42888d715fb1dfd8d27d6c0780253105c1aeb263
[deliverable/linux.git] / include / linux / sched.h
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3
4 #include <uapi/linux/sched.h>
5
6 #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 <linux/cputime.h>
33
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/shm.h>
37 #include <linux/signal.h>
38 #include <linux/compiler.h>
39 #include <linux/completion.h>
40 #include <linux/pid.h>
41 #include <linux/percpu.h>
42 #include <linux/topology.h>
43 #include <linux/proportions.h>
44 #include <linux/seccomp.h>
45 #include <linux/rcupdate.h>
46 #include <linux/rculist.h>
47 #include <linux/rtmutex.h>
48
49 #include <linux/time.h>
50 #include <linux/param.h>
51 #include <linux/resource.h>
52 #include <linux/timer.h>
53 #include <linux/hrtimer.h>
54 #include <linux/task_io_accounting.h>
55 #include <linux/latencytop.h>
56 #include <linux/cred.h>
57 #include <linux/llist.h>
58 #include <linux/uidgid.h>
59 #include <linux/gfp.h>
60
61 #include <asm/processor.h>
62
63 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
64
65 /*
66 * Extended scheduling parameters data structure.
67 *
68 * This is needed because the original struct sched_param can not be
69 * altered without introducing ABI issues with legacy applications
70 * (e.g., in sched_getparam()).
71 *
72 * However, the possibility of specifying more than just a priority for
73 * the tasks may be useful for a wide variety of application fields, e.g.,
74 * multimedia, streaming, automation and control, and many others.
75 *
76 * This variant (sched_attr) is meant at describing a so-called
77 * sporadic time-constrained task. In such model a task is specified by:
78 * - the activation period or minimum instance inter-arrival time;
79 * - the maximum (or average, depending on the actual scheduling
80 * discipline) computation time of all instances, a.k.a. runtime;
81 * - the deadline (relative to the actual activation time) of each
82 * instance.
83 * Very briefly, a periodic (sporadic) task asks for the execution of
84 * some specific computation --which is typically called an instance--
85 * (at most) every period. Moreover, each instance typically lasts no more
86 * than the runtime and must be completed by time instant t equal to
87 * the instance activation time + the deadline.
88 *
89 * This is reflected by the actual fields of the sched_attr structure:
90 *
91 * @size size of the structure, for fwd/bwd compat.
92 *
93 * @sched_policy task's scheduling policy
94 * @sched_flags for customizing the scheduler behaviour
95 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
96 * @sched_priority task's static priority (SCHED_FIFO/RR)
97 * @sched_deadline representative of the task's deadline
98 * @sched_runtime representative of the task's runtime
99 * @sched_period representative of the task's period
100 *
101 * Given this task model, there are a multiplicity of scheduling algorithms
102 * and policies, that can be used to ensure all the tasks will make their
103 * timing constraints.
104 *
105 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
106 * only user of this new interface. More information about the algorithm
107 * available in the scheduling class file or in Documentation/.
108 */
109 struct sched_attr {
110 u32 size;
111
112 u32 sched_policy;
113 u64 sched_flags;
114
115 /* SCHED_NORMAL, SCHED_BATCH */
116 s32 sched_nice;
117
118 /* SCHED_FIFO, SCHED_RR */
119 u32 sched_priority;
120
121 /* SCHED_DEADLINE */
122 u64 sched_runtime;
123 u64 sched_deadline;
124 u64 sched_period;
125 };
126
127 struct exec_domain;
128 struct futex_pi_state;
129 struct robust_list_head;
130 struct bio_list;
131 struct fs_struct;
132 struct perf_event_context;
133 struct blk_plug;
134 struct filename;
135
136 #define VMACACHE_BITS 2
137 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
138 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
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 void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
173
174 extern void calc_global_load(unsigned long ticks);
175 extern void update_cpu_load_nohz(void);
176
177 extern unsigned long get_parent_ip(unsigned long addr);
178
179 extern void dump_cpu_task(int cpu);
180
181 struct seq_file;
182 struct cfs_rq;
183 struct task_group;
184 #ifdef CONFIG_SCHED_DEBUG
185 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
186 extern void proc_sched_set_task(struct task_struct *p);
187 extern void
188 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
189 #endif
190
191 /*
192 * Task state bitmask. NOTE! These bits are also
193 * encoded in fs/proc/array.c: get_task_state().
