xref: /linux/include/linux/sched.h (revision 4413e16d9d21673bb5048a2e542f1aaa00015c2e)
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3 
4 /*
5  * cloning flags:
6  */
7 #define CSIGNAL		0x000000ff	/* signal mask to be sent at exit */
8 #define CLONE_VM	0x00000100	/* set if VM shared between processes */
9 #define CLONE_FS	0x00000200	/* set if fs info shared between processes */
10 #define CLONE_FILES	0x00000400	/* set if open files shared between processes */
11 #define CLONE_SIGHAND	0x00000800	/* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE	0x00002000	/* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK	0x00004000	/* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT	0x00008000	/* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD	0x00010000	/* Same thread group? */
16 #define CLONE_NEWNS	0x00020000	/* New namespace group? */
17 #define CLONE_SYSVSEM	0x00040000	/* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS	0x00080000	/* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID	0x00100000	/* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID	0x00200000	/* clear the TID in the child */
21 #define CLONE_DETACHED		0x00400000	/* Unused, ignored */
22 #define CLONE_UNTRACED		0x00800000	/* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID	0x01000000	/* set the TID in the child */
24 /* 0x02000000 was previously the unused CLONE_STOPPED (Start in stopped state)
25    and is now available for re-use. */
26 #define CLONE_NEWUTS		0x04000000	/* New utsname group? */
27 #define CLONE_NEWIPC		0x08000000	/* New ipcs */
28 #define CLONE_NEWUSER		0x10000000	/* New user namespace */
29 #define CLONE_NEWPID		0x20000000	/* New pid namespace */
30 #define CLONE_NEWNET		0x40000000	/* New network namespace */
31 #define CLONE_IO		0x80000000	/* Clone io context */
32 
33 /*
34  * Scheduling policies
35  */
36 #define SCHED_NORMAL		0
37 #define SCHED_FIFO		1
38 #define SCHED_RR		2
39 #define SCHED_BATCH		3
40 /* SCHED_ISO: reserved but not implemented yet */
41 #define SCHED_IDLE		5
42 /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
43 #define SCHED_RESET_ON_FORK     0x40000000
44 
45 #ifdef __KERNEL__
46 
47 struct sched_param {
48 	int sched_priority;
49 };
50 
51 #include <asm/param.h>	/* for HZ */
52 
53 #include <linux/capability.h>
54 #include <linux/threads.h>
55 #include <linux/kernel.h>
56 #include <linux/types.h>
57 #include <linux/timex.h>
58 #include <linux/jiffies.h>
59 #include <linux/rbtree.h>
60 #include <linux/thread_info.h>
61 #include <linux/cpumask.h>
62 #include <linux/errno.h>
63 #include <linux/nodemask.h>
64 #include <linux/mm_types.h>
65 
66 #include <asm/page.h>
67 #include <asm/ptrace.h>
68 #include <asm/cputime.h>
69 
70 #include <linux/smp.h>
71 #include <linux/sem.h>
72 #include <linux/signal.h>
73 #include <linux/compiler.h>
74 #include <linux/completion.h>
75 #include <linux/pid.h>
76 #include <linux/percpu.h>
77 #include <linux/topology.h>
78 #include <linux/proportions.h>
79 #include <linux/seccomp.h>
80 #include <linux/rcupdate.h>
81 #include <linux/rculist.h>
82 #include <linux/rtmutex.h>
83 
84 #include <linux/time.h>
85 #include <linux/param.h>
86 #include <linux/resource.h>
87 #include <linux/timer.h>
88 #include <linux/hrtimer.h>
89 #include <linux/task_io_accounting.h>
90 #include <linux/latencytop.h>
91 #include <linux/cred.h>
92 #include <linux/llist.h>
93 #include <linux/uidgid.h>
94 
95 #include <asm/processor.h>
96 
97 struct exec_domain;
98 struct futex_pi_state;
99 struct robust_list_head;
100 struct bio_list;
101 struct fs_struct;
102 struct perf_event_context;
103 struct blk_plug;
104 
105 /*
106  * List of flags we want to share for kernel threads,
107  * if only because they are not used by them anyway.
108  */
109 #define CLONE_KERNEL	(CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
110 
111 /*
112  * These are the constant used to fake the fixed-point load-average
113  * counting. Some notes:
114  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
115  *    a load-average precision of 10 bits integer + 11 bits fractional
116  *  - if you want to count load-averages more often, you need more
117  *    precision, or rounding will get you. With 2-second counting freq,
118  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
119  *    11 bit fractions.
120  */
121 extern unsigned long avenrun[];		/* Load averages */
122 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
123 
124 #define FSHIFT		11		/* nr of bits of precision */
125 #define FIXED_1		(1<<FSHIFT)	/* 1.0 as fixed-point */
126 #define LOAD_FREQ	(5*HZ+1)	/* 5 sec intervals */
127 #define EXP_1		1884		/* 1/exp(5sec/1min) as fixed-point */
128 #define EXP_5		2014		/* 1/exp(5sec/5min) */
129 #define EXP_15		2037		/* 1/exp(5sec/15min) */
130 
131 #define CALC_LOAD(load,exp,n) \
132 	load *= exp; \
133 	load += n*(FIXED_1-exp); \
134 	load >>= FSHIFT;
135 
136 extern unsigned long total_forks;
137 extern int nr_threads;
138 DECLARE_PER_CPU(unsigned long, process_counts);
139 extern int nr_processes(void);
140 extern unsigned long nr_running(void);
141 extern unsigned long nr_uninterruptible(void);
142 extern unsigned long nr_iowait(void);
143 extern unsigned long nr_iowait_cpu(int cpu);
144 extern unsigned long this_cpu_load(void);
145 
146 
147 extern void calc_global_load(unsigned long ticks);
148 extern void update_cpu_load_nohz(void);
149 
150 extern unsigned long get_parent_ip(unsigned long addr);
151 
152 struct seq_file;
153 struct cfs_rq;
154 struct task_group;
155 #ifdef CONFIG_SCHED_DEBUG
156 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
157 extern void proc_sched_set_task(struct task_struct *p);
158 extern void
159 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
160 #else
161 static inline void
162 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
163 {
164 }
165 static inline void proc_sched_set_task(struct task_struct *p)
166 {
167 }
168 static inline void
169 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
170 {
171 }
172 #endif
173 
174 /*
175  * Task state bitmask. NOTE! These bits are also
176  * encoded in fs/proc/array.c: get_task_state().
177  *
178  * We have two separate sets of flags: task->state
179  * is about runnability, while task->exit_state are
180  * about the task exiting. Confusing, but this way
181  * modifying one set can't modify the other one by
182  * mistake.
183  */
184 #define TASK_RUNNING		0
185 #define TASK_INTERRUPTIBLE	1
186 #define TASK_UNINTERRUPTIBLE	2
187 #define __TASK_STOPPED		4
188 #define __TASK_TRACED		8
189 /* in tsk->exit_state */
190 #define EXIT_ZOMBIE		16
191 #define EXIT_DEAD		32
192 /* in tsk->state again */
193 #define TASK_DEAD		64
194 #define TASK_WAKEKILL		128
195 #define TASK_WAKING		256
196 #define TASK_STATE_MAX		512
197 
198 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
199 
200 extern char ___assert_task_state[1 - 2*!!(
201 		sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
202 
203 /* Convenience macros for the sake of set_task_state */
204 #define TASK_KILLABLE		(TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
205 #define TASK_STOPPED		(TASK_WAKEKILL | __TASK_STOPPED)
206 #define TASK_TRACED		(TASK_WAKEKILL | __TASK_TRACED)
207 
208 /* Convenience macros for the sake of wake_up */
209 #define TASK_NORMAL		(TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
210 #define TASK_ALL		(TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
211 
212 /* get_task_state() */
213 #define TASK_REPORT		(TASK_RUNNING | TASK_INTERRUPTIBLE | \
214 				 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
215 				 __TASK_TRACED)
216 
217 #define task_is_traced(task)	((task->state & __TASK_TRACED) != 0)
218 #define task_is_stopped(task)	((task->state & __TASK_STOPPED) != 0)
219 #define task_is_dead(task)	((task)->exit_state != 0)
220 #define task_is_stopped_or_traced(task)	\
221 			((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
222 #define task_contributes_to_load(task)	\
223 				((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
224 				 (task->flags & PF_FROZEN) == 0)
225 
226 #define __set_task_state(tsk, state_value)		\
227 	do { (tsk)->state = (state_value); } while (0)
228 #define set_task_state(tsk, state_value)		\
229 	set_mb((tsk)->state, (state_value))
230 
231 /*
232  * set_current_state() includes a barrier so that the write of current->state
233  * is correctly serialised wrt the caller's subsequent test of whether to
234  * actually sleep:
235  *
236  *	set_current_state(TASK_UNINTERRUPTIBLE);
237  *	if (do_i_need_to_sleep())
238  *		schedule();
239  *
240  * If the caller does not need such serialisation then use __set_current_state()
241  */
242 #define __set_current_state(state_value)			\
243 	do { current->state = (state_value); } while (0)
244 #define set_current_state(state_value)		\
245 	set_mb(current->state, (state_value))
246 
247 /* Task command name length */
248 #define TASK_COMM_LEN 16
249 
250 #include <linux/spinlock.h>
251 
252 /*
253  * This serializes "schedule()" and also protects
254  * the run-queue from deletions/modifications (but
255  * _adding_ to the beginning of the run-queue has
256  * a separate lock).
257  */
258 extern rwlock_t tasklist_lock;
259 extern spinlock_t mmlist_lock;
260 
261 struct task_struct;
262 
263 #ifdef CONFIG_PROVE_RCU
264 extern int lockdep_tasklist_lock_is_held(void);
265 #endif /* #ifdef CONFIG_PROVE_RCU */
266 
267 extern void sched_init(void);
268 extern void sched_init_smp(void);
269 extern asmlinkage void schedule_tail(struct task_struct *prev);
270 extern void init_idle(struct task_struct *idle, int cpu);
271 extern void init_idle_bootup_task(struct task_struct *idle);
272 
273 extern int runqueue_is_locked(int cpu);
274 
275 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
276 extern void nohz_balance_enter_idle(int cpu);
277 extern void set_cpu_sd_state_idle(void);
278 extern int get_nohz_timer_target(void);
279 #else
280 static inline void nohz_balance_enter_idle(int cpu) { }
281 static inline void set_cpu_sd_state_idle(void) { }
282 #endif
283 
284 /*
285  * Only dump TASK_* tasks. (0 for all tasks)
286  */
287 extern void show_state_filter(unsigned long state_filter);
288 
289 static inline void show_state(void)
290 {
291 	show_state_filter(0);
292 }
293 
294 extern void show_regs(struct pt_regs *);
295 
296 /*
297  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
298  * task), SP is the stack pointer of the first frame that should be shown in the back
299  * trace (or NULL if the entire call-chain of the task should be shown).
300  */
301 extern void show_stack(struct task_struct *task, unsigned long *sp);
302 
303 void io_schedule(void);
304 long io_schedule_timeout(long timeout);
305 
306 extern void cpu_init (void);
307 extern void trap_init(void);
308 extern void update_process_times(int user);
309 extern void scheduler_tick(void);
310 
311 extern void sched_show_task(struct task_struct *p);
312 
313 #ifdef CONFIG_LOCKUP_DETECTOR
314 extern void touch_softlockup_watchdog(void);
315 extern void touch_softlockup_watchdog_sync(void);
316 extern void touch_all_softlockup_watchdogs(void);
317 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
318 				  void __user *buffer,
319 				  size_t *lenp, loff_t *ppos);
320 extern unsigned int  softlockup_panic;
321 void lockup_detector_init(void);
322 #else
323 static inline void touch_softlockup_watchdog(void)
324 {
325 }
326 static inline void touch_softlockup_watchdog_sync(void)
327 {
328 }
329 static inline void touch_all_softlockup_watchdogs(void)
330 {
331 }
332 static inline void lockup_detector_init(void)
333 {
334 }
335 #endif
336 
337 #ifdef CONFIG_DETECT_HUNG_TASK
338 extern unsigned int  sysctl_hung_task_panic;
339 extern unsigned long sysctl_hung_task_check_count;
340 extern unsigned long sysctl_hung_task_timeout_secs;
341 extern unsigned long sysctl_hung_task_warnings;
342 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
343 					 void __user *buffer,
344 					 size_t *lenp, loff_t *ppos);
345 #else
346 /* Avoid need for ifdefs elsewhere in the code */
347 enum { sysctl_hung_task_timeout_secs = 0 };
348 #endif
349 
350 /* Attach to any functions which should be ignored in wchan output. */
351 #define __sched		__attribute__((__section__(".sched.text")))
352 
353 /* Linker adds these: start and end of __sched functions */
354 extern char __sched_text_start[], __sched_text_end[];
355 
356 /* Is this address in the __sched functions? */
357 extern int in_sched_functions(unsigned long addr);
358 
359 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
360 extern signed long schedule_timeout(signed long timeout);
361 extern signed long schedule_timeout_interruptible(signed long timeout);
362 extern signed long schedule_timeout_killable(signed long timeout);
363 extern signed long schedule_timeout_uninterruptible(signed long timeout);
364 asmlinkage void schedule(void);
365 extern void schedule_preempt_disabled(void);
366 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
367 
368 struct nsproxy;
369 struct user_namespace;
370 
371 /*
372  * Default maximum number of active map areas, this limits the number of vmas
373  * per mm struct. Users can overwrite this number by sysctl but there is a
374  * problem.
