]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blob - include/linux/sched.h
sched: Implement task_nice() as static inline function
[mirror_ubuntu-zesty-kernel.git] / include / linux / sched.h
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3
4 #include <uapi/linux/sched.h>
5
6 #include <linux/sched/prio.h>
7
8
9 struct sched_param {
10 int sched_priority;
11 };
12
13 #include <asm/param.h> /* for HZ */
14
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt_mask.h>
29
30 #include <asm/page.h>
31 #include <asm/ptrace.h>
32 #include <asm/cputime.h>
33
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/signal.h>
37 #include <linux/compiler.h>
38 #include <linux/completion.h>
39 #include <linux/pid.h>
40 #include <linux/percpu.h>
41 #include <linux/topology.h>
42 #include <linux/proportions.h>
43 #include <linux/seccomp.h>
44 #include <linux/rcupdate.h>
45 #include <linux/rculist.h>
46 #include <linux/rtmutex.h>
47
48 #include <linux/time.h>
49 #include <linux/param.h>
50 #include <linux/resource.h>
51 #include <linux/timer.h>
52 #include <linux/hrtimer.h>
53 #include <linux/task_io_accounting.h>
54 #include <linux/latencytop.h>
55 #include <linux/cred.h>
56 #include <linux/llist.h>
57 #include <linux/uidgid.h>
58 #include <linux/gfp.h>
59
60 #include <asm/processor.h>
61
62 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
63
64 /*
65 * Extended scheduling parameters data structure.
66 *
67 * This is needed because the original struct sched_param can not be
68 * altered without introducing ABI issues with legacy applications
69 * (e.g., in sched_getparam()).
70 *
71 * However, the possibility of specifying more than just a priority for
72 * the tasks may be useful for a wide variety of application fields, e.g.,
73 * multimedia, streaming, automation and control, and many others.
74 *
75 * This variant (sched_attr) is meant at describing a so-called
76 * sporadic time-constrained task. In such model a task is specified by:
77 * - the activation period or minimum instance inter-arrival time;
78 * - the maximum (or average, depending on the actual scheduling
79 * discipline) computation time of all instances, a.k.a. runtime;
80 * - the deadline (relative to the actual activation time) of each
81 * instance.
82 * Very briefly, a periodic (sporadic) task asks for the execution of
83 * some specific computation --which is typically called an instance--
84 * (at most) every period. Moreover, each instance typically lasts no more
85 * than the runtime and must be completed by time instant t equal to
86 * the instance activation time + the deadline.
87 *
88 * This is reflected by the actual fields of the sched_attr structure:
89 *
90 * @size size of the structure, for fwd/bwd compat.
91 *
92 * @sched_policy task's scheduling policy
93 * @sched_flags for customizing the scheduler behaviour
94 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
95 * @sched_priority task's static priority (SCHED_FIFO/RR)
96 * @sched_deadline representative of the task's deadline
97 * @sched_runtime representative of the task's runtime
98 * @sched_period representative of the task's period
99 *
100 * Given this task model, there are a multiplicity of scheduling algorithms
101 * and policies, that can be used to ensure all the tasks will make their
102 * timing constraints.
103 *
104 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
105 * only user of this new interface. More information about the algorithm
106 * available in the scheduling class file or in Documentation/.
107 */
108 struct sched_attr {
109 u32 size;
110
111 u32 sched_policy;
112 u64 sched_flags;
113
114 /* SCHED_NORMAL, SCHED_BATCH */
115 s32 sched_nice;
116
117 /* SCHED_FIFO, SCHED_RR */
118 u32 sched_priority;
119
120 /* SCHED_DEADLINE */
121 u64 sched_runtime;
122 u64 sched_deadline;
123 u64 sched_period;
124 };
125
126 struct exec_domain;
127 struct futex_pi_state;
128 struct robust_list_head;
129 struct bio_list;
130 struct fs_struct;
131 struct perf_event_context;
132 struct blk_plug;
133
134 /*
135 * List of flags we want to share for kernel threads,
136 * if only because they are not used by them anyway.
137 */
138 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
139
140 /*
141 * These are the constant used to fake the fixed-point load-average
142 * counting. Some notes:
143 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
144 * a load-average precision of 10 bits integer + 11 bits fractional
145 * - if you want to count load-averages more often, you need more
146 * precision, or rounding will get you. With 2-second counting freq,
147 * the EXP_n values would be 1981, 2034 and 2043 if still using only
148 * 11 bit fractions.
149 */
150 extern unsigned long avenrun[]; /* Load averages */
151 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
152
153 #define FSHIFT 11 /* nr of bits of precision */
154 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
155 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
156 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
157 #define EXP_5 2014 /* 1/exp(5sec/5min) */
158 #define EXP_15 2037 /* 1/exp(5sec/15min) */
159
160 #define CALC_LOAD(load,exp,n) \
161 load *= exp; \
162 load += n*(FIXED_1-exp); \
163 load >>= FSHIFT;
164
165 extern unsigned long total_forks;
166 extern int nr_threads;
167 DECLARE_PER_CPU(unsigned long, process_counts);
168 extern int nr_processes(void);
169 extern unsigned long nr_running(void);
170 extern unsigned long nr_iowait(void);
171 extern unsigned long nr_iowait_cpu(int cpu);
172 extern unsigned long this_cpu_load(void);
173
174
175 extern void calc_global_load(unsigned long ticks);
176 extern void update_cpu_load_nohz(void);
177
178 extern unsigned long get_parent_ip(unsigned long addr);
179
180 extern void dump_cpu_task(int cpu);
181
182 struct seq_file;
183 struct cfs_rq;
184 struct task_group;
185 #ifdef CONFIG_SCHED_DEBUG
186 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
187 extern void proc_sched_set_task(struct task_struct *p);
188 extern void
189 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
190 #endif
191
192 /*
193 * Task state bitmask. NOTE! These bits are also
194 * encoded in fs/proc/array.c: get_task_state().
195 *
196 * We have two separate sets of flags: task->state
197 * is about runnability, while task->exit_state are
198 * about the task exiting. Confusing, but this way
199 * modifying one set can't modify the other one by
200 * mistake.
201 */
202 #define TASK_RUNNING 0
203 #define TASK_INTERRUPTIBLE 1
204 #define TASK_UNINTERRUPTIBLE 2
205 #define __TASK_STOPPED 4
206 #define __TASK_TRACED 8
207 /* in tsk->exit_state */
208 #define EXIT_ZOMBIE 16
209 #define EXIT_DEAD 32
210 /* in tsk->state again */
211 #define TASK_DEAD 64
212 #define TASK_WAKEKILL 128
213 #define TASK_WAKING 256
214 #define TASK_PARKED 512
215 #define TASK_STATE_MAX 1024
216
217 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
218
219 extern char ___assert_task_state[1 - 2*!!(
220 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
221
222 /* Convenience macros for the sake of set_task_state */
223 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
224 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
225 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
226
227 /* Convenience macros for the sake of wake_up */
228 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
229 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
230
231 /* get_task_state() */
232 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
233 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
234 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
235
236 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
237 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
238 #define task_is_stopped_or_traced(task) \
239 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
240 #define task_contributes_to_load(task) \
241 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
242 (task->flags & PF_FROZEN) == 0)
243
244 #define __set_task_state(tsk, state_value) \
245 do { (tsk)->state = (state_value); } while (0)
246 #define set_task_state(tsk, state_value) \
247 set_mb((tsk)->state, (state_value))
248
249 /*
250 * set_current_state() includes a barrier so that the write of current->state
251 * is correctly serialised wrt the caller's subsequent test of whether to
252 * actually sleep:
253 *
254 * set_current_state(TASK_UNINTERRUPTIBLE);
255 * if (do_i_need_to_sleep())
256 * schedule();
257 *
258 * If the caller does not need such serialisation then use __set_current_state()
259 */
260 #define __set_current_state(state_value) \
261 do { current->state = (state_value); } while (0)
262 #define set_current_state(state_value) \
263 set_mb(current->state, (state_value))
264
265 /* Task command name length */
266 #define TASK_COMM_LEN 16
267
268 #include <linux/spinlock.h>
269
270 /*
271 * This serializes "schedule()" and also protects
272 * the run-queue from deletions/modifications (but
273 * _adding_ to the beginning of the run-queue has
274 * a separate lock).
275 */
276 extern rwlock_t tasklist_lock;
277 extern spinlock_t mmlist_lock;
278
279 struct task_struct;
280
281 #ifdef CONFIG_PROVE_RCU
282 extern int lockdep_tasklist_lock_is_held(void);
283 #endif /* #ifdef CONFIG_PROVE_RCU */
284
285 extern void sched_init(void);
286 extern void sched_init_smp(void);
287 extern asmlinkage void schedule_tail(struct task_struct *prev);
288 extern void init_idle(struct task_struct *idle, int cpu);
289 extern void init_idle_bootup_task(struct task_struct *idle);
290
291 extern int runqueue_is_locked(int cpu);
292
293 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
294 extern void nohz_balance_enter_idle(int cpu);
295 extern void set_cpu_sd_state_idle(void);
296 extern int get_nohz_timer_target(void);
297 #else
298 static inline void nohz_balance_enter_idle(int cpu) { }
299 static inline void set_cpu_sd_state_idle(void) { }
300 #endif
301
302 /*
303 * Only dump TASK_* tasks. (0 for all tasks)
304 */
305 extern void show_state_filter(unsigned long state_filter);
306
307 static inline void show_state(void)
308 {
309 show_state_filter(0);
310 }
311
312 extern void show_regs(struct pt_regs *);
313
314 /*
315 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
316 * task), SP is the stack pointer of the first frame that should be shown in the back
317 * trace (or NULL if the entire call-chain of the task should be shown).
318 */
319 extern void show_stack(struct task_struct *task, unsigned long *sp);
320
321 void io_schedule(void);
322 long io_schedule_timeout(long timeout);
323
324 extern void cpu_init (void);
325 extern void trap_init(void);
326 extern void update_process_times(int user);
327 extern void scheduler_tick(void);
328
329 extern void sched_show_task(struct task_struct *p);
330
331 #ifdef CONFIG_LOCKUP_DETECTOR
332 extern void touch_softlockup_watchdog(void);
333 extern void touch_softlockup_watchdog_sync(void);
334 extern void touch_all_softlockup_watchdogs(void);
335 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
336 void __user *buffer,
337 size_t *lenp, loff_t *ppos);
338 extern unsigned int softlockup_panic;
339 void lockup_detector_init(void);
340 #else
341 static inline void touch_softlockup_watchdog(void)
342 {
343 }
344 static inline void touch_softlockup_watchdog_sync(void)
345 {
346 }
347 static inline void touch_all_softlockup_watchdogs(void)
348 {
349 }
350 static inline void lockup_detector_init(void)
351 {
352 }
353 #endif
354
355 #ifdef CONFIG_DETECT_HUNG_TASK
356 void reset_hung_task_detector(void);
357 #else
358 static inline void reset_hung_task_detector(void)
359 {
360 }
361 #endif
362
363 /* Attach to any functions which should be ignored in wchan output. */
364 #define __sched __attribute__((__section__(".sched.text")))
365
366 /* Linker adds these: start and end of __sched functions */
367 extern char __sched_text_start[], __sched_text_end[];
368
369 /* Is this address in the __sched functions? */
370 extern int in_sched_functions(unsigned long addr);
371
372 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
373 extern signed long schedule_timeout(signed long timeout);
374 extern signed long schedule_timeout_interruptible(signed long timeout);
375 extern signed long schedule_timeout_killable(signed long timeout);
376 extern signed long schedule_timeout_uninterruptible(signed long timeout);
377 asmlinkage void schedule(void);
378 extern void schedule_preempt_disabled(void);
379
380 struct nsproxy;
381 struct user_namespace;
382
383 #ifdef CONFIG_MMU
384 extern void arch_pick_mmap_layout(struct mm_struct *mm);
385 extern unsigned long
386 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
387 unsigned long, unsigned long);
388 extern unsigned long
389 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
390 unsigned long len, unsigned long pgoff,
391 unsigned long flags);
392 #else
393 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
394 #endif
395
396 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
397 #define SUID_DUMP_USER 1 /* Dump as user of process */
398 #define SUID_DUMP_ROOT 2 /* Dump as root */
399
400 /* mm flags */
401
402 /* for SUID_DUMP_* above */
403 #define MMF_DUMPABLE_BITS 2
404 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
405
406 extern void set_dumpable(struct mm_struct *mm, int value);
407 /*
408 * This returns the actual value of the suid_dumpable flag. For things
409 * that are using this for checking for privilege transitions, it must
410 * test against SUID_DUMP_USER rather than treating it as a boolean
411 * value.