194 *
195 * We have two separate sets of flags: task->state
196 * is about runnability, while task->exit_state are
197 * about the task exiting. Confusing, but this way
198 * modifying one set can't modify the other one by
199 * mistake.
200 */
201 #define TASK_RUNNING 0
202 #define TASK_INTERRUPTIBLE 1
203 #define TASK_UNINTERRUPTIBLE 2
204 #define __TASK_STOPPED 4
205 #define __TASK_TRACED 8
206 /* in tsk->exit_state */
207 #define EXIT_DEAD 16
208 #define EXIT_ZOMBIE 32
209 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
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 "RSDTtXZxKWP"
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(int pinned);
297 #else
298 static inline void nohz_balance_enter_idle(int cpu) { }
299 static inline void set_cpu_sd_state_idle(void) { }
300 static inline int get_nohz_timer_target(int pinned)
301 {
302 return smp_processor_id();
303 }
304 #endif
305
306 /*
307 * Only dump TASK_* tasks. (0 for all tasks)
308 */
309 extern void show_state_filter(unsigned long state_filter);
310
311 static inline void show_state(void)
312 {
313 show_state_filter(0);
314 }
315
316 extern void show_regs(struct pt_regs *);
317
318 /*
319 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
320 * task), SP is the stack pointer of the first frame that should be shown in the back
321 * trace (or NULL if the entire call-chain of the task should be shown).
322 */
323 extern void show_stack(struct task_struct *task, unsigned long *sp);
324
325 void io_schedule(void);
326 long io_schedule_timeout(long timeout);
327
328 extern void cpu_init (void);
329 extern void trap_init(void);
330 extern void update_process_times(int user);
331 extern void scheduler_tick(void);
332
333 extern void sched_show_task(struct task_struct *p);
334
335 #ifdef CONFIG_LOCKUP_DETECTOR
336 extern void touch_softlockup_watchdog(void);
337 extern void touch_softlockup_watchdog_sync(void);
338 extern void touch_all_softlockup_watchdogs(void);
339 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
340 void __user *buffer,
341 size_t *lenp, loff_t *ppos);
342 extern unsigned int softlockup_panic;
343 void lockup_detector_init(void);
344 #else
345 static inline void touch_softlockup_watchdog(void)
346 {
347 }
348 static inline void touch_softlockup_watchdog_sync(void)
349 {
350 }
351 static inline void touch_all_softlockup_watchdogs(void)
352 {
353 }
354 static inline void lockup_detector_init(void)
355 {
356 }
357 #endif
358
359 #ifdef CONFIG_DETECT_HUNG_TASK
360 void reset_hung_task_detector(void);
361 #else
362 static inline void reset_hung_task_detector(void)
363 {
364 }
365 #endif
366
367 /* Attach to any functions which should be ignored in wchan output. */
368 #define __sched __attribute__((__section__(".sched.text")))
369
370 /* Linker adds these: start and end of __sched functions */
371 extern char __sched_text_start[], __sched_text_end[];
372
373 /* Is this address in the __sched functions? */
374 extern int in_sched_functions(unsigned long addr);
375
376 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
377 extern signed long schedule_timeout(signed long timeout);
378 extern signed long schedule_timeout_interruptible(signed long timeout);
379 extern signed long schedule_timeout_killable(signed long timeout);
380 extern signed long schedule_timeout_uninterruptible(signed long timeout);
381 asmlinkage void schedule(void);
382 extern void schedule_preempt_disabled(void);
383
384 struct nsproxy;
385 struct user_namespace;
386
387 #ifdef CONFIG_MMU
388 extern void arch_pick_mmap_layout(struct mm_struct *mm);
389 extern unsigned long
390 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
391 unsigned long, unsigned long);
392 extern unsigned long
393 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
394 unsigned long len, unsigned long pgoff,
395 unsigned long flags);
396 #else
397 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
398 #endif
399
400 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
401 #define SUID_DUMP_USER 1 /* Dump as user of process */
402 #define SUID_DUMP_ROOT 2 /* Dump as root */
403
404 /* mm flags */
405
406 /* for SUID_DUMP_* above */
407 #define MMF_DUMPABLE_BITS 2
408 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
409
410 extern void set_dumpable(struct mm_struct *mm, int value);
411 /*
412 * This returns the actual value of the suid_dumpable flag. For things
413 * that are using this for checking for privilege transitions, it must
414 * test against SUID_DUMP_USER rather than treating it as a boolean
415 * value.