375  *
376  * When a program's coredump is generated as ELF format, a section is created
377  * per a vma. In ELF, the number of sections is represented in unsigned short.
378  * This means the number of sections should be smaller than 65535 at coredump.
379  * Because the kernel adds some informative sections to a image of program at
380  * generating coredump, we need some margin. The number of extra sections is
381  * 1-3 now and depends on arch. We use "5" as safe margin, here.
382  */
383 #define MAPCOUNT_ELF_CORE_MARGIN	(5)
384 #define DEFAULT_MAX_MAP_COUNT	(USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
385 
386 extern int sysctl_max_map_count;
387 
388 #include <linux/aio.h>
389 
390 #ifdef CONFIG_MMU
391 extern void arch_pick_mmap_layout(struct mm_struct *mm);
392 extern unsigned long
393 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
394 		       unsigned long, unsigned long);
395 extern unsigned long
396 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
397 			  unsigned long len, unsigned long pgoff,
398 			  unsigned long flags);
399 extern void arch_unmap_area(struct mm_struct *, unsigned long);
400 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
401 #else
402 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
403 #endif
404 
405 
406 extern void set_dumpable(struct mm_struct *mm, int value);
407 extern int get_dumpable(struct mm_struct *mm);
408 
409 /* get/set_dumpable() values */
410 #define SUID_DUMPABLE_DISABLED	0
411 #define SUID_DUMPABLE_ENABLED	1
412 #define SUID_DUMPABLE_SAFE	2
413 
414 /* mm flags */
415 /* dumpable bits */
416 #define MMF_DUMPABLE      0  /* core dump is permitted */
417 #define MMF_DUMP_SECURELY 1  /* core file is readable only by root */
418 
419 #define MMF_DUMPABLE_BITS 2
420 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
421 
422 /* coredump filter bits */
423 #define MMF_DUMP_ANON_PRIVATE	2
424 #define MMF_DUMP_ANON_SHARED	3
425 #define MMF_DUMP_MAPPED_PRIVATE	4
426 #define MMF_DUMP_MAPPED_SHARED	5
427 #define MMF_DUMP_ELF_HEADERS	6
428 #define MMF_DUMP_HUGETLB_PRIVATE 7
429 #define MMF_DUMP_HUGETLB_SHARED  8
430 
431 #define MMF_DUMP_FILTER_SHIFT	MMF_DUMPABLE_BITS
432 #define MMF_DUMP_FILTER_BITS	7
433 #define MMF_DUMP_FILTER_MASK \
434 	(((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
435 #define MMF_DUMP_FILTER_DEFAULT \
436 	((1 << MMF_DUMP_ANON_PRIVATE) |	(1 << MMF_DUMP_ANON_SHARED) |\
437 	 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
438 
439 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
440 # define MMF_DUMP_MASK_DEFAULT_ELF	(1 << MMF_DUMP_ELF_HEADERS)
441 #else
442 # define MMF_DUMP_MASK_DEFAULT_ELF	0
443 #endif
444 					/* leave room for more dump flags */
445 #define MMF_VM_MERGEABLE	16	/* KSM may merge identical pages */
446 #define MMF_VM_HUGEPAGE		17	/* set when VM_HUGEPAGE is set on vma */
447 #define MMF_EXE_FILE_CHANGED	18	/* see prctl_set_mm_exe_file() */
448 
449 #define MMF_HAS_UPROBES		19	/* has uprobes */
450 #define MMF_RECALC_UPROBES	20	/* MMF_HAS_UPROBES can be wrong */
451 
452 #define MMF_INIT_MASK		(MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
453 
454 struct sighand_struct {
455 	atomic_t		count;
456 	struct k_sigaction	action[_NSIG];
457 	spinlock_t		siglock;
458 	wait_queue_head_t	signalfd_wqh;
459 };
460 
461 struct pacct_struct {
462 	int			ac_flag;
463 	long			ac_exitcode;
464 	unsigned long		ac_mem;
465 	cputime_t		ac_utime, ac_stime;
466 	unsigned long		ac_minflt, ac_majflt;
467 };
468 
469 struct cpu_itimer {
470 	cputime_t expires;
471 	cputime_t incr;
472 	u32 error;
473 	u32 incr_error;
474 };
475 
476 /**
477  * struct task_cputime - collected CPU time counts
478  * @utime:		time spent in user mode, in &cputime_t units
479  * @stime:		time spent in kernel mode, in &cputime_t units
480  * @sum_exec_runtime:	total time spent on the CPU, in nanoseconds
481  *
482  * This structure groups together three kinds of CPU time that are
483  * tracked for threads and thread groups.  Most things considering
484  * CPU time want to group these counts together and treat all three
485  * of them in parallel.
486  */
487 struct task_cputime {
488 	cputime_t utime;
489 	cputime_t stime;
490 	unsigned long long sum_exec_runtime;
491 };
492 /* Alternate field names when used to cache expirations. */
493 #define prof_exp	stime
494 #define virt_exp	utime
495 #define sched_exp	sum_exec_runtime
496 
497 #define INIT_CPUTIME	\
498 	(struct task_cputime) {					\
499 		.utime = 0,					\
500 		.stime = 0,					\
501 		.sum_exec_runtime = 0,				\
502 	}
503 
504 /*
505  * Disable preemption until the scheduler is running.
506  * Reset by start_kernel()->sched_init()->init_idle().
507  *
508  * We include PREEMPT_ACTIVE to avoid cond_resched() from working
509  * before the scheduler is active -- see should_resched().
510  */
511 #define INIT_PREEMPT_COUNT	(1 + PREEMPT_ACTIVE)
512 
513 /**
514  * struct thread_group_cputimer - thread group interval timer counts
515  * @cputime:		thread group interval timers.
516  * @running:		non-zero when there are timers running and
517  * 			@cputime receives updates.
518  * @lock:		lock for fields in this struct.
519  *
520  * This structure contains the version of task_cputime, above, that is
521  * used for thread group CPU timer calculations.
522  */
523 struct thread_group_cputimer {
524 	struct task_cputime cputime;
525 	int running;
526 	raw_spinlock_t lock;
527 };
528 
529 #include <linux/rwsem.h>
530 struct autogroup;
531 
532 /*
533  * NOTE! "signal_struct" does not have its own
534  * locking, because a shared signal_struct always
535  * implies a shared sighand_struct, so locking
536  * sighand_struct is always a proper superset of
537  * the locking of signal_struct.
538  */
539 struct signal_struct {
540 	atomic_t		sigcnt;
541 	atomic_t		live;
542 	int			nr_threads;
543 
544 	wait_queue_head_t	wait_chldexit;	/* for wait4() */
545 
546 	/* current thread group signal load-balancing target: */
547 	struct task_struct	*curr_target;
548 
549 	/* shared signal handling: */
550 	struct sigpending	shared_pending;
551 
552 	/* thread group exit support */
553 	int			group_exit_code;
554 	/* overloaded:
555 	 * - notify group_exit_task when ->count is equal to notify_count
556 	 * - everyone except group_exit_task is stopped during signal delivery
557 	 *   of fatal signals, group_exit_task processes the signal.
558 	 */
559 	int			notify_count;
560 	struct task_struct	*group_exit_task;
561 
562 	/* thread group stop support, overloads group_exit_code too */
563 	int			group_stop_count;
564 	unsigned int		flags; /* see SIGNAL_* flags below */
565 
566 	/*
567 	 * PR_SET_CHILD_SUBREAPER marks a process, like a service
568 	 * manager, to re-parent orphan (double-forking) child processes
569 	 * to this process instead of 'init'. The service manager is
570 	 * able to receive SIGCHLD signals and is able to investigate
571 	 * the process until it calls wait(). All children of this
572 	 * process will inherit a flag if they should look for a
573 	 * child_subreaper process at exit.
574 	 */
575 	unsigned int		is_child_subreaper:1;
576 	unsigned int		has_child_subreaper:1;
577 
578 	/* POSIX.1b Interval Timers */
579 	struct list_head posix_timers;
580 
581 	/* ITIMER_REAL timer for the process */
582 	struct hrtimer real_timer;
583 	struct pid *leader_pid;
584 	ktime_t it_real_incr;
585 
586 	/*
587 	 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
588 	 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
589 	 * values are defined to 0 and 1 respectively
590 	 */
591 	struct cpu_itimer it[2];
592 
593 	/*
594 	 * Thread group totals for process CPU timers.
595 	 * See thread_group_cputimer(), et al, for details.
596 	 */
597 	struct thread_group_cputimer cputimer;
598 
599 	/* Earliest-expiration cache. */
600 	struct task_cputime cputime_expires;
601 
602 	struct list_head cpu_timers[3];
603 
604 	struct pid *tty_old_pgrp;
605 
606 	/* boolean value for session group leader */
607 	int leader;
608 
609 	struct tty_struct *tty; /* NULL if no tty */
610 
611 #ifdef CONFIG_SCHED_AUTOGROUP
612 	struct autogroup *autogroup;
613 #endif
614 	/*
615 	 * Cumulative resource counters for dead threads in the group,
616 	 * and for reaped dead child processes forked by this group.
617 	 * Live threads maintain their own counters and add to these
618 	 * in __exit_signal, except for the group leader.
619 	 */
620 	cputime_t utime, stime, cutime, cstime;
621 	cputime_t gtime;
622 	cputime_t cgtime;
623 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
624 	cputime_t prev_utime, prev_stime;
625 #endif
626 	unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
627 	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
628 	unsigned long inblock, oublock, cinblock, coublock;
629 	unsigned long maxrss, cmaxrss;
630 	struct task_io_accounting ioac;
631 
632 	/*
633 	 * Cumulative ns of schedule CPU time fo dead threads in the
634 	 * group, not including a zombie group leader, (This only differs
635 	 * from jiffies_to_ns(utime + stime) if sched_clock uses something
636 	 * other than jiffies.)
637 	 */
638 	unsigned long long sum_sched_runtime;
639 
640 	/*
641 	 * We don't bother to synchronize most readers of this at all,
642 	 * because there is no reader checking a limit that actually needs
643 	 * to get both rlim_cur and rlim_max atomically, and either one
644 	 * alone is a single word that can safely be read normally.
645 	 * getrlimit/setrlimit use task_lock(current->group_leader) to
646 	 * protect this instead of the siglock, because they really
647 	 * have no need to disable irqs.
648 	 */
649 	struct rlimit rlim[RLIM_NLIMITS];
650 
651 #ifdef CONFIG_BSD_PROCESS_ACCT
652 	struct pacct_struct pacct;	/* per-process accounting information */
653 #endif
654 #ifdef CONFIG_TASKSTATS
655 	struct taskstats *stats;
656 #endif
657 #ifdef CONFIG_AUDIT
658 	unsigned audit_tty;
659 	struct tty_audit_buf *tty_audit_buf;
660 #endif
661 #ifdef CONFIG_CGROUPS
662 	/*
663 	 * group_rwsem prevents new tasks from entering the threadgroup and
664 	 * member tasks from exiting,a more specifically, setting of
665 	 * PF_EXITING.  fork and exit paths are protected with this rwsem
666 	 * using threadgroup_change_begin/end().  Users which require
667 	 * threadgroup to remain stable should use threadgroup_[un]lock()
668 	 * which also takes care of exec path.  Currently, cgroup is the
669 	 * only user.
670 	 */
671 	struct rw_semaphore group_rwsem;
672 #endif
673 
674 	int oom_adj;		/* OOM kill score adjustment (bit shift) */
675 	int oom_score_adj;	/* OOM kill score adjustment */
676 	int oom_score_adj_min;	/* OOM kill score adjustment minimum value.
