4781332f2e1153b50c93712bd5797ad91b808b0d
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / include / linux / sched.h
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3
4 #include <uapi/linux/sched.h>
5
6
7 struct sched_param {
8 int sched_priority;
9 };
10
11 #include <asm/param.h> /* for HZ */
12
13 #include <linux/capability.h>
14 #include <linux/threads.h>
15 #include <linux/kernel.h>
16 #include <linux/types.h>
17 #include <linux/timex.h>
18 #include <linux/jiffies.h>
19 #include <linux/rbtree.h>
20 #include <linux/thread_info.h>
21 #include <linux/cpumask.h>
22 #include <linux/errno.h>
23 #include <linux/nodemask.h>
24 #include <linux/mm_types.h>
25
26 #include <asm/page.h>
27 #include <asm/ptrace.h>
28 #include <asm/cputime.h>
29
30 #include <linux/smp.h>
31 #include <linux/sem.h>
32 #include <linux/signal.h>
33 #include <linux/compiler.h>
34 #include <linux/completion.h>
35 #include <linux/pid.h>
36 #include <linux/percpu.h>
37 #include <linux/topology.h>
38 #include <linux/proportions.h>
39 #include <linux/seccomp.h>
40 #include <linux/rcupdate.h>
41 #include <linux/rculist.h>
42 #include <linux/rtmutex.h>
43
44 #include <linux/time.h>
45 #include <linux/param.h>
46 #include <linux/resource.h>
47 #include <linux/timer.h>
48 #include <linux/hrtimer.h>
49 #include <linux/task_io_accounting.h>
50 #include <linux/latencytop.h>
51 #include <linux/cred.h>
52 #include <linux/llist.h>
53 #include <linux/uidgid.h>
54 #include <linux/gfp.h>
55
56 #include <asm/processor.h>
57
58 struct exec_domain;
59 struct futex_pi_state;
60 struct robust_list_head;
61 struct bio_list;
62 struct fs_struct;
63 struct perf_event_context;
64 struct blk_plug;
65
66 /*
67 * List of flags we want to share for kernel threads,
68 * if only because they are not used by them anyway.
69 */
70 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
71
72 /*
73 * These are the constant used to fake the fixed-point load-average
74 * counting. Some notes:
75 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
76 * a load-average precision of 10 bits integer + 11 bits fractional
77 * - if you want to count load-averages more often, you need more
78 * precision, or rounding will get you. With 2-second counting freq,
79 * the EXP_n values would be 1981, 2034 and 2043 if still using only
80 * 11 bit fractions.
81 */
82 extern unsigned long avenrun[]; /* Load averages */
83 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
84
85 #define FSHIFT 11 /* nr of bits of precision */
86 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
87 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
88 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
89 #define EXP_5 2014 /* 1/exp(5sec/5min) */
90 #define EXP_15 2037 /* 1/exp(5sec/15min) */
91
92 #define CALC_LOAD(load,exp,n) \
93 load *= exp; \
94 load += n*(FIXED_1-exp); \
95 load >>= FSHIFT;
96
97 extern unsigned long total_forks;
98 extern int nr_threads;
99 DECLARE_PER_CPU(unsigned long, process_counts);
100 extern int nr_processes(void);
101 extern unsigned long nr_running(void);
102 extern unsigned long nr_iowait(void);
103 extern unsigned long nr_iowait_cpu(int cpu);
104 extern unsigned long this_cpu_load(void);
105
106
107 extern void calc_global_load(unsigned long ticks);
108 extern void update_cpu_load_nohz(void);
109
110 /* Notifier for when a task gets migrated to a new CPU */
111 struct task_migration_notifier {
112 struct task_struct *task;
113 int from_cpu;
114 int to_cpu;
115 };
116 extern void register_task_migration_notifier(struct notifier_block *n);
117
118 extern unsigned long get_parent_ip(unsigned long addr);
119
120 extern void dump_cpu_task(int cpu);
121
122 struct seq_file;
123 struct cfs_rq;
124 struct task_group;
125 #ifdef CONFIG_SCHED_DEBUG
126 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
127 extern void proc_sched_set_task(struct task_struct *p);
128 extern void
129 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
130 #endif
131
132 /*
133 * Task state bitmask. NOTE! These bits are also
134 * encoded in fs/proc/array.c: get_task_state().
135 *
136 * We have two separate sets of flags: task->state
137 * is about runnability, while task->exit_state are
138 * about the task exiting. Confusing, but this way
139 * modifying one set can't modify the other one by
140 * mistake.
141 */
142 #define TASK_RUNNING 0
143 #define TASK_INTERRUPTIBLE 1
144 #define TASK_UNINTERRUPTIBLE 2
145 #define __TASK_STOPPED 4
146 #define __TASK_TRACED 8
147 /* in tsk->exit_state */
148 #define EXIT_ZOMBIE 16
149 #define EXIT_DEAD 32
150 /* in tsk->state again */
151 #define TASK_DEAD 64
152 #define TASK_WAKEKILL 128
153 #define TASK_WAKING 256
154 #define TASK_PARKED 512
155 #define TASK_STATE_MAX 1024
156
157 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
158
159 extern char ___assert_task_state[1 - 2*!!(
160 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
161
162 /* Convenience macros for the sake of set_task_state */
163 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
164 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
165 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
166
167 /* Convenience macros for the sake of wake_up */
168 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
169 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
170
171 /* get_task_state() */
172 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
173 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
174 __TASK_TRACED)
175
176 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
177 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
178 #define task_is_dead(task) ((task)->exit_state != 0)
179 #define task_is_stopped_or_traced(task) \
180 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
181 #define task_contributes_to_load(task) \
182 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
183 (task->flags & PF_FROZEN) == 0)
184
185 #define __set_task_state(tsk, state_value) \
186 do { (tsk)->state = (state_value); } while (0)
187 #define set_task_state(tsk, state_value) \
188 set_mb((tsk)->state, (state_value))
189
190 /*
191 * set_current_state() includes a barrier so that the write of current->state
192 * is correctly serialised wrt the caller's subsequent test of whether to
193 * actually sleep:
194 *
195 * set_current_state(TASK_UNINTERRUPTIBLE);
196 * if (do_i_need_to_sleep())
197 * schedule();
198 *
199 * If the caller does not need such serialisation then use __set_current_state()
200 */
201 #define __set_current_state(state_value) \
202 do { current->state = (state_value); } while (0)
203 #define set_current_state(state_value) \
204 set_mb(current->state, (state_value))
205
206 /* Task command name length */
207 #define TASK_COMM_LEN 16
208
209 #include <linux/spinlock.h>
210
211 /*
212 * This serializes "schedule()" and also protects
213 * the run-queue from deletions/modifications (but
214 * _adding_ to the beginning of the run-queue has
215 * a separate lock).
216 */
217 extern rwlock_t tasklist_lock;
218 extern spinlock_t mmlist_lock;
219
220 struct task_struct;
221
222 #ifdef CONFIG_PROVE_RCU
223 extern int lockdep_tasklist_lock_is_held(void);
224 #endif /* #ifdef CONFIG_PROVE_RCU */
225
226 extern void sched_init(void);
227 extern void sched_init_smp(void);
228 extern asmlinkage void schedule_tail(struct task_struct *prev);
229 extern void init_idle(struct task_struct *idle, int cpu);
230 extern void init_idle_bootup_task(struct task_struct *idle);
231
232 extern int runqueue_is_locked(int cpu);
233
234 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
235 extern void nohz_balance_enter_idle(int cpu);
236 extern void set_cpu_sd_state_idle(void);
237 extern int get_nohz_timer_target(void);
238 #else
239 static inline void nohz_balance_enter_idle(int cpu) { }
240 static inline void set_cpu_sd_state_idle(void) { }
241 #endif
242
243 /*
244 * Only dump TASK_* tasks. (0 for all tasks)
245 */
246 extern void show_state_filter(unsigned long state_filter);
247
248 static inline void show_state(void)
249 {
250 show_state_filter(0);
251 }
252
253 extern void show_regs(struct pt_regs *);
254
255 /*
256 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
257 * task), SP is the stack pointer of the first frame that should be shown in the back
258 * trace (or NULL if the entire call-chain of the task should be shown).