416 */
417 static inline int __get_dumpable(unsigned long mm_flags)
418 {
419 return mm_flags & MMF_DUMPABLE_MASK;
420 }
421
422 static inline int get_dumpable(struct mm_struct *mm)
423 {
424 return __get_dumpable(mm->flags);
425 }
426
427 /* coredump filter bits */
428 #define MMF_DUMP_ANON_PRIVATE 2
429 #define MMF_DUMP_ANON_SHARED 3
430 #define MMF_DUMP_MAPPED_PRIVATE 4
431 #define MMF_DUMP_MAPPED_SHARED 5
432 #define MMF_DUMP_ELF_HEADERS 6
433 #define MMF_DUMP_HUGETLB_PRIVATE 7
434 #define MMF_DUMP_HUGETLB_SHARED 8
435
436 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
437 #define MMF_DUMP_FILTER_BITS 7
438 #define MMF_DUMP_FILTER_MASK \
439 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
440 #define MMF_DUMP_FILTER_DEFAULT \
441 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
442 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
443
444 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
445 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
446 #else
447 # define MMF_DUMP_MASK_DEFAULT_ELF 0
448 #endif
449 /* leave room for more dump flags */
450 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
451 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
452 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
453
454 #define MMF_HAS_UPROBES 19 /* has uprobes */
455 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
456
457 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
458
459 struct sighand_struct {
460 atomic_t count;
461 struct k_sigaction action[_NSIG];
462 spinlock_t siglock;
463 wait_queue_head_t signalfd_wqh;
464 };
465
466 struct pacct_struct {
467 int ac_flag;
468 long ac_exitcode;
469 unsigned long ac_mem;
470 cputime_t ac_utime, ac_stime;
471 unsigned long ac_minflt, ac_majflt;
472 };
473
474 struct cpu_itimer {
475 cputime_t expires;
476 cputime_t incr;
477 u32 error;
478 u32 incr_error;
479 };
480
481 /**
482 * struct cputime - snaphsot of system and user cputime
483 * @utime: time spent in user mode
484 * @stime: time spent in system mode
485 *
486 * Gathers a generic snapshot of user and system time.
487 */
488 struct cputime {
489 cputime_t utime;
490 cputime_t stime;
491 };
492
493 /**
494 * struct task_cputime - collected CPU time counts
495 * @utime: time spent in user mode, in &cputime_t units
496 * @stime: time spent in kernel mode, in &cputime_t units
497 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
498 *
499 * This is an extension of struct cputime that includes the total runtime
500 * spent by the task from the scheduler point of view.
501 *
502 * As a result, this structure groups together three kinds of CPU time
503 * that are tracked for threads and thread groups. Most things considering
504 * CPU time want to group these counts together and treat all three
505 * of them in parallel.