412 */
413 static inline int __get_dumpable(unsigned long mm_flags)
414 {
415 return mm_flags & MMF_DUMPABLE_MASK;
416 }
417
418 static inline int get_dumpable(struct mm_struct *mm)
419 {
420 return __get_dumpable(mm->flags);
421 }
422
423 /* coredump filter bits */
424 #define MMF_DUMP_ANON_PRIVATE 2
425 #define MMF_DUMP_ANON_SHARED 3
426 #define MMF_DUMP_MAPPED_PRIVATE 4
427 #define MMF_DUMP_MAPPED_SHARED 5
428 #define MMF_DUMP_ELF_HEADERS 6
429 #define MMF_DUMP_HUGETLB_PRIVATE 7
430 #define MMF_DUMP_HUGETLB_SHARED 8
431
432 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
433 #define MMF_DUMP_FILTER_BITS 7
434 #define MMF_DUMP_FILTER_MASK \
435 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
436 #define MMF_DUMP_FILTER_DEFAULT \
437 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
438 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
439
440 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
441 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
442 #else
443 # define MMF_DUMP_MASK_DEFAULT_ELF 0
444 #endif
445 /* leave room for more dump flags */
446 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
447 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
448 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
449
450 #define MMF_HAS_UPROBES 19 /* has uprobes */
451 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
452
453 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
454
455 struct sighand_struct {
456 atomic_t count;
457 struct k_sigaction action[_NSIG];
458 spinlock_t siglock;
459 wait_queue_head_t signalfd_wqh;
460 };
461
462 struct pacct_struct {
463 int ac_flag;
464 long ac_exitcode;
465 unsigned long ac_mem;
466 cputime_t ac_utime, ac_stime;
467 unsigned long ac_minflt, ac_majflt;
468 };
469
470 struct cpu_itimer {
471 cputime_t expires;
472 cputime_t incr;
473 u32 error;
474 u32 incr_error;
475 };
476
477 /**
478 * struct cputime - snaphsot of system and user cputime
479 * @utime: time spent in user mode
480 * @stime: time spent in system mode
481 *
482 * Gathers a generic snapshot of user and system time.
483 */
484 struct cputime {
485 cputime_t utime;
486 cputime_t stime;
487 };
488
489 /**
490 * struct task_cputime - collected CPU time counts
491 * @utime: time spent in user mode, in &cputime_t units
492 * @stime: time spent in kernel mode, in &cputime_t units
493 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
494 *
495 * This is an extension of struct cputime that includes the total runtime
496 * spent by the task from the scheduler point of view.
497 *
498 * As a result, this structure groups together three kinds of CPU time
499 * that are tracked for threads and thread groups. Most things considering
500 * CPU time want to group these counts together and treat all three
501 * of them in parallel.
502 */
503 struct task_cputime {
504 cputime_t utime;
505 cputime_t stime;
506 unsigned long long sum_exec_runtime;
507 };
508 /* Alternate field names when used to cache expirations. */
509 #define prof_exp stime
510 #define virt_exp utime
511 #define sched_exp sum_exec_runtime
512
513 #define INIT_CPUTIME \
514 (struct task_cputime) { \
515 .utime = 0, \
516 .stime = 0, \
517 .sum_exec_runtime = 0, \
518 }
519
520 #ifdef CONFIG_PREEMPT_COUNT
521 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
522 #else
523 #define PREEMPT_DISABLED PREEMPT_ENABLED
524 #endif
525
526 /*
527 * Disable preemption until the scheduler is running.
528 * Reset by start_kernel()->sched_init()->init_idle().
529 *
530 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
531 * before the scheduler is active -- see should_resched().
532 */
533 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
534
535 /**
536 * struct thread_group_cputimer - thread group interval timer counts
537 * @cputime: thread group interval timers.
538 * @running: non-zero when there are timers running and
539 * @cputime receives updates.
540 * @lock: lock for fields in this struct.
541 *
542 * This structure contains the version of task_cputime, above, that is
543 * used for thread group CPU timer calculations.
544 */
545 struct thread_group_cputimer {
546 struct task_cputime cputime;
547 int running;
548 raw_spinlock_t lock;
549 };
550
551 #include <linux/rwsem.h>
552 struct autogroup;
553
554 /*
555 * NOTE! "signal_struct" does not have its own
556 * locking, because a shared signal_struct always
557 * implies a shared sighand_struct, so locking
558 * sighand_struct is always a proper superset of
559 * the locking of signal_struct.
560 */
561 struct signal_struct {
562 atomic_t sigcnt;
563 atomic_t live;
564 int nr_threads;
565 struct list_head thread_head;
566
567 wait_queue_head_t wait_chldexit; /* for wait4() */
568
569 /* current thread group signal load-balancing target: */
570 struct task_struct *curr_target;
571
572 /* shared signal handling: */
573 struct sigpending shared_pending;
574
575 /* thread group exit support */
576 int group_exit_code;
577 /* overloaded:
578 * - notify group_exit_task when ->count is equal to notify_count
579 * - everyone except group_exit_task is stopped during signal delivery
580 * of fatal signals, group_exit_task processes the signal.
581 */
582 int notify_count;
583 struct task_struct *group_exit_task;
584
585 /* thread group stop support, overloads group_exit_code too */
586 int group_stop_count;
587 unsigned int flags; /* see SIGNAL_* flags below */
588
589 /*
590 * PR_SET_CHILD_SUBREAPER marks a process, like a service
591 * manager, to re-parent orphan (double-forking) child processes
592 * to this process instead of 'init'. The service manager is
593 * able to receive SIGCHLD signals and is able to investigate
594 * the process until it calls wait(). All children of this
595 * process will inherit a flag if they should look for a
596 * child_subreaper process at exit.
597 */
598 unsigned int is_child_subreaper:1;
599 unsigned int has_child_subreaper:1;
600
601 /* POSIX.1b Interval Timers */
602 int posix_timer_id;
603 struct list_head posix_timers;
604
605 /* ITIMER_REAL timer for the process */
606 struct hrtimer real_timer;
607 struct pid *leader_pid;
608 ktime_t it_real_incr;
609
610 /*
611 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
612 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
613 * values are defined to 0 and 1 respectively
614 */
615 struct cpu_itimer it[2];
616
617 /*
618 * Thread group totals for process CPU timers.
619 * See thread_group_cputimer(), et al, for details.
620 */
621 struct thread_group_cputimer cputimer;
622
623 /* Earliest-expiration cache. */
624 struct task_cputime cputime_expires;
625
626 struct list_head cpu_timers[3];
627
628 struct pid *tty_old_pgrp;
629
630 /* boolean value for session group leader */
631 int leader;
632
633 struct tty_struct *tty; /* NULL if no tty */
634
635 #ifdef CONFIG_SCHED_AUTOGROUP
636 struct autogroup *autogroup;
637 #endif
638 /*
639 * Cumulative resource counters for dead threads in the group,
640 * and for reaped dead child processes forked by this group.
641 * Live threads maintain their own counters and add to these
642 * in __exit_signal, except for the group leader.
643 */
644 cputime_t utime, stime, cutime, cstime;
645 cputime_t gtime;
646 cputime_t cgtime;
647 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
648 struct cputime prev_cputime;
649 #endif
650 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
651 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
652 unsigned long inblock, oublock, cinblock, coublock;
653 unsigned long maxrss, cmaxrss;
654 struct task_io_accounting ioac;
655
656 /*
657 * Cumulative ns of schedule CPU time fo dead threads in the
658 * group, not including a zombie group leader, (This only differs
659 * from jiffies_to_ns(utime + stime) if sched_clock uses something
660 * other than jiffies.)
661 */
662 unsigned long long sum_sched_runtime;
663
664 /*
665 * We don't bother to synchronize most readers of this at all,
666 * because there is no reader checking a limit that actually needs
667 * to get both rlim_cur and rlim_max atomically, and either one
668 * alone is a single word that can safely be read normally.
669 * getrlimit/setrlimit use task_lock(current->group_leader) to
670 * protect this instead of the siglock, because they really
671 * have no need to disable irqs.
672 */
673 struct rlimit rlim[RLIM_NLIMITS];
674
675 #ifdef CONFIG_BSD_PROCESS_ACCT
676 struct pacct_struct pacct; /* per-process accounting information */
677 #endif
678 #ifdef CONFIG_TASKSTATS
679 struct taskstats *stats;
680 #endif
681 #ifdef CONFIG_AUDIT
682 unsigned audit_tty;
683 unsigned audit_tty_log_passwd;
684 struct tty_audit_buf *tty_audit_buf;
685 #endif
686 #ifdef CONFIG_CGROUPS
687 /*
688 * group_rwsem prevents new tasks from entering the threadgroup and
689 * member tasks from exiting,a more specifically, setting of
690 * PF_EXITING. fork and exit paths are protected with this rwsem
691 * using threadgroup_change_begin/end(). Users which require
692 * threadgroup to remain stable should use threadgroup_[un]lock()
693 * which also takes care of exec path. Currently, cgroup is the
694 * only user.
695 */
696 struct rw_semaphore group_rwsem;
697 #endif
698
699 oom_flags_t oom_flags;
700 short oom_score_adj; /* OOM kill score adjustment */
701 short oom_score_adj_min; /* OOM kill score adjustment min value.
702 * Only settable by CAP_SYS_RESOURCE. */
703
704 struct mutex cred_guard_mutex; /* guard against foreign influences on
705 * credential calculations
706 * (notably. ptrace) */
707 };
708
709 /*
710 * Bits in flags field of signal_struct.
711 */
712 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
713 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
714 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
715 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
716 /*
717 * Pending notifications to parent.