677 				 * Only settable by CAP_SYS_RESOURCE. */
678 
679 	struct mutex cred_guard_mutex;	/* guard against foreign influences on
680 					 * credential calculations
681 					 * (notably. ptrace) */
682 };
683 
684 /*
685  * Bits in flags field of signal_struct.
686  */
687 #define SIGNAL_STOP_STOPPED	0x00000001 /* job control stop in effect */
688 #define SIGNAL_STOP_CONTINUED	0x00000002 /* SIGCONT since WCONTINUED reap */
689 #define SIGNAL_GROUP_EXIT	0x00000004 /* group exit in progress */
690 /*
691  * Pending notifications to parent.
692  */
693 #define SIGNAL_CLD_STOPPED	0x00000010
694 #define SIGNAL_CLD_CONTINUED	0x00000020
695 #define SIGNAL_CLD_MASK		(SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
696 
697 #define SIGNAL_UNKILLABLE	0x00000040 /* for init: ignore fatal signals */
698 
699 /* If true, all threads except ->group_exit_task have pending SIGKILL */
700 static inline int signal_group_exit(const struct signal_struct *sig)
701 {
702 	return	(sig->flags & SIGNAL_GROUP_EXIT) ||
703 		(sig->group_exit_task != NULL);
704 }
705 
706 /*
707  * Some day this will be a full-fledged user tracking system..
708  */
709 struct user_struct {
710 	atomic_t __count;	/* reference count */
711 	atomic_t processes;	/* How many processes does this user have? */
712 	atomic_t files;		/* How many open files does this user have? */
713 	atomic_t sigpending;	/* How many pending signals does this user have? */
714 #ifdef CONFIG_INOTIFY_USER
715 	atomic_t inotify_watches; /* How many inotify watches does this user have? */
716 	atomic_t inotify_devs;	/* How many inotify devs does this user have opened? */
717 #endif
718 #ifdef CONFIG_FANOTIFY
719 	atomic_t fanotify_listeners;
720 #endif
721 #ifdef CONFIG_EPOLL
722 	atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
723 #endif
724 #ifdef CONFIG_POSIX_MQUEUE
725 	/* protected by mq_lock	*/
726 	unsigned long mq_bytes;	/* How many bytes can be allocated to mqueue? */
727 #endif
728 	unsigned long locked_shm; /* How many pages of mlocked shm ? */
729 
730 #ifdef CONFIG_KEYS
731 	struct key *uid_keyring;	/* UID specific keyring */
732 	struct key *session_keyring;	/* UID's default session keyring */
733 #endif
734 
735 	/* Hash table maintenance information */
736 	struct hlist_node uidhash_node;
737 	kuid_t uid;
738 
739 #ifdef CONFIG_PERF_EVENTS
740 	atomic_long_t locked_vm;
741 #endif
742 };
743 
744 extern int uids_sysfs_init(void);
745 
746 extern struct user_struct *find_user(kuid_t);
747 
748 extern struct user_struct root_user;
749 #define INIT_USER (&root_user)
750 
751 
752 struct backing_dev_info;
753 struct reclaim_state;
754 
755 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
756 struct sched_info {
757 	/* cumulative counters */
758 	unsigned long pcount;	      /* # of times run on this cpu */
759 	unsigned long long run_delay; /* time spent waiting on a runqueue */
760 
761 	/* timestamps */
762 	unsigned long long last_arrival,/* when we last ran on a cpu */
763 			   last_queued;	/* when we were last queued to run */
764 };
765 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
766 
767 #ifdef CONFIG_TASK_DELAY_ACCT
768 struct task_delay_info {
769 	spinlock_t	lock;
770 	unsigned int	flags;	/* Private per-task flags */
771 
772 	/* For each stat XXX, add following, aligned appropriately
773 	 *
774 	 * struct timespec XXX_start, XXX_end;
775 	 * u64 XXX_delay;
776 	 * u32 XXX_count;
777 	 *
778 	 * Atomicity of updates to XXX_delay, XXX_count protected by
779 	 * single lock above (split into XXX_lock if contention is an issue).
780 	 */
781 
782 	/*
783 	 * XXX_count is incremented on every XXX operation, the delay
784 	 * associated with the operation is added to XXX_delay.
785 	 * XXX_delay contains the accumulated delay time in nanoseconds.
786 	 */
787 	struct timespec blkio_start, blkio_end;	/* Shared by blkio, swapin */
788 	u64 blkio_delay;	/* wait for sync block io completion */
789 	u64 swapin_delay;	/* wait for swapin block io completion */
790 	u32 blkio_count;	/* total count of the number of sync block */
791 				/* io operations performed */
792 	u32 swapin_count;	/* total count of the number of swapin block */
793 				/* io operations performed */
794 
795 	struct timespec freepages_start, freepages_end;
796 	u64 freepages_delay;	/* wait for memory reclaim */
797 	u32 freepages_count;	/* total count of memory reclaim */
798 };
799 #endif	/* CONFIG_TASK_DELAY_ACCT */
800 
801 static inline int sched_info_on(void)
802 {
803 #ifdef CONFIG_SCHEDSTATS
804 	return 1;
805 #elif defined(CONFIG_TASK_DELAY_ACCT)
806 	extern int delayacct_on;
807 	return delayacct_on;
808 #else
809 	return 0;
810 #endif
811 }
812 
813 enum cpu_idle_type {
814 	CPU_IDLE,
815 	CPU_NOT_IDLE,
816 	CPU_NEWLY_IDLE,
817 	CPU_MAX_IDLE_TYPES
818 };
819 
820 /*
821  * Increase resolution of nice-level calculations for 64-bit architectures.
822  * The extra resolution improves shares distribution and load balancing of
823  * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
824  * hierarchies, especially on larger systems. This is not a user-visible change
825  * and does not change the user-interface for setting shares/weights.
826  *
827  * We increase resolution only if we have enough bits to allow this increased
828  * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
829  * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
830  * increased costs.
831  */
832 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load  */
833 # define SCHED_LOAD_RESOLUTION	10
834 # define scale_load(w)		((w) << SCHED_LOAD_RESOLUTION)
835 # define scale_load_down(w)	((w) >> SCHED_LOAD_RESOLUTION)
836 #else
837 # define SCHED_LOAD_RESOLUTION	0
838 # define scale_load(w)		(w)
839 # define scale_load_down(w)	(w)
840 #endif
841 
842 #define SCHED_LOAD_SHIFT	(10 + SCHED_LOAD_RESOLUTION)
843 #define SCHED_LOAD_SCALE	(1L << SCHED_LOAD_SHIFT)
844 
845 /*
846  * Increase resolution of cpu_power calculations
847  */
848 #define SCHED_POWER_SHIFT	10
849 #define SCHED_POWER_SCALE	(1L << SCHED_POWER_SHIFT)
850 
851 /*
852  * sched-domains (multiprocessor balancing) declarations:
853  */
854 #ifdef CONFIG_SMP
855 #define SD_LOAD_BALANCE		0x0001	/* Do load balancing on this domain. */
856 #define SD_BALANCE_NEWIDLE	0x0002	/* Balance when about to become idle */
857 #define SD_BALANCE_EXEC		0x0004	/* Balance on exec */
858 #define SD_BALANCE_FORK		0x0008	/* Balance on fork, clone */
859 #define SD_BALANCE_WAKE		0x0010  /* Balance on wakeup */
860 #define SD_WAKE_AFFINE		0x0020	/* Wake task to waking CPU */
861 #define SD_SHARE_CPUPOWER	0x0080	/* Domain members share cpu power */
862 #define SD_SHARE_PKG_RESOURCES	0x0200	/* Domain members share cpu pkg resources */
863 #define SD_SERIALIZE		0x0400	/* Only a single load balancing instance */
864 #define SD_ASYM_PACKING		0x0800  /* Place busy groups earlier in the domain */
865 #define SD_PREFER_SIBLING	0x1000	/* Prefer to place tasks in a sibling domain */
866 #define SD_OVERLAP		0x2000	/* sched_domains of this level overlap */
867 
868 extern int __weak arch_sd_sibiling_asym_packing(void);
869 
870 struct sched_group_power {
871 	atomic_t ref;
872 	/*
873 	 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
874 	 * single CPU.
875 	 */
876 	unsigned int power, power_orig;
877 	unsigned long next_update;
878 	/*
879 	 * Number of busy cpus in this group.
880 	 */
881 	atomic_t nr_busy_cpus;
882 
883 	unsigned long cpumask[0]; /* iteration mask */
884 };
885 
886 struct sched_group {
887 	struct sched_group *next;	/* Must be a circular list */
888 	atomic_t ref;
889 
890 	unsigned int group_weight;
891 	struct sched_group_power *sgp;
892 
893 	/*
894 	 * The CPUs this group covers.
895 	 *
896 	 * NOTE: this field is variable length. (Allocated dynamically
897 	 * by attaching extra space to the end of the structure,
898 	 * depending on how many CPUs the kernel has booted up with)
899 	 */
900 	unsigned long cpumask[0];
901 };
902 
903 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
904 {
905 	return to_cpumask(sg->cpumask);
906 }
907 
908 /*
909  * cpumask masking which cpus in the group are allowed to iterate up the domain
910  * tree.
911  */
912 static inline struct cpumask *sched_group_mask(struct sched_group *sg)
913 {
914 	return to_cpumask(sg->sgp->cpumask);
915 }
916 
917 /**
918  * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
919  * @group: The group whose first cpu is to be returned.
920  */
921 static inline unsigned int group_first_cpu(struct sched_group *group)
922 {
923 	return cpumask_first(sched_group_cpus(group));
924 }
925 
926 struct sched_domain_attr {
927 	int relax_domain_level;
928 };
929 
930 #define SD_ATTR_INIT	(struct sched_domain_attr) {	\
931 	.relax_domain_level = -1,			\
932 }
933 
934 extern int sched_domain_level_max;
935 
936 struct sched_domain {
937 	/* These fields must be setup */
938 	struct sched_domain *parent;	/* top domain must be null terminated */
939 	struct sched_domain *child;	/* bottom domain must be null terminated */
940 	struct sched_group *groups;	/* the balancing groups of the domain */
941 	unsigned long min_interval;	/* Minimum balance interval ms */
942 	unsigned long max_interval;	/* Maximum balance interval ms */
943 	unsigned int busy_factor;	/* less balancing by factor if busy */
944 	unsigned int imbalance_pct;	/* No balance until over watermark */
945 	unsigned int cache_nice_tries;	/* Leave cache hot tasks for # tries */
946 	unsigned int busy_idx;
947 	unsigned int idle_idx;
948 	unsigned int newidle_idx;
949 	unsigned int wake_idx;
950 	unsigned int forkexec_idx;
951 	unsigned int smt_gain;
952 	int flags;			/* See SD_* */
953 	int level;
954 
955 	/* Runtime fields. */
956 	unsigned long last_balance;	/* init to jiffies. units in jiffies */
957 	unsigned int balance_interval;	/* initialise to 1. units in ms. */
958 	unsigned int nr_balance_failed; /* initialise to 0 */
959 
960 	u64 last_update;
961 
962 #ifdef CONFIG_SCHEDSTATS
963 	/* load_balance() stats */
964 	unsigned int lb_count[CPU_MAX_IDLE_TYPES];
965 	unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
966 	unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
967 	unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
968 	unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
969 	unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
970 	unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
971 	unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
972 
973 	/* Active load balancing */
974 	unsigned int alb_count;
975 	unsigned int alb_failed;
976 	unsigned int alb_pushed;
977 
978 	/* SD_BALANCE_EXEC stats */
979 	unsigned int sbe_count;
980 	unsigned int sbe_balanced;
981 	unsigned int sbe_pushed;
982 
983 	/* SD_BALANCE_FORK stats */
984 	unsigned int sbf_count;
985 	unsigned int sbf_balanced;
986 	unsigned int sbf_pushed;
987 
988 	/* try_to_wake_up() stats */
989 	unsigned int ttwu_wake_remote;
990 	unsigned int ttwu_move_affine;
991 	unsigned int ttwu_move_balance;
992 #endif
993 #ifdef CONFIG_SCHED_DEBUG
994 	char *name;
995 #endif
996 	union {
997 		void *private;		/* used during construction */
998 		struct rcu_head rcu;	/* used during destruction */
999 	};
1000 
1001 	unsigned int span_weight;
1002 	/*
1003 	 * Span of all CPUs in this domain.