506 */
507 struct task_cputime {
508 cputime_t utime;
509 cputime_t stime;
510 unsigned long long sum_exec_runtime;
511 };
512 /* Alternate field names when used to cache expirations. */
513 #define prof_exp stime
514 #define virt_exp utime
515 #define sched_exp sum_exec_runtime
516
517 #define INIT_CPUTIME \
518 (struct task_cputime) { \
519 .utime = 0, \
520 .stime = 0, \
521 .sum_exec_runtime = 0, \
522 }
523
524 #ifdef CONFIG_PREEMPT_COUNT
525 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
526 #else
527 #define PREEMPT_DISABLED PREEMPT_ENABLED
528 #endif
529
530 /*
531 * Disable preemption until the scheduler is running.
532 * Reset by start_kernel()->sched_init()->init_idle().
533 *
534 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
535 * before the scheduler is active -- see should_resched().
536 */
537 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
538
539 /**
540 * struct thread_group_cputimer - thread group interval timer counts
541 * @cputime: thread group interval timers.
542 * @running: non-zero when there are timers running and
543 * @cputime receives updates.
544 * @lock: lock for fields in this struct.
545 *
546 * This structure contains the version of task_cputime, above, that is
547 * used for thread group CPU timer calculations.
548 */
549 struct thread_group_cputimer {
550 struct task_cputime cputime;
551 int running;
552 raw_spinlock_t lock;
553 };
554
555 #include <linux/rwsem.h>
556 struct autogroup;
557
558 /*
559 * NOTE! "signal_struct" does not have its own
560 * locking, because a shared signal_struct always
561 * implies a shared sighand_struct, so locking
562 * sighand_struct is always a proper superset of
563 * the locking of signal_struct.
564 */
565 struct signal_struct {
566 atomic_t sigcnt;
567 atomic_t live;
568 int nr_threads;
569 struct list_head thread_head;
570
571 wait_queue_head_t wait_chldexit; /* for wait4() */
572
573 /* current thread group signal load-balancing target: */
574 struct task_struct *curr_target;
575
576 /* shared signal handling: */
577 struct sigpending shared_pending;
578
579 /* thread group exit support */
580 int group_exit_code;
581 /* overloaded:
582 * - notify group_exit_task when ->count is equal to notify_count
583 * - everyone except group_exit_task is stopped during signal delivery
584 * of fatal signals, group_exit_task processes the signal.
585 */
586 int notify_count;
587 struct task_struct *group_exit_task;
588
589 /* thread group stop support, overloads group_exit_code too */
590 int group_stop_count;
591 unsigned int flags; /* see SIGNAL_* flags below */
592
593 /*
594 * PR_SET_CHILD_SUBREAPER marks a process, like a service
595 * manager, to re-parent orphan (double-forking) child processes
596 * to this process instead of 'init'. The service manager is
597 * able to receive SIGCHLD signals and is able to investigate
598 * the process until it calls wait(). All children of this
599 * process will inherit a flag if they should look for a
600 * child_subreaper process at exit.
601 */
602 unsigned int is_child_subreaper:1;
603 unsigned int has_child_subreaper:1;
604
605 /* POSIX.1b Interval Timers */
606 int posix_timer_id;
607 struct list_head posix_timers;
608
609 /* ITIMER_REAL timer for the process */
610 struct hrtimer real_timer;
611 struct pid *leader_pid;
612 ktime_t it_real_incr;
613
614 /*
615 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
616 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
617 * values are defined to 0 and 1 respectively
618 */
619 struct cpu_itimer it[2];
620
621 /*
622 * Thread group totals for process CPU timers.
623 * See thread_group_cputimer(), et al, for details.
624 */
625 struct thread_group_cputimer cputimer;
626
627 /* Earliest-expiration cache. */
628 struct task_cputime cputime_expires;
629
630 struct list_head cpu_timers[3];
631
632 struct pid *tty_old_pgrp;
633
634 /* boolean value for session group leader */
635 int leader;
636
637 struct tty_struct *tty; /* NULL if no tty */
638
639 #ifdef CONFIG_SCHED_AUTOGROUP
640 struct autogroup *autogroup;
641 #endif
642 /*
643 * Cumulative resource counters for dead threads in the group,
644 * and for reaped dead child processes forked by this group.