718 */
719 #define SIGNAL_CLD_STOPPED 0x00000010
720 #define SIGNAL_CLD_CONTINUED 0x00000020
721 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
722
723 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
724
725 /* If true, all threads except ->group_exit_task have pending SIGKILL */
726 static inline int signal_group_exit(const struct signal_struct *sig)
727 {
728 return (sig->flags & SIGNAL_GROUP_EXIT) ||
729 (sig->group_exit_task != NULL);
730 }
731
732 /*
733 * Some day this will be a full-fledged user tracking system..
734 */
735 struct user_struct {
736 atomic_t __count; /* reference count */
737 atomic_t processes; /* How many processes does this user have? */
738 atomic_t files; /* How many open files does this user have? */
739 atomic_t sigpending; /* How many pending signals does this user have? */
740 #ifdef CONFIG_INOTIFY_USER
741 atomic_t inotify_watches; /* How many inotify watches does this user have? */
742 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
743 #endif
744 #ifdef CONFIG_FANOTIFY
745 atomic_t fanotify_listeners;
746 #endif
747 #ifdef CONFIG_EPOLL
748 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
749 #endif
750 #ifdef CONFIG_POSIX_MQUEUE
751 /* protected by mq_lock */
752 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
753 #endif
754 unsigned long locked_shm; /* How many pages of mlocked shm ? */
755
756 #ifdef CONFIG_KEYS
757 struct key *uid_keyring; /* UID specific keyring */
758 struct key *session_keyring; /* UID's default session keyring */
759 #endif
760
761 /* Hash table maintenance information */
762 struct hlist_node uidhash_node;
763 kuid_t uid;
764
765 #ifdef CONFIG_PERF_EVENTS
766 atomic_long_t locked_vm;
767 #endif
768 };
769
770 extern int uids_sysfs_init(void);
771
772 extern struct user_struct *find_user(kuid_t);
773
774 extern struct user_struct root_user;
775 #define INIT_USER (&root_user)
776
777
778 struct backing_dev_info;
779 struct reclaim_state;
780
781 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
782 struct sched_info {
783 /* cumulative counters */
784 unsigned long pcount; /* # of times run on this cpu */
785 unsigned long long run_delay; /* time spent waiting on a runqueue */
786
787 /* timestamps */
788 unsigned long long last_arrival,/* when we last ran on a cpu */
789 last_queued; /* when we were last queued to run */
790 };
791 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
792
793 #ifdef CONFIG_TASK_DELAY_ACCT
794 struct task_delay_info {
795 spinlock_t lock;
796 unsigned int flags; /* Private per-task flags */
797
798 /* For each stat XXX, add following, aligned appropriately
799 *
800 * struct timespec XXX_start, XXX_end;
801 * u64 XXX_delay;
802 * u32 XXX_count;
803 *
804 * Atomicity of updates to XXX_delay, XXX_count protected by
805 * single lock above (split into XXX_lock if contention is an issue).
806 */
807
808 /*
809 * XXX_count is incremented on every XXX operation, the delay
810 * associated with the operation is added to XXX_delay.
811 * XXX_delay contains the accumulated delay time in nanoseconds.
812 */
813 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
814 u64 blkio_delay; /* wait for sync block io completion */
815 u64 swapin_delay; /* wait for swapin block io completion */
816 u32 blkio_count; /* total count of the number of sync block */
817 /* io operations performed */
818 u32 swapin_count; /* total count of the number of swapin block */
819 /* io operations performed */
820
821 struct timespec freepages_start, freepages_end;
822 u64 freepages_delay; /* wait for memory reclaim */
823 u32 freepages_count; /* total count of memory reclaim */
824 };
825 #endif /* CONFIG_TASK_DELAY_ACCT */
826
827 static inline int sched_info_on(void)
828 {
829 #ifdef CONFIG_SCHEDSTATS
830 return 1;
831 #elif defined(CONFIG_TASK_DELAY_ACCT)
832 extern int delayacct_on;
833 return delayacct_on;
834 #else
835 return 0;
836 #endif
837 }
838
839 enum cpu_idle_type {
840 CPU_IDLE,
841 CPU_NOT_IDLE,
842 CPU_NEWLY_IDLE,
843 CPU_MAX_IDLE_TYPES
844 };
845
846 /*
847 * Increase resolution of cpu_power calculations
848 */
849 #define SCHED_POWER_SHIFT 10
850 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
851
852 /*
853 * sched-domains (multiprocessor balancing) declarations:
854 */
855 #ifdef CONFIG_SMP
856 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
857 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
858 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
859 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
860 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
861 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
862 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
863 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
864 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
865 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
866 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
867 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
868 #define SD_NUMA 0x4000 /* cross-node balancing */
869
870 extern int __weak arch_sd_sibiling_asym_packing(void);
871
872 struct sched_domain_attr {
873 int relax_domain_level;
874 };
875
876 #define SD_ATTR_INIT (struct sched_domain_attr) { \
877 .relax_domain_level = -1, \
878 }
879
880 extern int sched_domain_level_max;
881
882 struct sched_group;
883
884 struct sched_domain {
885 /* These fields must be setup */
886 struct sched_domain *parent; /* top domain must be null terminated */
887 struct sched_domain *child; /* bottom domain must be null terminated */
888 struct sched_group *groups; /* the balancing groups of the domain */
889 unsigned long min_interval; /* Minimum balance interval ms */
890 unsigned long max_interval; /* Maximum balance interval ms */
891 unsigned int busy_factor; /* less balancing by factor if busy */
892 unsigned int imbalance_pct; /* No balance until over watermark */
893 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
894 unsigned int busy_idx;
895 unsigned int idle_idx;
896 unsigned int newidle_idx;
897 unsigned int wake_idx;
898 unsigned int forkexec_idx;
899 unsigned int smt_gain;
900
901 int nohz_idle; /* NOHZ IDLE status */
902 int flags; /* See SD_* */
903 int level;
904
905 /* Runtime fields. */
906 unsigned long last_balance; /* init to jiffies. units in jiffies */
907 unsigned int balance_interval; /* initialise to 1. units in ms. */
908 unsigned int nr_balance_failed; /* initialise to 0 */
909
910 /* idle_balance() stats */
911 u64 max_newidle_lb_cost;
912 unsigned long next_decay_max_lb_cost;
913
914 #ifdef CONFIG_SCHEDSTATS
915 /* load_balance() stats */
916 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
917 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
918 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
919 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
920 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
921 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
922 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
923 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
924
925 /* Active load balancing */
926 unsigned int alb_count;
927 unsigned int alb_failed;
928 unsigned int alb_pushed;
929
930 /* SD_BALANCE_EXEC stats */
931 unsigned int sbe_count;
932 unsigned int sbe_balanced;
933 unsigned int sbe_pushed;
934
935 /* SD_BALANCE_FORK stats */
936 unsigned int sbf_count;
937 unsigned int sbf_balanced;
938 unsigned int sbf_pushed;
939
940 /* try_to_wake_up() stats */
941 unsigned int ttwu_wake_remote;
942 unsigned int ttwu_move_affine;
943 unsigned int ttwu_move_balance;
944 #endif
945 #ifdef CONFIG_SCHED_DEBUG
946 char *name;
947 #endif
948 union {
949 void *private; /* used during construction */
950 struct rcu_head rcu; /* used during destruction */
951 };
952
953 unsigned int span_weight;
954 /*
955 * Span of all CPUs in this domain.
956 *
957 * NOTE: this field is variable length. (Allocated dynamically
958 * by attaching extra space to the end of the structure,
959 * depending on how many CPUs the kernel has booted up with)
960 */
961 unsigned long span[0];
962 };
963
964 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
965 {
966 return to_cpumask(sd->span);
967 }
968
969 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
970 struct sched_domain_attr *dattr_new);
971
972 /* Allocate an array of sched domains, for partition_sched_domains(). */
973 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
974 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
975
976 bool cpus_share_cache(int this_cpu, int that_cpu);
977
978 #else /* CONFIG_SMP */
979
980 struct sched_domain_attr;
981
982 static inline void
983 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
984 struct sched_domain_attr *dattr_new)
985 {
986 }
987
988 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
989 {
990 return true;
991 }
992
993 #endif /* !CONFIG_SMP */
994
995
996 struct io_context; /* See blkdev.h */
997
998
999 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1000 extern void prefetch_stack(struct task_struct *t);
1001 #else
1002 static inline void prefetch_stack(struct task_struct *t) { }
1003 #endif
1004
1005 struct audit_context; /* See audit.c */
1006 struct mempolicy;
1007 struct pipe_inode_info;
1008 struct uts_namespace;
1009
1010 struct load_weight {
1011 unsigned long weight;
1012 u32 inv_weight;
1013 };
1014
1015 struct sched_avg {
1016 /*
1017 * These sums represent an infinite geometric series and so are bound
1018 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1019 * choices of y < 1-2^(-32)*1024.
1020 */
1021 u32 runnable_avg_sum, runnable_avg_period;
1022 u64 last_runnable_update;
1023 s64 decay_count;
1024 unsigned long load_avg_contrib;
1025 };
1026
1027 #ifdef CONFIG_SCHEDSTATS
1028 struct sched_statistics {
1029 u64 wait_start;
1030 u64 wait_max;
1031 u64 wait_count;
1032 u64 wait_sum;
1033 u64 iowait_count;
1034 u64 iowait_sum;
1035
1036 u64 sleep_start;
1037 u64 sleep_max;
1038 s64 sum_sleep_runtime;
1039
1040 u64 block_start;
1041 u64 block_max;
1042 u64 exec_max;
1043 u64 slice_max;
1044
1045 u64 nr_migrations_cold;
1046 u64 nr_failed_migrations_affine;
1047 u64 nr_failed_migrations_running;
1048 u64 nr_failed_migrations_hot;
1049 u64 nr_forced_migrations;
1050
1051 u64 nr_wakeups;
1052 u64 nr_wakeups_sync;
1053 u64 nr_wakeups_migrate;
1054 u64 nr_wakeups_local;
1055 u64 nr_wakeups_remote;
1056 u64 nr_wakeups_affine;
1057 u64 nr_wakeups_affine_attempts;
1058 u64 nr_wakeups_passive;
1059 u64 nr_wakeups_idle;
1060 };
1061 #endif
1062
1063 struct sched_entity {
1064 struct load_weight load; /* for load-balancing */
1065 struct rb_node run_node;
1066 struct list_head group_node;
1067 unsigned int on_rq;
1068
1069 u64 exec_start;
1070 u64 sum_exec_runtime;
1071 u64 vruntime;
1072 u64 prev_sum_exec_runtime;
1073
1074 u64 nr_migrations;
1075
1076 #ifdef CONFIG_SCHEDSTATS
1077 struct sched_statistics statistics;
1078 #endif
1079
1080 #ifdef CONFIG_FAIR_GROUP_SCHED
1081 struct sched_entity *parent;
1082 /* rq on which this entity is (to be) queued: */
1083 struct cfs_rq *cfs_rq;
1084 /* rq "owned" by this entity/group: */
1085 struct cfs_rq *my_q;
1086 #endif
1087
1088 #ifdef CONFIG_SMP
1089 /* Per-entity load-tracking */
1090 struct sched_avg avg;
1091 #endif
1092 };
1093
1094 struct sched_rt_entity {
1095 struct list_head run_list;
1096 unsigned long timeout;
1097 unsigned long watchdog_stamp;
1098 unsigned int time_slice;
1099
1100 struct sched_rt_entity *back;
1101 #ifdef CONFIG_RT_GROUP_SCHED
1102 struct sched_rt_entity *parent;
1103 /* rq on which this entity is (to be) queued: */
1104 struct rt_rq *rt_rq;
1105 /* rq "owned" by this entity/group: */
1106 struct rt_rq *my_q;
1107 #endif
1108 };
1109
1110 struct sched_dl_entity {
1111 struct rb_node rb_node;
1112
1113 /*
1114 * Original scheduling parameters. Copied here from sched_attr
1115 * during sched_setscheduler2(), they will remain the same until
1116 * the next sched_setscheduler2().
