgroup_lock().
2452  */
2453 static inline void threadgroup_unlock(struct task_struct *tsk)
2454 {
2455 	up_write(&tsk->signal->group_rwsem);
2456 	mutex_unlock(&tsk->signal->cred_guard_mutex);
2457 }
2458 #else
2459 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2460 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2461 static inline void threadgroup_lock(struct task_struct *tsk) {}
2462 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2463 #endif
2464 
2465 #ifndef __HAVE_THREAD_FUNCTIONS
2466 
2467 #define task_thread_info(task)	((struct thread_info *)(task)->stack)
2468 #define task_stack_page(task)	((task)->stack)
2469 
2470 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2471 {
2472 	*task_thread_info(p) = *task_thread_info(org);
2473 	task_thread_info(p)->task = p;
2474 }
2475 
2476 static inline unsigned long *end_of_stack(struct task_struct *p)
2477 {
2478 	return (unsigned long *)(task_thread_info(p) + 1);
2479 }
2480 
2481 #endif
2482 
2483 static inline int object_is_on_stack(void *obj)
2484 {
2485 	void *stack = task_stack_page(current);
2486 
2487 	return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2488 }
2489 
2490 extern void thread_info_cache_init(void);
2491 
2492 #ifdef CONFIG_DEBUG_STACK_USAGE
2493 static inline unsigned long stack_not_used(struct task_struct *p)
2494 {
2495 	unsigned long *n = end_of_stack(p);
2496 
2497 	do { 	/* Skip over canary */
2498 		n++;
2499 	} while (!*n);
2500 
2501 	return (unsigned long)n - (unsigned long)end_of_stack(p);
2502 }
2503 #endif
2504 
2505 /* set thread flags in other task's structures
2506  * - see asm/thread_info.h for TIF_xxxx flags available
2507  */
2508 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2509 {
2510 	set_ti_thread_flag(task_thread_info(tsk), flag);
2511 }
2512 
2513 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2514 {
2515 	clear_ti_thread_flag(task_thread_info(tsk), flag);
2516 }
2517 
2518 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2519 {
2520 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2521 }
2522 
2523 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2524 {
2525 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2526 }
2527 
2528 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2529 {
2530 	return test_ti_thread_flag(task_thread_info(tsk), flag);
2531 }
2532 
2533 static inline void set_tsk_need_resched(struct task_struct *tsk)
2534 {
2535 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2536 }
2537 
2538 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2539 {
2540 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2541 }
2542 
2543 static inline int test_tsk_need_resched(struct task_struct *tsk)
2544 {
2545 	return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2546 }
2547 
2548 static inline int restart_syscall(void)
2549 {
2550 	set_tsk_thread_flag(current, TIF_SIGPENDING);
2551 	return -ERESTARTNOINTR;
2552 }
2553 
2554 static inline int signal_pending(struct task_struct *p)
2555 {
2556 	return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2557 }
2558 
2559 static inline int __fatal_signal_pending(struct task_struct *p)
2560 {
2561 	return unlikely(sigismember(&p->pending.signal, SIGKILL));
2562 }
2563 
2564 static inline int fatal_signal_pending(struct task_struct *p)
2565 {
2566 	return signal_pending(p) && __fatal_signal_pending(p);
2567 }
2568 
2569 static inline int signal_pending_state(long state, struct task_struct *p)
2570 {
2571 	if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2572 		return 0;
2573 	if (!signal_pending(p))
2574 		return 0;
2575 
2576 	return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2577 }
2578 
2579 static inline int need_resched(void)
2580 {
2581 	return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2582 }
2583 
2584 /*
2585  * cond_resched() and cond_resched_lock(): latency reduction via
2586  * explicit rescheduling in places that are safe. The return
2587  * value indicates whether a reschedule was done in fact.
2588  * cond_resched_lock() will drop the spinlock before scheduling,
2589  * cond_resched_softirq() will enable bhs before scheduling.
2590  */
2591 extern int _cond_resched(void);
2592 
2593 #define cond_resched() ({			\
2594 	__might_sleep(__FILE__, __LINE__, 0);	\
2595 	_cond_resched();			\
2596 })
2597 
2598 extern int __cond_resched_lock(spinlock_t *lock);
2599 
2600 #ifdef CONFIG_PREEMPT_COUNT
2601 #define PREEMPT_LOCK_OFFSET	PREEMPT_OFFSET
2602 #else
2603 #define PREEMPT_LOCK_OFFSET	0
2604 #endif
2605 
2606 #define cond_resched_lock(lock) ({				\
2607 	__might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);	\
2608 	__cond_resched_lock(lock);				\
2609 })
2610 
2611 extern int __cond_resched_softirq(void);
2612 
2613 #define cond_resched_softirq() ({					\
2614 	__might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET);	\
2615 	__cond_resched_softirq();					\
2616 })
2617 
2618 /*
2619  * Does a critical section need to be broken due to another
2620  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2621  * but a general need for low latency)
2622  */
2623 static inline int spin_needbreak(spinlock_t *lock)
2624 {
2625 #ifdef CONFIG_PREEMPT
2626 	return spin_is_contended(lock);
2627 #else
2628 	return 0;
2629 #endif
2630 }
2631 
2632 /*
2633  * Thread group CPU time accounting.