1004 	 *
1005 	 * NOTE: this field is variable length. (Allocated dynamically
1006 	 * by attaching extra space to the end of the structure,
1007 	 * depending on how many CPUs the kernel has booted up with)
1008 	 */
1009 	unsigned long span[0];
1010 };
1011 
1012 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1013 {
1014 	return to_cpumask(sd->span);
1015 }
1016 
1017 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1018 				    struct sched_domain_attr *dattr_new);
1019 
1020 /* Allocate an array of sched domains, for partition_sched_domains(). */
1021 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1022 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1023 
1024 /* Test a flag in parent sched domain */
1025 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1026 {
1027 	if (sd->parent && (sd->parent->flags & flag))
1028 		return 1;
1029 
1030 	return 0;
1031 }
1032 
1033 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1034 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1035 
1036 bool cpus_share_cache(int this_cpu, int that_cpu);
1037 
1038 #else /* CONFIG_SMP */
1039 
1040 struct sched_domain_attr;
1041 
1042 static inline void
1043 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1044 			struct sched_domain_attr *dattr_new)
1045 {
1046 }
1047 
1048 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1049 {
1050 	return true;
1051 }
1052 
1053 #endif	/* !CONFIG_SMP */
1054 
1055 
1056 struct io_context;			/* See blkdev.h */
1057 
1058 
1059 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1060 extern void prefetch_stack(struct task_struct *t);
1061 #else
1062 static inline void prefetch_stack(struct task_struct *t) { }
1063 #endif
1064 
1065 struct audit_context;		/* See audit.c */
1066 struct mempolicy;
1067 struct pipe_inode_info;
1068 struct uts_namespace;
1069 
1070 struct rq;
1071 struct sched_domain;
1072 
1073 /*
1074  * wake flags
1075  */
1076 #define WF_SYNC		0x01		/* waker goes to sleep after wakup */
1077 #define WF_FORK		0x02		/* child wakeup after fork */
1078 #define WF_MIGRATED	0x04		/* internal use, task got migrated */
1079 
1080 #define ENQUEUE_WAKEUP		1
1081 #define ENQUEUE_HEAD		2
1082 #ifdef CONFIG_SMP
1083 #define ENQUEUE_WAKING		4	/* sched_class::task_waking was called */
1084 #else
1085 #define ENQUEUE_WAKING		0
1086 #endif
1087 
1088 #define DEQUEUE_SLEEP		1
1089 
1090 struct sched_class {
1091 	const struct sched_class *next;
1092 
1093 	void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1094 	void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1095 	void (*yield_task) (struct rq *rq);
1096 	bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1097 
1098 	void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1099 
1100 	struct task_struct * (*pick_next_task) (struct rq *rq);
1101 	void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1102 
1103 #ifdef CONFIG_SMP
1104 	int  (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1105 
1106 	void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1107 	void (*post_schedule) (struct rq *this_rq);
1108 	void (*task_waking) (struct task_struct *task);
1109 	void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1110 
1111 	void (*set_cpus_allowed)(struct task_struct *p,
1112 				 const struct cpumask *newmask);
1113 
1114 	void (*rq_online)(struct rq *rq);
1115 	void (*rq_offline)(struct rq *rq);
1116 #endif
1117 
1118 	void (*set_curr_task) (struct rq *rq);
1119 	void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1120 	void (*task_fork) (struct task_struct *p);
1121 
1122 	void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1123 	void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1124 	void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1125 			     int oldprio);
1126 
1127 	unsigned int (*get_rr_interval) (struct rq *rq,
1128 					 struct task_struct *task);
1129 
1130 #ifdef CONFIG_FAIR_GROUP_SCHED
1131 	void (*task_move_group) (struct task_struct *p, int on_rq);
1132 #endif
1133 };
1134 
1135 struct load_weight {
1136 	unsigned long weight, inv_weight;
1137 };
1138 
1139 #ifdef CONFIG_SCHEDSTATS
1140 struct sched_statistics {
1141 	u64			wait_start;
1142 	u64			wait_max;
1143 	u64			wait_count;
1144 	u64			wait_sum;
1145 	u64			iowait_count;
1146 	u64			iowait_sum;
1147 
1148 	u64			sleep_start;
1149 	u64			sleep_max;
1150 	s64			sum_sleep_runtime;
1151 
1152 	u64			block_start;
1153 	u64			block_max;
1154 	u64			exec_max;
1155 	u64			slice_max;
1156 
1157 	u64			nr_migrations_cold;
1158 	u64			nr_failed_migrations_affine;
1159 	u64			nr_failed_migrations_running;
1160 	u64			nr_failed_migrations_hot;
1161 	u64			nr_forced_migrations;
1162 
1163 	u64			nr_wakeups;
1164 	u64			nr_wakeups_sync;
1165 	u64			nr_wakeups_migrate;
1166 	u64			nr_wakeups_local;
1167 	u64			nr_wakeups_remote;
1168 	u64			nr_wakeups_affine;
1169 	u64			nr_wakeups_affine_attempts;
1170 	u64			nr_wakeups_passive;
1171 	u64			nr_wakeups_idle;
1172 };
1173 #endif
1174 
1175 struct sched_entity {
1176 	struct load_weight	load;		/* for load-balancing */
1177 	struct rb_node		run_node;
1178 	struct list_head	group_node;
1179 	unsigned int		on_rq;
1180 
1181 	u64			exec_start;
1182 	u64			sum_exec_runtime;
1183 	u64			vruntime;
1184 	u64			prev_sum_exec_runtime;
1185 
1186 	u64			nr_migrations;
1187 
1188 #ifdef CONFIG_SCHEDSTATS
1189 	struct sched_statistics statistics;
1190 #endif
1191 
1192 #ifdef CONFIG_FAIR_GROUP_SCHED
1193 	struct sched_entity	*parent;
1194 	/* rq on which this entity is (to be) queued: */
1195 	struct cfs_rq		*cfs_rq;
1196 	/* rq "owned" by this entity/group: */
1197 	struct cfs_rq		*my_q;
1198 #endif
1199 };
1200 
1201 struct sched_rt_entity {
1202 	struct list_head run_list;
1203 	unsigned long timeout;
1204 	unsigned int time_slice;
1205 
1206 	struct sched_rt_entity *back;
1207 #ifdef CONFIG_RT_GROUP_SCHED
1208 	struct sched_rt_entity	*parent;
1209 	/* rq on which this entity is (to be) queued: */
1210 	struct rt_rq		*rt_rq;
1211 	/* rq "owned" by this entity/group: */
1212 	struct rt_rq		*my_q;
1213 #endif
1214 };
1215 
1216 /*
1217  * default timeslice is 100 msecs (used only for SCHED_RR tasks).
1218  * Timeslices get refilled after they expire.
1219  */
1220 #define RR_TIMESLICE		(100 * HZ / 1000)
1221 
1222 struct rcu_node;
1223 
1224 enum perf_event_task_context {
1225 	perf_invalid_context = -1,
1226 	perf_hw_context = 0,
1227 	perf_sw_context,
1228 	perf_nr_task_contexts,
1229 };
1230 
1231 struct task_struct {
1232 	volatile long state;	/* -1 unrunnable, 0 runnable, >0 stopped */
1233 	void *stack;
1234 	atomic_t usage;
1235 	unsigned int flags;	/* per process flags, defined below */
1236 	unsigned int ptrace;
1237 
1238 #ifdef CONFIG_SMP
1239 	struct llist_node wake_entry;
1240 	int on_cpu;
1241 #endif
1242 	int on_rq;
1243 
1244 	int prio, static_prio, normal_prio;
1245 	unsigned int rt_priority;
1246 	const struct sched_class *sched_class;
1247 	struct sched_entity se;
1248 	struct sched_rt_entity rt;
1249 #ifdef CONFIG_CGROUP_SCHED
1250 	struct task_group *sched_task_group;
1251 #endif
1252 
1253 #ifdef CONFIG_PREEMPT_NOTIFIERS
1254 	/* list of struct preempt_notifier: */
1255 	struct hlist_head preempt_notifiers;
1256 #endif
1257 
1258 	/*
1259 	 * fpu_counter contains the number of consecutive context switches
1260 	 * that the FPU is used. If this is over a threshold, the lazy fpu
1261 	 * saving becomes unlazy to save the trap. This is an unsigned char
1262 	 * so that after 256 times the counter wraps and the behavior turns
1263 	 * lazy again; this to deal with bursty apps that only use FPU for
1264 	 * a short time
1265 	 */
1266 	unsigned char fpu_counter;
1267 #ifdef CONFIG_BLK_DEV_IO_TRACE
1268 	unsigned int btrace_seq;
1269 #endif
1270 
1271 	unsigned int policy;
1272 	int nr_cpus_allowed;
1273 	cpumask_t cpus_allowed;
1274 
1275 #ifdef CONFIG_PREEMPT_RCU
1276 	int rcu_read_lock_nesting;
1277 	char rcu_read_unlock_special;
1278 	struct list_head rcu_node_entry;
1279 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1280 #ifdef CONFIG_TREE_PREEMPT_RCU
1281 	struct rcu_node *rcu_blocked_node;
1282 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1283 #ifdef CONFIG_RCU_BOOST
1284 	struct rt_mutex *rcu_boost_mutex;
1285 #endif /* #ifdef CONFIG_RCU_BOOST */
1286 
1287 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1288 	struct sched_info sched_info;
1289 #endif
1290 
1291 	struct list_head tasks;
1292 #ifdef CONFIG_SMP
1293 	struct plist_node pushable_tasks;
1294 #endif
1295 
1296 	struct mm_struct *mm, *active_mm;
1297 #ifdef CONFIG_COMPAT_BRK
1298 	unsigned brk_randomized:1;
1299 #endif
1300 #if defined(SPLIT_RSS_COUNTING)
1301 	struct task_rss_stat	rss_stat;
1302 #endif
1303 /* task state */
1304 	int exit_state;
1305 	int exit_code, exit_signal;
1306 	int pdeath_signal;  /*  The signal sent when the parent dies  */
1307 	unsigned int jobctl;	/* JOBCTL_*, siglock protected */
1308 	/* ??? */
1309 	unsigned int personality;
1310 	unsigned did_exec:1;
1311 	unsigned in_execve:1;	/* Tell the LSMs that the process is doing an
1312 				 * execve */
1313 	unsigned in_iowait:1;
1314 
1315 	/* task may not gain privileges */
1316 	unsigned no_new_privs:1;
1317 
1318 	/* Revert to default priority/policy when forking */
1319 	unsigned sched_reset_on_fork:1;
1320 	unsigned sched_contributes_to_load:1;
1321 
1322 	pid_t pid;
1323 	pid_t tgid;
1324 
1325 #ifdef CONFIG_CC_STACKPROTECTOR
1326 	/* Canary value for the -fstack-protector gcc feature */
1327 	unsigned long stack_canary;
1328 #endif
1329 	/*
1330 	 * pointers to (original) parent process, youngest child, younger sibling,
1331 	 * older sibling, respectively.  (p->father can be replaced with
1332 	 * p->real_parent->pid)
1333 	 */
1334 	struct task_struct __rcu *real_parent; /* real parent process */
1335 	struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1336 	/*
1337 	 * children/sibling forms the list of my natural children
1338 	 */
1339 	struct list_head children;	/* list of my children */
1340 	struct list_head sibling;	/* linkage in my parent's children list */
1341 	struct task_struct *group_leader;	/* threadgroup leader */
1342 
1343 	/*
1344 	 * ptraced is the list of tasks this task is using ptrace on.
1345 	 * This includes both natural children and PTRACE_ATTACH targets.
