cputime: Safely read cputime of full dynticks CPUs
[deliverable/linux.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
55 #include <asm/processor.h>
56
57 struct exec_domain;
58 struct futex_pi_state;
59 struct robust_list_head;
60 struct bio_list;
61 struct fs_struct;
62 struct perf_event_context;
63 struct blk_plug;
64
65 /*
66 * List of flags we want to share for kernel threads,
67 * if only because they are not used by them anyway.
68 */
69 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
70
71 /*
72 * These are the constant used to fake the fixed-point load-average
73 * counting. Some notes:
74 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
75 * a load-average precision of 10 bits integer + 11 bits fractional
76 * - if you want to count load-averages more often, you need more
77 * precision, or rounding will get you. With 2-second counting freq,
78 * the EXP_n values would be 1981, 2034 and 2043 if still using only
79 * 11 bit fractions.
80 */
81 extern unsigned long avenrun[]; /* Load averages */
82 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
83
84 #define FSHIFT 11 /* nr of bits of precision */
85 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
86 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
87 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
88 #define EXP_5 2014 /* 1/exp(5sec/5min) */
89 #define EXP_15 2037 /* 1/exp(5sec/15min) */
90
91 #define CALC_LOAD(load,exp,n) \
92 load *= exp; \
93 load += n*(FIXED_1-exp); \
94 load >>= FSHIFT;
95
96 extern unsigned long total_forks;
97 extern int nr_threads;
98 DECLARE_PER_CPU(unsigned long, process_counts);
99 extern int nr_processes(void);
100 extern unsigned long nr_running(void);
101 extern unsigned long nr_uninterruptible(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 #else
131 static inline void
132 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
133 {
134 }
135 static inline void proc_sched_set_task(struct task_struct *p)
136 {
137 }
138 static inline void
139 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
140 {
141 }
142 #endif
143
144 /*
145 * Task state bitmask. NOTE! These bits are also
146 * encoded in fs/proc/array.c: get_task_state().
147 *
148 * We have two separate sets of flags: task->state
149 * is about runnability, while task->exit_state are
150 * about the task exiting. Confusing, but this way
151 * modifying one set can't modify the other one by
152 * mistake.
153 */
154 #define TASK_RUNNING 0
155 #define TASK_INTERRUPTIBLE 1
156 #define TASK_UNINTERRUPTIBLE 2
157 #define __TASK_STOPPED 4
158 #define __TASK_TRACED 8
159 /* in tsk->exit_state */
160 #define EXIT_ZOMBIE 16
161 #define EXIT_DEAD 32
162 /* in tsk->state again */
163 #define TASK_DEAD 64
164 #define TASK_WAKEKILL 128
165 #define TASK_WAKING 256
166 #define TASK_STATE_MAX 512
167
168 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
169
170 extern char ___assert_task_state[1 - 2*!!(
171 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
172
173 /* Convenience macros for the sake of set_task_state */
174 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
175 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
176 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
177
178 /* Convenience macros for the sake of wake_up */
179 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
180 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
181
182 /* get_task_state() */
183 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
184 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
185 __TASK_TRACED)
186
187 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
188 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
189 #define task_is_dead(task) ((task)->exit_state != 0)
190 #define task_is_stopped_or_traced(task) \
191 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
192 #define task_contributes_to_load(task) \
193 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
194 (task->flags & PF_FROZEN) == 0)
195
196 #define __set_task_state(tsk, state_value) \
197 do { (tsk)->state = (state_value); } while (0)
198 #define set_task_state(tsk, state_value) \
199 set_mb((tsk)->state, (state_value))
200
201 /*
202 * set_current_state() includes a barrier so that the write of current->state
203 * is correctly serialised wrt the caller's subsequent test of whether to
204 * actually sleep:
205 *
206 * set_current_state(TASK_UNINTERRUPTIBLE);
207 * if (do_i_need_to_sleep())
208 * schedule();
209 *
210 * If the caller does not need such serialisation then use __set_current_state()
211 */
212 #define __set_current_state(state_value) \
213 do { current->state = (state_value); } while (0)
214 #define set_current_state(state_value) \
215 set_mb(current->state, (state_value))
216
217 /* Task command name length */
218 #define TASK_COMM_LEN 16
219
220 #include <linux/spinlock.h>
221
222 /*
223 * This serializes "schedule()" and also protects
224 * the run-queue from deletions/modifications (but
225 * _adding_ to the beginning of the run-queue has
226 * a separate lock).
227 */
228 extern rwlock_t tasklist_lock;
229 extern spinlock_t mmlist_lock;
230
231 struct task_struct;
232
233 #ifdef CONFIG_PROVE_RCU
234 extern int lockdep_tasklist_lock_is_held(void);
235 #endif /* #ifdef CONFIG_PROVE_RCU */
236
237 extern void sched_init(void);
238 extern void sched_init_smp(void);
239 extern asmlinkage void schedule_tail(struct task_struct *prev);
240 extern void init_idle(struct task_struct *idle, int cpu);
241 extern void init_idle_bootup_task(struct task_struct *idle);
242
243 extern int runqueue_is_locked(int cpu);
244
245 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
246 extern void nohz_balance_enter_idle(int cpu);
247 extern void set_cpu_sd_state_idle(void);
248 extern int get_nohz_timer_target(void);
249 #else
250 static inline void nohz_balance_enter_idle(int cpu) { }
251 static inline void set_cpu_sd_state_idle(void) { }
252 #endif
253
254 /*
255 * Only dump TASK_* tasks. (0 for all tasks)
256 */
257 extern void show_state_filter(unsigned long state_filter);
258
259 static inline void show_state(void)
260 {
261 show_state_filter(0);
262 }
263
264 extern void show_regs(struct pt_regs *);
265
266 /*
267 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
268 * task), SP is the stack pointer of the first frame that should be shown in the back
269 * trace (or NULL if the entire call-chain of the task should be shown).
270 */
271 extern void show_stack(struct task_struct *task, unsigned long *sp);
272
273 void io_schedule(void);
274 long io_schedule_timeout(long timeout);
275
276 extern void cpu_init (void);
277 extern void trap_init(void);
278 extern void update_process_times(int user);
279 extern void scheduler_tick(void);
280
281 extern void sched_show_task(struct task_struct *p);
282
283 #ifdef CONFIG_LOCKUP_DETECTOR
284 extern void touch_softlockup_watchdog(void);
285 extern void touch_softlockup_watchdog_sync(void);
286 extern void touch_all_softlockup_watchdogs(void);
287 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
288 void __user *buffer,
289 size_t *lenp, loff_t *ppos);
290 extern unsigned int softlockup_panic;
291 void lockup_detector_init(void);
292 #else
293 static inline void touch_softlockup_watchdog(void)
294 {
295 }
296 static inline void touch_softlockup_watchdog_sync(void)
297 {
298 }
299 static inline void touch_all_softlockup_watchdogs(void)
300 {
301 }
302 static inline void lockup_detector_init(void)
303 {
304 }
305 #endif
306
307 #ifdef CONFIG_DETECT_HUNG_TASK
308 extern unsigned int sysctl_hung_task_panic;
309 extern unsigned long sysctl_hung_task_check_count;
310 extern unsigned long sysctl_hung_task_timeout_secs;
311 extern unsigned long sysctl_hung_task_warnings;
312 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
313 void __user *buffer,
314 size_t *lenp, loff_t *ppos);
315 #else
316 /* Avoid need for ifdefs elsewhere in the code */
317 enum { sysctl_hung_task_timeout_secs = 0 };
318 #endif
319
320 /* Attach to any functions which should be ignored in wchan output. */
321 #define __sched __attribute__((__section__(".sched.text")))
322
323 /* Linker adds these: start and end of __sched functions */
324 extern char __sched_text_start[], __sched_text_end[];
325
326 /* Is this address in the __sched functions? */
327 extern int in_sched_functions(unsigned long addr);
328
329 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
330 extern signed long schedule_timeout(signed long timeout);
331 extern signed long schedule_timeout_interruptible(signed long timeout);
332 extern signed long schedule_timeout_killable(signed long timeout);
333 extern signed long schedule_timeout_uninterruptible(signed long timeout);
334 asmlinkage void schedule(void);
335 extern void schedule_preempt_disabled(void);
336 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
337
338 struct nsproxy;
339 struct user_namespace;
340
341 /*
342 * Default maximum number of active map areas, this limits the number of vmas
343 * per mm struct. Users can overwrite this number by sysctl but there is a
344 * problem.
345 *
346 * When a program's coredump is generated as ELF format, a section is created
347 * per a vma. In ELF, the number of sections is represented in unsigned short.
348 * This means the number of sections should be smaller than 65535 at coredump.
349 * Because the kernel adds some informative sections to a image of program at
350 * generating coredump, we need some margin. The number of extra sections is
351 * 1-3 now and depends on arch. We use "5" as safe margin, here.
352 */
353 #define MAPCOUNT_ELF_CORE_MARGIN (5)
354 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
355
356 extern int sysctl_max_map_count;
357
358 #include <linux/aio.h>
359
360 #ifdef CONFIG_MMU
361 extern void arch_pick_mmap_layout(struct mm_struct *mm);
362 extern unsigned long
363 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
364 unsigned long, unsigned long);
365 extern unsigned long
366 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
367 unsigned long len, unsigned long pgoff,
368 unsigned long flags);
369 extern void arch_unmap_area(struct mm_struct *, unsigned long);
370 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
371 #else
372 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
373 #endif
374
375
376 extern void set_dumpable(struct mm_struct *mm, int value);
377 extern int get_dumpable(struct mm_struct *mm);
378
379 /* get/set_dumpable() values */
380 #define SUID_DUMPABLE_DISABLED 0
381 #define SUID_DUMPABLE_ENABLED 1
382 #define SUID_DUMPABLE_SAFE 2
383
384 /* mm flags */
385 /* dumpable bits */
386 #define MMF_DUMPABLE 0 /* core dump is permitted */
387 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
388
389 #define MMF_DUMPABLE_BITS 2
390 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
391
392 /* coredump filter bits */
393 #define MMF_DUMP_ANON_PRIVATE 2
394 #define MMF_DUMP_ANON_SHARED 3
395 #define MMF_DUMP_MAPPED_PRIVATE 4
396 #define MMF_DUMP_MAPPED_SHARED 5
397 #define MMF_DUMP_ELF_HEADERS 6
398 #define MMF_DUMP_HUGETLB_PRIVATE 7
399 #define MMF_DUMP_HUGETLB_SHARED 8
400
401 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
402 #define MMF_DUMP_FILTER_BITS 7
403 #define MMF_DUMP_FILTER_MASK \
404 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
405 #define MMF_DUMP_FILTER_DEFAULT \
406 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
407 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
408
409 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
410 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
411 #else
412 # define MMF_DUMP_MASK_DEFAULT_ELF 0
413 #endif
414 /* leave room for more dump flags */
415 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
416 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
417 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
418
419 #define MMF_HAS_UPROBES 19 /* has uprobes */
420 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
421
422 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
423
424 struct sighand_struct {
425 atomic_t count;
426 struct k_sigaction action[_NSIG];
427 spinlock_t siglock;
428 wait_queue_head_t signalfd_wqh;
429 };
430
431 struct pacct_struct {
432 int ac_flag;
433 long ac_exitcode;
434 unsigned long ac_mem;
435 cputime_t ac_utime, ac_stime;
436 unsigned long ac_minflt, ac_majflt;
437 };
438
439 struct cpu_itimer {
440 cputime_t expires;
441 cputime_t incr;
442 u32 error;
443 u32 incr_error;
444 };
445
446 /**
447 * struct cputime - snaphsot of system and user cputime
448 * @utime: time spent in user mode
449 * @stime: time spent in system mode
450 *
451 * Gathers a generic snapshot of user and system time.