1117 */
1118 u64 dl_runtime; /* maximum runtime for each instance */
1119 u64 dl_deadline; /* relative deadline of each instance */
1120 u64 dl_period; /* separation of two instances (period) */
1121 u64 dl_bw; /* dl_runtime / dl_deadline */
1122
1123 /*
1124 * Actual scheduling parameters. Initialized with the values above,
1125 * they are continously updated during task execution. Note that
1126 * the remaining runtime could be < 0 in case we are in overrun.
1127 */
1128 s64 runtime; /* remaining runtime for this instance */
1129 u64 deadline; /* absolute deadline for this instance */
1130 unsigned int flags; /* specifying the scheduler behaviour */
1131
1132 /*
1133 * Some bool flags:
1134 *
1135 * @dl_throttled tells if we exhausted the runtime. If so, the
1136 * task has to wait for a replenishment to be performed at the
1137 * next firing of dl_timer.
1138 *
1139 * @dl_new tells if a new instance arrived. If so we must
1140 * start executing it with full runtime and reset its absolute
1141 * deadline;
1142 *
1143 * @dl_boosted tells if we are boosted due to DI. If so we are
1144 * outside bandwidth enforcement mechanism (but only until we
1145 * exit the critical section).
1146 */
1147 int dl_throttled, dl_new, dl_boosted;
1148
1149 /*
1150 * Bandwidth enforcement timer. Each -deadline task has its
1151 * own bandwidth to be enforced, thus we need one timer per task.
1152 */
1153 struct hrtimer dl_timer;
1154 };
1155
1156 struct rcu_node;
1157
1158 enum perf_event_task_context {
1159 perf_invalid_context = -1,
1160 perf_hw_context = 0,
1161 perf_sw_context,
1162 perf_nr_task_contexts,
1163 };
1164
1165 struct task_struct {
1166 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1167 void *stack;
1168 atomic_t usage;
1169 unsigned int flags; /* per process flags, defined below */
1170 unsigned int ptrace;
1171
1172 #ifdef CONFIG_SMP
1173 struct llist_node wake_entry;
1174 int on_cpu;
1175 struct task_struct *last_wakee;
1176 unsigned long wakee_flips;
1177 unsigned long wakee_flip_decay_ts;
1178
1179 int wake_cpu;
1180 #endif
1181 int on_rq;
1182
1183 int prio, static_prio, normal_prio;
1184 unsigned int rt_priority;
1185 const struct sched_class *sched_class;
1186 struct sched_entity se;
1187 struct sched_rt_entity rt;
1188 #ifdef CONFIG_CGROUP_SCHED
1189 struct task_group *sched_task_group;
1190 #endif
1191 struct sched_dl_entity dl;
1192
1193 #ifdef CONFIG_PREEMPT_NOTIFIERS
1194 /* list of struct preempt_notifier: */
1195 struct hlist_head preempt_notifiers;
1196 #endif
1197
1198 #ifdef CONFIG_BLK_DEV_IO_TRACE
1199 unsigned int btrace_seq;
1200 #endif
1201
1202 unsigned int policy;
1203 int nr_cpus_allowed;
1204 cpumask_t cpus_allowed;
1205
1206 #ifdef CONFIG_PREEMPT_RCU
1207 int rcu_read_lock_nesting;
1208 char rcu_read_unlock_special;
1209 struct list_head rcu_node_entry;
1210 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1211 #ifdef CONFIG_TREE_PREEMPT_RCU
1212 struct rcu_node *rcu_blocked_node;
1213 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1214 #ifdef CONFIG_RCU_BOOST
1215 struct rt_mutex *rcu_boost_mutex;
1216 #endif /* #ifdef CONFIG_RCU_BOOST */
1217
1218 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1219 struct sched_info sched_info;
1220 #endif
1221
1222 struct list_head tasks;
1223 #ifdef CONFIG_SMP
1224 struct plist_node pushable_tasks;
1225 struct rb_node pushable_dl_tasks;
1226 #endif
1227
1228 struct mm_struct *mm, *active_mm;
1229 #ifdef CONFIG_COMPAT_BRK
1230 unsigned brk_randomized:1;
1231 #endif
1232 #if defined(SPLIT_RSS_COUNTING)
1233 struct task_rss_stat rss_stat;
1234 #endif
1235 /* task state */
1236 int exit_state;
1237 int exit_code, exit_signal;
1238 int pdeath_signal; /* The signal sent when the parent dies */
1239 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1240
1241 /* Used for emulating ABI behavior of previous Linux versions */
1242 unsigned int personality;
1243
1244 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1245 * execve */
1246 unsigned in_iowait:1;
1247
1248 /* task may not gain privileges */
1249 unsigned no_new_privs:1;
1250
1251 /* Revert to default priority/policy when forking */
1252 unsigned sched_reset_on_fork:1;
1253 unsigned sched_contributes_to_load:1;
1254
1255 pid_t pid;
1256 pid_t tgid;
1257
1258 #ifdef CONFIG_CC_STACKPROTECTOR
1259 /* Canary value for the -fstack-protector gcc feature */
1260 unsigned long stack_canary;
1261 #endif
1262 /*
1263 * pointers to (original) parent process, youngest child, younger sibling,
1264 * older sibling, respectively. (p->father can be replaced with
1265 * p->real_parent->pid)
1266 */
1267 struct task_struct __rcu *real_parent; /* real parent process */
1268 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1269 /*
1270 * children/sibling forms the list of my natural children
1271 */
1272 struct list_head children; /* list of my children */
1273 struct list_head sibling; /* linkage in my parent's children list */
1274 struct task_struct *group_leader; /* threadgroup leader */
1275
1276 /*
1277 * ptraced is the list of tasks this task is using ptrace on.
1278 * This includes both natural children and PTRACE_ATTACH targets.
1279 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1280 */
1281 struct list_head ptraced;
1282 struct list_head ptrace_entry;
1283
1284 /* PID/PID hash table linkage. */
1285 struct pid_link pids[PIDTYPE_MAX];
1286 struct list_head thread_group;
1287 struct list_head thread_node;
1288
1289 struct completion *vfork_done; /* for vfork() */
1290 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1291 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1292
1293 cputime_t utime, stime, utimescaled, stimescaled;
1294 cputime_t gtime;
1295 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1296 struct cputime prev_cputime;
1297 #endif
1298 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1299 seqlock_t vtime_seqlock;
1300 unsigned long long vtime_snap;
1301 enum {
1302 VTIME_SLEEPING = 0,
1303 VTIME_USER,
1304 VTIME_SYS,
1305 } vtime_snap_whence;
1306 #endif
1307 unsigned long nvcsw, nivcsw; /* context switch counts */
1308 struct timespec start_time; /* monotonic time */
1309 struct timespec real_start_time; /* boot based time */
1310 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1311 unsigned long min_flt, maj_flt;
1312
1313 struct task_cputime cputime_expires;
1314 struct list_head cpu_timers[3];
1315
1316 /* process credentials */
1317 const struct cred __rcu *real_cred; /* objective and real subjective task
1318 * credentials (COW) */
1319 const struct cred __rcu *cred; /* effective (overridable) subjective task
1320 * credentials (COW) */
1321 char comm[TASK_COMM_LEN]; /* executable name excluding path
1322 - access with [gs]et_task_comm (which lock
1323 it with task_lock())
1324 - initialized normally by setup_new_exec */
1325 /* file system info */
1326 int link_count, total_link_count;
1327 #ifdef CONFIG_SYSVIPC
1328 /* ipc stuff */
1329 struct sysv_sem sysvsem;
1330 #endif
1331 #ifdef CONFIG_DETECT_HUNG_TASK
1332 /* hung task detection */
1333 unsigned long last_switch_count;
1334 #endif
1335 /* CPU-specific state of this task */
1336 struct thread_struct thread;
1337 /* filesystem information */
1338 struct fs_struct *fs;
1339 /* open file information */
1340 struct files_struct *files;
1341 /* namespaces */
1342 struct nsproxy *nsproxy;
1343 /* signal handlers */
1344 struct signal_struct *signal;
1345 struct sighand_struct *sighand;
1346
1347 sigset_t blocked, real_blocked;
1348 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1349 struct sigpending pending;
1350
1351 unsigned long sas_ss_sp;
1352 size_t sas_ss_size;
1353 int (*notifier)(void *priv);
1354 void *notifier_data;
1355 sigset_t *notifier_mask;
1356 struct callback_head *task_works;
1357
1358 struct audit_context *audit_context;
1359 #ifdef CONFIG_AUDITSYSCALL
1360 kuid_t loginuid;
1361 unsigned int sessionid;
1362 #endif
1363 struct seccomp seccomp;
1364
1365 /* Thread group tracking */
1366 u32 parent_exec_id;
1367 u32 self_exec_id;
1368 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1369 * mempolicy */
1370 spinlock_t alloc_lock;
1371
1372 /* Protection of the PI data structures: */
1373 raw_spinlock_t pi_lock;
1374
1375 #ifdef CONFIG_RT_MUTEXES
1376 /* PI waiters blocked on a rt_mutex held by this task */
1377 struct rb_root pi_waiters;
1378 struct rb_node *pi_waiters_leftmost;
1379 /* Deadlock detection and priority inheritance handling */
1380 struct rt_mutex_waiter *pi_blocked_on;
1381 /* Top pi_waiters task */
1382 struct task_struct *pi_top_task;
1383 #endif
1384
1385 #ifdef CONFIG_DEBUG_MUTEXES
1386 /* mutex deadlock detection */
1387 struct mutex_waiter *blocked_on;
1388 #endif
1389 #ifdef CONFIG_TRACE_IRQFLAGS
1390 unsigned int irq_events;
1391 unsigned long hardirq_enable_ip;
1392 unsigned long hardirq_disable_ip;
1393 unsigned int hardirq_enable_event;
1394 unsigned int hardirq_disable_event;
1395 int hardirqs_enabled;
1396 int hardirq_context;
1397 unsigned long softirq_disable_ip;
1398 unsigned long softirq_enable_ip;
1399 unsigned int softirq_disable_event;
1400 unsigned int softirq_enable_event;
1401 int softirqs_enabled;
1402 int softirq_context;
1403 #endif
1404 #ifdef CONFIG_LOCKDEP
1405 # define MAX_LOCK_DEPTH 48UL
1406 u64 curr_chain_key;
1407 int lockdep_depth;
1408 unsigned int lockdep_recursion;
1409 struct held_lock held_locks[MAX_LOCK_DEPTH];
1410 gfp_t lockdep_reclaim_gfp;
1411 #endif
1412
1413 /* journalling filesystem info */
1414 void *journal_info;
1415
1416 /* stacked block device info */
1417 struct bio_list *bio_list;
1418
1419 #ifdef CONFIG_BLOCK
1420 /* stack plugging */
1421 struct blk_plug *plug;
1422 #endif
1423
1424 /* VM state */
1425 struct reclaim_state *reclaim_state;
1426
1427 struct backing_dev_info *backing_dev_info;
1428
1429 struct io_context *io_context;
1430
1431 unsigned long ptrace_message;
1432 siginfo_t *last_siginfo; /* For ptrace use. */
1433 struct task_io_accounting ioac;
1434 #if defined(CONFIG_TASK_XACCT)
1435 u64 acct_rss_mem1; /* accumulated rss usage */
1436 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1437 cputime_t acct_timexpd; /* stime + utime since last update */
1438 #endif
1439 #ifdef CONFIG_CPUSETS
1440 nodemask_t mems_allowed; /* Protected by alloc_lock */
1441 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1442 int cpuset_mem_spread_rotor;
1443 int cpuset_slab_spread_rotor;
1444 #endif
1445 #ifdef CONFIG_CGROUPS
1446 /* Control Group info protected by css_set_lock */
1447 struct css_set __rcu *cgroups;
1448 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1449 struct list_head cg_list;
1450 #endif
1451 #ifdef CONFIG_FUTEX
1452 struct robust_list_head __user *robust_list;
1453 #ifdef CONFIG_COMPAT
1454 struct compat_robust_list_head __user *compat_robust_list;
1455 #endif
1456 struct list_head pi_state_list;
1457 struct futex_pi_state *pi_state_cache;
1458 #endif
1459 #ifdef CONFIG_PERF_EVENTS
1460 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1461 struct mutex perf_event_mutex;
1462 struct list_head perf_event_list;
1463 #endif
1464 #ifdef CONFIG_NUMA
1465 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1466 short il_next;
1467 short pref_node_fork;
1468 #endif
1469 #ifdef CONFIG_NUMA_BALANCING
1470 int numa_scan_seq;
1471 unsigned int numa_scan_period;
1472 unsigned int numa_scan_period_max;
1473 int numa_preferred_nid;
1474 unsigned long numa_migrate_retry;
1475 u64 node_stamp; /* migration stamp */
1476 u64 last_task_numa_placement;
1477 u64 last_sum_exec_runtime;
1478 struct callback_head numa_work;
1479
1480 struct list_head numa_entry;
1481 struct numa_group *numa_group;
1482
1483 /*
1484 * Exponential decaying average of faults on a per-node basis.