1346 	 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1347 	 */
1348 	struct list_head ptraced;
1349 	struct list_head ptrace_entry;
1350 
1351 	/* PID/PID hash table linkage. */
1352 	struct pid_link pids[PIDTYPE_MAX];
1353 	struct list_head thread_group;
1354 
1355 	struct completion *vfork_done;		/* for vfork() */
1356 	int __user *set_child_tid;		/* CLONE_CHILD_SETTID */
1357 	int __user *clear_child_tid;		/* CLONE_CHILD_CLEARTID */
1358 
1359 	cputime_t utime, stime, utimescaled, stimescaled;
1360 	cputime_t gtime;
1361 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1362 	cputime_t prev_utime, prev_stime;
1363 #endif
1364 	unsigned long nvcsw, nivcsw; /* context switch counts */
1365 	struct timespec start_time; 		/* monotonic time */
1366 	struct timespec real_start_time;	/* boot based time */
1367 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1368 	unsigned long min_flt, maj_flt;
1369 
1370 	struct task_cputime cputime_expires;
1371 	struct list_head cpu_timers[3];
1372 
1373 /* process credentials */
1374 	const struct cred __rcu *real_cred; /* objective and real subjective task
1375 					 * credentials (COW) */
1376 	const struct cred __rcu *cred;	/* effective (overridable) subjective task
1377 					 * credentials (COW) */
1378 	char comm[TASK_COMM_LEN]; /* executable name excluding path
1379 				     - access with [gs]et_task_comm (which lock
1380 				       it with task_lock())
1381 				     - initialized normally by setup_new_exec */
1382 /* file system info */
1383 	int link_count, total_link_count;
1384 #ifdef CONFIG_SYSVIPC
1385 /* ipc stuff */
1386 	struct sysv_sem sysvsem;
1387 #endif
1388 #ifdef CONFIG_DETECT_HUNG_TASK
1389 /* hung task detection */
1390 	unsigned long last_switch_count;
1391 #endif
1392 /* CPU-specific state of this task */
1393 	struct thread_struct thread;
1394 /* filesystem information */
1395 	struct fs_struct *fs;
1396 /* open file information */
1397 	struct files_struct *files;
1398 /* namespaces */
1399 	struct nsproxy *nsproxy;
1400 /* signal handlers */
1401 	struct signal_struct *signal;
1402 	struct sighand_struct *sighand;
1403 
1404 	sigset_t blocked, real_blocked;
1405 	sigset_t saved_sigmask;	/* restored if set_restore_sigmask() was used */
1406 	struct sigpending pending;
1407 
1408 	unsigned long sas_ss_sp;
1409 	size_t sas_ss_size;
1410 	int (*notifier)(void *priv);
1411 	void *notifier_data;
1412 	sigset_t *notifier_mask;
1413 	struct callback_head *task_works;
1414 
1415 	struct audit_context *audit_context;
1416 #ifdef CONFIG_AUDITSYSCALL
1417 	uid_t loginuid;
1418 	unsigned int sessionid;
1419 #endif
1420 	struct seccomp seccomp;
1421 
1422 /* Thread group tracking */
1423    	u32 parent_exec_id;
1424    	u32 self_exec_id;
1425 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1426  * mempolicy */
1427 	spinlock_t alloc_lock;
1428 
1429 	/* Protection of the PI data structures: */
1430 	raw_spinlock_t pi_lock;
1431 
1432 #ifdef CONFIG_RT_MUTEXES
1433 	/* PI waiters blocked on a rt_mutex held by this task */
1434 	struct plist_head pi_waiters;
1435 	/* Deadlock detection and priority inheritance handling */
1436 	struct rt_mutex_waiter *pi_blocked_on;
1437 #endif
1438 
1439 #ifdef CONFIG_DEBUG_MUTEXES
1440 	/* mutex deadlock detection */
1441 	struct mutex_waiter *blocked_on;
1442 #endif
1443 #ifdef CONFIG_TRACE_IRQFLAGS
1444 	unsigned int irq_events;
1445 	unsigned long hardirq_enable_ip;
1446 	unsigned long hardirq_disable_ip;
1447 	unsigned int hardirq_enable_event;
1448 	unsigned int hardirq_disable_event;
1449 	int hardirqs_enabled;
1450 	int hardirq_context;
1451 	unsigned long softirq_disable_ip;
1452 	unsigned long softirq_enable_ip;
1453 	unsigned int softirq_disable_event;
1454 	unsigned int softirq_enable_event;
1455 	int softirqs_enabled;
1456 	int softirq_context;
1457 #endif
1458 #ifdef CONFIG_LOCKDEP
1459 # define MAX_LOCK_DEPTH 48UL
1460 	u64 curr_chain_key;
1461 	int lockdep_depth;
1462 	unsigned int lockdep_recursion;
1463 	struct held_lock held_locks[MAX_LOCK_DEPTH];
1464 	gfp_t lockdep_reclaim_gfp;
1465 #endif
1466 
1467 /* journalling filesystem info */
1468 	void *journal_info;
1469 
1470 /* stacked block device info */
1471 	struct bio_list *bio_list;
1472 
1473 #ifdef CONFIG_BLOCK
1474 /* stack plugging */
1475 	struct blk_plug *plug;
1476 #endif
1477 
1478 /* VM state */
1479 	struct reclaim_state *reclaim_state;
1480 
1481 	struct backing_dev_info *backing_dev_info;
1482 
1483 	struct io_context *io_context;
1484 
1485 	unsigned long ptrace_message;
1486 	siginfo_t *last_siginfo; /* For ptrace use.  */
1487 	struct task_io_accounting ioac;
1488 #if defined(CONFIG_TASK_XACCT)
1489 	u64 acct_rss_mem1;	/* accumulated rss usage */
1490 	u64 acct_vm_mem1;	/* accumulated virtual memory usage */
1491 	cputime_t acct_timexpd;	/* stime + utime since last update */
1492 #endif
1493 #ifdef CONFIG_CPUSETS
1494 	nodemask_t mems_allowed;	/* Protected by alloc_lock */
1495 	seqcount_t mems_allowed_seq;	/* Seqence no to catch updates */
1496 	int cpuset_mem_spread_rotor;
1497 	int cpuset_slab_spread_rotor;
1498 #endif
1499 #ifdef CONFIG_CGROUPS
1500 	/* Control Group info protected by css_set_lock */
1501 	struct css_set __rcu *cgroups;
1502 	/* cg_list protected by css_set_lock and tsk->alloc_lock */
1503 	struct list_head cg_list;
1504 #endif
1505 #ifdef CONFIG_FUTEX
1506 	struct robust_list_head __user *robust_list;
1507 #ifdef CONFIG_COMPAT
1508 	struct compat_robust_list_head __user *compat_robust_list;
1509 #endif
1510 	struct list_head pi_state_list;
1511 	struct futex_pi_state *pi_state_cache;
1512 #endif
1513 #ifdef CONFIG_PERF_EVENTS
1514 	struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1515 	struct mutex perf_event_mutex;
1516 	struct list_head perf_event_list;
1517 #endif
1518 #ifdef CONFIG_NUMA
1519 	struct mempolicy *mempolicy;	/* Protected by alloc_lock */
1520 	short il_next;
1521 	short pref_node_fork;
1522 #endif
1523 	struct rcu_head rcu;
1524 
1525 	/*
1526 	 * cache last used pipe for splice
1527 	 */
1528 	struct pipe_inode_info *splice_pipe;
1529 #ifdef	CONFIG_TASK_DELAY_ACCT
1530 	struct task_delay_info *delays;
1531 #endif
1532 #ifdef CONFIG_FAULT_INJECTION
1533 	int make_it_fail;
1534 #endif
1535 	/*
1536 	 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1537 	 * balance_dirty_pages() for some dirty throttling pause
1538 	 */
1539 	int nr_dirtied;
1540 	int nr_dirtied_pause;
1541 	unsigned long dirty_paused_when; /* start of a write-and-pause period */
1542 
1543 #ifdef CONFIG_LATENCYTOP
1544 	int latency_record_count;
1545 	struct latency_record latency_record[LT_SAVECOUNT];
1546 #endif
1547 	/*
1548 	 * time slack values; these are used to round up poll() and
1549 	 * select() etc timeout values. These are in nanoseconds.
1550 	 */
1551 	unsigned long timer_slack_ns;
1552 	unsigned long default_timer_slack_ns;
1553 
1554 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1555 	/* Index of current stored address in ret_stack */
1556 	int curr_ret_stack;
1557 	/* Stack of return addresses for return function tracing */
1558 	struct ftrace_ret_stack	*ret_stack;
1559 	/* time stamp for last schedule */
1560 	unsigned long long ftrace_timestamp;
1561 	/*
1562 	 * Number of functions that haven't been traced
1563 	 * because of depth overrun.
1564 	 */
1565 	atomic_t trace_overrun;
1566 	/* Pause for the tracing */
1567 	atomic_t tracing_graph_pause;
1568 #endif
1569 #ifdef CONFIG_TRACING
1570 	/* state flags for use by tracers */
1571 	unsigned long trace;
1572 	/* bitmask and counter of trace recursion */
1573 	unsigned long trace_recursion;
1574 #endif /* CONFIG_TRACING */
1575 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1576 	struct memcg_batch_info {
1577 		int do_batch;	/* incremented when batch uncharge started */
1578 		struct mem_cgroup *memcg; /* target memcg of uncharge */
1579 		unsigned long nr_pages;	/* uncharged usage */
1580 		unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1581 	} memcg_batch;
1582 #endif
1583 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1584 	atomic_t ptrace_bp_refcnt;
1585 #endif
1586 #ifdef CONFIG_UPROBES
1587 	struct uprobe_task *utask;
1588 #endif
1589 };
1590 
1591 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1592 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1593 
1594 /*
1595  * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1596  * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1597  * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1598  * values are inverted: lower p->prio value means higher priority.
1599  *
1600  * The MAX_USER_RT_PRIO value allows the actual maximum
1601  * RT priority to be separate from the value exported to
1602  * user-space.  This allows kernel threads to set their
1603  * priority to a value higher than any user task. Note:
1604  * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1605  */
1606 
1607 #define MAX_USER_RT_PRIO	100
1608 #define MAX_RT_PRIO		MAX_USER_RT_PRIO
1609 
1610 #define MAX_PRIO		(MAX_RT_PRIO + 40)
1611 #define DEFAULT_PRIO		(MAX_RT_PRIO + 20)
1612 
1613 static inline int rt_prio(int prio)
1614 {
1615 	if (unlikely(prio < MAX_RT_PRIO))
1616 		return 1;
1617 	return 0;
1618 }
1619 
1620 static inline int rt_task(struct task_struct *p)
1621 {
1622 	return rt_prio(p->prio);
1623 }
1624 
1625 static inline struct pid *task_pid(struct task_struct *task)
1626 {
1627 	return task->pids[PIDTYPE_PID].pid;
1628 }
1629 
1630 static inline struct pid *task_tgid(struct task_struct *task)
1631 {
1632 	return task->group_leader->pids[PIDTYPE_PID].pid;
1633 }
1634 
1635 /*
1636  * Without tasklist or rcu lock it is not safe to dereference
1637  * the result of task_pgrp/task_session even if task == current,
1638  * we can race with another thread doing sys_setsid/sys_setpgid.
1639  */
1640 static inline struct pid *task_pgrp(struct task_struct *task)
1641 {
1642 	return task->group_leader->pids[PIDTYPE_PGID].pid;
1643 }
1644 
1645 static inline struct pid *task_session(struct task_struct *task)
1646 {
1647 	return task->group_leader->pids[PIDTYPE_SID].pid;
1648 }
1649 
1650 struct pid_namespace;
1651 
1652 /*
1653  * the helpers to get the task's different pids as they are seen
1654  * from various namespaces
1655  *
1656  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
1657  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
1658  *                     current.
1659  * task_xid_nr_ns()  : id seen from the ns specified;
1660  *
1661  * set_task_vxid()   : assigns a virtual id to a task;
1662  *
1663  * see also pid_nr() etc in include/linux/pid.h
1664  */
1665 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1666 			struct pid_namespace *ns);
1667 
1668 static inline pid_t task_pid_nr(struct task_struct *tsk)
1669 {
1670 	return tsk->pid;
1671 }
1672 
1673 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1674 					struct pid_namespace *ns)
1675 {
1676 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1677 }
1678 
1679 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1680 {
1681 	return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1682 }
1683 
1684 
1685 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1686 {
1687 	return tsk->tgid;
1688 }
1689 
1690 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1691 
1692 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1693 {
1694 	return pid_vnr(task_tgid(tsk));
1695 }
1696 
1697 
1698 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1699 					struct pid_namespace *ns)
1700 {
1701 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1702 }
1703 
1704 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1705 {
1706 	return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1707 }
1708 
1709 
1710 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1711 					struct pid_namespace *ns)
1712 {
1713 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1714 }
1715 
1716 static inline pid_t task_session_vnr(struct task_struct *tsk)
1717 {
1718 	return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1719 }
1720 
1721 /* obsolete, do not use */
1722 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1723 {
1724 	return task_pgrp_nr_ns(tsk, &init_pid_ns);
1725 }
1726 
1727 /**
1728  * pid_alive - check that a task structure is not stale
1729  * @p: Task structure to be checked.
1730  *
1731  * Test if a process is not yet dead (at most zombie state)
1732  * If pid_alive fails, then pointers within the task structure
1733  * can be stale and must not be dereferenced.