452 */
453 struct cputime {
454 cputime_t utime;
455 cputime_t stime;
456 };
457
458 /**
459 * struct task_cputime - collected CPU time counts
460 * @utime: time spent in user mode, in &cputime_t units
461 * @stime: time spent in kernel mode, in &cputime_t units
462 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
463 *
464 * This is an extension of struct cputime that includes the total runtime
465 * spent by the task from the scheduler point of view.
466 *
467 * As a result, this structure groups together three kinds of CPU time
468 * that are tracked for threads and thread groups. Most things considering
469 * CPU time want to group these counts together and treat all three
470 * of them in parallel.
471 */
472 struct task_cputime {
473 cputime_t utime;
474 cputime_t stime;
475 unsigned long long sum_exec_runtime;
476 };
477 /* Alternate field names when used to cache expirations. */
478 #define prof_exp stime
479 #define virt_exp utime
480 #define sched_exp sum_exec_runtime
481
482 #define INIT_CPUTIME \
483 (struct task_cputime) { \
484 .utime = 0, \
485 .stime = 0, \
486 .sum_exec_runtime = 0, \
487 }
488
489 /*
490 * Disable preemption until the scheduler is running.
491 * Reset by start_kernel()->sched_init()->init_idle().
492 *
493 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
494 * before the scheduler is active -- see should_resched().
495 */
496 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
497
498 /**
499 * struct thread_group_cputimer - thread group interval timer counts
500 * @cputime: thread group interval timers.
501 * @running: non-zero when there are timers running and
502 * @cputime receives updates.
503 * @lock: lock for fields in this struct.
504 *
505 * This structure contains the version of task_cputime, above, that is
506 * used for thread group CPU timer calculations.
507 */
508 struct thread_group_cputimer {
509 struct task_cputime cputime;
510 int running;
511 raw_spinlock_t lock;
512 };
513
514 #include <linux/rwsem.h>
515 struct autogroup;
516
517 /*
518 * NOTE! "signal_struct" does not have its own
519 * locking, because a shared signal_struct always
520 * implies a shared sighand_struct, so locking
521 * sighand_struct is always a proper superset of
522 * the locking of signal_struct.
523 */
524 struct signal_struct {
525 atomic_t sigcnt;
526 atomic_t live;
527 int nr_threads;
528
529 wait_queue_head_t wait_chldexit; /* for wait4() */
530
531 /* current thread group signal load-balancing target: */
532 struct task_struct *curr_target;
533
534 /* shared signal handling: */
535 struct sigpending shared_pending;
536
537 /* thread group exit support */
538 int group_exit_code;
539 /* overloaded:
540 * - notify group_exit_task when ->count is equal to notify_count
541 * - everyone except group_exit_task is stopped during signal delivery
542 * of fatal signals, group_exit_task processes the signal.
543 */
544 int notify_count;
545 struct task_struct *group_exit_task;
546
547 /* thread group stop support, overloads group_exit_code too */
548 int group_stop_count;
549 unsigned int flags; /* see SIGNAL_* flags below */
550
551 /*
552 * PR_SET_CHILD_SUBREAPER marks a process, like a service
553 * manager, to re-parent orphan (double-forking) child processes
554 * to this process instead of 'init'. The service manager is
555 * able to receive SIGCHLD signals and is able to investigate
556 * the process until it calls wait(). All children of this
557 * process will inherit a flag if they should look for a
558 * child_subreaper process at exit.
559 */
560 unsigned int is_child_subreaper:1;
561 unsigned int has_child_subreaper:1;
562
563 /* POSIX.1b Interval Timers */
564 struct list_head posix_timers;
565
566 /* ITIMER_REAL timer for the process */
567 struct hrtimer real_timer;
568 struct pid *leader_pid;
569 ktime_t it_real_incr;
570
571 /*
572 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
573 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
574 * values are defined to 0 and 1 respectively
575 */
576 struct cpu_itimer it[2];
577
578 /*
579 * Thread group totals for process CPU timers.
580 * See thread_group_cputimer(), et al, for details.
581 */
582 struct thread_group_cputimer cputimer;
583
584 /* Earliest-expiration cache. */
585 struct task_cputime cputime_expires;
586
587 struct list_head cpu_timers[3];
588
589 struct pid *tty_old_pgrp;
590
591 /* boolean value for session group leader */
592 int leader;
593
594 struct tty_struct *tty; /* NULL if no tty */
595
596 #ifdef CONFIG_SCHED_AUTOGROUP
597 struct autogroup *autogroup;
598 #endif
599 /*
600 * Cumulative resource counters for dead threads in the group,
601 * and for reaped dead child processes forked by this group.
602 * Live threads maintain their own counters and add to these
603 * in __exit_signal, except for the group leader.
604 */
605 cputime_t utime, stime, cutime, cstime;
606 cputime_t gtime;
607 cputime_t cgtime;
608 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
609 struct cputime prev_cputime;
610 #endif
611 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
612 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
613 unsigned long inblock, oublock, cinblock, coublock;
614 unsigned long maxrss, cmaxrss;
615 struct task_io_accounting ioac;
616
617 /*
618 * Cumulative ns of schedule CPU time fo dead threads in the
619 * group, not including a zombie group leader, (This only differs
620 * from jiffies_to_ns(utime + stime) if sched_clock uses something
621 * other than jiffies.)
622 */
623 unsigned long long sum_sched_runtime;
624
625 /*
626 * We don't bother to synchronize most readers of this at all,
627 * because there is no reader checking a limit that actually needs
628 * to get both rlim_cur and rlim_max atomically, and either one
629 * alone is a single word that can safely be read normally.
630 * getrlimit/setrlimit use task_lock(current->group_leader) to
631 * protect this instead of the siglock, because they really
632 * have no need to disable irqs.
633 */
634 struct rlimit rlim[RLIM_NLIMITS];
635
636 #ifdef CONFIG_BSD_PROCESS_ACCT
637 struct pacct_struct pacct; /* per-process accounting information */
638 #endif
639 #ifdef CONFIG_TASKSTATS
640 struct taskstats *stats;
641 #endif
642 #ifdef CONFIG_AUDIT
643 unsigned audit_tty;
644 struct tty_audit_buf *tty_audit_buf;
645 #endif
646 #ifdef CONFIG_CGROUPS
647 /*
648 * group_rwsem prevents new tasks from entering the threadgroup and
649 * member tasks from exiting,a more specifically, setting of
650 * PF_EXITING. fork and exit paths are protected with this rwsem
651 * using threadgroup_change_begin/end(). Users which require
652 * threadgroup to remain stable should use threadgroup_[un]lock()
653 * which also takes care of exec path. Currently, cgroup is the
654 * only user.
655 */
656 struct rw_semaphore group_rwsem;
657 #endif
658
659 oom_flags_t oom_flags;
660 short oom_score_adj; /* OOM kill score adjustment */
661 short oom_score_adj_min; /* OOM kill score adjustment min value.
662 * Only settable by CAP_SYS_RESOURCE. */
663
664 struct mutex cred_guard_mutex; /* guard against foreign influences on
665 * credential calculations
666 * (notably. ptrace) */
667 };
668
669 /*
670 * Bits in flags field of signal_struct.
671 */
672 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
673 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
674 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
675 /*
676 * Pending notifications to parent.
677 */
678 #define SIGNAL_CLD_STOPPED 0x00000010
679 #define SIGNAL_CLD_CONTINUED 0x00000020
680 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
681
682 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
683
684 /* If true, all threads except ->group_exit_task have pending SIGKILL */
685 static inline int signal_group_exit(const struct signal_struct *sig)
686 {
687 return (sig->flags & SIGNAL_GROUP_EXIT) ||
688 (sig->group_exit_task != NULL);
689 }
690
691 /*
692 * Some day this will be a full-fledged user tracking system..
693 */
694 struct user_struct {
695 atomic_t __count; /* reference count */
696 atomic_t processes; /* How many processes does this user have? */
697 atomic_t files; /* How many open files does this user have? */
698 atomic_t sigpending; /* How many pending signals does this user have? */
699 #ifdef CONFIG_INOTIFY_USER
700 atomic_t inotify_watches; /* How many inotify watches does this user have? */
701 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
702 #endif
703 #ifdef CONFIG_FANOTIFY
704 atomic_t fanotify_listeners;
705 #endif
706 #ifdef CONFIG_EPOLL
707 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
708 #endif
709 #ifdef CONFIG_POSIX_MQUEUE
710 /* protected by mq_lock */
711 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
712 #endif
713 unsigned long locked_shm; /* How many pages of mlocked shm ? */
714
715 #ifdef CONFIG_KEYS
716 struct key *uid_keyring; /* UID specific keyring */
717 struct key *session_keyring; /* UID's default session keyring */
718 #endif
719
720 /* Hash table maintenance information */
721 struct hlist_node uidhash_node;
722 kuid_t uid;
723
724 #ifdef CONFIG_PERF_EVENTS
725 atomic_long_t locked_vm;
726 #endif
727 };
728
729 extern int uids_sysfs_init(void);
730
731 extern struct user_struct *find_user(kuid_t);
732
733 extern struct user_struct root_user;
734 #define INIT_USER (&root_user)
735
736
737 struct backing_dev_info;
738 struct reclaim_state;
739
740 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
741 struct sched_info {
742 /* cumulative counters */
743 unsigned long pcount; /* # of times run on this cpu */
744 unsigned long long run_delay; /* time spent waiting on a runqueue */
745
746 /* timestamps */
747 unsigned long long last_arrival,/* when we last ran on a cpu */
748 last_queued; /* when we were last queued to run */
749 };
750 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
751
752 #ifdef CONFIG_TASK_DELAY_ACCT
753 struct task_delay_info {
754 spinlock_t lock;
755 unsigned int flags; /* Private per-task flags */
756
757 /* For each stat XXX, add following, aligned appropriately
758 *
759 * struct timespec XXX_start, XXX_end;
760 * u64 XXX_delay;
761 * u32 XXX_count;
762 *
763 * Atomicity of updates to XXX_delay, XXX_count protected by
764 * single lock above (split into XXX_lock if contention is an issue).