259 */
260 extern void show_stack(struct task_struct *task, unsigned long *sp);
261
262 void io_schedule(void);
263 long io_schedule_timeout(long timeout);
264
265 extern void cpu_init (void);
266 extern void trap_init(void);
267 extern void update_process_times(int user);
268 extern void scheduler_tick(void);
269
270 extern void sched_show_task(struct task_struct *p);
271
272 #ifdef CONFIG_LOCKUP_DETECTOR
273 extern void touch_softlockup_watchdog(void);
274 extern void touch_softlockup_watchdog_sync(void);
275 extern void touch_all_softlockup_watchdogs(void);
276 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
277 void __user *buffer,
278 size_t *lenp, loff_t *ppos);
279 extern unsigned int softlockup_panic;
280 void lockup_detector_init(void);
281 #else
282 static inline void touch_softlockup_watchdog(void)
283 {
284 }
285 static inline void touch_softlockup_watchdog_sync(void)
286 {
287 }
288 static inline void touch_all_softlockup_watchdogs(void)
289 {
290 }
291 static inline void lockup_detector_init(void)
292 {
293 }
294 #endif
295
296 /* Attach to any functions which should be ignored in wchan output. */
297 #define __sched __attribute__((__section__(".sched.text")))
298
299 /* Linker adds these: start and end of __sched functions */
300 extern char __sched_text_start[], __sched_text_end[];
301
302 /* Is this address in the __sched functions? */
303 extern int in_sched_functions(unsigned long addr);
304
305 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
306 extern signed long schedule_timeout(signed long timeout);
307 extern signed long schedule_timeout_interruptible(signed long timeout);
308 extern signed long schedule_timeout_killable(signed long timeout);
309 extern signed long schedule_timeout_uninterruptible(signed long timeout);
310 asmlinkage void schedule(void);
311 extern void schedule_preempt_disabled(void);
312
313 struct nsproxy;
314 struct user_namespace;
315
316 #ifdef CONFIG_MMU
317 extern void arch_pick_mmap_layout(struct mm_struct *mm);
318 extern unsigned long
319 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
320 unsigned long, unsigned long);
321 extern unsigned long
322 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
323 unsigned long len, unsigned long pgoff,
324 unsigned long flags);
325 extern void arch_unmap_area(struct mm_struct *, unsigned long);
326 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
327 #else
328 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
329 #endif
330
331
332 extern void set_dumpable(struct mm_struct *mm, int value);
333 extern int get_dumpable(struct mm_struct *mm);
334
335 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
336 #define SUID_DUMP_USER 1 /* Dump as user of process */
337 #define SUID_DUMP_ROOT 2 /* Dump as root */
338
339 /* mm flags */
340 /* dumpable bits */
341 #define MMF_DUMPABLE 0 /* core dump is permitted */
342 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
343
344 #define MMF_DUMPABLE_BITS 2
345 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
346
347 /* coredump filter bits */
348 #define MMF_DUMP_ANON_PRIVATE 2
349 #define MMF_DUMP_ANON_SHARED 3
350 #define MMF_DUMP_MAPPED_PRIVATE 4
351 #define MMF_DUMP_MAPPED_SHARED 5
352 #define MMF_DUMP_ELF_HEADERS 6
353 #define MMF_DUMP_HUGETLB_PRIVATE 7
354 #define MMF_DUMP_HUGETLB_SHARED 8
355
356 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
357 #define MMF_DUMP_FILTER_BITS 7
358 #define MMF_DUMP_FILTER_MASK \
359 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
360 #define MMF_DUMP_FILTER_DEFAULT \
361 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
362 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
363
364 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
365 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
366 #else
367 # define MMF_DUMP_MASK_DEFAULT_ELF 0
368 #endif
369 /* leave room for more dump flags */
370 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
371 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
372 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
373
374 #define MMF_HAS_UPROBES 19 /* has uprobes */
375 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
376
377 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
378
379 struct sighand_struct {
380 atomic_t count;
381 struct k_sigaction action[_NSIG];
382 spinlock_t siglock;
383 wait_queue_head_t signalfd_wqh;
384 };
385
386 struct pacct_struct {
387 int ac_flag;
388 long ac_exitcode;
389 unsigned long ac_mem;
390 cputime_t ac_utime, ac_stime;
391 unsigned long ac_minflt, ac_majflt;
392 };
393
394 struct cpu_itimer {
395 cputime_t expires;
396 cputime_t incr;
397 u32 error;
398 u32 incr_error;
399 };
400
401 /**
402 * struct cputime - snaphsot of system and user cputime
403 * @utime: time spent in user mode
404 * @stime: time spent in system mode
405 *
406 * Gathers a generic snapshot of user and system time.
407 */
408 struct cputime {
409 cputime_t utime;
410 cputime_t stime;
411 };
412
413 /**
414 * struct task_cputime - collected CPU time counts
415 * @utime: time spent in user mode, in &cputime_t units
416 * @stime: time spent in kernel mode, in &cputime_t units
417 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
418 *
419 * This is an extension of struct cputime that includes the total runtime
420 * spent by the task from the scheduler point of view.
421 *
422 * As a result, this structure groups together three kinds of CPU time
423 * that are tracked for threads and thread groups. Most things considering
424 * CPU time want to group these counts together and treat all three
425 * of them in parallel.
426 */
427 struct task_cputime {
428 cputime_t utime;
429 cputime_t stime;
430 unsigned long long sum_exec_runtime;
431 };
432 /* Alternate field names when used to cache expirations. */
433 #define prof_exp stime
434 #define virt_exp utime
435 #define sched_exp sum_exec_runtime
436
437 #define INIT_CPUTIME \
438 (struct task_cputime) { \
439 .utime = 0, \
440 .stime = 0, \
441 .sum_exec_runtime = 0, \
442 }
443
444 /*
445 * Disable preemption until the scheduler is running.
446 * Reset by start_kernel()->sched_init()->init_idle().
447 *
448 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
449 * before the scheduler is active -- see should_resched().
450 */
451 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
452
453 /**
454 * struct thread_group_cputimer - thread group interval timer counts
455 * @cputime: thread group interval timers.
456 * @running: non-zero when there are timers running and
457 * @cputime receives updates.
458 * @lock: lock for fields in this struct.
459 *
460 * This structure contains the version of task_cputime, above, that is
461 * used for thread group CPU timer calculations.
462 */
463 struct thread_group_cputimer {
464 struct task_cputime cputime;
465 int running;
466 raw_spinlock_t lock;
467 };
468
469 #include <linux/rwsem.h>
470 struct autogroup;
471
472 /*
473 * NOTE! "signal_struct" does not have its own
474 * locking, because a shared signal_struct always
475 * implies a shared sighand_struct, so locking
476 * sighand_struct is always a proper superset of
477 * the locking of signal_struct.
478 */
479 struct signal_struct {
480 atomic_t sigcnt;
481 atomic_t live;
482 int nr_threads;
483 struct list_head thread_head;
484
485 wait_queue_head_t wait_chldexit; /* for wait4() */
486
487 /* current thread group signal load-balancing target: */
488 struct task_struct *curr_target;
489
490 /* shared signal handling: */
491 struct sigpending shared_pending;
492
493 /* thread group exit support */
494 int group_exit_code;
495 /* overloaded:
496 * - notify group_exit_task when ->count is equal to notify_count
497 * - everyone except group_exit_task is stopped during signal delivery
498 * of fatal signals, group_exit_task processes the signal.
499 */
500 int notify_count;
501 struct task_struct *group_exit_task;
502
503 /* thread group stop support, overloads group_exit_code too */
504 int group_stop_count;
505 unsigned int flags; /* see SIGNAL_* flags below */
506
507 /*
508 * PR_SET_CHILD_SUBREAPER marks a process, like a service
509 * manager, to re-parent orphan (double-forking) child processes
510 * to this process instead of 'init'. The service manager is
511 * able to receive SIGCHLD signals and is able to investigate
512 * the process until it calls wait(). All children of this
513 * process will inherit a flag if they should look for a
514 * child_subreaper process at exit.
515 */
516 unsigned int is_child_subreaper:1;
517 unsigned int has_child_subreaper:1;
518
519 /* POSIX.1b Interval Timers */
520 int posix_timer_id;
521 struct list_head posix_timers;
522
523 /* ITIMER_REAL timer for the process */
524 struct hrtimer real_timer;
525 struct pid *leader_pid;
526 ktime_t it_real_incr;
527
528 /*
529 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
530 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
531 * values are defined to 0 and 1 respectively
532 */
533 struct cpu_itimer it[2];
534
535 /*
536 * Thread group totals for process CPU timers.
537 * See thread_group_cputimer(), et al, for details.
538 */
539 struct thread_group_cputimer cputimer;
540
541 /* Earliest-expiration cache. */
542 struct task_cputime cputime_expires;
543
544 struct list_head cpu_timers[3];
545
546 struct pid *tty_old_pgrp;
547
548 /* boolean value for session group leader */
549 int leader;
550
551 struct tty_struct *tty; /* NULL if no tty */
552
553 #ifdef CONFIG_SCHED_AUTOGROUP
554 struct autogroup *autogroup;
555 #endif
556 /*
557 * Cumulative resource counters for dead threads in the group,
558 * and for reaped dead child processes forked by this group.
559 * Live threads maintain their own counters and add to these
560 * in __exit_signal, except for the group leader.
561 */
562 cputime_t utime, stime, cutime, cstime;
563 cputime_t gtime;
564 cputime_t cgtime;
565 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
566 struct cputime prev_cputime;
567 #endif
568 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
569 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
570 unsigned long inblock, oublock, cinblock, coublock;
571 unsigned long maxrss, cmaxrss;
572 struct task_io_accounting ioac;
573
574 /*
575 * Cumulative ns of schedule CPU time fo dead threads in the
576 * group, not including a zombie group leader, (This only differs
577 * from jiffies_to_ns(utime + stime) if sched_clock uses something
578 * other than jiffies.)
579 */
580 unsigned long long sum_sched_runtime;
581
582 /*
583 * We don't bother to synchronize most readers of this at all,
584 * because there is no reader checking a limit that actually needs
585 * to get both rlim_cur and rlim_max atomically, and either one
586 * alone is a single word that can safely be read normally.
587 * getrlimit/setrlimit use task_lock(current->group_leader) to
588 * protect this instead of the siglock, because they really
589 * have no need to disable irqs.
590 */
591 struct rlimit rlim[RLIM_NLIMITS];
592
593 #ifdef CONFIG_BSD_PROCESS_ACCT
594 struct pacct_struct pacct; /* per-process accounting information */
595 #endif
596 #ifdef CONFIG_TASKSTATS
597 struct taskstats *stats;
598 #endif
599 #ifdef CONFIG_AUDIT
600 unsigned audit_tty;
601 unsigned audit_tty_log_passwd;
602 struct tty_audit_buf *tty_audit_buf;
603 #endif
604 #ifdef CONFIG_CGROUPS
605 /*
606 * group_rwsem prevents new tasks from entering the threadgroup and
607 * member tasks from exiting,a more specifically, setting of
608 * PF_EXITING. fork and exit paths are protected with this rwsem
609 * using threadgroup_change_begin/end(). Users which require
610 * threadgroup to remain stable should use threadgroup_[un]lock()
611 * which also takes care of exec path. Currently, cgroup is the
612 * only user.
613 */
614 struct rw_semaphore group_rwsem;
615 #endif
616
617 oom_flags_t oom_flags;
618 short oom_score_adj; /* OOM kill score adjustment */
619 short oom_score_adj_min; /* OOM kill score adjustment min value.
620 * Only settable by CAP_SYS_RESOURCE. */
621
622 struct mutex cred_guard_mutex; /* guard against foreign influences on
623 * credential calculations
624 * (notably. ptrace) */
625 };
626
627 /*
628 * Bits in flags field of signal_struct.
629 */
630 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
631 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
632 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
633 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
634 /*
635 * Pending notifications to parent.
636 */
637 #define SIGNAL_CLD_STOPPED 0x00000010
638 #define SIGNAL_CLD_CONTINUED 0x00000020
639 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
640
641 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
642
643 /* If true, all threads except ->group_exit_task have pending SIGKILL */
644 static inline int signal_group_exit(const struct signal_struct *sig)
645 {
646 return (sig->flags & SIGNAL_GROUP_EXIT) ||
647 (sig->group_exit_task != NULL);
648 }
649
650 /*
651 * Some day this will be a full-fledged user tracking system..