1734  */
1735 static inline int pid_alive(struct task_struct *p)
1736 {
1737 	return p->pids[PIDTYPE_PID].pid != NULL;
1738 }
1739 
1740 /**
1741  * is_global_init - check if a task structure is init
1742  * @tsk: Task structure to be checked.
1743  *
1744  * Check if a task structure is the first user space task the kernel created.
1745  */
1746 static inline int is_global_init(struct task_struct *tsk)
1747 {
1748 	return tsk->pid == 1;
1749 }
1750 
1751 /*
1752  * is_container_init:
1753  * check whether in the task is init in its own pid namespace.
1754  */
1755 extern int is_container_init(struct task_struct *tsk);
1756 
1757 extern struct pid *cad_pid;
1758 
1759 extern void free_task(struct task_struct *tsk);
1760 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1761 
1762 extern void __put_task_struct(struct task_struct *t);
1763 
1764 static inline void put_task_struct(struct task_struct *t)
1765 {
1766 	if (atomic_dec_and_test(&t->usage))
1767 		__put_task_struct(t);
1768 }
1769 
1770 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1771 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1772 
1773 /*
1774  * Per process flags
1775  */
1776 #define PF_EXITING	0x00000004	/* getting shut down */
1777 #define PF_EXITPIDONE	0x00000008	/* pi exit done on shut down */
1778 #define PF_VCPU		0x00000010	/* I'm a virtual CPU */
1779 #define PF_WQ_WORKER	0x00000020	/* I'm a workqueue worker */
1780 #define PF_FORKNOEXEC	0x00000040	/* forked but didn't exec */
1781 #define PF_MCE_PROCESS  0x00000080      /* process policy on mce errors */
1782 #define PF_SUPERPRIV	0x00000100	/* used super-user privileges */
1783 #define PF_DUMPCORE	0x00000200	/* dumped core */
1784 #define PF_SIGNALED	0x00000400	/* killed by a signal */
1785 #define PF_MEMALLOC	0x00000800	/* Allocating memory */
1786 #define PF_NPROC_EXCEEDED 0x00001000	/* set_user noticed that RLIMIT_NPROC was exceeded */
1787 #define PF_USED_MATH	0x00002000	/* if unset the fpu must be initialized before use */
1788 #define PF_NOFREEZE	0x00008000	/* this thread should not be frozen */
1789 #define PF_FROZEN	0x00010000	/* frozen for system suspend */
1790 #define PF_FSTRANS	0x00020000	/* inside a filesystem transaction */
1791 #define PF_KSWAPD	0x00040000	/* I am kswapd */
1792 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
1793 #define PF_KTHREAD	0x00200000	/* I am a kernel thread */
1794 #define PF_RANDOMIZE	0x00400000	/* randomize virtual address space */
1795 #define PF_SWAPWRITE	0x00800000	/* Allowed to write to swap */
1796 #define PF_SPREAD_PAGE	0x01000000	/* Spread page cache over cpuset */
1797 #define PF_SPREAD_SLAB	0x02000000	/* Spread some slab caches over cpuset */
1798 #define PF_THREAD_BOUND	0x04000000	/* Thread bound to specific cpu */
1799 #define PF_MCE_EARLY    0x08000000      /* Early kill for mce process policy */
1800 #define PF_MEMPOLICY	0x10000000	/* Non-default NUMA mempolicy */
1801 #define PF_MUTEX_TESTER	0x20000000	/* Thread belongs to the rt mutex tester */
1802 #define PF_FREEZER_SKIP	0x40000000	/* Freezer should not count it as freezable */
1803 
1804 /*
1805  * Only the _current_ task can read/write to tsk->flags, but other
1806  * tasks can access tsk->flags in readonly mode for example
1807  * with tsk_used_math (like during threaded core dumping).
1808  * There is however an exception to this rule during ptrace
1809  * or during fork: the ptracer task is allowed to write to the
1810  * child->flags of its traced child (same goes for fork, the parent
1811  * can write to the child->flags), because we're guaranteed the
1812  * child is not running and in turn not changing child->flags
1813  * at the same time the parent does it.
1814  */
1815 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1816 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1817 #define clear_used_math() clear_stopped_child_used_math(current)
1818 #define set_used_math() set_stopped_child_used_math(current)
1819 #define conditional_stopped_child_used_math(condition, child) \
1820 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1821 #define conditional_used_math(condition) \
1822 	conditional_stopped_child_used_math(condition, current)
1823 #define copy_to_stopped_child_used_math(child) \
1824 	do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1825 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1826 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1827 #define used_math() tsk_used_math(current)
1828 
1829 /*
1830  * task->jobctl flags
1831  */
1832 #define JOBCTL_STOP_SIGMASK	0xffff	/* signr of the last group stop */
1833 
1834 #define JOBCTL_STOP_DEQUEUED_BIT 16	/* stop signal dequeued */
1835 #define JOBCTL_STOP_PENDING_BIT	17	/* task should stop for group stop */
1836 #define JOBCTL_STOP_CONSUME_BIT	18	/* consume group stop count */
1837 #define JOBCTL_TRAP_STOP_BIT	19	/* trap for STOP */
1838 #define JOBCTL_TRAP_NOTIFY_BIT	20	/* trap for NOTIFY */
1839 #define JOBCTL_TRAPPING_BIT	21	/* switching to TRACED */
1840 #define JOBCTL_LISTENING_BIT	22	/* ptracer is listening for events */
1841 
1842 #define JOBCTL_STOP_DEQUEUED	(1 << JOBCTL_STOP_DEQUEUED_BIT)
1843 #define JOBCTL_STOP_PENDING	(1 << JOBCTL_STOP_PENDING_BIT)
1844 #define JOBCTL_STOP_CONSUME	(1 << JOBCTL_STOP_CONSUME_BIT)
1845 #define JOBCTL_TRAP_STOP	(1 << JOBCTL_TRAP_STOP_BIT)
1846 #define JOBCTL_TRAP_NOTIFY	(1 << JOBCTL_TRAP_NOTIFY_BIT)
1847 #define JOBCTL_TRAPPING		(1 << JOBCTL_TRAPPING_BIT)
1848 #define JOBCTL_LISTENING	(1 << JOBCTL_LISTENING_BIT)
1849 
1850 #define JOBCTL_TRAP_MASK	(JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1851 #define JOBCTL_PENDING_MASK	(JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1852 
1853 extern bool task_set_jobctl_pending(struct task_struct *task,
1854 				    unsigned int mask);
1855 extern void task_clear_jobctl_trapping(struct task_struct *task);
1856 extern void task_clear_jobctl_pending(struct task_struct *task,
1857 				      unsigned int mask);
1858 
1859 #ifdef CONFIG_PREEMPT_RCU
1860 
1861 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1862 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1863 
1864 static inline void rcu_copy_process(struct task_struct *p)
1865 {
1866 	p->rcu_read_lock_nesting = 0;
1867 	p->rcu_read_unlock_special = 0;
1868 #ifdef CONFIG_TREE_PREEMPT_RCU
1869 	p->rcu_blocked_node = NULL;
1870 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1871 #ifdef CONFIG_RCU_BOOST
1872 	p->rcu_boost_mutex = NULL;
1873 #endif /* #ifdef CONFIG_RCU_BOOST */
1874 	INIT_LIST_HEAD(&p->rcu_node_entry);
1875 }
1876 
1877 #else
1878 
1879 static inline void rcu_copy_process(struct task_struct *p)
1880 {
1881 }
1882 
1883 #endif
1884 
1885 static inline void rcu_switch(struct task_struct *prev,
1886 			      struct task_struct *next)
1887 {
1888 #ifdef CONFIG_RCU_USER_QS
1889 	rcu_user_hooks_switch(prev, next);
1890 #endif
1891 }
1892 
1893 static inline void tsk_restore_flags(struct task_struct *task,
1894 				unsigned long orig_flags, unsigned long flags)
1895 {
1896 	task->flags &= ~flags;
1897 	task->flags |= orig_flags & flags;
1898 }
1899 
1900 #ifdef CONFIG_SMP
1901 extern void do_set_cpus_allowed(struct task_struct *p,
1902 			       const struct cpumask *new_mask);
1903 
1904 extern int set_cpus_allowed_ptr(struct task_struct *p,
1905 				const struct cpumask *new_mask);
1906 #else
1907 static inline void do_set_cpus_allowed(struct task_struct *p,
1908 				      const struct cpumask *new_mask)
1909 {
1910 }
1911 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1912 				       const struct cpumask *new_mask)
1913 {
1914 	if (!cpumask_test_cpu(0, new_mask))
1915 		return -EINVAL;
1916 	return 0;
1917 }
1918 #endif
1919 
1920 #ifdef CONFIG_NO_HZ
1921 void calc_load_enter_idle(void);
1922 void calc_load_exit_idle(void);
1923 #else
1924 static inline void calc_load_enter_idle(void) { }
1925 static inline void calc_load_exit_idle(void) { }
1926 #endif /* CONFIG_NO_HZ */
1927 
1928 #ifndef CONFIG_CPUMASK_OFFSTACK
1929 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1930 {
1931 	return set_cpus_allowed_ptr(p, &new_mask);
1932 }
1933 #endif
1934 
1935 /*
1936  * Do not use outside of architecture code which knows its limitations.
1937  *
1938  * sched_clock() has no promise of monotonicity or bounded drift between
1939  * CPUs, use (which you should not) requires disabling IRQs.
1940  *
1941  * Please use one of the three interfaces below.
1942  */
1943 extern unsigned long long notrace sched_clock(void);
1944 /*
1945  * See the comment in kernel/sched/clock.c
1946  */
1947 extern u64 cpu_clock(int cpu);
1948 extern u64 local_clock(void);
1949 extern u64 sched_clock_cpu(int cpu);
1950 
1951 
1952 extern void sched_clock_init(void);
1953 
1954 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1955 static inline void sched_clock_tick(void)
1956 {
1957 }
1958 
1959 static inline void sched_clock_idle_sleep_event(void)
1960 {
1961 }
1962 
1963 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1964 {
1965 }
1966 #else
1967 /*
1968  * Architectures can set this to 1 if they have specified
1969  * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1970  * but then during bootup it turns out that sched_clock()
1971  * is reliable after all:
1972  */
1973 extern int sched_clock_stable;
1974 
1975 extern void sched_clock_tick(void);
1976 extern void sched_clock_idle_sleep_event(void);
1977 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1978 #endif
1979 
1980 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1981 /*
1982  * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1983  * The reason for this explicit opt-in is not to have perf penalty with
1984  * slow sched_clocks.