765 */
766
767 /*
768 * XXX_count is incremented on every XXX operation, the delay
769 * associated with the operation is added to XXX_delay.
770 * XXX_delay contains the accumulated delay time in nanoseconds.
771 */
772 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
773 u64 blkio_delay; /* wait for sync block io completion */
774 u64 swapin_delay; /* wait for swapin block io completion */
775 u32 blkio_count; /* total count of the number of sync block */
776 /* io operations performed */
777 u32 swapin_count; /* total count of the number of swapin block */
778 /* io operations performed */
779
780 struct timespec freepages_start, freepages_end;
781 u64 freepages_delay; /* wait for memory reclaim */
782 u32 freepages_count; /* total count of memory reclaim */
783 };
784 #endif /* CONFIG_TASK_DELAY_ACCT */
785
786 static inline int sched_info_on(void)
787 {
788 #ifdef CONFIG_SCHEDSTATS
789 return 1;
790 #elif defined(CONFIG_TASK_DELAY_ACCT)
791 extern int delayacct_on;
792 return delayacct_on;
793 #else
794 return 0;
795 #endif
796 }
797
798 enum cpu_idle_type {
799 CPU_IDLE,
800 CPU_NOT_IDLE,
801 CPU_NEWLY_IDLE,
802 CPU_MAX_IDLE_TYPES
803 };
804
805 /*
806 * Increase resolution of nice-level calculations for 64-bit architectures.
807 * The extra resolution improves shares distribution and load balancing of
808 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
809 * hierarchies, especially on larger systems. This is not a user-visible change
810 * and does not change the user-interface for setting shares/weights.
811 *
812 * We increase resolution only if we have enough bits to allow this increased
813 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
814 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
815 * increased costs.
816 */
817 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load */
818 # define SCHED_LOAD_RESOLUTION 10
819 # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
820 # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
821 #else
822 # define SCHED_LOAD_RESOLUTION 0
823 # define scale_load(w) (w)
824 # define scale_load_down(w) (w)
825 #endif
826
827 #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
828 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
829
830 /*
831 * Increase resolution of cpu_power calculations
832 */
833 #define SCHED_POWER_SHIFT 10
834 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
835
836 /*
837 * sched-domains (multiprocessor balancing) declarations:
838 */
839 #ifdef CONFIG_SMP
840 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
841 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
842 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
843 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
844 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
845 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
846 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
847 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
848 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
849 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
850 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
851 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
852
853 extern int __weak arch_sd_sibiling_asym_packing(void);
854
855 struct sched_group_power {
856 atomic_t ref;
857 /*
858 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
859 * single CPU.
860 */
861 unsigned int power, power_orig;
862 unsigned long next_update;
863 /*
864 * Number of busy cpus in this group.
865 */
866 atomic_t nr_busy_cpus;
867
868 unsigned long cpumask[0]; /* iteration mask */
869 };
870
871 struct sched_group {
872 struct sched_group *next; /* Must be a circular list */
873 atomic_t ref;
874
875 unsigned int group_weight;
876 struct sched_group_power *sgp;
877
878 /*
879 * The CPUs this group covers.
880 *
881 * NOTE: this field is variable length. (Allocated dynamically
882 * by attaching extra space to the end of the structure,
883 * depending on how many CPUs the kernel has booted up with)
884 */
885 unsigned long cpumask[0];
886 };
887
888 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
889 {
890 return to_cpumask(sg->cpumask);
891 }
892
893 /*
894 * cpumask masking which cpus in the group are allowed to iterate up the domain
895 * tree.
896 */
897 static inline struct cpumask *sched_group_mask(struct sched_group *sg)
898 {
899 return to_cpumask(sg->sgp->cpumask);
900 }
901
902 /**
903 * group_first_cpu - Returns the first cpu in the cpumask of a sched_group.
904 * @group: The group whose first cpu is to be returned.
905 */
906 static inline unsigned int group_first_cpu(struct sched_group *group)
907 {
908 return cpumask_first(sched_group_cpus(group));
909 }
910
911 struct sched_domain_attr {
912 int relax_domain_level;
913 };
914
915 #define SD_ATTR_INIT (struct sched_domain_attr) { \
916 .relax_domain_level = -1, \
917 }
918
919 extern int sched_domain_level_max;
920
921 struct sched_domain {
922 /* These fields must be setup */
923 struct sched_domain *parent; /* top domain must be null terminated */
924 struct sched_domain *child; /* bottom domain must be null terminated */
925 struct sched_group *groups; /* the balancing groups of the domain */
926 unsigned long min_interval; /* Minimum balance interval ms */
927 unsigned long max_interval; /* Maximum balance interval ms */
928 unsigned int busy_factor; /* less balancing by factor if busy */
929 unsigned int imbalance_pct; /* No balance until over watermark */
930 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
931 unsigned int busy_idx;
932 unsigned int idle_idx;
933 unsigned int newidle_idx;
934 unsigned int wake_idx;
935 unsigned int forkexec_idx;
936 unsigned int smt_gain;
937 int flags; /* See SD_* */
938 int level;
939
940 /* Runtime fields. */
941 unsigned long last_balance; /* init to jiffies. units in jiffies */
942 unsigned int balance_interval; /* initialise to 1. units in ms. */
943 unsigned int nr_balance_failed; /* initialise to 0 */
944
945 u64 last_update;
946
947 #ifdef CONFIG_SCHEDSTATS
948 /* load_balance() stats */
949 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
950 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
951 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
952 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
953 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
954 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
955 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
956 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
957
958 /* Active load balancing */
959 unsigned int alb_count;
960 unsigned int alb_failed;
961 unsigned int alb_pushed;
962
963 /* SD_BALANCE_EXEC stats */
964 unsigned int sbe_count;
965 unsigned int sbe_balanced;
966 unsigned int sbe_pushed;
967
968 /* SD_BALANCE_FORK stats */
969 unsigned int sbf_count;
970 unsigned int sbf_balanced;
971 unsigned int sbf_pushed;
972
973 /* try_to_wake_up() stats */
974 unsigned int ttwu_wake_remote;
975 unsigned int ttwu_move_affine;
976 unsigned int ttwu_move_balance;
977 #endif
978 #ifdef CONFIG_SCHED_DEBUG
979 char *name;
980 #endif
981 union {
982 void *private; /* used during construction */
983 struct rcu_head rcu; /* used during destruction */
984 };
985
986 unsigned int span_weight;
987 /*
988 * Span of all CPUs in this domain.
989 *
990 * NOTE: this field is variable length. (Allocated dynamically
991 * by attaching extra space to the end of the structure,
992 * depending on how many CPUs the kernel has booted up with)
993 */
994 unsigned long span[0];
995 };
996
997 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
998 {
999 return to_cpumask(sd->span);
1000 }
1001
1002 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1003 struct sched_domain_attr *dattr_new);
1004
1005 /* Allocate an array of sched domains, for partition_sched_domains(). */
1006 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1007 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1008
1009 /* Test a flag in parent sched domain */
1010 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1011 {
1012 if (sd->parent && (sd->parent->flags & flag))
1013 return 1;
1014
1015 return 0;
1016 }
1017
1018 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1019 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1020
1021 bool cpus_share_cache(int this_cpu, int that_cpu);
1022
1023 #else /* CONFIG_SMP */
1024
1025 struct sched_domain_attr;
1026
1027 static inline void
1028 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1029 struct sched_domain_attr *dattr_new)
1030 {
1031 }
1032
1033 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1034 {
1035 return true;
1036 }
1037
1038 #endif /* !CONFIG_SMP */
1039
1040
1041 struct io_context; /* See blkdev.h */
1042
1043
1044 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1045 extern void prefetch_stack(struct task_struct *t);
1046 #else
1047 static inline void prefetch_stack(struct task_struct *t) { }
1048 #endif
1049
1050 struct audit_context; /* See audit.c */
1051 struct mempolicy;
1052 struct pipe_inode_info;
1053 struct uts_namespace;
1054
1055 struct rq;
1056 struct sched_domain;
1057
1058 /*
1059 * wake flags
1060 */
1061 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1062 #define WF_FORK 0x02 /* child wakeup after fork */
1063 #define WF_MIGRATED 0x04 /* internal use, task got migrated */
1064
1065 #define ENQUEUE_WAKEUP 1
1066 #define ENQUEUE_HEAD 2
1067 #ifdef CONFIG_SMP
1068 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1069 #else
1070 #define ENQUEUE_WAKING 0
1071 #endif
1072
1073 #define DEQUEUE_SLEEP 1
1074
1075 struct sched_class {
1076 const struct sched_class *next;
1077
1078 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1079 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1080 void (*yield_task) (struct rq *rq);
1081 bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1082
1083 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1084
1085 struct task_struct * (*pick_next_task) (struct rq *rq);
1086 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1087
1088 #ifdef CONFIG_SMP
1089 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1090 void (*migrate_task_rq)(struct task_struct *p, int next_cpu);
1091
1092 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1093 void (*post_schedule) (struct rq *this_rq);
1094 void (*task_waking) (struct task_struct *task);
1095 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1096
1097 void (*set_cpus_allowed)(struct task_struct *p,
1098 const struct cpumask *newmask);
1099
1100 void (*rq_online)(struct rq *rq);
1101 void (*rq_offline)(struct rq *rq);
1102 #endif
1103
1104 void (*set_curr_task) (struct rq *rq);
1105 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1106 void (*task_fork) (struct task_struct *p);
1107
1108 void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1109 void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1110 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1111 int oldprio);
1112
1113 unsigned int (*get_rr_interval) (struct rq *rq,
1114 struct task_struct *task);
1115
1116 #ifdef CONFIG_FAIR_GROUP_SCHED
1117 void (*task_move_group) (struct task_struct *p, int on_rq);
1118 #endif
1119 };
1120
1121 struct load_weight {
1122 unsigned long weight, inv_weight;
1123 };
1124
1125 struct sched_avg {
1126 /*
1127 * These sums represent an infinite geometric series and so are bound
1128 * above by 1024/(1-y). Thus we only need a u32 to store them for for all
1129 * choices of y < 1-2^(-32)*1024.