652 */
653 struct user_struct {
654 atomic_t __count; /* reference count */
655 atomic_t processes; /* How many processes does this user have? */
656 atomic_t files; /* How many open files does this user have? */
657 atomic_t sigpending; /* How many pending signals does this user have? */
658 #ifdef CONFIG_INOTIFY_USER
659 atomic_t inotify_watches; /* How many inotify watches does this user have? */
660 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
661 #endif
662 #ifdef CONFIG_FANOTIFY
663 atomic_t fanotify_listeners;
664 #endif
665 #ifdef CONFIG_EPOLL
666 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
667 #endif
668 #ifdef CONFIG_POSIX_MQUEUE
669 /* protected by mq_lock */
670 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
671 #endif
672 unsigned long locked_shm; /* How many pages of mlocked shm ? */
673 unsigned long unix_inflight; /* How many files in flight in unix sockets */
674
675 #ifdef CONFIG_KEYS
676 struct key *uid_keyring; /* UID specific keyring */
677 struct key *session_keyring; /* UID's default session keyring */
678 #endif
679
680 /* Hash table maintenance information */
681 struct hlist_node uidhash_node;
682 kuid_t uid;
683
684 #ifdef CONFIG_PERF_EVENTS
685 atomic_long_t locked_vm;
686 #endif
687 };
688
689 extern int uids_sysfs_init(void);
690
691 extern struct user_struct *find_user(kuid_t);
692
693 extern struct user_struct root_user;
694 #define INIT_USER (&root_user)
695
696
697 struct backing_dev_info;
698 struct reclaim_state;
699
700 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
701 struct sched_info {
702 /* cumulative counters */
703 unsigned long pcount; /* # of times run on this cpu */
704 unsigned long long run_delay; /* time spent waiting on a runqueue */
705
706 /* timestamps */
707 unsigned long long last_arrival,/* when we last ran on a cpu */
708 last_queued; /* when we were last queued to run */
709 };
710 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
711
712 #ifdef CONFIG_TASK_DELAY_ACCT
713 struct task_delay_info {
714 spinlock_t lock;
715 unsigned int flags; /* Private per-task flags */
716
717 /* For each stat XXX, add following, aligned appropriately
718 *
719 * struct timespec XXX_start, XXX_end;
720 * u64 XXX_delay;
721 * u32 XXX_count;
722 *
723 * Atomicity of updates to XXX_delay, XXX_count protected by
724 * single lock above (split into XXX_lock if contention is an issue).
725 */
726
727 /*
728 * XXX_count is incremented on every XXX operation, the delay
729 * associated with the operation is added to XXX_delay.
730 * XXX_delay contains the accumulated delay time in nanoseconds.
731 */
732 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
733 u64 blkio_delay; /* wait for sync block io completion */
734 u64 swapin_delay; /* wait for swapin block io completion */
735 u32 blkio_count; /* total count of the number of sync block */
736 /* io operations performed */
737 u32 swapin_count; /* total count of the number of swapin block */
738 /* io operations performed */
739
740 struct timespec freepages_start, freepages_end;
741 u64 freepages_delay; /* wait for memory reclaim */
742 u32 freepages_count; /* total count of memory reclaim */
743 };
744 #endif /* CONFIG_TASK_DELAY_ACCT */
745
746 static inline int sched_info_on(void)
747 {
748 #ifdef CONFIG_SCHEDSTATS
749 return 1;
750 #elif defined(CONFIG_TASK_DELAY_ACCT)
751 extern int delayacct_on;
752 return delayacct_on;
753 #else
754 return 0;
755 #endif
756 }
757
758 enum cpu_idle_type {
759 CPU_IDLE,
760 CPU_NOT_IDLE,
761 CPU_NEWLY_IDLE,
762 CPU_MAX_IDLE_TYPES
763 };
764
765 /*
766 * Increase resolution of cpu_power calculations
767 */
768 #define SCHED_POWER_SHIFT 10
769 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
770
771 /*
772 * sched-domains (multiprocessor balancing) declarations:
773 */
774 #ifdef CONFIG_SMP
775 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
776 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
777 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
778 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
779 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
780 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
781 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
782 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
783 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
784 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
785 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
786 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
787
788 extern int __weak arch_sd_sibiling_asym_packing(void);
789
790 struct sched_domain_attr {
791 int relax_domain_level;
792 };
793
794 #define SD_ATTR_INIT (struct sched_domain_attr) { \
795 .relax_domain_level = -1, \
796 }
797
798 extern int sched_domain_level_max;
799
800 struct sched_group;
801
802 struct sched_domain {
803 /* These fields must be setup */
804 struct sched_domain *parent; /* top domain must be null terminated */
805 struct sched_domain *child; /* bottom domain must be null terminated */
806 struct sched_group *groups; /* the balancing groups of the domain */
807 unsigned long min_interval; /* Minimum balance interval ms */
808 unsigned long max_interval; /* Maximum balance interval ms */
809 unsigned int busy_factor; /* less balancing by factor if busy */
810 unsigned int imbalance_pct; /* No balance until over watermark */
811 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
812 unsigned int busy_idx;
813 unsigned int idle_idx;
814 unsigned int newidle_idx;
815 unsigned int wake_idx;
816 unsigned int forkexec_idx;
817 unsigned int smt_gain;
818
819 int nohz_idle; /* NOHZ IDLE status */
820 int flags; /* See SD_* */
821 int level;
822
823 /* Runtime fields. */
824 unsigned long last_balance; /* init to jiffies. units in jiffies */
825 unsigned int balance_interval; /* initialise to 1. units in ms. */
826 unsigned int nr_balance_failed; /* initialise to 0 */
827
828 u64 last_update;
829
830 #ifdef CONFIG_SCHEDSTATS
831 /* load_balance() stats */
832 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
833 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
834 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
835 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
836 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
837 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
838 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
839 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
840
841 /* Active load balancing */
842 unsigned int alb_count;
843 unsigned int alb_failed;
844 unsigned int alb_pushed;
845
846 /* SD_BALANCE_EXEC stats */
847 unsigned int sbe_count;
848 unsigned int sbe_balanced;
849 unsigned int sbe_pushed;
850
851 /* SD_BALANCE_FORK stats */
852 unsigned int sbf_count;
853 unsigned int sbf_balanced;
854 unsigned int sbf_pushed;
855
856 /* try_to_wake_up() stats */
857 unsigned int ttwu_wake_remote;
858 unsigned int ttwu_move_affine;
859 unsigned int ttwu_move_balance;
860 #endif
861 #ifdef CONFIG_SCHED_DEBUG
862 char *name;
863 #endif
864 union {
865 void *private; /* used during construction */
866 struct rcu_head rcu; /* used during destruction */
867 };
868
869 unsigned int span_weight;
870 /*
871 * Span of all CPUs in this domain.
872 *
873 * NOTE: this field is variable length. (Allocated dynamically
874 * by attaching extra space to the end of the structure,
875 * depending on how many CPUs the kernel has booted up with)
876 */
877 unsigned long span[0];
878 };
879
880 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
881 {
882 return to_cpumask(sd->span);
883 }
884
885 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
886 struct sched_domain_attr *dattr_new);
887
888 /* Allocate an array of sched domains, for partition_sched_domains(). */
889 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
890 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
891
892 bool cpus_share_cache(int this_cpu, int that_cpu);
893
894 #else /* CONFIG_SMP */
895
896 struct sched_domain_attr;
897
898 static inline void
899 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
900 struct sched_domain_attr *dattr_new)
901 {
902 }
903
904 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
905 {
906 return true;
907 }
908
909 #endif /* !CONFIG_SMP */
910
911
912 struct io_context; /* See blkdev.h */
913
914
915 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
916 extern void prefetch_stack(struct task_struct *t);
917 #else
918 static inline void prefetch_stack(struct task_struct *t) { }
919 #endif
920
921 struct audit_context; /* See audit.c */
922 struct mempolicy;
923 struct pipe_inode_info;
924 struct uts_namespace;
925
926 struct load_weight {
927 unsigned long weight, inv_weight;
928 };
929
930 struct sched_avg {
931 /*
932 * These sums represent an infinite geometric series and so are bound
933 * above by 1024/(1-y). Thus we only need a u32 to store them for for all
934 * choices of y < 1-2^(-32)*1024.
935 */
936 u32 runnable_avg_sum, runnable_avg_period;
937 u64 last_runnable_update;
938 s64 decay_count;
939 unsigned long load_avg_contrib;
940 };
941
942 #ifdef CONFIG_SCHEDSTATS
943 struct sched_statistics {
944 u64 wait_start;
945 u64 wait_max;
946 u64 wait_count;
947 u64 wait_sum;
948 u64 iowait_count;
949 u64 iowait_sum;
950
951 u64 sleep_start;
952 u64 sleep_max;
953 s64 sum_sleep_runtime;
954
955 u64 block_start;
956 u64 block_max;
957 u64 exec_max;
958 u64 slice_max;
959
960 u64 nr_migrations_cold;
961 u64 nr_failed_migrations_affine;
962 u64 nr_failed_migrations_running;
963 u64 nr_failed_migrations_hot;
964 u64 nr_forced_migrations;
965
966 u64 nr_wakeups;
967 u64 nr_wakeups_sync;
968 u64 nr_wakeups_migrate;
969 u64 nr_wakeups_local;
970 u64 nr_wakeups_remote;
971 u64 nr_wakeups_affine;
972 u64 nr_wakeups_affine_attempts;
973 u64 nr_wakeups_passive;
974 u64 nr_wakeups_idle;
975 };
976 #endif
977
978 struct sched_entity {
979 struct load_weight load; /* for load-balancing */
980 struct rb_node run_node;
981 struct list_head group_node;
982 unsigned int on_rq;
983
984 u64 exec_start;
985 u64 sum_exec_runtime;
986 u64 vruntime;
987 u64 prev_sum_exec_runtime;
988
989 u64 nr_migrations;
990
991 #ifdef CONFIG_SCHEDSTATS
992 struct sched_statistics statistics;
993 #endif
994
995 #ifdef CONFIG_FAIR_GROUP_SCHED
996 struct sched_entity *parent;
997 /* rq on which this entity is (to be) queued: */
998 struct cfs_rq *cfs_rq;
999 /* rq "owned" by this entity/group: */
1000 struct cfs_rq *my_q;
1001 #endif
1002
1003 /*
1004 * Load-tracking only depends on SMP, FAIR_GROUP_SCHED dependency below may be
1005 * removed when useful for applications beyond shares distribution (e.g.