1985  */
1986 extern void enable_sched_clock_irqtime(void);
1987 extern void disable_sched_clock_irqtime(void);
1988 #else
1989 static inline void enable_sched_clock_irqtime(void) {}
1990 static inline void disable_sched_clock_irqtime(void) {}
1991 #endif
1992 
1993 extern unsigned long long
1994 task_sched_runtime(struct task_struct *task);
1995 
1996 /* sched_exec is called by processes performing an exec */
1997 #ifdef CONFIG_SMP
1998 extern void sched_exec(void);
1999 #else
2000 #define sched_exec()   {}
2001 #endif
2002 
2003 extern void sched_clock_idle_sleep_event(void);
2004 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2005 
2006 #ifdef CONFIG_HOTPLUG_CPU
2007 extern void idle_task_exit(void);
2008 #else
2009 static inline void idle_task_exit(void) {}
2010 #endif
2011 
2012 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
2013 extern void wake_up_idle_cpu(int cpu);
2014 #else
2015 static inline void wake_up_idle_cpu(int cpu) { }
2016 #endif
2017 
2018 extern unsigned int sysctl_sched_latency;
2019 extern unsigned int sysctl_sched_min_granularity;
2020 extern unsigned int sysctl_sched_wakeup_granularity;
2021 extern unsigned int sysctl_sched_child_runs_first;
2022 
2023 enum sched_tunable_scaling {
2024 	SCHED_TUNABLESCALING_NONE,
2025 	SCHED_TUNABLESCALING_LOG,
2026 	SCHED_TUNABLESCALING_LINEAR,
2027 	SCHED_TUNABLESCALING_END,
2028 };
2029 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
2030 
2031 #ifdef CONFIG_SCHED_DEBUG
2032 extern unsigned int sysctl_sched_migration_cost;
2033 extern unsigned int sysctl_sched_nr_migrate;
2034 extern unsigned int sysctl_sched_time_avg;
2035 extern unsigned int sysctl_timer_migration;
2036 extern unsigned int sysctl_sched_shares_window;
2037 
2038 int sched_proc_update_handler(struct ctl_table *table, int write,
2039 		void __user *buffer, size_t *length,
2040 		loff_t *ppos);
2041 #endif
2042 #ifdef CONFIG_SCHED_DEBUG
2043 static inline unsigned int get_sysctl_timer_migration(void)
2044 {
2045 	return sysctl_timer_migration;
2046 }
2047 #else
2048 static inline unsigned int get_sysctl_timer_migration(void)
2049 {
2050 	return 1;
2051 }
2052 #endif
2053 extern unsigned int sysctl_sched_rt_period;
2054 extern int sysctl_sched_rt_runtime;
2055 
2056 int sched_rt_handler(struct ctl_table *table, int write,
2057 		void __user *buffer, size_t *lenp,
2058 		loff_t *ppos);
2059 
2060 #ifdef CONFIG_SCHED_AUTOGROUP
2061 extern unsigned int sysctl_sched_autogroup_enabled;
2062 
2063 extern void sched_autogroup_create_attach(struct task_struct *p);
2064 extern void sched_autogroup_detach(struct task_struct *p);
2065 extern void sched_autogroup_fork(struct signal_struct *sig);
2066 extern void sched_autogroup_exit(struct signal_struct *sig);
2067 #ifdef CONFIG_PROC_FS
2068 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2069 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2070 #endif
2071 #else
2072 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2073 static inline void sched_autogroup_detach(struct task_struct *p) { }
2074 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2075 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2076 #endif
2077 
2078 #ifdef CONFIG_CFS_BANDWIDTH
2079 extern unsigned int sysctl_sched_cfs_bandwidth_slice;
2080 #endif
2081 
2082 #ifdef CONFIG_RT_MUTEXES
2083 extern int rt_mutex_getprio(struct task_struct *p);
2084 extern void rt_mutex_setprio(struct task_struct *p, int prio);
2085 extern void rt_mutex_adjust_pi(struct task_struct *p);
2086 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2087 {
2088 	return tsk->pi_blocked_on != NULL;
2089 }
2090 #else
2091 static inline int rt_mutex_getprio(struct task_struct *p)
2092 {
2093 	return p->normal_prio;
2094 }
2095 # define rt_mutex_adjust_pi(p)		do { } while (0)
2096 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2097 {
2098 	return false;
2099 }
2100 #endif
2101 
2102 extern bool yield_to(struct task_struct *p, bool preempt);
2103 extern void set_user_nice(struct task_struct *p, long nice);
2104 extern int task_prio(const struct task_struct *p);
2105 extern int task_nice(const struct task_struct *p);
2106 extern int can_nice(const struct task_struct *p, const int nice);
2107 extern int task_curr(const struct task_struct *p);
2108 extern int idle_cpu(int cpu);
2109 extern int sched_setscheduler(struct task_struct *, int,
2110 			      const struct sched_param *);
2111 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2112 				      const struct sched_param *);
2113 extern struct task_struct *idle_task(int cpu);
2114 /**
2115  * is_idle_task - is the specified task an idle task?
2116  * @p: the task in question.
2117  */
2118 static inline bool is_idle_task(const struct task_struct *p)
2119 {
2120 	return p->pid == 0;
2121 }
2122 extern struct task_struct *curr_task(int cpu);
2123 extern void set_curr_task(int cpu, struct task_struct *p);
2124 
2125 void yield(void);
2126 
2127 /*
2128  * The default (Linux) execution domain.
2129  */
2130 extern struct exec_domain	default_exec_domain;
2131 
2132 union thread_union {
2133 	struct thread_info thread_info;
2134 	unsigned long stack[THREAD_SIZE/sizeof(long)];
2135 };
2136 
2137 #ifndef __HAVE_ARCH_KSTACK_END
2138 static inline int kstack_end(void *addr)
2139 {
2140 	/* Reliable end of stack detection:
2141 	 * Some APM bios versions misalign the stack
2142 	 */
2143 	return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2144 }
2145 #endif
2146 
2147 extern union thread_union init_thread_union;
2148 extern struct task_struct init_task;
2149 
2150 extern struct   mm_struct init_mm;
2151 
2152 extern struct pid_namespace init_pid_ns;
2153 
2154 /*
2155  * find a task by one of its numerical ids
2156  *
2157  * find_task_by_pid_ns():
2158  *      finds a task by its pid in the specified namespace
2159  * find_task_by_vpid():
2160  *      finds a task by its virtual pid
2161  *
2162  * see also find_vpid() etc in include/linux/pid.h
2163  */
2164 
2165 extern struct task_struct *find_task_by_vpid(pid_t nr);
2166 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2167 		struct pid_namespace *ns);
2168 
2169 extern void __set_special_pids(struct pid *pid);
2170 
2171 /* per-UID process charging. */
2172 extern struct user_struct * alloc_uid(kuid_t);
2173 static inline struct user_struct *get_uid(struct user_struct *u)
2174 {
2175 	atomic_inc(&u->__count);
2176 	return u;
2177 }
2178 extern void free_uid(struct user_struct *);
2179 
2180 #include <asm/current.h>
2181 
2182 extern void xtime_update(unsigned long ticks);
2183 
2184 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2185 extern int wake_up_process(struct task_struct *tsk);
2186 extern void wake_up_new_task(struct task_struct *tsk);
2187 #ifdef CONFIG_SMP
2188  extern void kick_process(struct task_struct *tsk);
2189 #else
2190  static inline void kick_process(struct task_struct *tsk) { }
2191 #endif
2192 extern void sched_fork(struct task_struct *p);
2193 extern void sched_dead(struct task_struct *p);
2194 
2195 extern void proc_caches_init(void);
2196 extern void flush_signals(struct task_struct *);
2197 extern void __flush_signals(struct task_struct *);
2198 extern void ignore_signals(struct task_struct *);
2199 extern void flush_signal_handlers(struct task_struct *, int force_default);
2200 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2201 
2202 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2203 {
2204 	unsigned long flags;
2205 	int ret;
2206 
2207 	spin_lock_irqsave(&tsk->sighand->siglock, flags);
2208 	ret = dequeue_signal(tsk, mask, info);
2209 	spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2210 
2211 	return ret;
2212 }
2213 
2214 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2215 			      sigset_t *mask);
2216 extern void unblock_all_signals(void);
2217 extern void release_task(struct task_struct * p);
2218 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2219 extern int force_sigsegv(int, struct task_struct *);
2220 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2221 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2222 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2223 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2224 				const struct cred *, u32);
2225 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2226 extern int kill_pid(struct pid *pid, int sig, int priv);
2227 extern int kill_proc_info(int, struct siginfo *, pid_t);
2228 extern __must_check bool do_notify_parent(struct task_struct *, int);
2229 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2230 extern void force_sig(int, struct task_struct *);
2231 extern int send_sig(int, struct task_struct *, int);
2232 extern int zap_other_threads(struct task_struct *p);
2233 extern struct sigqueue *sigqueue_alloc(void);
2234 extern void sigqueue_free(struct sigqueue *);
2235 extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
2236 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2237 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2238 
2239 static inline void restore_saved_sigmask(void)
2240 {
2241 	if (test_and_clear_restore_sigmask())
2242 		__set_current_blocked(&current->saved_sigmask);
2243 }
2244 
2245 static inline sigset_t *sigmask_to_save(void)
2246 {
2247 	sigset_t *res = &current->blocked;
2248 	if (unlikely(test_restore_sigmask()))
2249 		res = &current->saved_sigmask;
2250 	return res;
2251 }
2252 
2253 static inline int kill_cad_pid(int sig, int priv)
2254 {
2255 	return kill_pid(cad_pid, sig, priv);
2256 }
2257 
2258 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
2259 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2260 #define SEND_SIG_PRIV	((struct siginfo *) 1)
2261 #define SEND_SIG_FORCED	((struct siginfo *) 2)
2262 
2263 /*
2264  * True if we are on the alternate signal stack.
2265  */
2266 static inline int on_sig_stack(unsigned long sp)
2267 {
2268 #ifdef CONFIG_STACK_GROWSUP
2269 	return sp >= current->sas_ss_sp &&
2270 		sp - current->sas_ss_sp < current->sas_ss_size;
2271 #else
2272 	return sp > current->sas_ss_sp &&
2273 		sp - current->sas_ss_sp <= current->sas_ss_size;
2274 #endif
2275 }
2276 
2277 static inline int sas_ss_flags(unsigned long sp)
2278 {
2279 	return (current->sas_ss_size == 0 ? SS_DISABLE
2280 		: on_sig_stack(sp) ? SS_ONSTACK : 0);
2281 }
2282 
2283 /*
2284  * Routines for handling mm_structs
2285  */
2286 extern struct mm_struct * mm_alloc(void);
2287 
2288 /* mmdrop drops the mm and the page tables */
2289 extern void __mmdrop(struct mm_struct *);
2290 static inline void mmdrop(struct mm_struct * mm)
2291 {
2292 	if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2293 		__mmdrop(mm);
2294 }
2295 
2296 /* mmput gets rid of the mappings and all user-space */
2297 extern void mmput(struct mm_struct *);
2298 /* Grab a reference to a task's mm, if it is not already going away */
2299 extern struct mm_struct *get_task_mm(struct task_struct *task);
2300 /*
2301  * Grab a reference to a task's mm, if it is not already going away
2302  * and ptrace_may_access with the mode parameter passed to it
2303  * succeeds.
2304  */
2305 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2306 /* Remove the current tasks stale references to the old mm_struct */
2307 extern void mm_release(struct task_struct *, struct mm_struct *);
2308 /* Allocate a new mm structure and copy contents from tsk->mm */
2309 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2310 
2311 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2312 			struct task_struct *, struct pt_regs *);
2313 extern void flush_thread(void);
2314 extern void exit_thread(void);
2315 
2316 extern void exit_files(struct task_struct *);
2317 extern void __cleanup_sighand(struct sighand_struct *);
2318 
2319 extern void exit_itimers(struct signal_struct *);
2320 extern void flush_itimer_signals(void);
2321 
2322 extern void do_group_exit(int);
2323 
2324 extern void daemonize(const char *, ...);
2325 extern int allow_signal(int);
2326 extern int disallow_signal(int);
2327 
2328 extern int do_execve(const char *,
2329 		     const char __user * const __user *,
2330 		     const char __user * const __user *, struct pt_regs *);
2331 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2332 struct task_struct *fork_idle(int);
2333 
2334 extern void set_task_comm(struct task_struct *tsk, char *from);
2335 extern char *get_task_comm(char *to, struct task_struct *tsk);
2336 
2337 #ifdef CONFIG_SMP
2338 void scheduler_ipi(void);
2339 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2340 #else
2341 static inline void scheduler_ipi(void) { }
2342 static inline unsigned long wait_task_inactive(struct task_struct *p,
2343 					       long match_state)
2344 {
2345 	return 1;
2346 }
2347 #endif
2348 
2349 #define next_task(p) \
2350 	list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2351 
2352 #define for_each_process(p) \
2353 	for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2354 
2355 extern bool current_is_single_threaded(void);
2356 
2357 /*
2358  * Careful: do_each_thread/while_each_thread is a double loop so
2359  *          'break' will not work as expected - use goto instead.
2360  */
2361 #define do_each_thread(g, t) \
2362 	for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2363 
2364 #define while_each_thread(g, t) \
2365 	while ((t = next_thread(t)) != g)
2366 
2367 static inline int get_nr_threads(struct task_struct *tsk)
2368 {
2369 	return tsk->signal->nr_threads;
2370 }
2371 
2372 static inline bool thread_group_leader(struct task_struct *p)
2373 {
2374 	return p->exit_signal >= 0;
2375 }
2376 
2377 /* Do to the insanities of de_thread it is possible for a process
2378  * to have the pid of the thread group leader without actually being
2379  * the thread group leader.  For iteration through the pids in proc
2380  * all we care about is that we have a task with the appropriate
2381  * pid, we don't actually care if we have the right task.
2382  */
2383 static inline int has_group_leader_pid(struct task_struct *p)
2384 {
2385 	return p->pid == p->tgid;
2386 }
2387 
2388 static inline
2389 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2390 {
2391 	return p1->tgid == p2->tgid;
2392 }
2393 
2394 static inline struct task_struct *next_thread(const struct task_struct *p)
2395 {
2396 	return list_entry_rcu(p->thread_group.next,
2397 			      struct task_struct, thread_group);
2398 }
2399 
2400 static inline int thread_group_empty(struct task_struct *p)
2401 {
2402 	return list_empty(&p->thread_group);
2403 }
2404 
2405 #define delay_group_leader(p) \
2406 		(thread_group_leader(p) && !thread_group_empty(p))
2407 
2408 /*
2409  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2410  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
2411  * pins the final release of task.io_context.  Also protects ->cpuset and
2412  * ->cgroup.subsys[]. And ->vfork_done.