1130 */
1131 u32 runnable_avg_sum, runnable_avg_period;
1132 u64 last_runnable_update;
1133 s64 decay_count;
1134 unsigned long load_avg_contrib;
1135 };
1136
1137 #ifdef CONFIG_SCHEDSTATS
1138 struct sched_statistics {
1139 u64 wait_start;
1140 u64 wait_max;
1141 u64 wait_count;
1142 u64 wait_sum;
1143 u64 iowait_count;
1144 u64 iowait_sum;
1145
1146 u64 sleep_start;
1147 u64 sleep_max;
1148 s64 sum_sleep_runtime;
1149
1150 u64 block_start;
1151 u64 block_max;
1152 u64 exec_max;
1153 u64 slice_max;
1154
1155 u64 nr_migrations_cold;
1156 u64 nr_failed_migrations_affine;
1157 u64 nr_failed_migrations_running;
1158 u64 nr_failed_migrations_hot;
1159 u64 nr_forced_migrations;
1160
1161 u64 nr_wakeups;
1162 u64 nr_wakeups_sync;
1163 u64 nr_wakeups_migrate;
1164 u64 nr_wakeups_local;
1165 u64 nr_wakeups_remote;
1166 u64 nr_wakeups_affine;
1167 u64 nr_wakeups_affine_attempts;
1168 u64 nr_wakeups_passive;
1169 u64 nr_wakeups_idle;
1170 };
1171 #endif
1172
1173 struct sched_entity {
1174 struct load_weight load; /* for load-balancing */
1175 struct rb_node run_node;
1176 struct list_head group_node;
1177 unsigned int on_rq;
1178
1179 u64 exec_start;
1180 u64 sum_exec_runtime;
1181 u64 vruntime;
1182 u64 prev_sum_exec_runtime;
1183
1184 u64 nr_migrations;
1185
1186 #ifdef CONFIG_SCHEDSTATS
1187 struct sched_statistics statistics;
1188 #endif
1189
1190 #ifdef CONFIG_FAIR_GROUP_SCHED
1191 struct sched_entity *parent;
1192 /* rq on which this entity is (to be) queued: */
1193 struct cfs_rq *cfs_rq;
1194 /* rq "owned" by this entity/group: */
1195 struct cfs_rq *my_q;
1196 #endif
1197 /*
1198 * Load-tracking only depends on SMP, FAIR_GROUP_SCHED dependency below may be
1199 * removed when useful for applications beyond shares distribution (e.g.
1200 * load-balance).
1201 */
1202 #if defined(CONFIG_SMP) && defined(CONFIG_FAIR_GROUP_SCHED)
1203 /* Per-entity load-tracking */
1204 struct sched_avg avg;
1205 #endif
1206 };
1207
1208 struct sched_rt_entity {
1209 struct list_head run_list;
1210 unsigned long timeout;
1211 unsigned int time_slice;
1212
1213 struct sched_rt_entity *back;
1214 #ifdef CONFIG_RT_GROUP_SCHED
1215 struct sched_rt_entity *parent;
1216 /* rq on which this entity is (to be) queued: */
1217 struct rt_rq *rt_rq;
1218 /* rq "owned" by this entity/group: */
1219 struct rt_rq *my_q;
1220 #endif
1221 };
1222
1223 /*
1224 * default timeslice is 100 msecs (used only for SCHED_RR tasks).
1225 * Timeslices get refilled after they expire.
1226 */
1227 #define RR_TIMESLICE (100 * HZ / 1000)
1228
1229 struct rcu_node;
1230
1231 enum perf_event_task_context {
1232 perf_invalid_context = -1,
1233 perf_hw_context = 0,
1234 perf_sw_context,
1235 perf_nr_task_contexts,
1236 };
1237
1238 struct task_struct {
1239 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1240 void *stack;
1241 atomic_t usage;
1242 unsigned int flags; /* per process flags, defined below */
1243 unsigned int ptrace;
1244
1245 #ifdef CONFIG_SMP
1246 struct llist_node wake_entry;
1247 int on_cpu;
1248 #endif
1249 int on_rq;
1250
1251 int prio, static_prio, normal_prio;
1252 unsigned int rt_priority;
1253 const struct sched_class *sched_class;
1254 struct sched_entity se;
1255 struct sched_rt_entity rt;
1256 #ifdef CONFIG_CGROUP_SCHED
1257 struct task_group *sched_task_group;
1258 #endif
1259
1260 #ifdef CONFIG_PREEMPT_NOTIFIERS
1261 /* list of struct preempt_notifier: */
1262 struct hlist_head preempt_notifiers;
1263 #endif
1264
1265 /*
1266 * fpu_counter contains the number of consecutive context switches
1267 * that the FPU is used. If this is over a threshold, the lazy fpu
1268 * saving becomes unlazy to save the trap. This is an unsigned char
1269 * so that after 256 times the counter wraps and the behavior turns
1270 * lazy again; this to deal with bursty apps that only use FPU for
1271 * a short time
1272 */
1273 unsigned char fpu_counter;
1274 #ifdef CONFIG_BLK_DEV_IO_TRACE
1275 unsigned int btrace_seq;
1276 #endif
1277
1278 unsigned int policy;
1279 int nr_cpus_allowed;
1280 cpumask_t cpus_allowed;
1281
1282 #ifdef CONFIG_PREEMPT_RCU
1283 int rcu_read_lock_nesting;
1284 char rcu_read_unlock_special;
1285 struct list_head rcu_node_entry;
1286 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1287 #ifdef CONFIG_TREE_PREEMPT_RCU
1288 struct rcu_node *rcu_blocked_node;
1289 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1290 #ifdef CONFIG_RCU_BOOST
1291 struct rt_mutex *rcu_boost_mutex;
1292 #endif /* #ifdef CONFIG_RCU_BOOST */
1293
1294 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1295 struct sched_info sched_info;
1296 #endif
1297
1298 struct list_head tasks;
1299 #ifdef CONFIG_SMP
1300 struct plist_node pushable_tasks;
1301 #endif
1302
1303 struct mm_struct *mm, *active_mm;
1304 #ifdef CONFIG_COMPAT_BRK
1305 unsigned brk_randomized:1;
1306 #endif
1307 #if defined(SPLIT_RSS_COUNTING)
1308 struct task_rss_stat rss_stat;
1309 #endif
1310 /* task state */
1311 int exit_state;
1312 int exit_code, exit_signal;
1313 int pdeath_signal; /* The signal sent when the parent dies */
1314 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1315 /* ??? */
1316 unsigned int personality;
1317 unsigned did_exec:1;
1318 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1319 * execve */
1320 unsigned in_iowait:1;
1321
1322 /* task may not gain privileges */
1323 unsigned no_new_privs:1;
1324
1325 /* Revert to default priority/policy when forking */
1326 unsigned sched_reset_on_fork:1;
1327 unsigned sched_contributes_to_load:1;
1328
1329 pid_t pid;
1330 pid_t tgid;
1331
1332 #ifdef CONFIG_CC_STACKPROTECTOR
1333 /* Canary value for the -fstack-protector gcc feature */
1334 unsigned long stack_canary;
1335 #endif
1336 /*
1337 * pointers to (original) parent process, youngest child, younger sibling,
1338 * older sibling, respectively. (p->father can be replaced with
1339 * p->real_parent->pid)
1340 */
1341 struct task_struct __rcu *real_parent; /* real parent process */
1342 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1343 /*
1344 * children/sibling forms the list of my natural children
1345 */
1346 struct list_head children; /* list of my children */
1347 struct list_head sibling; /* linkage in my parent's children list */
1348 struct task_struct *group_leader; /* threadgroup leader */
1349
1350 /*
1351 * ptraced is the list of tasks this task is using ptrace on.
1352 * This includes both natural children and PTRACE_ATTACH targets.
1353 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1354 */
1355 struct list_head ptraced;
1356 struct list_head ptrace_entry;
1357
1358 /* PID/PID hash table linkage. */
1359 struct pid_link pids[PIDTYPE_MAX];
1360 struct list_head thread_group;
1361
1362 struct completion *vfork_done; /* for vfork() */
1363 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1364 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1365
1366 cputime_t utime, stime, utimescaled, stimescaled;
1367 cputime_t gtime;
1368 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1369 struct cputime prev_cputime;
1370 #endif
1371 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1372 seqlock_t vtime_seqlock;
1373 unsigned long long vtime_snap;
1374 enum {
1375 VTIME_SLEEPING = 0,
1376 VTIME_USER,
1377 VTIME_SYS,
1378 } vtime_snap_whence;
1379 #endif
1380 unsigned long nvcsw, nivcsw; /* context switch counts */
1381 struct timespec start_time; /* monotonic time */
1382 struct timespec real_start_time; /* boot based time */
1383 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1384 unsigned long min_flt, maj_flt;
1385
1386 struct task_cputime cputime_expires;
1387 struct list_head cpu_timers[3];
1388
1389 /* process credentials */
1390 const struct cred __rcu *real_cred; /* objective and real subjective task
1391 * credentials (COW) */
1392 const struct cred __rcu *cred; /* effective (overridable) subjective task
1393 * credentials (COW) */
1394 char comm[TASK_COMM_LEN]; /* executable name excluding path
1395 - access with [gs]et_task_comm (which lock
1396 it with task_lock())
1397 - initialized normally by setup_new_exec */
1398 /* file system info */
1399 int link_count, total_link_count;
1400 #ifdef CONFIG_SYSVIPC
1401 /* ipc stuff */
1402 struct sysv_sem sysvsem;
1403 #endif
1404 #ifdef CONFIG_DETECT_HUNG_TASK
1405 /* hung task detection */
1406 unsigned long last_switch_count;
1407 #endif
1408 /* CPU-specific state of this task */
1409 struct thread_struct thread;
1410 /* filesystem information */
1411 struct fs_struct *fs;
1412 /* open file information */
1413 struct files_struct *files;
1414 /* namespaces */
1415 struct nsproxy *nsproxy;
1416 /* signal handlers */
1417 struct signal_struct *signal;
1418 struct sighand_struct *sighand;
1419
1420 sigset_t blocked, real_blocked;
1421 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1422 struct sigpending pending;
1423
1424 unsigned long sas_ss_sp;
1425 size_t sas_ss_size;
1426 int (*notifier)(void *priv);
1427 void *notifier_data;
1428 sigset_t *notifier_mask;
1429 struct callback_head *task_works;
1430
1431 struct audit_context *audit_context;
1432 #ifdef CONFIG_AUDITSYSCALL
1433 kuid_t loginuid;
1434 unsigned int sessionid;
1435 #endif
1436 struct seccomp seccomp;
1437
1438 /* Thread group tracking */
1439 u32 parent_exec_id;
1440 u32 self_exec_id;
1441 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1442 * mempolicy */
1443 spinlock_t alloc_lock;
1444
1445 /* Protection of the PI data structures: */
1446 raw_spinlock_t pi_lock;
1447
1448 #ifdef CONFIG_RT_MUTEXES
1449 /* PI waiters blocked on a rt_mutex held by this task */
1450 struct plist_head pi_waiters;
1451 /* Deadlock detection and priority inheritance handling */
1452 struct rt_mutex_waiter *pi_blocked_on;
1453 #endif
1454
1455 #ifdef CONFIG_DEBUG_MUTEXES
1456 /* mutex deadlock detection */
1457 struct mutex_waiter *blocked_on;
1458 #endif
1459 #ifdef CONFIG_TRACE_IRQFLAGS
1460 unsigned int irq_events;