1006 * load-balance).
1007 */
1008 #if defined(CONFIG_SMP) && defined(CONFIG_FAIR_GROUP_SCHED)
1009 /* Per-entity load-tracking */
1010 struct sched_avg avg;
1011 #endif
1012 };
1013
1014 struct sched_rt_entity {
1015 struct list_head run_list;
1016 unsigned long timeout;
1017 unsigned long watchdog_stamp;
1018 unsigned int time_slice;
1019
1020 struct sched_rt_entity *back;
1021 #ifdef CONFIG_RT_GROUP_SCHED
1022 struct sched_rt_entity *parent;
1023 /* rq on which this entity is (to be) queued: */
1024 struct rt_rq *rt_rq;
1025 /* rq "owned" by this entity/group: */
1026 struct rt_rq *my_q;
1027 #endif
1028 };
1029
1030
1031 struct rcu_node;
1032
1033 enum perf_event_task_context {
1034 perf_invalid_context = -1,
1035 perf_hw_context = 0,
1036 perf_sw_context,
1037 perf_nr_task_contexts,
1038 };
1039
1040 struct task_struct {
1041 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1042 void *stack;
1043 atomic_t usage;
1044 unsigned int flags; /* per process flags, defined below */
1045 unsigned int ptrace;
1046
1047 #ifdef CONFIG_SMP
1048 struct llist_node wake_entry;
1049 int on_cpu;
1050 #endif
1051 int on_rq;
1052
1053 int prio, static_prio, normal_prio;
1054 unsigned int rt_priority;
1055 const struct sched_class *sched_class;
1056 struct sched_entity se;
1057 struct sched_rt_entity rt;
1058 #ifdef CONFIG_CGROUP_SCHED
1059 struct task_group *sched_task_group;
1060 #endif
1061
1062 #ifdef CONFIG_PREEMPT_NOTIFIERS
1063 /* list of struct preempt_notifier: */
1064 struct hlist_head preempt_notifiers;
1065 #endif
1066
1067 /*
1068 * fpu_counter contains the number of consecutive context switches
1069 * that the FPU is used. If this is over a threshold, the lazy fpu
1070 * saving becomes unlazy to save the trap. This is an unsigned char
1071 * so that after 256 times the counter wraps and the behavior turns
1072 * lazy again; this to deal with bursty apps that only use FPU for
1073 * a short time
1074 */
1075 unsigned char fpu_counter;
1076 #ifdef CONFIG_BLK_DEV_IO_TRACE
1077 unsigned int btrace_seq;
1078 #endif
1079
1080 unsigned int policy;
1081 int nr_cpus_allowed;
1082 cpumask_t cpus_allowed;
1083
1084 #ifdef CONFIG_PREEMPT_RCU
1085 int rcu_read_lock_nesting;
1086 char rcu_read_unlock_special;
1087 struct list_head rcu_node_entry;
1088 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1089 #ifdef CONFIG_TREE_PREEMPT_RCU
1090 struct rcu_node *rcu_blocked_node;
1091 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1092 #ifdef CONFIG_RCU_BOOST
1093 struct rt_mutex *rcu_boost_mutex;
1094 #endif /* #ifdef CONFIG_RCU_BOOST */
1095
1096 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1097 struct sched_info sched_info;
1098 #endif
1099
1100 struct list_head tasks;
1101 #ifdef CONFIG_SMP
1102 struct plist_node pushable_tasks;
1103 #endif
1104
1105 struct mm_struct *mm, *active_mm;
1106 #ifdef CONFIG_COMPAT_BRK
1107 unsigned brk_randomized:1;
1108 #endif
1109 #if defined(SPLIT_RSS_COUNTING)
1110 struct task_rss_stat rss_stat;
1111 #endif
1112 /* task state */
1113 int exit_state;
1114 int exit_code, exit_signal;
1115 int pdeath_signal; /* The signal sent when the parent dies */
1116 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1117
1118 /* Used for emulating ABI behavior of previous Linux versions */
1119 unsigned int personality;
1120
1121 unsigned did_exec:1;
1122 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1123 * execve */
1124 unsigned in_iowait:1;
1125
1126 /* task may not gain privileges */
1127 unsigned no_new_privs:1;
1128
1129 /* Revert to default priority/policy when forking */
1130 unsigned sched_reset_on_fork:1;
1131 unsigned sched_contributes_to_load:1;
1132
1133 pid_t pid;
1134 pid_t tgid;
1135
1136 #ifdef CONFIG_CC_STACKPROTECTOR
1137 /* Canary value for the -fstack-protector gcc feature */
1138 unsigned long stack_canary;
1139 #endif
1140 /*
1141 * pointers to (original) parent process, youngest child, younger sibling,
1142 * older sibling, respectively. (p->father can be replaced with
1143 * p->real_parent->pid)
1144 */
1145 struct task_struct __rcu *real_parent; /* real parent process */
1146 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1147 /*
1148 * children/sibling forms the list of my natural children
1149 */
1150 struct list_head children; /* list of my children */
1151 struct list_head sibling; /* linkage in my parent's children list */
1152 struct task_struct *group_leader; /* threadgroup leader */
1153
1154 /*
1155 * ptraced is the list of tasks this task is using ptrace on.
1156 * This includes both natural children and PTRACE_ATTACH targets.
1157 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1158 */
1159 struct list_head ptraced;
1160 struct list_head ptrace_entry;
1161
1162 /* PID/PID hash table linkage. */
1163 struct pid_link pids[PIDTYPE_MAX];
1164 struct list_head thread_group;
1165 struct list_head thread_node;
1166
1167 struct completion *vfork_done; /* for vfork() */
1168 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1169 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1170
1171 cputime_t utime, stime, utimescaled, stimescaled;
1172 cputime_t gtime;
1173 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1174 struct cputime prev_cputime;
1175 #endif
1176 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1177 seqlock_t vtime_seqlock;
1178 unsigned long long vtime_snap;
1179 enum {
1180 VTIME_SLEEPING = 0,
1181 VTIME_USER,
1182 VTIME_SYS,
1183 } vtime_snap_whence;
1184 #endif
1185 unsigned long nvcsw, nivcsw; /* context switch counts */
1186 struct timespec start_time; /* monotonic time */
1187 struct timespec real_start_time; /* boot based time */
1188 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1189 unsigned long min_flt, maj_flt;
1190
1191 struct task_cputime cputime_expires;
1192 struct list_head cpu_timers[3];
1193
1194 /* process credentials */
1195 const struct cred __rcu *real_cred; /* objective and real subjective task
1196 * credentials (COW) */
1197 const struct cred __rcu *cred; /* effective (overridable) subjective task
1198 * credentials (COW) */
1199 char comm[TASK_COMM_LEN]; /* executable name excluding path
1200 - access with [gs]et_task_comm (which lock
1201 it with task_lock())
1202 - initialized normally by setup_new_exec */
1203 /* file system info */
1204 int link_count, total_link_count;
1205 #ifdef CONFIG_SYSVIPC
1206 /* ipc stuff */
1207 struct sysv_sem sysvsem;
1208 #endif
1209 #ifdef CONFIG_DETECT_HUNG_TASK
1210 /* hung task detection */
1211 unsigned long last_switch_count;
1212 #endif
1213 /* CPU-specific state of this task */
1214 struct thread_struct thread;
1215 /* filesystem information */
1216 struct fs_struct *fs;
1217 /* open file information */
1218 struct files_struct *files;
1219 /* namespaces */
1220 struct nsproxy *nsproxy;
1221 /* signal handlers */
1222 struct signal_struct *signal;
1223 struct sighand_struct *sighand;
1224
1225 sigset_t blocked, real_blocked;
1226 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1227 struct sigpending pending;
1228
1229 unsigned long sas_ss_sp;
1230 size_t sas_ss_size;
1231 int (*notifier)(void *priv);
1232 void *notifier_data;
1233 sigset_t *notifier_mask;
1234 struct callback_head *task_works;
1235
1236 struct audit_context *audit_context;
1237 #ifdef CONFIG_AUDITSYSCALL
1238 kuid_t loginuid;
1239 unsigned int sessionid;
1240 #endif
1241 struct seccomp seccomp;
1242
1243 /* Thread group tracking */
1244 u32 parent_exec_id;
1245 u32 self_exec_id;
1246 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1247 * mempolicy */
1248 spinlock_t alloc_lock;
1249
1250 /* Protection of the PI data structures: */
1251 raw_spinlock_t pi_lock;
1252
1253 #ifdef CONFIG_RT_MUTEXES
1254 /* PI waiters blocked on a rt_mutex held by this task */
1255 struct plist_head pi_waiters;
1256 /* Deadlock detection and priority inheritance handling */
1257 struct rt_mutex_waiter *pi_blocked_on;
1258 #endif
1259
1260 #ifdef CONFIG_DEBUG_MUTEXES
1261 /* mutex deadlock detection */
1262 struct mutex_waiter *blocked_on;
1263 #endif
1264 #ifdef CONFIG_TRACE_IRQFLAGS
1265 unsigned int irq_events;
1266 unsigned long hardirq_enable_ip;
1267 unsigned long hardirq_disable_ip;
1268 unsigned int hardirq_enable_event;
1269 unsigned int hardirq_disable_event;
1270 int hardirqs_enabled;
1271 int hardirq_context;
1272 unsigned long softirq_disable_ip;
1273 unsigned long softirq_enable_ip;
1274 unsigned int softirq_disable_event;
1275 unsigned int softirq_enable_event;
1276 int softirqs_enabled;
1277 int softirq_context;
1278 #endif
1279 #ifdef CONFIG_LOCKDEP
1280 # define MAX_LOCK_DEPTH 48UL
1281 u64 curr_chain_key;
1282 int lockdep_depth;
1283 unsigned int lockdep_recursion;
1284 struct held_lock held_locks[MAX_LOCK_DEPTH];
1285 gfp_t lockdep_reclaim_gfp;
1286 #endif
1287
1288 /* journalling filesystem info */
1289 void *journal_info;
1290
1291 /* stacked block device info */
1292 struct bio_list *bio_list;
1293
1294 #ifdef CONFIG_BLOCK
1295 /* stack plugging */
1296 struct blk_plug *plug;
1297 #endif
1298
1299 /* VM state */
1300 struct reclaim_state *reclaim_state;
1301
1302 struct backing_dev_info *backing_dev_info;
1303
1304 struct io_context *io_context;
1305
1306 unsigned long ptrace_message;
1307 siginfo_t *last_siginfo; /* For ptrace use. */
1308 struct task_io_accounting ioac;
1309 #if defined(CONFIG_TASK_XACCT)
1310 u64 acct_rss_mem1; /* accumulated rss usage */
1311 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1312 cputime_t acct_timexpd; /* stime + utime since last update */
1313 #endif
1314 #ifdef CONFIG_CPUSETS
1315 nodemask_t mems_allowed; /* Protected by alloc_lock */
1316 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1317 int cpuset_mem_spread_rotor;
1318 int cpuset_slab_spread_rotor;
1319 #endif
1320 #ifdef CONFIG_CGROUPS
1321 /* Control Group info protected by css_set_lock */
1322 struct css_set __rcu *cgroups;
1323 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1324 struct list_head cg_list;
1325 #endif
1326 #ifdef CONFIG_FUTEX
1327 struct robust_list_head __user *robust_list;
1328 #ifdef CONFIG_COMPAT
1329 struct compat_robust_list_head __user *compat_robust_list;
1330 #endif
1331 struct list_head pi_state_list;
1332 struct futex_pi_state *pi_state_cache;
1333 #endif
1334 #ifdef CONFIG_PERF_EVENTS
1335 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1336 struct mutex perf_event_mutex;
1337 struct list_head perf_event_list;
1338 #endif
1339 #ifdef CONFIG_NUMA
1340 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1341 short il_next;
1342 short pref_node_fork;
1343 #endif
1344 #ifdef CONFIG_NUMA_BALANCING
1345 int numa_scan_seq;
1346 int numa_migrate_seq;
1347 unsigned int numa_scan_period;
1348 u64 node_stamp; /* migration stamp */
1349 struct callback_head numa_work;
1350 #endif /* CONFIG_NUMA_BALANCING */
1351
1352 struct rcu_head rcu;
1353
1354 /*
1355 * cache last used pipe for splice
1356 */
1357 struct pipe_inode_info *splice_pipe;
1358
1359 struct page_frag task_frag;
1360
1361 #ifdef CONFIG_TASK_DELAY_ACCT
1362 struct task_delay_info *delays;
1363 #endif
1364 #ifdef CONFIG_FAULT_INJECTION
1365 int make_it_fail;
1366 #endif
1367 /*
1368 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1369 * balance_dirty_pages() for some dirty throttling pause
1370 */
1371 int nr_dirtied;
1372 int nr_dirtied_pause;
1373 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1374
1375 #ifdef CONFIG_LATENCYTOP
1376 int latency_record_count;
1377 struct latency_record latency_record[LT_SAVECOUNT];
1378 #endif
1379 /*
1380 * time slack values; these are used to round up poll() and
1381 * select() etc timeout values. These are in nanoseconds.