1485 * Scheduling placement decisions are made based on the these counts.
1486 * The values remain static for the duration of a PTE scan
1487 */
1488 unsigned long *numa_faults_memory;
1489 unsigned long total_numa_faults;
1490
1491 /*
1492 * numa_faults_buffer records faults per node during the current
1493 * scan window. When the scan completes, the counts in
1494 * numa_faults_memory decay and these values are copied.
1495 */
1496 unsigned long *numa_faults_buffer_memory;
1497
1498 /*
1499 * Track the nodes the process was running on when a NUMA hinting
1500 * fault was incurred.
1501 */
1502 unsigned long *numa_faults_cpu;
1503 unsigned long *numa_faults_buffer_cpu;
1504
1505 /*
1506 * numa_faults_locality tracks if faults recorded during the last
1507 * scan window were remote/local. The task scan period is adapted
1508 * based on the locality of the faults with different weights
1509 * depending on whether they were shared or private faults
1510 */
1511 unsigned long numa_faults_locality[2];
1512
1513 unsigned long numa_pages_migrated;
1514 #endif /* CONFIG_NUMA_BALANCING */
1515
1516 struct rcu_head rcu;
1517
1518 /*
1519 * cache last used pipe for splice
1520 */
1521 struct pipe_inode_info *splice_pipe;
1522
1523 struct page_frag task_frag;
1524
1525 #ifdef CONFIG_TASK_DELAY_ACCT
1526 struct task_delay_info *delays;
1527 #endif
1528 #ifdef CONFIG_FAULT_INJECTION
1529 int make_it_fail;
1530 #endif
1531 /*
1532 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1533 * balance_dirty_pages() for some dirty throttling pause
1534 */
1535 int nr_dirtied;
1536 int nr_dirtied_pause;
1537 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1538
1539 #ifdef CONFIG_LATENCYTOP
1540 int latency_record_count;
1541 struct latency_record latency_record[LT_SAVECOUNT];
1542 #endif
1543 /*
1544 * time slack values; these are used to round up poll() and
1545 * select() etc timeout values. These are in nanoseconds.
1546 */
1547 unsigned long timer_slack_ns;
1548 unsigned long default_timer_slack_ns;
1549
1550 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1551 /* Index of current stored address in ret_stack */
1552 int curr_ret_stack;
1553 /* Stack of return addresses for return function tracing */
1554 struct ftrace_ret_stack *ret_stack;
1555 /* time stamp for last schedule */
1556 unsigned long long ftrace_timestamp;
1557 /*
1558 * Number of functions that haven't been traced
1559 * because of depth overrun.
1560 */
1561 atomic_t trace_overrun;
1562 /* Pause for the tracing */
1563 atomic_t tracing_graph_pause;
1564 #endif
1565 #ifdef CONFIG_TRACING
1566 /* state flags for use by tracers */
1567 unsigned long trace;
1568 /* bitmask and counter of trace recursion */
1569 unsigned long trace_recursion;
1570 #endif /* CONFIG_TRACING */
1571 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1572 struct memcg_batch_info {
1573 int do_batch; /* incremented when batch uncharge started */
1574 struct mem_cgroup *memcg; /* target memcg of uncharge */
1575 unsigned long nr_pages; /* uncharged usage */
1576 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1577 } memcg_batch;
1578 unsigned int memcg_kmem_skip_account;
1579 struct memcg_oom_info {
1580 struct mem_cgroup *memcg;
1581 gfp_t gfp_mask;
1582 int order;
1583 unsigned int may_oom:1;
1584 } memcg_oom;
1585 #endif
1586 #ifdef CONFIG_UPROBES
1587 struct uprobe_task *utask;
1588 #endif
1589 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1590 unsigned int sequential_io;
1591 unsigned int sequential_io_avg;
1592 #endif
1593 };
1594
1595 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1596 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1597
1598 #define TNF_MIGRATED 0x01
1599 #define TNF_NO_GROUP 0x02
1600 #define TNF_SHARED 0x04
1601 #define TNF_FAULT_LOCAL 0x08
1602
1603 #ifdef CONFIG_NUMA_BALANCING
1604 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1605 extern pid_t task_numa_group_id(struct task_struct *p);
1606 extern void set_numabalancing_state(bool enabled);
1607 extern void task_numa_free(struct task_struct *p);
1608 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1609 int src_nid, int dst_cpu);
1610 #else
1611 static inline void task_numa_fault(int last_node, int node, int pages,
1612 int flags)
1613 {
1614 }
1615 static inline pid_t task_numa_group_id(struct task_struct *p)
1616 {
1617 return 0;
1618 }
1619 static inline void set_numabalancing_state(bool enabled)
1620 {
1621 }
1622 static inline void task_numa_free(struct task_struct *p)
1623 {
1624 }
1625 static inline bool should_numa_migrate_memory(struct task_struct *p,
1626 struct page *page, int src_nid, int dst_cpu)
1627 {
1628 return true;
1629 }
1630 #endif
1631
1632 static inline struct pid *task_pid(struct task_struct *task)
1633 {
1634 return task->pids[PIDTYPE_PID].pid;
1635 }
1636
1637 static inline struct pid *task_tgid(struct task_struct *task)
1638 {
1639 return task->group_leader->pids[PIDTYPE_PID].pid;
1640 }
1641
1642 /*
1643 * Without tasklist or rcu lock it is not safe to dereference
1644 * the result of task_pgrp/task_session even if task == current,
1645 * we can race with another thread doing sys_setsid/sys_setpgid.
1646 */
1647 static inline struct pid *task_pgrp(struct task_struct *task)
1648 {
1649 return task->group_leader->pids[PIDTYPE_PGID].pid;
1650 }
1651
1652 static inline struct pid *task_session(struct task_struct *task)
1653 {
1654 return task->group_leader->pids[PIDTYPE_SID].pid;
1655 }
1656
1657 struct pid_namespace;
1658
1659 /*
1660 * the helpers to get the task's different pids as they are seen
1661 * from various namespaces
1662 *
1663 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1664 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1665 * current.
1666 * task_xid_nr_ns() : id seen from the ns specified;
1667 *
1668 * set_task_vxid() : assigns a virtual id to a task;
1669 *
1670 * see also pid_nr() etc in include/linux/pid.h
1671 */
1672 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1673 struct pid_namespace *ns);
1674
1675 static inline pid_t task_pid_nr(struct task_struct *tsk)
1676 {
1677 return tsk->pid;
1678 }
1679
1680 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1681 struct pid_namespace *ns)
1682 {
1683 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1684 }
1685
1686 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1687 {
1688 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1689 }
1690
1691
1692 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1693 {
1694 return tsk->tgid;
1695 }
1696
1697 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1698
1699 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1700 {
1701 return pid_vnr(task_tgid(tsk));
1702 }
1703
1704
1705 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1706 struct pid_namespace *ns)
1707 {
1708 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1709 }
1710
1711 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1712 {
1713 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1714 }
1715
1716
1717 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1718 struct pid_namespace *ns)
1719 {
1720 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1721 }
1722
1723 static inline pid_t task_session_vnr(struct task_struct *tsk)
1724 {
1725 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1726 }
1727
1728 /* obsolete, do not use */
1729 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1730 {
1731 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1732 }
1733
1734 /**
1735 * pid_alive - check that a task structure is not stale
1736 * @p: Task structure to be checked.
1737 *
1738 * Test if a process is not yet dead (at most zombie state)
1739 * If pid_alive fails, then pointers within the task structure
1740 * can be stale and must not be dereferenced.
1741 *
1742 * Return: 1 if the process is alive. 0 otherwise.
1743 */
1744 static inline int pid_alive(struct task_struct *p)
1745 {
1746 return p->pids[PIDTYPE_PID].pid != NULL;
1747 }
1748
1749 /**
1750 * is_global_init - check if a task structure is init
1751 * @tsk: Task structure to be checked.
1752 *
1753 * Check if a task structure is the first user space task the kernel created.
1754 *
1755 * Return: 1 if the task structure is init. 0 otherwise.