1461 unsigned long hardirq_enable_ip;
1462 unsigned long hardirq_disable_ip;
1463 unsigned int hardirq_enable_event;
1464 unsigned int hardirq_disable_event;
1465 int hardirqs_enabled;
1466 int hardirq_context;
1467 unsigned long softirq_disable_ip;
1468 unsigned long softirq_enable_ip;
1469 unsigned int softirq_disable_event;
1470 unsigned int softirq_enable_event;
1471 int softirqs_enabled;
1472 int softirq_context;
1473 #endif
1474 #ifdef CONFIG_LOCKDEP
1475 # define MAX_LOCK_DEPTH 48UL
1476 u64 curr_chain_key;
1477 int lockdep_depth;
1478 unsigned int lockdep_recursion;
1479 struct held_lock held_locks[MAX_LOCK_DEPTH];
1480 gfp_t lockdep_reclaim_gfp;
1481 #endif
1482
1483 /* journalling filesystem info */
1484 void *journal_info;
1485
1486 /* stacked block device info */
1487 struct bio_list *bio_list;
1488
1489 #ifdef CONFIG_BLOCK
1490 /* stack plugging */
1491 struct blk_plug *plug;
1492 #endif
1493
1494 /* VM state */
1495 struct reclaim_state *reclaim_state;
1496
1497 struct backing_dev_info *backing_dev_info;
1498
1499 struct io_context *io_context;
1500
1501 unsigned long ptrace_message;
1502 siginfo_t *last_siginfo; /* For ptrace use. */
1503 struct task_io_accounting ioac;
1504 #if defined(CONFIG_TASK_XACCT)
1505 u64 acct_rss_mem1; /* accumulated rss usage */
1506 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1507 cputime_t acct_timexpd; /* stime + utime since last update */
1508 #endif
1509 #ifdef CONFIG_CPUSETS
1510 nodemask_t mems_allowed; /* Protected by alloc_lock */
1511 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1512 int cpuset_mem_spread_rotor;
1513 int cpuset_slab_spread_rotor;
1514 #endif
1515 #ifdef CONFIG_CGROUPS
1516 /* Control Group info protected by css_set_lock */
1517 struct css_set __rcu *cgroups;
1518 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1519 struct list_head cg_list;
1520 #endif
1521 #ifdef CONFIG_FUTEX
1522 struct robust_list_head __user *robust_list;
1523 #ifdef CONFIG_COMPAT
1524 struct compat_robust_list_head __user *compat_robust_list;
1525 #endif
1526 struct list_head pi_state_list;
1527 struct futex_pi_state *pi_state_cache;
1528 #endif
1529 #ifdef CONFIG_PERF_EVENTS
1530 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1531 struct mutex perf_event_mutex;
1532 struct list_head perf_event_list;
1533 #endif
1534 #ifdef CONFIG_NUMA
1535 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1536 short il_next;
1537 short pref_node_fork;
1538 #endif
1539 #ifdef CONFIG_NUMA_BALANCING
1540 int numa_scan_seq;
1541 int numa_migrate_seq;
1542 unsigned int numa_scan_period;
1543 u64 node_stamp; /* migration stamp */
1544 struct callback_head numa_work;
1545 #endif /* CONFIG_NUMA_BALANCING */
1546
1547 struct rcu_head rcu;
1548
1549 /*
1550 * cache last used pipe for splice
1551 */
1552 struct pipe_inode_info *splice_pipe;
1553
1554 struct page_frag task_frag;
1555
1556 #ifdef CONFIG_TASK_DELAY_ACCT
1557 struct task_delay_info *delays;
1558 #endif
1559 #ifdef CONFIG_FAULT_INJECTION
1560 int make_it_fail;
1561 #endif
1562 /*
1563 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1564 * balance_dirty_pages() for some dirty throttling pause
1565 */
1566 int nr_dirtied;
1567 int nr_dirtied_pause;
1568 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1569
1570 #ifdef CONFIG_LATENCYTOP
1571 int latency_record_count;
1572 struct latency_record latency_record[LT_SAVECOUNT];
1573 #endif
1574 /*
1575 * time slack values; these are used to round up poll() and
1576 * select() etc timeout values. These are in nanoseconds.
1577 */
1578 unsigned long timer_slack_ns;
1579 unsigned long default_timer_slack_ns;
1580
1581 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1582 /* Index of current stored address in ret_stack */
1583 int curr_ret_stack;
1584 /* Stack of return addresses for return function tracing */
1585 struct ftrace_ret_stack *ret_stack;
1586 /* time stamp for last schedule */
1587 unsigned long long ftrace_timestamp;
1588 /*
1589 * Number of functions that haven't been traced
1590 * because of depth overrun.
1591 */
1592 atomic_t trace_overrun;
1593 /* Pause for the tracing */
1594 atomic_t tracing_graph_pause;
1595 #endif
1596 #ifdef CONFIG_TRACING
1597 /* state flags for use by tracers */
1598 unsigned long trace;
1599 /* bitmask and counter of trace recursion */
1600 unsigned long trace_recursion;
1601 #endif /* CONFIG_TRACING */
1602 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1603 struct memcg_batch_info {
1604 int do_batch; /* incremented when batch uncharge started */
1605 struct mem_cgroup *memcg; /* target memcg of uncharge */
1606 unsigned long nr_pages; /* uncharged usage */
1607 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1608 } memcg_batch;
1609 unsigned int memcg_kmem_skip_account;
1610 #endif
1611 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1612 atomic_t ptrace_bp_refcnt;
1613 #endif
1614 #ifdef CONFIG_UPROBES
1615 struct uprobe_task *utask;
1616 #endif
1617 };
1618
1619 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1620 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1621
1622 #ifdef CONFIG_NUMA_BALANCING
1623 extern void task_numa_fault(int node, int pages, bool migrated);
1624 extern void set_numabalancing_state(bool enabled);
1625 #else
1626 static inline void task_numa_fault(int node, int pages, bool migrated)
1627 {
1628 }
1629 static inline void set_numabalancing_state(bool enabled)
1630 {
1631 }
1632 #endif
1633
1634 /*
1635 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1636 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1637 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1638 * values are inverted: lower p->prio value means higher priority.
1639 *
1640 * The MAX_USER_RT_PRIO value allows the actual maximum
1641 * RT priority to be separate from the value exported to
1642 * user-space. This allows kernel threads to set their
1643 * priority to a value higher than any user task. Note:
1644 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1645 */
1646
1647 #define MAX_USER_RT_PRIO 100
1648 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1649
1650 #define MAX_PRIO (MAX_RT_PRIO + 40)
1651 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1652
1653 static inline int rt_prio(int prio)
1654 {
1655 if (unlikely(prio < MAX_RT_PRIO))
1656 return 1;
1657 return 0;
1658 }
1659
1660 static inline int rt_task(struct task_struct *p)
1661 {
1662 return rt_prio(p->prio);
1663 }
1664
1665 static inline struct pid *task_pid(struct task_struct *task)
1666 {
1667 return task->pids[PIDTYPE_PID].pid;
1668 }
1669
1670 static inline struct pid *task_tgid(struct task_struct *task)
1671 {
1672 return task->group_leader->pids[PIDTYPE_PID].pid;
1673 }
1674
1675 /*
1676 * Without tasklist or rcu lock it is not safe to dereference
1677 * the result of task_pgrp/task_session even if task == current,
1678 * we can race with another thread doing sys_setsid/sys_setpgid.
1679 */
1680 static inline struct pid *task_pgrp(struct task_struct *task)
1681 {
1682 return task->group_leader->pids[PIDTYPE_PGID].pid;
1683 }
1684
1685 static inline struct pid *task_session(struct task_struct *task)
1686 {
1687 return task->group_leader->pids[PIDTYPE_SID].pid;
1688 }
1689
1690 struct pid_namespace;
1691
1692 /*
1693 * the helpers to get the task's different pids as they are seen
1694 * from various namespaces
1695 *
1696 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1697 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1698 * current.
1699 * task_xid_nr_ns() : id seen from the ns specified;
1700 *
1701 * set_task_vxid() : assigns a virtual id to a task;
1702 *
1703 * see also pid_nr() etc in include/linux/pid.h
1704 */
1705 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1706 struct pid_namespace *ns);
1707
1708 static inline pid_t task_pid_nr(struct task_struct *tsk)
1709 {
1710 return tsk->pid;
1711 }
1712
1713 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1714 struct pid_namespace *ns)
1715 {
1716 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1717 }
1718
1719 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1720 {
1721 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1722 }
1723
1724
1725 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1726 {
1727 return tsk->tgid;
1728 }
1729
1730 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1731
1732 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1733 {
1734 return pid_vnr(task_tgid(tsk));
1735 }
1736
1737
1738 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1739 struct pid_namespace *ns)
1740 {
1741 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1742 }
1743
1744 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1745 {
1746 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1747 }
1748
1749
1750 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1751 struct pid_namespace *ns)
1752 {
1753 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1754 }
1755
1756 static inline pid_t task_session_vnr(struct task_struct *tsk)
1757 {
1758 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1759 }
1760
1761 /* obsolete, do not use */
1762 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1763 {
1764 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1765 }
1766
1767 /**
1768 * pid_alive - check that a task structure is not stale
1769 * @p: Task structure to be checked.
1770 *
1771 * Test if a process is not yet dead (at most zombie state)
1772 * If pid_alive fails, then pointers within the task structure
1773 * can be stale and must not be dereferenced.
1774 */
1775 static inline int pid_alive(struct task_struct *p)
1776 {
1777 return p->pids[PIDTYPE_PID].pid != NULL;
1778 }
1779
1780 /**
1781 * is_global_init - check if a task structure is init
1782 * @tsk: Task structure to be checked.
1783 *
1784 * Check if a task structure is the first user space task the kernel created.