1382 */
1383 unsigned long timer_slack_ns;
1384 unsigned long default_timer_slack_ns;
1385
1386 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1387 /* Index of current stored address in ret_stack */
1388 int curr_ret_stack;
1389 /* Stack of return addresses for return function tracing */
1390 struct ftrace_ret_stack *ret_stack;
1391 /* time stamp for last schedule */
1392 unsigned long long ftrace_timestamp;
1393 /*
1394 * Number of functions that haven't been traced
1395 * because of depth overrun.
1396 */
1397 atomic_t trace_overrun;
1398 /* Pause for the tracing */
1399 atomic_t tracing_graph_pause;
1400 #endif
1401 #ifdef CONFIG_TRACING
1402 /* state flags for use by tracers */
1403 unsigned long trace;
1404 /* bitmask and counter of trace recursion */
1405 unsigned long trace_recursion;
1406 #endif /* CONFIG_TRACING */
1407 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1408 struct memcg_batch_info {
1409 int do_batch; /* incremented when batch uncharge started */
1410 struct mem_cgroup *memcg; /* target memcg of uncharge */
1411 unsigned long nr_pages; /* uncharged usage */
1412 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1413 } memcg_batch;
1414 unsigned int memcg_kmem_skip_account;
1415 struct memcg_oom_info {
1416 struct mem_cgroup *memcg;
1417 gfp_t gfp_mask;
1418 int order;
1419 unsigned int may_oom:1;
1420 } memcg_oom;
1421 #endif
1422 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1423 atomic_t ptrace_bp_refcnt;
1424 #endif
1425 #ifdef CONFIG_UPROBES
1426 struct uprobe_task *utask;
1427 #endif
1428 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1429 unsigned int sequential_io;
1430 unsigned int sequential_io_avg;
1431 #endif
1432 };
1433
1434 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1435 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1436
1437 #ifdef CONFIG_NUMA_BALANCING
1438 extern void task_numa_fault(int node, int pages, bool migrated);
1439 extern void set_numabalancing_state(bool enabled);
1440 #else
1441 static inline void task_numa_fault(int node, int pages, bool migrated)
1442 {
1443 }
1444 static inline void set_numabalancing_state(bool enabled)
1445 {
1446 }
1447 #endif
1448
1449 static inline struct pid *task_pid(struct task_struct *task)
1450 {
1451 return task->pids[PIDTYPE_PID].pid;
1452 }
1453
1454 static inline struct pid *task_tgid(struct task_struct *task)
1455 {
1456 return task->group_leader->pids[PIDTYPE_PID].pid;
1457 }
1458
1459 /*
1460 * Without tasklist or rcu lock it is not safe to dereference
1461 * the result of task_pgrp/task_session even if task == current,
1462 * we can race with another thread doing sys_setsid/sys_setpgid.
1463 */
1464 static inline struct pid *task_pgrp(struct task_struct *task)
1465 {
1466 return task->group_leader->pids[PIDTYPE_PGID].pid;
1467 }
1468
1469 static inline struct pid *task_session(struct task_struct *task)
1470 {
1471 return task->group_leader->pids[PIDTYPE_SID].pid;
1472 }
1473
1474 struct pid_namespace;
1475
1476 /*
1477 * the helpers to get the task's different pids as they are seen
1478 * from various namespaces
1479 *
1480 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1481 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1482 * current.
1483 * task_xid_nr_ns() : id seen from the ns specified;
1484 *
1485 * set_task_vxid() : assigns a virtual id to a task;
1486 *
1487 * see also pid_nr() etc in include/linux/pid.h
1488 */
1489 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1490 struct pid_namespace *ns);
1491
1492 static inline pid_t task_pid_nr(struct task_struct *tsk)
1493 {
1494 return tsk->pid;
1495 }
1496
1497 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1498 struct pid_namespace *ns)
1499 {
1500 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1501 }
1502
1503 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1504 {
1505 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1506 }
1507
1508
1509 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1510 {
1511 return tsk->tgid;
1512 }
1513
1514 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1515
1516 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1517 {
1518 return pid_vnr(task_tgid(tsk));
1519 }
1520
1521
1522 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1523 struct pid_namespace *ns)
1524 {
1525 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1526 }
1527
1528 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1529 {
1530 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1531 }
1532
1533
1534 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1535 struct pid_namespace *ns)
1536 {
1537 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1538 }
1539
1540 static inline pid_t task_session_vnr(struct task_struct *tsk)
1541 {
1542 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1543 }
1544
1545 /* obsolete, do not use */
1546 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1547 {
1548 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1549 }
1550
1551 /**
1552 * pid_alive - check that a task structure is not stale
1553 * @p: Task structure to be checked.
1554 *
1555 * Test if a process is not yet dead (at most zombie state)
1556 * If pid_alive fails, then pointers within the task structure
1557 * can be stale and must not be dereferenced.
1558 */
1559 static inline int pid_alive(struct task_struct *p)
1560 {
1561 return p->pids[PIDTYPE_PID].pid != NULL;
1562 }
1563
1564 /**
1565 * is_global_init - check if a task structure is init
1566 * @tsk: Task structure to be checked.
1567 *
1568 * Check if a task structure is the first user space task the kernel created.
1569 */
1570 static inline int is_global_init(struct task_struct *tsk)
1571 {
1572 return tsk->pid == 1;
1573 }
1574
1575 extern struct pid *cad_pid;
1576
1577 extern void free_task(struct task_struct *tsk);
1578 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1579
1580 extern void __put_task_struct(struct task_struct *t);
1581
1582 static inline void put_task_struct(struct task_struct *t)
1583 {
1584 if (atomic_dec_and_test(&t->usage))
1585 __put_task_struct(t);
1586 }
1587
1588 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1589 extern void task_cputime(struct task_struct *t,
1590 cputime_t *utime, cputime_t *stime);
1591 extern void task_cputime_scaled(struct task_struct *t,
1592 cputime_t *utimescaled, cputime_t *stimescaled);
1593 extern cputime_t task_gtime(struct task_struct *t);
1594 #else
1595 static inline void task_cputime(struct task_struct *t,
1596 cputime_t *utime, cputime_t *stime)
1597 {
1598 if (utime)
1599 *utime = t->utime;
1600 if (stime)
1601 *stime = t->stime;
1602 }
1603
1604 static inline void task_cputime_scaled(struct task_struct *t,
1605 cputime_t *utimescaled,
1606 cputime_t *stimescaled)
1607 {
1608 if (utimescaled)
1609 *utimescaled = t->utimescaled;
1610 if (stimescaled)
1611 *stimescaled = t->stimescaled;
1612 }
1613
1614 static inline cputime_t task_gtime(struct task_struct *t)
1615 {
1616 return t->gtime;
1617 }
1618 #endif
1619 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1620 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1621
1622 /*
1623 * Per process flags
1624 */
1625 #define PF_EXITING 0x00000004 /* getting shut down */
1626 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1627 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1628 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1629 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1630 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1631 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1632 #define PF_DUMPCORE 0x00000200 /* dumped core */
1633 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1634 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1635 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1636 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1637 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1638 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1639 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1640 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1641 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1642 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1643 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1644 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1645 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1646 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1647 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1648 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1649 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1650 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1651 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1652 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1653 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1654
1655 /*
1656 * Only the _current_ task can read/write to tsk->flags, but other
1657 * tasks can access tsk->flags in readonly mode for example
1658 * with tsk_used_math (like during threaded core dumping).