2413  *
2414  * Nests both inside and outside of read_lock(&tasklist_lock).
2415  * It must not be nested with write_lock_irq(&tasklist_lock),
2416  * neither inside nor outside.
2417  */
2418 static inline void task_lock(struct task_struct *p)
2419 {
2420 	spin_lock(&p->alloc_lock);
2421 }
2422 
2423 static inline void task_unlock(struct task_struct *p)
2424 {
2425 	spin_unlock(&p->alloc_lock);
2426 }
2427 
2428 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2429 							unsigned long *flags);
2430 
2431 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2432 						       unsigned long *flags)
2433 {
2434 	struct sighand_struct *ret;
2435 
2436 	ret = __lock_task_sighand(tsk, flags);
2437 	(void)__cond_lock(&tsk->sighand->siglock, ret);
2438 	return ret;
2439 }
2440 
2441 static inline void unlock_task_sighand(struct task_struct *tsk,
2442 						unsigned long *flags)
2443 {
2444 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2445 }
2446 
2447 #ifdef CONFIG_CGROUPS
2448 static inline void threadgroup_change_begin(struct task_struct *tsk)
2449 {
2450 	down_read(&tsk->signal->group_rwsem);
2451 }
2452 static inline void threadgroup_change_end(struct task_struct *tsk)
2453 {
2454 	up_read(&tsk->signal->group_rwsem);
2455 }
2456 
2457 /**
2458  * threadgroup_lock - lock threadgroup
2459  * @tsk: member task of the threadgroup to lock
2460  *
2461  * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
2462  * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2463  * perform exec.  This is useful for cases where the threadgroup needs to
2464  * stay stable across blockable operations.
2465  *
2466  * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2467  * synchronization.  While held, no new task will be added to threadgroup
2468  * and no existing live task will have its PF_EXITING set.
2469  *
2470  * During exec, a task goes and puts its thread group through unusual
2471  * changes.  After de-threading, exclusive access is assumed to resources
2472  * which are usually shared by tasks in the same group - e.g. sighand may
2473  * be replaced with a new one.  Also, the exec'ing task takes over group
2474  * leader role including its pid.  Exclude these changes while locked by
2475  * grabbing cred_guard_mutex which is used to synchronize exec path.
2476  */
2477 static inline void threadgroup_lock(struct task_struct *tsk)
2478 {
2479 	/*
2480 	 * exec uses exit for de-threading nesting group_rwsem inside
2481 	 * cred_guard_mutex. Grab cred_guard_mutex first.
2482 	 */
2483 	mutex_lock(&tsk->signal->cred_guard_mutex);
2484 	down_write(&tsk->signal->group_rwsem);
2485 }
2486 
2487 /**
2488  * threadgroup_unlock - unlock threadgroup
2489  * @tsk: member task of the threadgroup to unlock
2490  *
2491  * Reverse threadgroup_lock().
2492  */
2493 static inline void threadgroup_unlock(struct task_struct *tsk)
2494 {
2495 	up_write(&tsk->signal->group_rwsem);
2496 	mutex_unlock(&tsk->signal->cred_guard_mutex);
2497 }
2498 #else
2499 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2500 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2501 static inline void threadgroup_lock(struct task_struct *tsk) {}
2502 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2503 #endif
2504 
2505 #ifndef __HAVE_THREAD_FUNCTIONS
2506 
2507 #define task_thread_info(task)	((struct thread_info *)(task)->stack)
2508 #define task_stack_page(task)	((task)->stack)
2509 
2510 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2511 {
2512 	*task_thread_info(p) = *task_thread_info(org);
2513 	task_thread_info(p)->task = p;
2514 }
2515 
2516 static inline unsigned long *end_of_stack(struct task_struct *p)
2517 {
2518 	return (unsigned long *)(task_thread_info(p) + 1);
2519 }
2520 
2521 #endif
2522 
2523 static inline int object_is_on_stack(void *obj)
2524 {
2525 	void *stack = task_stack_page(current);
2526 
2527 	return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2528 }
2529 
2530 extern void thread_info_cache_init(void);
2531 
2532 #ifdef CONFIG_DEBUG_STACK_USAGE
2533 static inline unsigned long stack_not_used(struct task_struct *p)
2534 {
2535 	unsigned long *n = end_of_stack(p);
2536 
2537 	do { 	/* Skip over canary */
2538 		n++;
2539 	} while (!*n);
2540 
2541 	return (unsigned long)n - (unsigned long)end_of_stack(p);
2542 }
2543 #endif
2544 
2545 /* set thread flags in other task's structures
2546  * - see asm/thread_info.h for TIF_xxxx flags available
2547  */
2548 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2549 {
2550 	set_ti_thread_flag(task_thread_info(tsk), flag);
2551 }
2552 
2553 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2554 {
2555 	clear_ti_thread_flag(task_thread_info(tsk), flag);
2556 }
2557 
2558 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2559 {
2560 	return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2561 }
2562 
2563 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2564 {
2565 	return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2566 }
2567 
2568 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2569 {
2570 	return test_ti_thread_flag(task_thread_info(tsk), flag);
2571 }
2572 
2573 static inline void set_tsk_need_resched(struct task_struct *tsk)
2574 {
2575 	set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2576 }
2577 
2578 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2579 {
2580 	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2581 }
2582 
2583 static inline int test_tsk_need_resched(struct task_struct *tsk)
2584 {
2585 	return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2586 }
2587 
2588 static inline int restart_syscall(void)
2589 {
2590 	set_tsk_thread_flag(current, TIF_SIGPENDING);
2591 	return -ERESTARTNOINTR;
2592 }
2593 
2594 static inline int signal_pending(struct task_struct *p)
2595 {
2596 	return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2597 }
2598 
2599 static inline int __fatal_signal_pending(struct task_struct *p)
2600 {
2601 	return unlikely(sigismember(&p->pending.signal, SIGKILL));
2602 }
2603 
2604 static inline int fatal_signal_pending(struct task_struct *p)
2605 {
2606 	return signal_pending(p) && __fatal_signal_pending(p);
2607 }
2608 
2609 static inline int signal_pending_state(long state, struct task_struct *p)
2610 {
2611 	if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2612 		return 0;
2613 	if (!signal_pending(p))
2614 		return 0;
2615 
2616 	return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2617 }
2618 
2619 static inline int need_resched(void)
2620 {
2621 	return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2622 }
2623 
2624 /*
2625  * cond_resched() and cond_resched_lock(): latency reduction via
2626  * explicit rescheduling in places that are safe. The return
2627  * value indicates whether a reschedule was done in fact.
2628  * cond_resched_lock() will drop the spinlock before scheduling,
2629  * cond_resched_softirq() will enable bhs before scheduling.
2630  */
2631 extern int _cond_resched(void);
2632 
2633 #define cond_resched() ({			\
2634 	__might_sleep(__FILE__, __LINE__, 0);	\
2635 	_cond_resched();			\
2636 })
2637 
2638 extern int __cond_resched_lock(spinlock_t *lock);
2639 
2640 #ifdef CONFIG_PREEMPT_COUNT
2641 #define PREEMPT_LOCK_OFFSET	PREEMPT_OFFSET
2642 #else
2643 #define PREEMPT_LOCK_OFFSET	0
2644 #endif
2645 
2646 #define cond_resched_lock(lock) ({				\
2647 	__might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);	\
2648 	__cond_resched_lock(lock);				\
2649 })
2650 
2651 extern int __cond_resched_softirq(void);
2652 
2653 #define cond_resched_softirq() ({					\
2654 	__might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET);	\
2655 	__cond_resched_softirq();					\
2656 })
2657 
2658 /*
2659  * Does a critical section need to be broken due to another
2660  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2661  * but a general need for low latency)
2662  */
2663 static inline int spin_needbreak(spinlock_t *lock)
2664 {
2665 #ifdef CONFIG_PREEMPT
2666 	return spin_is_contended(lock);
2667 #else
2668 	return 0;
2669 #endif
2670 }
2671 
2672 /*
2673  * Thread group CPU time accounting.
2674  */
2675 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2676 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2677 
2678 static inline void thread_group_cputime_init(struct signal_struct *sig)
2679 {
2680 	raw_spin_lock_init(&sig->cputimer.lock);
2681 }
2682 
2683 /*
2684  * Reevaluate whether the task has signals pending delivery.
2685  * Wake the task if so.
2686  * This is required every time the blocked sigset_t changes.
2687  * callers must hold sighand->siglock.
2688  */
2689 extern void recalc_sigpending_and_wake(struct task_struct *t);
2690 extern void recalc_sigpending(void);
2691 
2692 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2693 
2694 /*
2695  * Wrappers for p->thread_info->cpu access. No-op on UP.
2696  */
2697 #ifdef CONFIG_SMP
2698 
2699 static inline unsigned int task_cpu(const struct task_struct *p)
2700 {
2701 	return task_thread_info(p)->cpu;
2702 }
2703 
2704 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2705 
2706 #else
2707 
2708 static inline unsigned int task_cpu(const struct task_struct *p)
2709 {
2710 	return 0;
2711 }
2712 
2713 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2714 {
2715 }
2716 
2717 #endif /* CONFIG_SMP */
2718 
2719 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2720 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2721 
2722 extern void normalize_rt_tasks(void);
2723 
2724 #ifdef CONFIG_CGROUP_SCHED
2725 
2726 extern struct task_group root_task_group;
2727 
2728 extern struct task_group *sched_create_group(struct task_group *parent);
2729 extern void sched_destroy_group(struct task_group *tg);
2730 extern void sched_move_task(struct task_struct *tsk);
2731 #ifdef CONFIG_FAIR_GROUP_SCHED
2732 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2733 extern unsigned long sched_group_shares(struct task_group *tg);
2734 #endif
2735 #ifdef CONFIG_RT_GROUP_SCHED
2736 extern int sched_group_set_rt_runtime(struct task_group *tg,
2737 				      long rt_runtime_us);
2738 extern long sched_group_rt_runtime(struct task_group *tg);
2739 extern int sched_group_set_rt_period(struct task_group *tg,
2740 				      long rt_period_us);
2741 extern long sched_group_rt_period(struct task_group *tg);
2742 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2743 #endif
2744 #endif /* CONFIG_CGROUP_SCHED */
2745 
2746 extern int task_can_switch_user(struct user_struct *up,
2747 					struct task_struct *tsk);
2748 
2749 #ifdef CONFIG_TASK_XACCT
2750 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2751 {
2752 	tsk->ioac.rchar += amt;
2753 }
2754 
2755 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2756 {
2757 	tsk->ioac.wchar += amt;
2758 }
2759 
2760 static inline void inc_syscr(struct task_struct *tsk)
2761 {
2762 	tsk->ioac.syscr++;
2763 }
2764 
2765 static inline void inc_syscw(struct task_struct *tsk)
2766 {
2767 	tsk->ioac.syscw++;
2768 }
2769 #else
2770 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2771 {
2772 }
2773 
2774 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2775 {
2776 }
2777 
2778 static inline void inc_syscr(struct task_struct *tsk)
2779 {
2780 }
2781 
2782 static inline void inc_syscw(struct task_struct *tsk)
2783 {
2784 }
2785 #endif
2786 
2787 #ifndef TASK_SIZE_OF
2788 #define TASK_SIZE_OF(tsk)	TASK_SIZE
2789 #endif
2790 
2791 #ifdef CONFIG_MM_OWNER
2792 extern void mm_update_next_owner(struct mm_struct *mm);
2793 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2794 #else
2795 static inline void mm_update_next_owner(struct mm_struct *mm)
2796 {
2797 }
2798 
2799 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2800 {
2801 }
2802 #endif /* CONFIG_MM_OWNER */
2803 
2804 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2805 		unsigned int limit)
2806 {
2807 	return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2808 }
2809 
2810 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2811 		unsigned int limit)
2812 {
2813 	return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2814 }
2815 
2816 static inline unsigned long rlimit(unsigned int limit)
2817 {
2818 	return task_rlimit(current, limit);
2819 }
2820 
2821 static inline unsigned long rlimit_max(unsigned int limit)
2822 {
2823 	return task_rlimit_max(current, limit);
2824 }
2825 
2826 #endif /* __KERNEL__ */
2827 
2828 #endif
2829