1785 */
1786 static inline int is_global_init(struct task_struct *tsk)
1787 {
1788 return tsk->pid == 1;
1789 }
1790
1791 extern struct pid *cad_pid;
1792
1793 extern void free_task(struct task_struct *tsk);
1794 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1795
1796 extern void __put_task_struct(struct task_struct *t);
1797
1798 static inline void put_task_struct(struct task_struct *t)
1799 {
1800 if (atomic_dec_and_test(&t->usage))
1801 __put_task_struct(t);
1802 }
1803
1804 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1805 extern void task_cputime(struct task_struct *t,
1806 cputime_t *utime, cputime_t *stime);
1807 extern void task_cputime_scaled(struct task_struct *t,
1808 cputime_t *utimescaled, cputime_t *stimescaled);
1809 extern cputime_t task_gtime(struct task_struct *t);
1810 #else
1811 static inline void task_cputime(struct task_struct *t,
1812 cputime_t *utime, cputime_t *stime)
1813 {
1814 if (utime)
1815 *utime = t->utime;
1816 if (stime)
1817 *stime = t->stime;
1818 }
1819
1820 static inline void task_cputime_scaled(struct task_struct *t,
1821 cputime_t *utimescaled,
1822 cputime_t *stimescaled)
1823 {
1824 if (utimescaled)
1825 *utimescaled = t->utimescaled;
1826 if (stimescaled)
1827 *stimescaled = t->stimescaled;
1828 }
1829
1830 static inline cputime_t task_gtime(struct task_struct *t)
1831 {
1832 return t->gtime;
1833 }
1834 #endif
1835 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1836 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1837
1838 /*
1839 * Per process flags
1840 */
1841 #define PF_EXITING 0x00000004 /* getting shut down */
1842 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1843 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1844 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1845 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1846 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1847 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1848 #define PF_DUMPCORE 0x00000200 /* dumped core */
1849 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1850 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1851 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1852 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1853 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1854 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1855 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1856 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1857 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1858 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1859 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1860 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1861 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1862 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1863 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1864 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1865 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1866 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1867 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1868 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1869
1870 /*
1871 * Only the _current_ task can read/write to tsk->flags, but other
1872 * tasks can access tsk->flags in readonly mode for example
1873 * with tsk_used_math (like during threaded core dumping).
1874 * There is however an exception to this rule during ptrace
1875 * or during fork: the ptracer task is allowed to write to the
1876 * child->flags of its traced child (same goes for fork, the parent
1877 * can write to the child->flags), because we're guaranteed the
1878 * child is not running and in turn not changing child->flags
1879 * at the same time the parent does it.
1880 */
1881 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1882 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1883 #define clear_used_math() clear_stopped_child_used_math(current)
1884 #define set_used_math() set_stopped_child_used_math(current)
1885 #define conditional_stopped_child_used_math(condition, child) \
1886 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1887 #define conditional_used_math(condition) \
1888 conditional_stopped_child_used_math(condition, current)
1889 #define copy_to_stopped_child_used_math(child) \
1890 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1891 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1892 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1893 #define used_math() tsk_used_math(current)
1894
1895 /*
1896 * task->jobctl flags
1897 */
1898 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1899
1900 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1901 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1902 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1903 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1904 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1905 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1906 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1907
1908 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1909 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1910 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1911 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1912 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1913 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1914 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1915
1916 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1917 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1918
1919 extern bool task_set_jobctl_pending(struct task_struct *task,
1920 unsigned int mask);
1921 extern void task_clear_jobctl_trapping(struct task_struct *task);
1922 extern void task_clear_jobctl_pending(struct task_struct *task,
1923 unsigned int mask);
1924
1925 #ifdef CONFIG_PREEMPT_RCU
1926
1927 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1928 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1929
1930 static inline void rcu_copy_process(struct task_struct *p)
1931 {
1932 p->rcu_read_lock_nesting = 0;
1933 p->rcu_read_unlock_special = 0;
1934 #ifdef CONFIG_TREE_PREEMPT_RCU
1935 p->rcu_blocked_node = NULL;
1936 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1937 #ifdef CONFIG_RCU_BOOST
1938 p->rcu_boost_mutex = NULL;
1939 #endif /* #ifdef CONFIG_RCU_BOOST */
1940 INIT_LIST_HEAD(&p->rcu_node_entry);
1941 }
1942
1943 #else
1944
1945 static inline void rcu_copy_process(struct task_struct *p)
1946 {
1947 }
1948
1949 #endif
1950
1951 static inline void tsk_restore_flags(struct task_struct *task,
1952 unsigned long orig_flags, unsigned long flags)
1953 {
1954 task->flags &= ~flags;
1955 task->flags |= orig_flags & flags;
1956 }
1957
1958 #ifdef CONFIG_SMP
1959 extern void do_set_cpus_allowed(struct task_struct *p,
1960 const struct cpumask *new_mask);
1961
1962 extern int set_cpus_allowed_ptr(struct task_struct *p,
1963 const struct cpumask *new_mask);
1964 #else
1965 static inline void do_set_cpus_allowed(struct task_struct *p,
1966 const struct cpumask *new_mask)
1967 {
1968 }
1969 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1970 const struct cpumask *new_mask)
1971 {
1972 if (!cpumask_test_cpu(0, new_mask))
1973 return -EINVAL;
1974 return 0;
1975 }
1976 #endif
1977
1978 #ifdef CONFIG_NO_HZ
1979 void calc_load_enter_idle(void);
1980 void calc_load_exit_idle(void);
1981 #else
1982 static inline void calc_load_enter_idle(void) { }
1983 static inline void calc_load_exit_idle(void) { }
1984 #endif /* CONFIG_NO_HZ */
1985
1986 #ifndef CONFIG_CPUMASK_OFFSTACK
1987 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1988 {
1989 return set_cpus_allowed_ptr(p, &new_mask);
1990 }
1991 #endif
1992
1993 /*
1994 * Do not use outside of architecture code which knows its limitations.
1995 *
1996 * sched_clock() has no promise of monotonicity or bounded drift between
1997 * CPUs, use (which you should not) requires disabling IRQs.
1998 *
1999 * Please use one of the three interfaces below.
2000 */
2001 extern unsigned long long notrace sched_clock(void);
2002 /*
2003 * See the comment in kernel/sched/clock.c
2004 */
2005 extern u64 cpu_clock(int cpu);
2006 extern u64 local_clock(void);
2007 extern u64 sched_clock_cpu(int cpu);
2008
2009
2010 extern void sched_clock_init(void);
2011
2012 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2013 static inline void sched_clock_tick(void)
2014 {
2015 }
2016
2017 static inline void sched_clock_idle_sleep_event(void)
2018 {
2019 }
2020
2021 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2022 {
2023 }
2024 #else
2025 /*
2026 * Architectures can set this to 1 if they have specified
2027 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2028 * but then during bootup it turns out that sched_clock()
2029 * is reliable after all:
2030 */
2031 extern int sched_clock_stable;
2032
2033 extern void sched_clock_tick(void);
2034 extern void sched_clock_idle_sleep_event(void);
2035 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2036 #endif
2037
2038 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2039 /*
2040 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2041 * The reason for this explicit opt-in is not to have perf penalty with
2042 * slow sched_clocks.
2043 */
2044 extern void enable_sched_clock_irqtime(void);
2045 extern void disable_sched_clock_irqtime(void);
2046 #else
2047 static inline void enable_sched_clock_irqtime(void) {}
2048 static inline void disable_sched_clock_irqtime(void) {}
2049 #endif
2050
2051 extern unsigned long long
2052 task_sched_runtime(struct task_struct *task);
2053
2054 /* sched_exec is called by processes performing an exec */
2055 #ifdef CONFIG_SMP
2056 extern void sched_exec(void);
2057 #else
2058 #define sched_exec() {}
2059 #endif
2060
2061 extern void sched_clock_idle_sleep_event(void);
2062 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2063
2064 #ifdef CONFIG_HOTPLUG_CPU
2065 extern void idle_task_exit(void);
2066 #else
2067 static inline void idle_task_exit(void) {}
2068 #endif
2069
2070 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
2071 extern void wake_up_idle_cpu(int cpu);
2072 #else
2073 static inline void wake_up_idle_cpu(int cpu) { }
2074 #endif
2075
2076 extern unsigned int sysctl_sched_latency;
2077 extern unsigned int sysctl_sched_min_granularity;
2078 extern unsigned int sysctl_sched_wakeup_granularity;
2079 extern unsigned int sysctl_sched_child_runs_first;
2080
2081 enum sched_tunable_scaling {
2082 SCHED_TUNABLESCALING_NONE,
2083 SCHED_TUNABLESCALING_LOG,
2084 SCHED_TUNABLESCALING_LINEAR,
2085 SCHED_TUNABLESCALING_END,
2086 };
2087 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
2088
2089 extern unsigned int sysctl_numa_balancing_scan_delay;
2090 extern unsigned int sysctl_numa_balancing_scan_period_min;
2091 extern unsigned int sysctl_numa_balancing_scan_period_max;
2092 extern unsigned int sysctl_numa_balancing_scan_period_reset;
2093 extern unsigned int sysctl_numa_balancing_scan_size;
2094 extern unsigned int sysctl_numa_balancing_settle_count;
2095
2096 #ifdef CONFIG_SCHED_DEBUG
2097 extern unsigned int sysctl_sched_migration_cost;
2098 extern unsigned int sysctl_sched_nr_migrate;
2099 extern unsigned int sysctl_sched_time_avg;
2100 extern unsigned int sysctl_timer_migration;
2101 extern unsigned int sysctl_sched_shares_window;
2102
2103 int sched_proc_update_handler(struct ctl_table *table, int write,
2104 void __user *buffer, size_t *length,
2105 loff_t *ppos);
2106 #endif
2107 #ifdef CONFIG_SCHED_DEBUG
2108 static inline unsigned int get_sysctl_timer_migration(void)
2109 {
2110 return sysctl_timer_migration;
2111 }
2112 #else
2113 static inline unsigned int get_sysctl_timer_migration(void)
2114 {
2115 return 1;
2116 }
2117 #endif
2118 extern unsigned int sysctl_sched_rt_period;
2119 extern int sysctl_sched_rt_runtime;
2120
2121 int sched_rt_handler(struct ctl_table *table, int write,
2122 void __user *buffer, size_t *lenp,
2123 loff_t *ppos);
2124
2125 #ifdef CONFIG_SCHED_AUTOGROUP
2126 extern unsigned int sysctl_sched_autogroup_enabled;
2127
2128 extern void sched_autogroup_create_attach(struct task_struct *p);
2129 extern void sched_autogroup_detach(struct task_struct *p);
2130 extern void sched_autogroup_fork(struct signal_struct *sig);
2131 extern void sched_autogroup_exit(struct signal_struct *sig);
2132 #ifdef CONFIG_PROC_FS
2133 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2134 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2135 #endif
2136 #else
2137 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2138 static inline void sched_autogroup_detach(struct task_struct *p) { }
2139 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2140 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2141 #endif
2142
2143 #ifdef CONFIG_CFS_BANDWIDTH
2144 extern unsigned int sysctl_sched_cfs_bandwidth_slice;
2145 #endif
2146
2147 #ifdef CONFIG_RT_MUTEXES
2148 extern int rt_mutex_getprio(struct task_struct *p);
2149 extern void rt_mutex_setprio(struct task_struct *p, int prio);
2150 extern void rt_mutex_adjust_pi(struct task_struct *p);
2151 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2152 {
2153 return tsk->pi_blocked_on != NULL;
2154 }
2155 #else
2156 static inline int rt_mutex_getprio(struct task_struct *p)
2157 {
2158 return p->normal_prio;
2159 }
2160 # define rt_mutex_adjust_pi(p) do { } while (0)
2161 static inline bool tsk_is_pi_blocked(struct task_struct *tsk)
2162 {
2163 return false;
2164 }
2165 #endif
2166
2167 extern bool yield_to(struct task_struct *p, bool preempt);
2168 extern void set_user_nice(struct task_struct *p, long nice);
2169 extern int task_prio(const struct task_struct *p);
2170 extern int task_nice(const struct task_struct *p);
2171 extern int can_nice(const struct task_struct *p, const int nice);
2172 extern int task_curr(const struct task_struct *p);
2173 extern int idle_cpu(int cpu);
2174 extern int sched_setscheduler(struct task_struct *, int,
2175 const struct sched_param *);
2176 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2177 const struct sched_param *);
2178 extern struct task_struct *idle_task(int cpu);
2179 /**
2180 * is_idle_task - is the specified task an idle task?