1659 * There is however an exception to this rule during ptrace
1660 * or during fork: the ptracer task is allowed to write to the
1661 * child->flags of its traced child (same goes for fork, the parent
1662 * can write to the child->flags), because we're guaranteed the
1663 * child is not running and in turn not changing child->flags
1664 * at the same time the parent does it.
1665 */
1666 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1667 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1668 #define clear_used_math() clear_stopped_child_used_math(current)
1669 #define set_used_math() set_stopped_child_used_math(current)
1670 #define conditional_stopped_child_used_math(condition, child) \
1671 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1672 #define conditional_used_math(condition) \
1673 conditional_stopped_child_used_math(condition, current)
1674 #define copy_to_stopped_child_used_math(child) \
1675 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1676 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1677 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1678 #define used_math() tsk_used_math(current)
1679
1680 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
1681 * __GFP_FS is also cleared as it implies __GFP_IO.
1682 */
1683 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1684 {
1685 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1686 flags &= ~(__GFP_IO | __GFP_FS);
1687 return flags;
1688 }
1689
1690 static inline unsigned int memalloc_noio_save(void)
1691 {
1692 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1693 current->flags |= PF_MEMALLOC_NOIO;
1694 return flags;
1695 }
1696
1697 static inline void memalloc_noio_restore(unsigned int flags)
1698 {
1699 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1700 }
1701
1702 /*
1703 * task->jobctl flags
1704 */
1705 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1706
1707 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1708 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1709 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1710 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1711 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1712 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1713 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1714
1715 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1716 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1717 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1718 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1719 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1720 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1721 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1722
1723 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1724 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1725
1726 extern bool task_set_jobctl_pending(struct task_struct *task,
1727 unsigned int mask);
1728 extern void task_clear_jobctl_trapping(struct task_struct *task);
1729 extern void task_clear_jobctl_pending(struct task_struct *task,
1730 unsigned int mask);
1731
1732 #ifdef CONFIG_PREEMPT_RCU
1733
1734 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1735 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1736
1737 static inline void rcu_copy_process(struct task_struct *p)
1738 {
1739 p->rcu_read_lock_nesting = 0;
1740 p->rcu_read_unlock_special = 0;
1741 #ifdef CONFIG_TREE_PREEMPT_RCU
1742 p->rcu_blocked_node = NULL;
1743 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1744 #ifdef CONFIG_RCU_BOOST
1745 p->rcu_boost_mutex = NULL;
1746 #endif /* #ifdef CONFIG_RCU_BOOST */
1747 INIT_LIST_HEAD(&p->rcu_node_entry);
1748 }
1749
1750 #else
1751
1752 static inline void rcu_copy_process(struct task_struct *p)
1753 {
1754 }
1755
1756 #endif
1757
1758 static inline void tsk_restore_flags(struct task_struct *task,
1759 unsigned long orig_flags, unsigned long flags)
1760 {
1761 task->flags &= ~flags;
1762 task->flags |= orig_flags & flags;
1763 }
1764
1765 #ifdef CONFIG_SMP
1766 extern void do_set_cpus_allowed(struct task_struct *p,
1767 const struct cpumask *new_mask);
1768
1769 extern int set_cpus_allowed_ptr(struct task_struct *p,
1770 const struct cpumask *new_mask);
1771 #else
1772 static inline void do_set_cpus_allowed(struct task_struct *p,
1773 const struct cpumask *new_mask)
1774 {
1775 }
1776 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1777 const struct cpumask *new_mask)
1778 {
1779 if (!cpumask_test_cpu(0, new_mask))
1780 return -EINVAL;
1781 return 0;
1782 }
1783 #endif
1784
1785 #ifdef CONFIG_NO_HZ_COMMON
1786 void calc_load_enter_idle(void);
1787 void calc_load_exit_idle(void);
1788 #else
1789 static inline void calc_load_enter_idle(void) { }
1790 static inline void calc_load_exit_idle(void) { }
1791 #endif /* CONFIG_NO_HZ_COMMON */
1792
1793 #ifndef CONFIG_CPUMASK_OFFSTACK
1794 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1795 {
1796 return set_cpus_allowed_ptr(p, &new_mask);
1797 }
1798 #endif
1799
1800 /*
1801 * Do not use outside of architecture code which knows its limitations.
1802 *
1803 * sched_clock() has no promise of monotonicity or bounded drift between
1804 * CPUs, use (which you should not) requires disabling IRQs.
1805 *
1806 * Please use one of the three interfaces below.
1807 */
1808 extern unsigned long long notrace sched_clock(void);
1809 /*
1810 * See the comment in kernel/sched/clock.c
1811 */
1812 extern u64 cpu_clock(int cpu);
1813 extern u64 local_clock(void);
1814 extern u64 sched_clock_cpu(int cpu);
1815
1816
1817 extern void sched_clock_init(void);
1818
1819 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1820 static inline void sched_clock_tick(void)
1821 {
1822 }
1823
1824 static inline void sched_clock_idle_sleep_event(void)
1825 {
1826 }
1827
1828 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1829 {
1830 }
1831 #else
1832 /*
1833 * Architectures can set this to 1 if they have specified
1834 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1835 * but then during bootup it turns out that sched_clock()
1836 * is reliable after all:
1837 */
1838 extern int sched_clock_stable;
1839
1840 extern void sched_clock_tick(void);
1841 extern void sched_clock_idle_sleep_event(void);
1842 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1843 #endif
1844
1845 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1846 /*
1847 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1848 * The reason for this explicit opt-in is not to have perf penalty with
1849 * slow sched_clocks.
1850 */
1851 extern void enable_sched_clock_irqtime(void);
1852 extern void disable_sched_clock_irqtime(void);
1853 #else
1854 static inline void enable_sched_clock_irqtime(void) {}
1855 static inline void disable_sched_clock_irqtime(void) {}
1856 #endif
1857
1858 extern unsigned long long
1859 task_sched_runtime(struct task_struct *task);
1860
1861 /* sched_exec is called by processes performing an exec */
1862 #ifdef CONFIG_SMP
1863 extern void sched_exec(void);
1864 #else
1865 #define sched_exec() {}
1866 #endif
1867
1868 extern void sched_clock_idle_sleep_event(void);
1869 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1870
1871 #ifdef CONFIG_HOTPLUG_CPU
1872 extern void idle_task_exit(void);
1873 #else
1874 static inline void idle_task_exit(void) {}
1875 #endif
1876
1877 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1878 extern void wake_up_nohz_cpu(int cpu);
1879 #else
1880 static inline void wake_up_nohz_cpu(int cpu) { }
1881 #endif
1882
1883 #ifdef CONFIG_NO_HZ_FULL
1884 extern bool sched_can_stop_tick(void);
1885 extern u64 scheduler_tick_max_deferment(void);
1886 #else
1887 static inline bool sched_can_stop_tick(void) { return false; }
1888 #endif
1889
1890 #ifdef CONFIG_SCHED_AUTOGROUP
1891 extern void sched_autogroup_create_attach(struct task_struct *p);
1892 extern void sched_autogroup_detach(struct task_struct *p);
1893 extern void sched_autogroup_fork(struct signal_struct *sig);
1894 extern void sched_autogroup_exit(struct signal_struct *sig);
1895 #ifdef CONFIG_PROC_FS
1896 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1897 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
1898 #endif
1899 #else
1900 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1901 static inline void sched_autogroup_detach(struct task_struct *p) { }
1902 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1903 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1904 #endif
1905
1906 extern bool yield_to(struct task_struct *p, bool preempt);
1907 extern void set_user_nice(struct task_struct *p, long nice);
1908 extern int task_prio(const struct task_struct *p);
1909 extern int task_nice(const struct task_struct *p);
1910 extern int can_nice(const struct task_struct *p, const int nice);
1911 extern int task_curr(const struct task_struct *p);
1912 extern int idle_cpu(int cpu);
1913 extern int sched_setscheduler(struct task_struct *, int,
1914 const struct sched_param *);
1915 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1916 const struct sched_param *);
1917 extern struct task_struct *idle_task(int cpu);
1918 /**
1919 * is_idle_task - is the specified task an idle task?
1920 * @p: the task in question.
1921 */
1922 static inline bool is_idle_task(const struct task_struct *p)
1923 {
1924 return p->pid == 0;
1925 }
1926 extern struct task_struct *curr_task(int cpu);
1927 extern void set_curr_task(int cpu, struct task_struct *p);
1928
1929 void yield(void);
1930
1931 /*
1932 * The default (Linux) execution domain.