1756 */
1757 static inline int is_global_init(struct task_struct *tsk)
1758 {
1759 return tsk->pid == 1;
1760 }
1761
1762 extern struct pid *cad_pid;
1763
1764 extern void free_task(struct task_struct *tsk);
1765 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1766
1767 extern void __put_task_struct(struct task_struct *t);
1768
1769 static inline void put_task_struct(struct task_struct *t)
1770 {
1771 if (atomic_dec_and_test(&t->usage))
1772 __put_task_struct(t);
1773 }
1774
1775 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1776 extern void task_cputime(struct task_struct *t,
1777 cputime_t *utime, cputime_t *stime);
1778 extern void task_cputime_scaled(struct task_struct *t,
1779 cputime_t *utimescaled, cputime_t *stimescaled);
1780 extern cputime_t task_gtime(struct task_struct *t);
1781 #else
1782 static inline void task_cputime(struct task_struct *t,
1783 cputime_t *utime, cputime_t *stime)
1784 {
1785 if (utime)
1786 *utime = t->utime;
1787 if (stime)
1788 *stime = t->stime;
1789 }
1790
1791 static inline void task_cputime_scaled(struct task_struct *t,
1792 cputime_t *utimescaled,
1793 cputime_t *stimescaled)
1794 {
1795 if (utimescaled)
1796 *utimescaled = t->utimescaled;
1797 if (stimescaled)
1798 *stimescaled = t->stimescaled;
1799 }
1800
1801 static inline cputime_t task_gtime(struct task_struct *t)
1802 {
1803 return t->gtime;
1804 }
1805 #endif
1806 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1807 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1808
1809 /*
1810 * Per process flags
1811 */
1812 #define PF_EXITING 0x00000004 /* getting shut down */
1813 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1814 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1815 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1816 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1817 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1818 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1819 #define PF_DUMPCORE 0x00000200 /* dumped core */
1820 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1821 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1822 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1823 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1824 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1825 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1826 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1827 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1828 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1829 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1830 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1831 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1832 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1833 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1834 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1835 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1836 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1837 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1838 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1839 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1840 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1841 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1842
1843 /*
1844 * Only the _current_ task can read/write to tsk->flags, but other
1845 * tasks can access tsk->flags in readonly mode for example
1846 * with tsk_used_math (like during threaded core dumping).
1847 * There is however an exception to this rule during ptrace
1848 * or during fork: the ptracer task is allowed to write to the
1849 * child->flags of its traced child (same goes for fork, the parent
1850 * can write to the child->flags), because we're guaranteed the
1851 * child is not running and in turn not changing child->flags
1852 * at the same time the parent does it.
1853 */
1854 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1855 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1856 #define clear_used_math() clear_stopped_child_used_math(current)
1857 #define set_used_math() set_stopped_child_used_math(current)
1858 #define conditional_stopped_child_used_math(condition, child) \
1859 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1860 #define conditional_used_math(condition) \
1861 conditional_stopped_child_used_math(condition, current)
1862 #define copy_to_stopped_child_used_math(child) \
1863 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1864 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1865 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1866 #define used_math() tsk_used_math(current)
1867
1868 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1869 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1870 {
1871 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1872 flags &= ~__GFP_IO;
1873 return flags;
1874 }
1875
1876 static inline unsigned int memalloc_noio_save(void)
1877 {
1878 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1879 current->flags |= PF_MEMALLOC_NOIO;
1880 return flags;
1881 }
1882
1883 static inline void memalloc_noio_restore(unsigned int flags)
1884 {
1885 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1886 }
1887
1888 /*
1889 * task->jobctl flags
1890 */
1891 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1892
1893 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1894 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1895 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1896 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1897 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1898 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1899 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1900
1901 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1902 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1903 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1904 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1905 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1906 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1907 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1908
1909 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1910 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1911
1912 extern bool task_set_jobctl_pending(struct task_struct *task,
1913 unsigned int mask);
1914 extern void task_clear_jobctl_trapping(struct task_struct *task);
1915 extern void task_clear_jobctl_pending(struct task_struct *task,
1916 unsigned int mask);
1917
1918 #ifdef CONFIG_PREEMPT_RCU
1919
1920 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1921 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1922
1923 static inline void rcu_copy_process(struct task_struct *p)
1924 {
1925 p->rcu_read_lock_nesting = 0;
1926 p->rcu_read_unlock_special = 0;
1927 #ifdef CONFIG_TREE_PREEMPT_RCU
1928 p->rcu_blocked_node = NULL;
1929 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1930 #ifdef CONFIG_RCU_BOOST
1931 p->rcu_boost_mutex = NULL;
1932 #endif /* #ifdef CONFIG_RCU_BOOST */
1933 INIT_LIST_HEAD(&p->rcu_node_entry);
1934 }
1935
1936 #else
1937
1938 static inline void rcu_copy_process(struct task_struct *p)
1939 {
1940 }
1941
1942 #endif
1943
1944 static inline void tsk_restore_flags(struct task_struct *task,
1945 unsigned long orig_flags, unsigned long flags)
1946 {
1947 task->flags &= ~flags;
1948 task->flags |= orig_flags & flags;
1949 }
1950
1951 #ifdef CONFIG_SMP
1952 extern void do_set_cpus_allowed(struct task_struct *p,
1953 const struct cpumask *new_mask);
1954
1955 extern int set_cpus_allowed_ptr(struct task_struct *p,
1956 const struct cpumask *new_mask);
1957 #else
1958 static inline void do_set_cpus_allowed(struct task_struct *p,
1959 const struct cpumask *new_mask)
1960 {
1961 }
1962 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1963 const struct cpumask *new_mask)
1964 {
1965 if (!cpumask_test_cpu(0, new_mask))
1966 return -EINVAL;
1967 return 0;
1968 }
1969 #endif
1970
1971 #ifdef CONFIG_NO_HZ_COMMON
1972 void calc_load_enter_idle(void);
1973 void calc_load_exit_idle(void);
1974 #else
1975 static inline void calc_load_enter_idle(void) { }
1976 static inline void calc_load_exit_idle(void) { }
1977 #endif /* CONFIG_NO_HZ_COMMON */
1978
1979 #ifndef CONFIG_CPUMASK_OFFSTACK
1980 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1981 {
1982 return set_cpus_allowed_ptr(p, &new_mask);
1983 }
1984 #endif
1985
1986 /*
1987 * Do not use outside of architecture code which knows its limitations.
1988 *
1989 * sched_clock() has no promise of monotonicity or bounded drift between
1990 * CPUs, use (which you should not) requires disabling IRQs.
1991 *
1992 * Please use one of the three interfaces below.
1993 */
1994 extern unsigned long long notrace sched_clock(void);
1995 /*
1996 * See the comment in kernel/sched/clock.c
1997 */
1998 extern u64 cpu_clock(int cpu);
1999 extern u64 local_clock(void);
2000 extern u64 sched_clock_cpu(int cpu);
2001
2002
2003 extern void sched_clock_init(void);
2004
2005 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2006 static inline void sched_clock_tick(void)
2007 {
2008 }
2009
2010 static inline void sched_clock_idle_sleep_event(void)
2011 {
2012 }
2013
2014 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2015 {
2016 }
2017 #else
2018 /*
2019 * Architectures can set this to 1 if they have specified
2020 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2021 * but then during bootup it turns out that sched_clock()
2022 * is reliable after all:
2023 */
2024 extern int sched_clock_stable(void);
2025 extern void set_sched_clock_stable(void);
2026 extern void clear_sched_clock_stable(void);
2027
2028 extern void sched_clock_tick(void);
2029 extern void sched_clock_idle_sleep_event(void);
2030 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2031 #endif
2032
2033 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2034 /*
2035 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2036 * The reason for this explicit opt-in is not to have perf penalty with
2037 * slow sched_clocks.
2038 */
2039 extern void enable_sched_clock_irqtime(void);
2040 extern void disable_sched_clock_irqtime(void);
2041 #else
2042 static inline void enable_sched_clock_irqtime(void) {}
2043 static inline void disable_sched_clock_irqtime(void) {}
2044 #endif
2045
2046 extern unsigned long long
2047 task_sched_runtime(struct task_struct *task);
2048
2049 /* sched_exec is called by processes performing an exec */
2050 #ifdef CONFIG_SMP
2051 extern void sched_exec(void);
2052 #else
2053 #define sched_exec() {}
2054 #endif
2055
2056 extern void sched_clock_idle_sleep_event(void);
2057 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2058
2059 #ifdef CONFIG_HOTPLUG_CPU
2060 extern void idle_task_exit(void);
2061 #else
2062 static inline void idle_task_exit(void) {}
2063 #endif
2064
2065 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2066 extern void wake_up_nohz_cpu(int cpu);
2067 #else
2068 static inline void wake_up_nohz_cpu(int cpu) { }
2069 #endif
2070
2071 #ifdef CONFIG_NO_HZ_FULL
2072 extern bool sched_can_stop_tick(void);
2073 extern u64 scheduler_tick_max_deferment(void);
2074 #else
2075 static inline bool sched_can_stop_tick(void) { return false; }
2076 #endif
2077
2078 #ifdef CONFIG_SCHED_AUTOGROUP
2079 extern void sched_autogroup_create_attach(struct task_struct *p);
2080 extern void sched_autogroup_detach(struct task_struct *p);
2081 extern void sched_autogroup_fork(struct signal_struct *sig);
2082 extern void sched_autogroup_exit(struct signal_struct *sig);
2083 #ifdef CONFIG_PROC_FS
2084 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2085 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2086 #endif
2087 #else
2088 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2089 static inline void sched_autogroup_detach(struct task_struct *p) { }
2090 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2091 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2092 #endif
2093
2094 extern bool yield_to(struct task_struct *p, bool preempt);
2095 extern void set_user_nice(struct task_struct *p, long nice);
2096 extern int task_prio(const struct task_struct *p);
2097 /**
2098 * task_nice - return the nice value of a given task.
2099 * @p: the task in question.
2100 *
2101 * Return: The nice value [ -20 ... 0 ... 19 ].
2102 */
2103 static inline int task_nice(const struct task_struct *p)
2104 {
2105 return PRIO_TO_NICE((p)->static_prio);
2106 }
2107 extern int can_nice(const struct task_struct *p, const int nice);
2108 extern int task_curr(const struct task_struct *p);
2109 extern int idle_cpu(int cpu);
2110 extern int sched_setscheduler(struct task_struct *, int,
2111 const struct sched_param *);
2112 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2113 const struct sched_param *);
2114 extern int sched_setattr(struct task_struct *,
2115 const struct sched_attr *);
2116 extern struct task_struct *idle_task(int cpu);
2117 /**
2118 * is_idle_task - is the specified task an idle task?
2119 * @p: the task in question.
2120 *
2121 * Return: 1 if @p is an idle task. 0 otherwise.
2122 */
2123 static inline bool is_idle_task(const struct task_struct *p)
2124 {
2125 return p->pid == 0;
2126 }
2127 extern struct task_struct *curr_task(int cpu);
2128 extern void set_curr_task(int cpu, struct task_struct *p);
2129
2130 void yield(void);
2131
2132 /*
2133 * The default (Linux) execution domain.