2181 * @p: the task in question.
2182 */
2183 static inline bool is_idle_task(const struct task_struct *p)
2184 {
2185 return p->pid == 0;
2186 }
2187 extern struct task_struct *curr_task(int cpu);
2188 extern void set_curr_task(int cpu, struct task_struct *p);
2189
2190 void yield(void);
2191
2192 /*
2193 * The default (Linux) execution domain.
2194 */
2195 extern struct exec_domain default_exec_domain;
2196
2197 union thread_union {
2198 struct thread_info thread_info;
2199 unsigned long stack[THREAD_SIZE/sizeof(long)];
2200 };
2201
2202 #ifndef __HAVE_ARCH_KSTACK_END
2203 static inline int kstack_end(void *addr)
2204 {
2205 /* Reliable end of stack detection:
2206 * Some APM bios versions misalign the stack
2207 */
2208 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2209 }
2210 #endif
2211
2212 extern union thread_union init_thread_union;
2213 extern struct task_struct init_task;
2214
2215 extern struct mm_struct init_mm;
2216
2217 extern struct pid_namespace init_pid_ns;
2218
2219 /*
2220 * find a task by one of its numerical ids
2221 *
2222 * find_task_by_pid_ns():
2223 * finds a task by its pid in the specified namespace
2224 * find_task_by_vpid():
2225 * finds a task by its virtual pid
2226 *
2227 * see also find_vpid() etc in include/linux/pid.h
2228 */
2229
2230 extern struct task_struct *find_task_by_vpid(pid_t nr);
2231 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2232 struct pid_namespace *ns);
2233
2234 extern void __set_special_pids(struct pid *pid);
2235
2236 /* per-UID process charging. */
2237 extern struct user_struct * alloc_uid(kuid_t);
2238 static inline struct user_struct *get_uid(struct user_struct *u)
2239 {
2240 atomic_inc(&u->__count);
2241 return u;
2242 }
2243 extern void free_uid(struct user_struct *);
2244
2245 #include <asm/current.h>
2246
2247 extern void xtime_update(unsigned long ticks);
2248
2249 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2250 extern int wake_up_process(struct task_struct *tsk);
2251 extern void wake_up_new_task(struct task_struct *tsk);
2252 #ifdef CONFIG_SMP
2253 extern void kick_process(struct task_struct *tsk);
2254 #else
2255 static inline void kick_process(struct task_struct *tsk) { }
2256 #endif
2257 extern void sched_fork(struct task_struct *p);
2258 extern void sched_dead(struct task_struct *p);
2259
2260 extern void proc_caches_init(void);
2261 extern void flush_signals(struct task_struct *);
2262 extern void __flush_signals(struct task_struct *);
2263 extern void ignore_signals(struct task_struct *);
2264 extern void flush_signal_handlers(struct task_struct *, int force_default);
2265 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2266
2267 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2268 {
2269 unsigned long flags;
2270 int ret;
2271
2272 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2273 ret = dequeue_signal(tsk, mask, info);
2274 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2275
2276 return ret;
2277 }
2278
2279 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2280 sigset_t *mask);
2281 extern void unblock_all_signals(void);
2282 extern void release_task(struct task_struct * p);
2283 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2284 extern int force_sigsegv(int, struct task_struct *);
2285 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2286 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2287 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2288 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2289 const struct cred *, u32);
2290 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2291 extern int kill_pid(struct pid *pid, int sig, int priv);
2292 extern int kill_proc_info(int, struct siginfo *, pid_t);
2293 extern __must_check bool do_notify_parent(struct task_struct *, int);
2294 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2295 extern void force_sig(int, struct task_struct *);
2296 extern int send_sig(int, struct task_struct *, int);
2297 extern int zap_other_threads(struct task_struct *p);
2298 extern struct sigqueue *sigqueue_alloc(void);
2299 extern void sigqueue_free(struct sigqueue *);
2300 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2301 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2302 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2303
2304 static inline void restore_saved_sigmask(void)
2305 {
2306 if (test_and_clear_restore_sigmask())
2307 __set_current_blocked(&current->saved_sigmask);
2308 }
2309
2310 static inline sigset_t *sigmask_to_save(void)
2311 {
2312 sigset_t *res = &current->blocked;
2313 if (unlikely(test_restore_sigmask()))
2314 res = &current->saved_sigmask;
2315 return res;
2316 }
2317
2318 static inline int kill_cad_pid(int sig, int priv)
2319 {
2320 return kill_pid(cad_pid, sig, priv);
2321 }
2322
2323 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2324 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2325 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2326 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2327
2328 /*
2329 * True if we are on the alternate signal stack.
2330 */
2331 static inline int on_sig_stack(unsigned long sp)
2332 {
2333 #ifdef CONFIG_STACK_GROWSUP
2334 return sp >= current->sas_ss_sp &&
2335 sp - current->sas_ss_sp < current->sas_ss_size;
2336 #else
2337 return sp > current->sas_ss_sp &&
2338 sp - current->sas_ss_sp <= current->sas_ss_size;
2339 #endif
2340 }
2341
2342 static inline int sas_ss_flags(unsigned long sp)
2343 {
2344 return (current->sas_ss_size == 0 ? SS_DISABLE
2345 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2346 }
2347
2348 /*
2349 * Routines for handling mm_structs
2350 */
2351 extern struct mm_struct * mm_alloc(void);
2352
2353 /* mmdrop drops the mm and the page tables */
2354 extern void __mmdrop(struct mm_struct *);
2355 static inline void mmdrop(struct mm_struct * mm)
2356 {
2357 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2358 __mmdrop(mm);
2359 }
2360
2361 /* mmput gets rid of the mappings and all user-space */
2362 extern void mmput(struct mm_struct *);
2363 /* Grab a reference to a task's mm, if it is not already going away */
2364 extern struct mm_struct *get_task_mm(struct task_struct *task);
2365 /*
2366 * Grab a reference to a task's mm, if it is not already going away
2367 * and ptrace_may_access with the mode parameter passed to it
2368 * succeeds.
2369 */
2370 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2371 /* Remove the current tasks stale references to the old mm_struct */
2372 extern void mm_release(struct task_struct *, struct mm_struct *);
2373 /* Allocate a new mm structure and copy contents from tsk->mm */
2374 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2375
2376 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2377 struct task_struct *);
2378 extern void flush_thread(void);
2379 extern void exit_thread(void);
2380
2381 extern void exit_files(struct task_struct *);
2382 extern void __cleanup_sighand(struct sighand_struct *);
2383
2384 extern void exit_itimers(struct signal_struct *);
2385 extern void flush_itimer_signals(void);
2386
2387 extern void do_group_exit(int);
2388
2389 extern int allow_signal(int);
2390 extern int disallow_signal(int);
2391
2392 extern int do_execve(const char *,
2393 const char __user * const __user *,
2394 const char __user * const __user *);
2395 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2396 struct task_struct *fork_idle(int);
2397 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2398
2399 extern void set_task_comm(struct task_struct *tsk, char *from);
2400 extern char *get_task_comm(char *to, struct task_struct *tsk);
2401
2402 #ifdef CONFIG_SMP
2403 void scheduler_ipi(void);
2404 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2405 #else
2406 static inline void scheduler_ipi(void) { }
2407 static inline unsigned long wait_task_inactive(struct task_struct *p,
2408 long match_state)
2409 {
2410 return 1;
2411 }
2412 #endif
2413
2414 #define next_task(p) \
2415 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2416
2417 #define for_each_process(p) \
2418 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2419
2420 extern bool current_is_single_threaded(void);
2421
2422 /*
2423 * Careful: do_each_thread/while_each_thread is a double loop so
2424 * 'break' will not work as expected - use goto instead.
2425 */
2426 #define do_each_thread(g, t) \
2427 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2428
2429 #define while_each_thread(g, t) \
2430 while ((t = next_thread(t)) != g)
2431
2432 static inline int get_nr_threads(struct task_struct *tsk)
2433 {
2434 return tsk->signal->nr_threads;
2435 }
2436
2437 static inline bool thread_group_leader(struct task_struct *p)
2438 {
2439 return p->exit_signal >= 0;
2440 }
2441
2442 /* Do to the insanities of de_thread it is possible for a process
2443 * to have the pid of the thread group leader without actually being
2444 * the thread group leader. For iteration through the pids in proc
2445 * all we care about is that we have a task with the appropriate
2446 * pid, we don't actually care if we have the right task.
2447 */
2448 static inline int has_group_leader_pid(struct task_struct *p)
2449 {
2450 return p->pid == p->tgid;
2451 }
2452
2453 static inline
2454 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2455 {
2456 return p1->tgid == p2->tgid;
2457 }
2458
2459 static inline struct task_struct *next_thread(const struct task_struct *p)
2460 {
2461 return list_entry_rcu(p->thread_group.next,
2462 struct task_struct, thread_group);
2463 }
2464
2465 static inline int thread_group_empty(struct task_struct *p)
2466 {
2467 return list_empty(&p->thread_group);
2468 }
2469
2470 #define delay_group_leader(p) \
2471 (thread_group_leader(p) && !thread_group_empty(p))
2472
2473 /*
2474 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2475 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2476 * pins the final release of task.io_context. Also protects ->cpuset and
2477 * ->cgroup.subsys[]. And ->vfork_done.
2478 *
2479 * Nests both inside and outside of read_lock(&tasklist_lock).
2480 * It must not be nested with write_lock_irq(&tasklist_lock),
2481 * neither inside nor outside.