2634  */
2635 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2636 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2637 
2638 static inline void thread_group_cputime_init(struct signal_struct *sig)
2639 {
2640 	raw_spin_lock_init(&sig->cputimer.lock);
2641 }
2642 
2643 /*
2644  * Reevaluate whether the task has signals pending delivery.
2645  * Wake the task if so.
2646  * This is required every time the blocked sigset_t changes.
2647  * callers must hold sighand->siglock.
2648  */
2649 extern void recalc_sigpending_and_wake(struct task_struct *t);
2650 extern void recalc_sigpending(void);
2651 
2652 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2653 
2654 /*
2655  * Wrappers for p->thread_info->cpu access. No-op on UP.
2656  */
2657 #ifdef CONFIG_SMP
2658 
2659 static inline unsigned int task_cpu(const struct task_struct *p)
2660 {
2661 	return task_thread_info(p)->cpu;
2662 }
2663 
2664 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2665 
2666 #else
2667 
2668 static inline unsigned int task_cpu(const struct task_struct *p)
2669 {
2670 	return 0;
2671 }
2672 
2673 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2674 {
2675 }
2676 
2677 #endif /* CONFIG_SMP */
2678 
2679 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2680 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2681 
2682 extern void normalize_rt_tasks(void);
2683 
2684 #ifdef CONFIG_CGROUP_SCHED
2685 
2686 extern struct task_group root_task_group;
2687 
2688 extern struct task_group *sched_create_group(struct task_group *parent);
2689 extern void sched_destroy_group(struct task_group *tg);
2690 extern void sched_move_task(struct task_struct *tsk);
2691 #ifdef CONFIG_FAIR_GROUP_SCHED
2692 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2693 extern unsigned long sched_group_shares(struct task_group *tg);
2694 #endif
2695 #ifdef CONFIG_RT_GROUP_SCHED
2696 extern int sched_group_set_rt_runtime(struct task_group *tg,
2697 				      long rt_runtime_us);
2698 extern long sched_group_rt_runtime(struct task_group *tg);
2699 extern int sched_group_set_rt_period(struct task_group *tg,
2700 				      long rt_period_us);
2701 extern long sched_group_rt_period(struct task_group *tg);
2702 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2703 #endif
2704 #endif
2705 
2706 extern int task_can_switch_user(struct user_struct *up,
2707 					struct task_struct *tsk);
2708 
2709 #ifdef CONFIG_TASK_XACCT
2710 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2711 {
2712 	tsk->ioac.rchar += amt;
2713 }
2714 
2715 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2716 {
2717 	tsk->ioac.wchar += amt;
2718 }
2719 
2720 static inline void inc_syscr(struct task_struct *tsk)
2721 {
2722 	tsk->ioac.syscr++;
2723 }
2724 
2725 static inline void inc_syscw(struct task_struct *tsk)
2726 {
2727 	tsk->ioac.syscw++;
2728 }
2729 #else
2730 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2731 {
2732 }
2733 
2734 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2735 {
2736 }
2737 
2738 static inline void inc_syscr(struct task_struct *tsk)
2739 {
2740 }
2741 
2742 static inline void inc_syscw(struct task_struct *tsk)
2743 {
2744 }
2745 #endif
2746 
2747 #ifndef TASK_SIZE_OF
2748 #define TASK_SIZE_OF(tsk)	TASK_SIZE
2749 #endif
2750 
2751 #ifdef CONFIG_MM_OWNER
2752 extern void mm_update_next_owner(struct mm_struct *mm);
2753 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2754 #else
2755 static inline void mm_update_next_owner(struct mm_struct *mm)
2756 {
2757 }
2758 
2759 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2760 {
2761 }
2762 #endif /* CONFIG_MM_OWNER */
2763 
2764 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2765 		unsigned int limit)
2766 {
2767 	return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2768 }
2769 
2770 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2771 		unsigned int limit)
2772 {
2773 	return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2774 }
2775 
2776 static inline unsigned long rlimit(unsigned int limit)
2777 {
2778 	return task_rlimit(current, limit);
2779 }
2780 
2781 static inline unsigned long rlimit_max(unsigned int limit)
2782 {
2783 	return task_rlimit_max(current, limit);
2784 }
2785 
2786 #endif /* __KERNEL__ */
2787 
2788 #endif
2789