1933 */
1934 extern struct exec_domain default_exec_domain;
1935
1936 union thread_union {
1937 struct thread_info thread_info;
1938 unsigned long stack[THREAD_SIZE/sizeof(long)];
1939 };
1940
1941 #ifndef __HAVE_ARCH_KSTACK_END
1942 static inline int kstack_end(void *addr)
1943 {
1944 /* Reliable end of stack detection:
1945 * Some APM bios versions misalign the stack
1946 */
1947 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1948 }
1949 #endif
1950
1951 extern union thread_union init_thread_union;
1952 extern struct task_struct init_task;
1953
1954 extern struct mm_struct init_mm;
1955
1956 extern struct pid_namespace init_pid_ns;
1957
1958 /*
1959 * find a task by one of its numerical ids
1960 *
1961 * find_task_by_pid_ns():
1962 * finds a task by its pid in the specified namespace
1963 * find_task_by_vpid():
1964 * finds a task by its virtual pid
1965 *
1966 * see also find_vpid() etc in include/linux/pid.h
1967 */
1968
1969 extern struct task_struct *find_task_by_vpid(pid_t nr);
1970 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1971 struct pid_namespace *ns);
1972
1973 extern void __set_special_pids(struct pid *pid);
1974
1975 /* per-UID process charging. */
1976 extern struct user_struct * alloc_uid(kuid_t);
1977 static inline struct user_struct *get_uid(struct user_struct *u)
1978 {
1979 atomic_inc(&u->__count);
1980 return u;
1981 }
1982 extern void free_uid(struct user_struct *);
1983
1984 #include <asm/current.h>
1985
1986 extern void xtime_update(unsigned long ticks);
1987
1988 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1989 extern int wake_up_process(struct task_struct *tsk);
1990 extern void wake_up_new_task(struct task_struct *tsk);
1991 #ifdef CONFIG_SMP
1992 extern void kick_process(struct task_struct *tsk);
1993 #else
1994 static inline void kick_process(struct task_struct *tsk) { }
1995 #endif
1996 extern void sched_fork(struct task_struct *p);
1997 extern void sched_dead(struct task_struct *p);
1998
1999 extern void proc_caches_init(void);
2000 extern void flush_signals(struct task_struct *);
2001 extern void __flush_signals(struct task_struct *);
2002 extern void ignore_signals(struct task_struct *);
2003 extern void flush_signal_handlers(struct task_struct *, int force_default);
2004 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2005
2006 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2007 {
2008 unsigned long flags;
2009 int ret;
2010
2011 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2012 ret = dequeue_signal(tsk, mask, info);
2013 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2014
2015 return ret;
2016 }
2017
2018 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2019 sigset_t *mask);
2020 extern void unblock_all_signals(void);
2021 extern void release_task(struct task_struct * p);
2022 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2023 extern int force_sigsegv(int, struct task_struct *);
2024 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2025 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2026 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2027 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2028 const struct cred *, u32);
2029 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2030 extern int kill_pid(struct pid *pid, int sig, int priv);
2031 extern int kill_proc_info(int, struct siginfo *, pid_t);
2032 extern __must_check bool do_notify_parent(struct task_struct *, int);
2033 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2034 extern void force_sig(int, struct task_struct *);
2035 extern int send_sig(int, struct task_struct *, int);
2036 extern int zap_other_threads(struct task_struct *p);
2037 extern struct sigqueue *sigqueue_alloc(void);
2038 extern void sigqueue_free(struct sigqueue *);
2039 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2040 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2041
2042 static inline void restore_saved_sigmask(void)
2043 {
2044 if (test_and_clear_restore_sigmask())
2045 __set_current_blocked(&current->saved_sigmask);
2046 }
2047
2048 static inline sigset_t *sigmask_to_save(void)
2049 {
2050 sigset_t *res = &current->blocked;
2051 if (unlikely(test_restore_sigmask()))
2052 res = &current->saved_sigmask;
2053 return res;
2054 }
2055
2056 static inline int kill_cad_pid(int sig, int priv)
2057 {
2058 return kill_pid(cad_pid, sig, priv);
2059 }
2060
2061 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2062 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2063 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2064 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2065
2066 /*
2067 * True if we are on the alternate signal stack.
2068 */
2069 static inline int on_sig_stack(unsigned long sp)
2070 {
2071 #ifdef CONFIG_STACK_GROWSUP
2072 return sp >= current->sas_ss_sp &&
2073 sp - current->sas_ss_sp < current->sas_ss_size;
2074 #else
2075 return sp > current->sas_ss_sp &&
2076 sp - current->sas_ss_sp <= current->sas_ss_size;
2077 #endif
2078 }
2079
2080 static inline int sas_ss_flags(unsigned long sp)
2081 {
2082 return (current->sas_ss_size == 0 ? SS_DISABLE
2083 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2084 }
2085
2086 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2087 {
2088 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2089 #ifdef CONFIG_STACK_GROWSUP
2090 return current->sas_ss_sp;
2091 #else
2092 return current->sas_ss_sp + current->sas_ss_size;
2093 #endif
2094 return sp;
2095 }
2096
2097 /*
2098 * Routines for handling mm_structs
2099 */
2100 extern struct mm_struct * mm_alloc(void);
2101
2102 /* mmdrop drops the mm and the page tables */
2103 extern void __mmdrop(struct mm_struct *);
2104 static inline void mmdrop(struct mm_struct * mm)
2105 {
2106 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2107 __mmdrop(mm);
2108 }
2109
2110 /* mmput gets rid of the mappings and all user-space */
2111 extern void mmput(struct mm_struct *);
2112 /* Grab a reference to a task's mm, if it is not already going away */
2113 extern struct mm_struct *get_task_mm(struct task_struct *task);
2114 /*
2115 * Grab a reference to a task's mm, if it is not already going away
2116 * and ptrace_may_access with the mode parameter passed to it
2117 * succeeds.
2118 */
2119 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2120 /* Remove the current tasks stale references to the old mm_struct */
2121 extern void mm_release(struct task_struct *, struct mm_struct *);
2122 /* Allocate a new mm structure and copy contents from tsk->mm */
2123 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2124
2125 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2126 struct task_struct *);
2127 extern void flush_thread(void);
2128 extern void exit_thread(void);
2129
2130 extern void exit_files(struct task_struct *);
2131 extern void __cleanup_sighand(struct sighand_struct *);
2132
2133 extern void exit_itimers(struct signal_struct *);
2134 extern void flush_itimer_signals(void);
2135
2136 extern void do_group_exit(int);
2137
2138 extern int allow_signal(int);
2139 extern int disallow_signal(int);
2140
2141 extern int do_execve(const char *,
2142 const char __user * const __user *,
2143 const char __user * const __user *);
2144 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2145 struct task_struct *fork_idle(int);
2146 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2147
2148 extern void set_task_comm(struct task_struct *tsk, char *from);
2149 extern char *get_task_comm(char *to, struct task_struct *tsk);
2150
2151 #ifdef CONFIG_SMP
2152 void scheduler_ipi(void);
2153 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2154 #else
2155 static inline void scheduler_ipi(void) { }
2156 static inline unsigned long wait_task_inactive(struct task_struct *p,
2157 long match_state)
2158 {
2159 return 1;
2160 }
2161 #endif
2162
2163 #define next_task(p) \
2164 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2165
2166 #define for_each_process(p) \
2167 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2168
2169 extern bool current_is_single_threaded(void);
2170
2171 /*
2172 * Careful: do_each_thread/while_each_thread is a double loop so
2173 * 'break' will not work as expected - use goto instead.
2174 */
2175 #define do_each_thread(g, t) \
2176 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2177
2178 #define while_each_thread(g, t) \
2179 while ((t = next_thread(t)) != g)
2180
2181 #define __for_each_thread(signal, t) \
2182 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2183
2184 #define for_each_thread(p, t) \
2185 __for_each_thread((p)->signal, t)
2186
2187 /* Careful: this is a double loop, 'break' won't work as expected. */
2188 #define for_each_process_thread(p, t) \
2189 for_each_process(p) for_each_thread(p, t)
2190
2191 static inline int get_nr_threads(struct task_struct *tsk)
2192 {
2193 return tsk->signal->nr_threads;
2194 }
2195
2196 static inline bool thread_group_leader(struct task_struct *p)
2197 {
2198 return p->exit_signal >= 0;
2199 }
2200
2201 /* Do to the insanities of de_thread it is possible for a process
2202 * to have the pid of the thread group leader without actually being
2203 * the thread group leader. For iteration through the pids in proc
2204 * all we care about is that we have a task with the appropriate
2205 * pid, we don't actually care if we have the right task.
2206 */
2207 static inline bool has_group_leader_pid(struct task_struct *p)
2208 {
2209 return task_pid(p) == p->signal->leader_pid;
2210 }
2211
2212 static inline
2213 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2214 {
2215 return p1->signal == p2->signal;
2216 }
2217
2218 static inline struct task_struct *next_thread(const struct task_struct *p)
2219 {
2220 return list_entry_rcu(p->thread_group.next,
2221 struct task_struct, thread_group);
2222 }
2223
2224 static inline int thread_group_empty(struct task_struct *p)
2225 {
2226 return list_empty(&p->thread_group);
2227 }
2228
2229 #define delay_group_leader(p) \
2230 (thread_group_leader(p) && !thread_group_empty(p))
2231
2232 /*
2233 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2234 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2235 * pins the final release of task.io_context. Also protects ->cpuset and
2236 * ->cgroup.subsys[]. And ->vfork_done.
2237 *
2238 * Nests both inside and outside of read_lock(&tasklist_lock).
2239 * It must not be nested with write_lock_irq(&tasklist_lock),
2240 * neither inside nor outside.
2241 */
2242 static inline void task_lock(struct task_struct *p)
2243 {
2244 spin_lock(&p->alloc_lock);
2245 }
2246
2247 static inline void task_unlock(struct task_struct *p)
2248 {
2249 spin_unlock(&p->alloc_lock);
2250 }
2251
2252 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2253 unsigned long *flags);
2254
2255 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2256 unsigned long *flags)
2257 {
2258 struct sighand_struct *ret;
2259
2260 ret = __lock_task_sighand(tsk, flags);
2261 (void)__cond_lock(&tsk->sighand->siglock, ret);
2262 return ret;
2263 }
2264
2265 static inline void unlock_task_sighand(struct task_struct *tsk,
2266 unsigned long *flags)
2267 {
2268 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2269 }
2270
2271 #ifdef CONFIG_CGROUPS
2272 static inline void threadgroup_change_begin(struct task_struct *tsk)
2273 {
2274 down_read(&tsk->signal->group_rwsem);
2275 }
2276 static inline void threadgroup_change_end(struct task_struct *tsk)
2277 {
2278 up_read(&tsk->signal->group_rwsem);
2279 }
2280
2281 /**
2282 * threadgroup_lock - lock threadgroup
2283 * @tsk: member task of the threadgroup to lock
2284 *
2285 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2286 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2287 * change ->group_leader/pid. This is useful for cases where the threadgroup
2288 * needs to stay stable across blockable operations.
2289 *
2290 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2291 * synchronization. While held, no new task will be added to threadgroup
2292 * and no existing live task will have its PF_EXITING set.
2293 *
2294 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2295 * sub-thread becomes a new leader.
2296 */
2297 static inline void threadgroup_lock(struct task_struct *tsk)
2298 {
2299 down_write(&tsk->signal->group_rwsem);
2300 }
2301
2302 /**
2303 * threadgroup_unlock - unlock threadgroup
2304 * @tsk: member task of the threadgroup to unlock
2305 *
2306 * Reverse threadgroup_lock().