2482 */
2483 static inline void task_lock(struct task_struct *p)
2484 {
2485 spin_lock(&p->alloc_lock);
2486 }
2487
2488 static inline void task_unlock(struct task_struct *p)
2489 {
2490 spin_unlock(&p->alloc_lock);
2491 }
2492
2493 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2494 unsigned long *flags);
2495
2496 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2497 unsigned long *flags)
2498 {
2499 struct sighand_struct *ret;
2500
2501 ret = __lock_task_sighand(tsk, flags);
2502 (void)__cond_lock(&tsk->sighand->siglock, ret);
2503 return ret;
2504 }
2505
2506 static inline void unlock_task_sighand(struct task_struct *tsk,
2507 unsigned long *flags)
2508 {
2509 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2510 }
2511
2512 #ifdef CONFIG_CGROUPS
2513 static inline void threadgroup_change_begin(struct task_struct *tsk)
2514 {
2515 down_read(&tsk->signal->group_rwsem);
2516 }
2517 static inline void threadgroup_change_end(struct task_struct *tsk)
2518 {
2519 up_read(&tsk->signal->group_rwsem);
2520 }
2521
2522 /**
2523 * threadgroup_lock - lock threadgroup
2524 * @tsk: member task of the threadgroup to lock
2525 *
2526 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2527 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2528 * perform exec. This is useful for cases where the threadgroup needs to
2529 * stay stable across blockable operations.
2530 *
2531 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2532 * synchronization. While held, no new task will be added to threadgroup
2533 * and no existing live task will have its PF_EXITING set.
2534 *
2535 * During exec, a task goes and puts its thread group through unusual
2536 * changes. After de-threading, exclusive access is assumed to resources
2537 * which are usually shared by tasks in the same group - e.g. sighand may
2538 * be replaced with a new one. Also, the exec'ing task takes over group
2539 * leader role including its pid. Exclude these changes while locked by
2540 * grabbing cred_guard_mutex which is used to synchronize exec path.
2541 */
2542 static inline void threadgroup_lock(struct task_struct *tsk)
2543 {
2544 /*
2545 * exec uses exit for de-threading nesting group_rwsem inside
2546 * cred_guard_mutex. Grab cred_guard_mutex first.
2547 */
2548 mutex_lock(&tsk->signal->cred_guard_mutex);
2549 down_write(&tsk->signal->group_rwsem);
2550 }
2551
2552 /**
2553 * threadgroup_unlock - unlock threadgroup
2554 * @tsk: member task of the threadgroup to unlock
2555 *
2556 * Reverse threadgroup_lock().
2557 */
2558 static inline void threadgroup_unlock(struct task_struct *tsk)
2559 {
2560 up_write(&tsk->signal->group_rwsem);
2561 mutex_unlock(&tsk->signal->cred_guard_mutex);
2562 }
2563 #else
2564 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2565 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2566 static inline void threadgroup_lock(struct task_struct *tsk) {}
2567 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2568 #endif
2569
2570 #ifndef __HAVE_THREAD_FUNCTIONS
2571
2572 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2573 #define task_stack_page(task) ((task)->stack)
2574
2575 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2576 {
2577 *task_thread_info(p) = *task_thread_info(org);
2578 task_thread_info(p)->task = p;
2579 }
2580
2581 static inline unsigned long *end_of_stack(struct task_struct *p)
2582 {
2583 return (unsigned long *)(task_thread_info(p) + 1);
2584 }
2585
2586 #endif
2587
2588 static inline int object_is_on_stack(void *obj)
2589 {
2590 void *stack = task_stack_page(current);
2591
2592 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2593 }
2594
2595 extern void thread_info_cache_init(void);
2596
2597 #ifdef CONFIG_DEBUG_STACK_USAGE
2598 static inline unsigned long stack_not_used(struct task_struct *p)
2599 {
2600 unsigned long *n = end_of_stack(p);
2601
2602 do { /* Skip over canary */
2603 n++;
2604 } while (!*n);
2605
2606 return (unsigned long)n - (unsigned long)end_of_stack(p);
2607 }
2608 #endif
2609
2610 /* set thread flags in other task's structures
2611 * - see asm/thread_info.h for TIF_xxxx flags available
2612 */
2613 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2614 {
2615 set_ti_thread_flag(task_thread_info(tsk), flag);
2616 }
2617
2618 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2619 {
2620 clear_ti_thread_flag(task_thread_info(tsk), flag);
2621 }
2622
2623 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2624 {
2625 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2626 }
2627
2628 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2629 {
2630 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2631 }
2632
2633 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2634 {
2635 return test_ti_thread_flag(task_thread_info(tsk), flag);
2636 }
2637
2638 static inline void set_tsk_need_resched(struct task_struct *tsk)
2639 {
2640 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2641 }
2642
2643 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2644 {
2645 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2646 }
2647
2648 static inline int test_tsk_need_resched(struct task_struct *tsk)
2649 {
2650 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2651 }
2652
2653 static inline int restart_syscall(void)
2654 {
2655 set_tsk_thread_flag(current, TIF_SIGPENDING);
2656 return -ERESTARTNOINTR;
2657 }
2658
2659 static inline int signal_pending(struct task_struct *p)
2660 {
2661 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2662 }
2663
2664 static inline int __fatal_signal_pending(struct task_struct *p)
2665 {
2666 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2667 }
2668
2669 static inline int fatal_signal_pending(struct task_struct *p)
2670 {
2671 return signal_pending(p) && __fatal_signal_pending(p);
2672 }
2673
2674 static inline int signal_pending_state(long state, struct task_struct *p)
2675 {
2676 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2677 return 0;
2678 if (!signal_pending(p))
2679 return 0;
2680
2681 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2682 }
2683
2684 static inline int need_resched(void)
2685 {
2686 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2687 }
2688
2689 /*
2690 * cond_resched() and cond_resched_lock(): latency reduction via
2691 * explicit rescheduling in places that are safe. The return
2692 * value indicates whether a reschedule was done in fact.
2693 * cond_resched_lock() will drop the spinlock before scheduling,
2694 * cond_resched_softirq() will enable bhs before scheduling.
2695 */
2696 extern int _cond_resched(void);
2697
2698 #define cond_resched() ({ \
2699 __might_sleep(__FILE__, __LINE__, 0); \
2700 _cond_resched(); \
2701 })
2702
2703 extern int __cond_resched_lock(spinlock_t *lock);
2704
2705 #ifdef CONFIG_PREEMPT_COUNT
2706 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2707 #else
2708 #define PREEMPT_LOCK_OFFSET 0
2709 #endif
2710
2711 #define cond_resched_lock(lock) ({ \
2712 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2713 __cond_resched_lock(lock); \
2714 })
2715
2716 extern int __cond_resched_softirq(void);
2717
2718 #define cond_resched_softirq() ({ \
2719 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2720 __cond_resched_softirq(); \
2721 })
2722
2723 /*
2724 * Does a critical section need to be broken due to another
2725 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2726 * but a general need for low latency)
2727 */
2728 static inline int spin_needbreak(spinlock_t *lock)
2729 {
2730 #ifdef CONFIG_PREEMPT
2731 return spin_is_contended(lock);
2732 #else
2733 return 0;
2734 #endif
2735 }
2736
2737 /*
2738 * Thread group CPU time accounting.
2739 */
2740 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2741 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2742
2743 static inline void thread_group_cputime_init(struct signal_struct *sig)
2744 {
2745 raw_spin_lock_init(&sig->cputimer.lock);
2746 }
2747
2748 /*
2749 * Reevaluate whether the task has signals pending delivery.
2750 * Wake the task if so.
2751 * This is required every time the blocked sigset_t changes.
2752 * callers must hold sighand->siglock.
2753 */
2754 extern void recalc_sigpending_and_wake(struct task_struct *t);
2755 extern void recalc_sigpending(void);
2756
2757 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2758
2759 /*
2760 * Wrappers for p->thread_info->cpu access. No-op on UP.
2761 */
2762 #ifdef CONFIG_SMP
2763
2764 static inline unsigned int task_cpu(const struct task_struct *p)
2765 {
2766 return task_thread_info(p)->cpu;
2767 }
2768
2769 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2770
2771 #else
2772
2773 static inline unsigned int task_cpu(const struct task_struct *p)
2774 {
2775 return 0;
2776 }
2777
2778 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2779 {
2780 }
2781
2782 #endif /* CONFIG_SMP */
2783
2784 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2785 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2786
2787 extern void normalize_rt_tasks(void);
2788
2789 #ifdef CONFIG_CGROUP_SCHED
2790
2791 extern struct task_group root_task_group;
2792
2793 extern struct task_group *sched_create_group(struct task_group *parent);
2794 extern void sched_destroy_group(struct task_group *tg);
2795 extern void sched_move_task(struct task_struct *tsk);
2796 #ifdef CONFIG_FAIR_GROUP_SCHED
2797 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2798 extern unsigned long sched_group_shares(struct task_group *tg);
2799 #endif
2800 #ifdef CONFIG_RT_GROUP_SCHED
2801 extern int sched_group_set_rt_runtime(struct task_group *tg,
2802 long rt_runtime_us);
2803 extern long sched_group_rt_runtime(struct task_group *tg);
2804 extern int sched_group_set_rt_period(struct task_group *tg,
2805 long rt_period_us);
2806 extern long sched_group_rt_period(struct task_group *tg);
2807 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2808 #endif
2809 #endif /* CONFIG_CGROUP_SCHED */
2810
2811 extern int task_can_switch_user(struct user_struct *up,
2812 struct task_struct *tsk);
2813
2814 #ifdef CONFIG_TASK_XACCT
2815 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2816 {
2817 tsk->ioac.rchar += amt;
2818 }
2819
2820 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2821 {
2822 tsk->ioac.wchar += amt;
2823 }
2824
2825 static inline void inc_syscr(struct task_struct *tsk)
2826 {
2827 tsk->ioac.syscr++;
2828 }
2829
2830 static inline void inc_syscw(struct task_struct *tsk)
2831 {
2832 tsk->ioac.syscw++;
2833 }
2834 #else
2835 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2836 {
2837 }
2838
2839 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2840 {
2841 }
2842
2843 static inline void inc_syscr(struct task_struct *tsk)
2844 {
2845 }
2846
2847 static inline void inc_syscw(struct task_struct *tsk)
2848 {
2849 }
2850 #endif
2851
2852 #ifndef TASK_SIZE_OF
2853 #define TASK_SIZE_OF(tsk) TASK_SIZE
2854 #endif
2855
2856 #ifdef CONFIG_MM_OWNER
2857 extern void mm_update_next_owner(struct mm_struct *mm);
2858 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2859 #else
2860 static inline void mm_update_next_owner(struct mm_struct *mm)
2861 {
2862 }
2863
2864 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2865 {
2866 }
2867 #endif /* CONFIG_MM_OWNER */
2868
2869 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2870 unsigned int limit)
2871 {
2872 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2873 }
2874
2875 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2876 unsigned int limit)
2877 {
2878 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2879 }
2880
2881 static inline unsigned long rlimit(unsigned int limit)
2882 {
2883 return task_rlimit(current, limit);
2884 }
2885
2886 static inline unsigned long rlimit_max(unsigned int limit)
2887 {
2888 return task_rlimit_max(current, limit);
2889 }
2890
2891 #endif
This page took 0.098774 seconds and 5 git commands to generate.