2134 */
2135 extern struct exec_domain default_exec_domain;
2136
2137 union thread_union {
2138 struct thread_info thread_info;
2139 unsigned long stack[THREAD_SIZE/sizeof(long)];
2140 };
2141
2142 #ifndef __HAVE_ARCH_KSTACK_END
2143 static inline int kstack_end(void *addr)
2144 {
2145 /* Reliable end of stack detection:
2146 * Some APM bios versions misalign the stack
2147 */
2148 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2149 }
2150 #endif
2151
2152 extern union thread_union init_thread_union;
2153 extern struct task_struct init_task;
2154
2155 extern struct mm_struct init_mm;
2156
2157 extern struct pid_namespace init_pid_ns;
2158
2159 /*
2160 * find a task by one of its numerical ids
2161 *
2162 * find_task_by_pid_ns():
2163 * finds a task by its pid in the specified namespace
2164 * find_task_by_vpid():
2165 * finds a task by its virtual pid
2166 *
2167 * see also find_vpid() etc in include/linux/pid.h
2168 */
2169
2170 extern struct task_struct *find_task_by_vpid(pid_t nr);
2171 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2172 struct pid_namespace *ns);
2173
2174 /* per-UID process charging. */
2175 extern struct user_struct * alloc_uid(kuid_t);
2176 static inline struct user_struct *get_uid(struct user_struct *u)
2177 {
2178 atomic_inc(&u->__count);
2179 return u;
2180 }
2181 extern void free_uid(struct user_struct *);
2182
2183 #include <asm/current.h>
2184
2185 extern void xtime_update(unsigned long ticks);
2186
2187 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2188 extern int wake_up_process(struct task_struct *tsk);
2189 extern void wake_up_new_task(struct task_struct *tsk);
2190 #ifdef CONFIG_SMP
2191 extern void kick_process(struct task_struct *tsk);
2192 #else
2193 static inline void kick_process(struct task_struct *tsk) { }
2194 #endif
2195 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2196 extern void sched_dead(struct task_struct *p);
2197
2198 extern void proc_caches_init(void);
2199 extern void flush_signals(struct task_struct *);
2200 extern void __flush_signals(struct task_struct *);
2201 extern void ignore_signals(struct task_struct *);
2202 extern void flush_signal_handlers(struct task_struct *, int force_default);
2203 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2204
2205 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2206 {
2207 unsigned long flags;
2208 int ret;
2209
2210 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2211 ret = dequeue_signal(tsk, mask, info);
2212 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2213
2214 return ret;
2215 }
2216
2217 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2218 sigset_t *mask);
2219 extern void unblock_all_signals(void);
2220 extern void release_task(struct task_struct * p);
2221 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2222 extern int force_sigsegv(int, struct task_struct *);
2223 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2224 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2225 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2226 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2227 const struct cred *, u32);
2228 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2229 extern int kill_pid(struct pid *pid, int sig, int priv);
2230 extern int kill_proc_info(int, struct siginfo *, pid_t);
2231 extern __must_check bool do_notify_parent(struct task_struct *, int);
2232 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2233 extern void force_sig(int, struct task_struct *);
2234 extern int send_sig(int, struct task_struct *, int);
2235 extern int zap_other_threads(struct task_struct *p);
2236 extern struct sigqueue *sigqueue_alloc(void);
2237 extern void sigqueue_free(struct sigqueue *);
2238 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2239 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2240
2241 static inline void restore_saved_sigmask(void)
2242 {
2243 if (test_and_clear_restore_sigmask())
2244 __set_current_blocked(&current->saved_sigmask);
2245 }
2246
2247 static inline sigset_t *sigmask_to_save(void)
2248 {
2249 sigset_t *res = &current->blocked;
2250 if (unlikely(test_restore_sigmask()))
2251 res = &current->saved_sigmask;
2252 return res;
2253 }
2254
2255 static inline int kill_cad_pid(int sig, int priv)
2256 {
2257 return kill_pid(cad_pid, sig, priv);
2258 }
2259
2260 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2261 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2262 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2263 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2264
2265 /*
2266 * True if we are on the alternate signal stack.
2267 */
2268 static inline int on_sig_stack(unsigned long sp)
2269 {
2270 #ifdef CONFIG_STACK_GROWSUP
2271 return sp >= current->sas_ss_sp &&
2272 sp - current->sas_ss_sp < current->sas_ss_size;
2273 #else
2274 return sp > current->sas_ss_sp &&
2275 sp - current->sas_ss_sp <= current->sas_ss_size;
2276 #endif
2277 }
2278
2279 static inline int sas_ss_flags(unsigned long sp)
2280 {
2281 return (current->sas_ss_size == 0 ? SS_DISABLE
2282 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2283 }
2284
2285 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2286 {
2287 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2288 #ifdef CONFIG_STACK_GROWSUP
2289 return current->sas_ss_sp;
2290 #else
2291 return current->sas_ss_sp + current->sas_ss_size;
2292 #endif
2293 return sp;
2294 }
2295
2296 /*
2297 * Routines for handling mm_structs
2298 */
2299 extern struct mm_struct * mm_alloc(void);
2300
2301 /* mmdrop drops the mm and the page tables */
2302 extern void __mmdrop(struct mm_struct *);
2303 static inline void mmdrop(struct mm_struct * mm)
2304 {
2305 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2306 __mmdrop(mm);
2307 }
2308
2309 /* mmput gets rid of the mappings and all user-space */
2310 extern void mmput(struct mm_struct *);
2311 /* Grab a reference to a task's mm, if it is not already going away */
2312 extern struct mm_struct *get_task_mm(struct task_struct *task);
2313 /*
2314 * Grab a reference to a task's mm, if it is not already going away
2315 * and ptrace_may_access with the mode parameter passed to it
2316 * succeeds.
2317 */
2318 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2319 /* Remove the current tasks stale references to the old mm_struct */
2320 extern void mm_release(struct task_struct *, struct mm_struct *);
2321
2322 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2323 struct task_struct *);
2324 extern void flush_thread(void);
2325 extern void exit_thread(void);
2326
2327 extern void exit_files(struct task_struct *);
2328 extern void __cleanup_sighand(struct sighand_struct *);
2329
2330 extern void exit_itimers(struct signal_struct *);
2331 extern void flush_itimer_signals(void);
2332
2333 extern void do_group_exit(int);
2334
2335 extern int allow_signal(int);
2336 extern int disallow_signal(int);
2337
2338 extern int do_execve(const char *,
2339 const char __user * const __user *,
2340 const char __user * const __user *);
2341 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2342 struct task_struct *fork_idle(int);
2343 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2344
2345 extern void set_task_comm(struct task_struct *tsk, char *from);
2346 extern char *get_task_comm(char *to, struct task_struct *tsk);
2347
2348 #ifdef CONFIG_SMP
2349 void scheduler_ipi(void);
2350 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2351 #else
2352 static inline void scheduler_ipi(void) { }
2353 static inline unsigned long wait_task_inactive(struct task_struct *p,
2354 long match_state)
2355 {
2356 return 1;
2357 }
2358 #endif
2359
2360 #define next_task(p) \
2361 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2362
2363 #define for_each_process(p) \
2364 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2365
2366 extern bool current_is_single_threaded(void);
2367
2368 /*
2369 * Careful: do_each_thread/while_each_thread is a double loop so
2370 * 'break' will not work as expected - use goto instead.
2371 */
2372 #define do_each_thread(g, t) \
2373 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2374
2375 #define while_each_thread(g, t) \
2376 while ((t = next_thread(t)) != g)
2377
2378 #define __for_each_thread(signal, t) \
2379 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2380
2381 #define for_each_thread(p, t) \
2382 __for_each_thread((p)->signal, t)
2383
2384 /* Careful: this is a double loop, 'break' won't work as expected. */
2385 #define for_each_process_thread(p, t) \
2386 for_each_process(p) for_each_thread(p, t)
2387
2388 static inline int get_nr_threads(struct task_struct *tsk)
2389 {
2390 return tsk->signal->nr_threads;
2391 }
2392
2393 static inline bool thread_group_leader(struct task_struct *p)
2394 {
2395 return p->exit_signal >= 0;
2396 }
2397
2398 /* Do to the insanities of de_thread it is possible for a process
2399 * to have the pid of the thread group leader without actually being
2400 * the thread group leader. For iteration through the pids in proc
2401 * all we care about is that we have a task with the appropriate
2402 * pid, we don't actually care if we have the right task.
2403 */
2404 static inline bool has_group_leader_pid(struct task_struct *p)
2405 {
2406 return task_pid(p) == p->signal->leader_pid;
2407 }
2408
2409 static inline
2410 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2411 {
2412 return p1->signal == p2->signal;
2413 }
2414
2415 static inline struct task_struct *next_thread(const struct task_struct *p)
2416 {
2417 return list_entry_rcu(p->thread_group.next,
2418 struct task_struct, thread_group);
2419 }
2420
2421 static inline int thread_group_empty(struct task_struct *p)
2422 {
2423 return list_empty(&p->thread_group);
2424 }
2425
2426 #define delay_group_leader(p) \
2427 (thread_group_leader(p) && !thread_group_empty(p))
2428
2429 /*
2430 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2431 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2432 * pins the final release of task.io_context. Also protects ->cpuset and
2433 * ->cgroup.subsys[]. And ->vfork_done.
2434 *
2435 * Nests both inside and outside of read_lock(&tasklist_lock).
2436 * It must not be nested with write_lock_irq(&tasklist_lock),
2437 * neither inside nor outside.
2438 */
2439 static inline void task_lock(struct task_struct *p)
2440 {
2441 spin_lock(&p->alloc_lock);
2442 }
2443
2444 static inline void task_unlock(struct task_struct *p)
2445 {
2446 spin_unlock(&p->alloc_lock);
2447 }
2448
2449 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2450 unsigned long *flags);
2451
2452 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2453 unsigned long *flags)
2454 {
2455 struct sighand_struct *ret;
2456
2457 ret = __lock_task_sighand(tsk, flags);
2458 (void)__cond_lock(&tsk->sighand->siglock, ret);
2459 return ret;
2460 }
2461
2462 static inline void unlock_task_sighand(struct task_struct *tsk,
2463 unsigned long *flags)
2464 {
2465 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2466 }
2467
2468 #ifdef CONFIG_CGROUPS
2469 static inline void threadgroup_change_begin(struct task_struct *tsk)
2470 {
2471 down_read(&tsk->signal->group_rwsem);
2472 }
2473 static inline void threadgroup_change_end(struct task_struct *tsk)
2474 {
2475 up_read(&tsk->signal->group_rwsem);
2476 }
2477
2478 /**
2479 * threadgroup_lock - lock threadgroup
2480 * @tsk: member task of the threadgroup to lock
2481 *
2482 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2483 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2484 * change ->group_leader/pid. This is useful for cases where the threadgroup
2485 * needs to stay stable across blockable operations.
2486 *
2487 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2488 * synchronization. While held, no new task will be added to threadgroup
2489 * and no existing live task will have its PF_EXITING set.
2490 *
2491 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2492 * sub-thread becomes a new leader.
2493 */
2494 static inline void threadgroup_lock(struct task_struct *tsk)
2495 {
2496 down_write(&tsk->signal->group_rwsem);
2497 }
2498
2499 /**
2500 * threadgroup_unlock - unlock threadgroup
2501 * @tsk: member task of the threadgroup to unlock
2502 *
2503 * Reverse threadgroup_lock().