2307 */
2308 static inline void threadgroup_unlock(struct task_struct *tsk)
2309 {
2310 up_write(&tsk->signal->group_rwsem);
2311 }
2312 #else
2313 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2314 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2315 static inline void threadgroup_lock(struct task_struct *tsk) {}
2316 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2317 #endif
2318
2319 #ifndef __HAVE_THREAD_FUNCTIONS
2320
2321 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2322 #define task_stack_page(task) ((task)->stack)
2323
2324 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2325 {
2326 *task_thread_info(p) = *task_thread_info(org);
2327 task_thread_info(p)->task = p;
2328 }
2329
2330 static inline unsigned long *end_of_stack(struct task_struct *p)
2331 {
2332 return (unsigned long *)(task_thread_info(p) + 1);
2333 }
2334
2335 #endif
2336
2337 static inline int object_is_on_stack(void *obj)
2338 {
2339 void *stack = task_stack_page(current);
2340
2341 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2342 }
2343
2344 extern void thread_info_cache_init(void);
2345
2346 #ifdef CONFIG_DEBUG_STACK_USAGE
2347 static inline unsigned long stack_not_used(struct task_struct *p)
2348 {
2349 unsigned long *n = end_of_stack(p);
2350
2351 do { /* Skip over canary */
2352 n++;
2353 } while (!*n);
2354
2355 return (unsigned long)n - (unsigned long)end_of_stack(p);
2356 }
2357 #endif
2358
2359 /* set thread flags in other task's structures
2360 * - see asm/thread_info.h for TIF_xxxx flags available
2361 */
2362 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2363 {
2364 set_ti_thread_flag(task_thread_info(tsk), flag);
2365 }
2366
2367 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2368 {
2369 clear_ti_thread_flag(task_thread_info(tsk), flag);
2370 }
2371
2372 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2373 {
2374 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2375 }
2376
2377 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2378 {
2379 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2380 }
2381
2382 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2383 {
2384 return test_ti_thread_flag(task_thread_info(tsk), flag);
2385 }
2386
2387 static inline void set_tsk_need_resched(struct task_struct *tsk)
2388 {
2389 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2390 }
2391
2392 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2393 {
2394 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2395 }
2396
2397 static inline int test_tsk_need_resched(struct task_struct *tsk)
2398 {
2399 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2400 }
2401
2402 static inline int restart_syscall(void)
2403 {
2404 set_tsk_thread_flag(current, TIF_SIGPENDING);
2405 return -ERESTARTNOINTR;
2406 }
2407
2408 static inline int signal_pending(struct task_struct *p)
2409 {
2410 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2411 }
2412
2413 static inline int __fatal_signal_pending(struct task_struct *p)
2414 {
2415 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2416 }
2417
2418 static inline int fatal_signal_pending(struct task_struct *p)
2419 {
2420 return signal_pending(p) && __fatal_signal_pending(p);
2421 }
2422
2423 static inline int signal_pending_state(long state, struct task_struct *p)
2424 {
2425 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2426 return 0;
2427 if (!signal_pending(p))
2428 return 0;
2429
2430 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2431 }
2432
2433 static inline int need_resched(void)
2434 {
2435 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2436 }
2437
2438 /*
2439 * cond_resched() and cond_resched_lock(): latency reduction via
2440 * explicit rescheduling in places that are safe. The return
2441 * value indicates whether a reschedule was done in fact.
2442 * cond_resched_lock() will drop the spinlock before scheduling,
2443 * cond_resched_softirq() will enable bhs before scheduling.
2444 */
2445 extern int _cond_resched(void);
2446
2447 #define cond_resched() ({ \
2448 __might_sleep(__FILE__, __LINE__, 0); \
2449 _cond_resched(); \
2450 })
2451
2452 extern int __cond_resched_lock(spinlock_t *lock);
2453
2454 #ifdef CONFIG_PREEMPT_COUNT
2455 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2456 #else
2457 #define PREEMPT_LOCK_OFFSET 0
2458 #endif
2459
2460 #define cond_resched_lock(lock) ({ \
2461 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2462 __cond_resched_lock(lock); \
2463 })
2464
2465 extern int __cond_resched_softirq(void);
2466
2467 #define cond_resched_softirq() ({ \
2468 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2469 __cond_resched_softirq(); \
2470 })
2471
2472 /*
2473 * Does a critical section need to be broken due to another
2474 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2475 * but a general need for low latency)
2476 */
2477 static inline int spin_needbreak(spinlock_t *lock)
2478 {
2479 #ifdef CONFIG_PREEMPT
2480 return spin_is_contended(lock);
2481 #else
2482 return 0;
2483 #endif
2484 }
2485
2486 /*
2487 * Idle thread specific functions to determine the need_resched
2488 * polling state. We have two versions, one based on TS_POLLING in
2489 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2490 * thread_info.flags
2491 */
2492 #ifdef TS_POLLING
2493 static inline int tsk_is_polling(struct task_struct *p)
2494 {
2495 return task_thread_info(p)->status & TS_POLLING;
2496 }
2497 static inline void __current_set_polling(void)
2498 {
2499 current_thread_info()->status |= TS_POLLING;
2500 }
2501
2502 static inline bool __must_check current_set_polling_and_test(void)
2503 {
2504 __current_set_polling();
2505
2506 /*
2507 * Polling state must be visible before we test NEED_RESCHED,
2508 * paired by resched_task()
2509 */
2510 smp_mb();
2511
2512 return unlikely(tif_need_resched());
2513 }
2514
2515 static inline void __current_clr_polling(void)
2516 {
2517 current_thread_info()->status &= ~TS_POLLING;
2518 }
2519
2520 static inline bool __must_check current_clr_polling_and_test(void)
2521 {
2522 __current_clr_polling();
2523
2524 /*
2525 * Polling state must be visible before we test NEED_RESCHED,
2526 * paired by resched_task()
2527 */
2528 smp_mb();
2529
2530 return unlikely(tif_need_resched());
2531 }
2532 #elif defined(TIF_POLLING_NRFLAG)
2533 static inline int tsk_is_polling(struct task_struct *p)
2534 {
2535 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2536 }
2537
2538 static inline void __current_set_polling(void)
2539 {
2540 set_thread_flag(TIF_POLLING_NRFLAG);
2541 }
2542
2543 static inline bool __must_check current_set_polling_and_test(void)
2544 {
2545 __current_set_polling();
2546
2547 /*
2548 * Polling state must be visible before we test NEED_RESCHED,
2549 * paired by resched_task()
2550 *
2551 * XXX: assumes set/clear bit are identical barrier wise.
2552 */
2553 smp_mb__after_clear_bit();
2554
2555 return unlikely(tif_need_resched());
2556 }
2557
2558 static inline void __current_clr_polling(void)
2559 {
2560 clear_thread_flag(TIF_POLLING_NRFLAG);
2561 }
2562
2563 static inline bool __must_check current_clr_polling_and_test(void)
2564 {
2565 __current_clr_polling();
2566
2567 /*
2568 * Polling state must be visible before we test NEED_RESCHED,
2569 * paired by resched_task()
2570 */
2571 smp_mb__after_clear_bit();
2572
2573 return unlikely(tif_need_resched());
2574 }
2575
2576 #else
2577 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2578 static inline void __current_set_polling(void) { }
2579 static inline void __current_clr_polling(void) { }
2580
2581 static inline bool __must_check current_set_polling_and_test(void)
2582 {
2583 return unlikely(tif_need_resched());
2584 }
2585 static inline bool __must_check current_clr_polling_and_test(void)
2586 {
2587 return unlikely(tif_need_resched());
2588 }
2589 #endif
2590
2591 /*
2592 * Thread group CPU time accounting.
2593 */
2594 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2595 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2596
2597 static inline void thread_group_cputime_init(struct signal_struct *sig)
2598 {
2599 raw_spin_lock_init(&sig->cputimer.lock);
2600 }
2601
2602 /*
2603 * Reevaluate whether the task has signals pending delivery.
2604 * Wake the task if so.
2605 * This is required every time the blocked sigset_t changes.
2606 * callers must hold sighand->siglock.
2607 */
2608 extern void recalc_sigpending_and_wake(struct task_struct *t);
2609 extern void recalc_sigpending(void);
2610
2611 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2612
2613 static inline void signal_wake_up(struct task_struct *t, bool resume)
2614 {
2615 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2616 }
2617 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2618 {
2619 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2620 }
2621
2622 /*
2623 * Wrappers for p->thread_info->cpu access. No-op on UP.
2624 */
2625 #ifdef CONFIG_SMP
2626
2627 static inline unsigned int task_cpu(const struct task_struct *p)
2628 {
2629 return task_thread_info(p)->cpu;
2630 }
2631
2632 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2633
2634 #else
2635
2636 static inline unsigned int task_cpu(const struct task_struct *p)
2637 {
2638 return 0;
2639 }
2640
2641 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2642 {
2643 }
2644
2645 #endif /* CONFIG_SMP */
2646
2647 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2648 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2649
2650 #ifdef CONFIG_CGROUP_SCHED
2651 extern struct task_group root_task_group;
2652 #endif /* CONFIG_CGROUP_SCHED */
2653
2654 extern int task_can_switch_user(struct user_struct *up,
2655 struct task_struct *tsk);
2656
2657 #ifdef CONFIG_TASK_XACCT
2658 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2659 {
2660 tsk->ioac.rchar += amt;
2661 }
2662
2663 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2664 {
2665 tsk->ioac.wchar += amt;
2666 }
2667
2668 static inline void inc_syscr(struct task_struct *tsk)
2669 {
2670 tsk->ioac.syscr++;
2671 }
2672
2673 static inline void inc_syscw(struct task_struct *tsk)
2674 {
2675 tsk->ioac.syscw++;
2676 }
2677 #else
2678 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2679 {
2680 }
2681
2682 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2683 {
2684 }
2685
2686 static inline void inc_syscr(struct task_struct *tsk)
2687 {
2688 }
2689
2690 static inline void inc_syscw(struct task_struct *tsk)
2691 {
2692 }
2693 #endif
2694
2695 #ifndef TASK_SIZE_OF
2696 #define TASK_SIZE_OF(tsk) TASK_SIZE
2697 #endif
2698
2699 #ifdef CONFIG_MM_OWNER
2700 extern void mm_update_next_owner(struct mm_struct *mm);
2701 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2702 #else
2703 static inline void mm_update_next_owner(struct mm_struct *mm)
2704 {
2705 }
2706
2707 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2708 {
2709 }
2710 #endif /* CONFIG_MM_OWNER */
2711
2712 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2713 unsigned int limit)
2714 {
2715 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2716 }
2717
2718 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2719 unsigned int limit)
2720 {
2721 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2722 }
2723
2724 static inline unsigned long rlimit(unsigned int limit)
2725 {
2726 return task_rlimit(current, limit);
2727 }
2728
2729 static inline unsigned long rlimit_max(unsigned int limit)
2730 {
2731 return task_rlimit_max(current, limit);
2732 }
2733
2734 #endif