2504 */
2505 static inline void threadgroup_unlock(struct task_struct *tsk)
2506 {
2507 up_write(&tsk->signal->group_rwsem);
2508 }
2509 #else
2510 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2511 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2512 static inline void threadgroup_lock(struct task_struct *tsk) {}
2513 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2514 #endif
2515
2516 #ifndef __HAVE_THREAD_FUNCTIONS
2517
2518 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2519 #define task_stack_page(task) ((task)->stack)
2520
2521 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2522 {
2523 *task_thread_info(p) = *task_thread_info(org);
2524 task_thread_info(p)->task = p;
2525 }
2526
2527 static inline unsigned long *end_of_stack(struct task_struct *p)
2528 {
2529 return (unsigned long *)(task_thread_info(p) + 1);
2530 }
2531
2532 #endif
2533
2534 static inline int object_is_on_stack(void *obj)
2535 {
2536 void *stack = task_stack_page(current);
2537
2538 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2539 }
2540
2541 extern void thread_info_cache_init(void);
2542
2543 #ifdef CONFIG_DEBUG_STACK_USAGE
2544 static inline unsigned long stack_not_used(struct task_struct *p)
2545 {
2546 unsigned long *n = end_of_stack(p);
2547
2548 do { /* Skip over canary */
2549 n++;
2550 } while (!*n);
2551
2552 return (unsigned long)n - (unsigned long)end_of_stack(p);
2553 }
2554 #endif
2555
2556 /* set thread flags in other task's structures
2557 * - see asm/thread_info.h for TIF_xxxx flags available
2558 */
2559 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2560 {
2561 set_ti_thread_flag(task_thread_info(tsk), flag);
2562 }
2563
2564 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2565 {
2566 clear_ti_thread_flag(task_thread_info(tsk), flag);
2567 }
2568
2569 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2570 {
2571 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2572 }
2573
2574 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2575 {
2576 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2577 }
2578
2579 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2580 {
2581 return test_ti_thread_flag(task_thread_info(tsk), flag);
2582 }
2583
2584 static inline void set_tsk_need_resched(struct task_struct *tsk)
2585 {
2586 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2587 }
2588
2589 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2590 {
2591 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2592 }
2593
2594 static inline int test_tsk_need_resched(struct task_struct *tsk)
2595 {
2596 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2597 }
2598
2599 static inline int restart_syscall(void)
2600 {
2601 set_tsk_thread_flag(current, TIF_SIGPENDING);
2602 return -ERESTARTNOINTR;
2603 }
2604
2605 static inline int signal_pending(struct task_struct *p)
2606 {
2607 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2608 }
2609
2610 static inline int __fatal_signal_pending(struct task_struct *p)
2611 {
2612 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2613 }
2614
2615 static inline int fatal_signal_pending(struct task_struct *p)
2616 {
2617 return signal_pending(p) && __fatal_signal_pending(p);
2618 }
2619
2620 static inline int signal_pending_state(long state, struct task_struct *p)
2621 {
2622 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2623 return 0;
2624 if (!signal_pending(p))
2625 return 0;
2626
2627 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2628 }
2629
2630 /*
2631 * cond_resched() and cond_resched_lock(): latency reduction via
2632 * explicit rescheduling in places that are safe. The return
2633 * value indicates whether a reschedule was done in fact.
2634 * cond_resched_lock() will drop the spinlock before scheduling,
2635 * cond_resched_softirq() will enable bhs before scheduling.
2636 */
2637 extern int _cond_resched(void);
2638
2639 #define cond_resched() ({ \
2640 __might_sleep(__FILE__, __LINE__, 0); \
2641 _cond_resched(); \
2642 })
2643
2644 extern int __cond_resched_lock(spinlock_t *lock);
2645
2646 #ifdef CONFIG_PREEMPT_COUNT
2647 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2648 #else
2649 #define PREEMPT_LOCK_OFFSET 0
2650 #endif
2651
2652 #define cond_resched_lock(lock) ({ \
2653 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2654 __cond_resched_lock(lock); \
2655 })
2656
2657 extern int __cond_resched_softirq(void);
2658
2659 #define cond_resched_softirq() ({ \
2660 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2661 __cond_resched_softirq(); \
2662 })
2663
2664 static inline void cond_resched_rcu(void)
2665 {
2666 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2667 rcu_read_unlock();
2668 cond_resched();
2669 rcu_read_lock();
2670 #endif
2671 }
2672
2673 /*
2674 * Does a critical section need to be broken due to another
2675 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2676 * but a general need for low latency)
2677 */
2678 static inline int spin_needbreak(spinlock_t *lock)
2679 {
2680 #ifdef CONFIG_PREEMPT
2681 return spin_is_contended(lock);
2682 #else
2683 return 0;
2684 #endif
2685 }
2686
2687 /*
2688 * Idle thread specific functions to determine the need_resched
2689 * polling state. We have two versions, one based on TS_POLLING in
2690 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2691 * thread_info.flags
2692 */
2693 #ifdef TS_POLLING
2694 static inline int tsk_is_polling(struct task_struct *p)
2695 {
2696 return task_thread_info(p)->status & TS_POLLING;
2697 }
2698 static inline void __current_set_polling(void)
2699 {
2700 current_thread_info()->status |= TS_POLLING;
2701 }
2702
2703 static inline bool __must_check current_set_polling_and_test(void)
2704 {
2705 __current_set_polling();
2706
2707 /*
2708 * Polling state must be visible before we test NEED_RESCHED,
2709 * paired by resched_task()
2710 */
2711 smp_mb();
2712
2713 return unlikely(tif_need_resched());
2714 }
2715
2716 static inline void __current_clr_polling(void)
2717 {
2718 current_thread_info()->status &= ~TS_POLLING;
2719 }
2720
2721 static inline bool __must_check current_clr_polling_and_test(void)
2722 {
2723 __current_clr_polling();
2724
2725 /*
2726 * Polling state must be visible before we test NEED_RESCHED,
2727 * paired by resched_task()
2728 */
2729 smp_mb();
2730
2731 return unlikely(tif_need_resched());
2732 }
2733 #elif defined(TIF_POLLING_NRFLAG)
2734 static inline int tsk_is_polling(struct task_struct *p)
2735 {
2736 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2737 }
2738
2739 static inline void __current_set_polling(void)
2740 {
2741 set_thread_flag(TIF_POLLING_NRFLAG);
2742 }
2743
2744 static inline bool __must_check current_set_polling_and_test(void)
2745 {
2746 __current_set_polling();
2747
2748 /*
2749 * Polling state must be visible before we test NEED_RESCHED,
2750 * paired by resched_task()
2751 *
2752 * XXX: assumes set/clear bit are identical barrier wise.
2753 */
2754 smp_mb__after_clear_bit();
2755
2756 return unlikely(tif_need_resched());
2757 }
2758
2759 static inline void __current_clr_polling(void)
2760 {
2761 clear_thread_flag(TIF_POLLING_NRFLAG);
2762 }
2763
2764 static inline bool __must_check current_clr_polling_and_test(void)
2765 {
2766 __current_clr_polling();
2767
2768 /*
2769 * Polling state must be visible before we test NEED_RESCHED,
2770 * paired by resched_task()
2771 */
2772 smp_mb__after_clear_bit();
2773
2774 return unlikely(tif_need_resched());
2775 }
2776
2777 #else
2778 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2779 static inline void __current_set_polling(void) { }
2780 static inline void __current_clr_polling(void) { }
2781
2782 static inline bool __must_check current_set_polling_and_test(void)
2783 {
2784 return unlikely(tif_need_resched());
2785 }
2786 static inline bool __must_check current_clr_polling_and_test(void)
2787 {
2788 return unlikely(tif_need_resched());
2789 }
2790 #endif
2791
2792 static inline void current_clr_polling(void)
2793 {
2794 __current_clr_polling();
2795
2796 /*
2797 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2798 * Once the bit is cleared, we'll get IPIs with every new
2799 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2800 * fold.
2801 */
2802 smp_mb(); /* paired with resched_task() */
2803
2804 preempt_fold_need_resched();
2805 }
2806
2807 static __always_inline bool need_resched(void)
2808 {
2809 return unlikely(tif_need_resched());
2810 }
2811
2812 /*
2813 * Thread group CPU time accounting.
2814 */
2815 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2816 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2817
2818 static inline void thread_group_cputime_init(struct signal_struct *sig)
2819 {
2820 raw_spin_lock_init(&sig->cputimer.lock);
2821 }
2822
2823 /*
2824 * Reevaluate whether the task has signals pending delivery.
2825 * Wake the task if so.
2826 * This is required every time the blocked sigset_t changes.
2827 * callers must hold sighand->siglock.
2828 */
2829 extern void recalc_sigpending_and_wake(struct task_struct *t);
2830 extern void recalc_sigpending(void);
2831
2832 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2833
2834 static inline void signal_wake_up(struct task_struct *t, bool resume)
2835 {
2836 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2837 }
2838 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2839 {
2840 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2841 }
2842
2843 /*
2844 * Wrappers for p->thread_info->cpu access. No-op on UP.
2845 */
2846 #ifdef CONFIG_SMP
2847
2848 static inline unsigned int task_cpu(const struct task_struct *p)
2849 {
2850 return task_thread_info(p)->cpu;
2851 }
2852
2853 static inline int task_node(const struct task_struct *p)
2854 {
2855 return cpu_to_node(task_cpu(p));
2856 }
2857
2858 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2859
2860 #else
2861
2862 static inline unsigned int task_cpu(const struct task_struct *p)
2863 {
2864 return 0;
2865 }
2866
2867 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2868 {
2869 }
2870
2871 #endif /* CONFIG_SMP */
2872
2873 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2874 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2875
2876 #ifdef CONFIG_CGROUP_SCHED
2877 extern struct task_group root_task_group;
2878 #endif /* CONFIG_CGROUP_SCHED */
2879
2880 extern int task_can_switch_user(struct user_struct *up,
2881 struct task_struct *tsk);
2882
2883 #ifdef CONFIG_TASK_XACCT
2884 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2885 {
2886 tsk->ioac.rchar += amt;
2887 }
2888
2889 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2890 {
2891 tsk->ioac.wchar += amt;
2892 }
2893
2894 static inline void inc_syscr(struct task_struct *tsk)
2895 {
2896 tsk->ioac.syscr++;
2897 }
2898
2899 static inline void inc_syscw(struct task_struct *tsk)
2900 {
2901 tsk->ioac.syscw++;
2902 }
2903 #else
2904 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2905 {
2906 }
2907
2908 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2909 {
2910 }
2911
2912 static inline void inc_syscr(struct task_struct *tsk)
2913 {
2914 }
2915
2916 static inline void inc_syscw(struct task_struct *tsk)
2917 {
2918 }
2919 #endif
2920
2921 #ifndef TASK_SIZE_OF
2922 #define TASK_SIZE_OF(tsk) TASK_SIZE
2923 #endif
2924
2925 #ifdef CONFIG_MM_OWNER
2926 extern void mm_update_next_owner(struct mm_struct *mm);
2927 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2928 #else
2929 static inline void mm_update_next_owner(struct mm_struct *mm)
2930 {
2931 }
2932
2933 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2934 {
2935 }
2936 #endif /* CONFIG_MM_OWNER */
2937
2938 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2939 unsigned int limit)
2940 {
2941 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2942 }
2943
2944 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2945 unsigned int limit)
2946 {
2947 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2948 }
2949
2950 static inline unsigned long rlimit(unsigned int limit)
2951 {
2952 return task_rlimit(current, limit);
2953 }
2954
2955 static inline unsigned long rlimit_max(unsigned int limit)
2956 {
2957 return task_rlimit_max(current, limit);
2958 }
2959
2960 #endif