645 * Live threads maintain their own counters and add to these
646 * in __exit_signal, except for the group leader.
647 */
648 cputime_t utime, stime, cutime, cstime;
649 cputime_t gtime;
650 cputime_t cgtime;
651 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
652 struct cputime prev_cputime;
653 #endif
654 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
655 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
656 unsigned long inblock, oublock, cinblock, coublock;
657 unsigned long maxrss, cmaxrss;
658 struct task_io_accounting ioac;
659
660 /*
661 * Cumulative ns of schedule CPU time fo dead threads in the
662 * group, not including a zombie group leader, (This only differs
663 * from jiffies_to_ns(utime + stime) if sched_clock uses something
664 * other than jiffies.)
665 */
666 unsigned long long sum_sched_runtime;
667
668 /*
669 * We don't bother to synchronize most readers of this at all,
670 * because there is no reader checking a limit that actually needs
671 * to get both rlim_cur and rlim_max atomically, and either one
672 * alone is a single word that can safely be read normally.
673 * getrlimit/setrlimit use task_lock(current->group_leader) to
674 * protect this instead of the siglock, because they really
675 * have no need to disable irqs.
676 */
677 struct rlimit rlim[RLIM_NLIMITS];
678
679 #ifdef CONFIG_BSD_PROCESS_ACCT
680 struct pacct_struct pacct; /* per-process accounting information */
681 #endif
682 #ifdef CONFIG_TASKSTATS
683 struct taskstats *stats;
684 #endif
685 #ifdef CONFIG_AUDIT
686 unsigned audit_tty;
687 unsigned audit_tty_log_passwd;
688 struct tty_audit_buf *tty_audit_buf;
689 #endif
690 #ifdef CONFIG_CGROUPS
691 /*
692 * group_rwsem prevents new tasks from entering the threadgroup and
693 * member tasks from exiting,a more specifically, setting of
694 * PF_EXITING. fork and exit paths are protected with this rwsem
695 * using threadgroup_change_begin/end(). Users which require
696 * threadgroup to remain stable should use threadgroup_[un]lock()
697 * which also takes care of exec path. Currently, cgroup is the
698 * only user.
699 */
700 struct rw_semaphore group_rwsem;
701 #endif
702
703 oom_flags_t oom_flags;
704 short oom_score_adj; /* OOM kill score adjustment */
705 short oom_score_adj_min; /* OOM kill score adjustment min value.
706 * Only settable by CAP_SYS_RESOURCE. */
707
708 struct mutex cred_guard_mutex; /* guard against foreign influences on
709 * credential calculations
710 * (notably. ptrace) */
711 };
712
713 /*
714 * Bits in flags field of signal_struct.
715 */
716 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
717 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
718 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
719 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
720 /*
721 * Pending notifications to parent.
722 */
723 #define SIGNAL_CLD_STOPPED 0x00000010
724 #define SIGNAL_CLD_CONTINUED 0x00000020
725 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
726
727 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
728
729 /* If true, all threads except ->group_exit_task have pending SIGKILL */
730 static inline int signal_group_exit(const struct signal_struct *sig)
731 {
732 return (sig->flags & SIGNAL_GROUP_EXIT) ||
733 (sig->group_exit_task != NULL);
734 }
735
736 /*
737 * Some day this will be a full-fledged user tracking system..
738 */
739 struct user_struct {
740 atomic_t __count; /* reference count */
741 atomic_t processes; /* How many processes does this user have? */
742 atomic_t sigpending; /* How many pending signals does this user have? */
743 #ifdef CONFIG_INOTIFY_USER
744 atomic_t inotify_watches; /* How many inotify watches does this user have? */
745 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
746 #endif
747 #ifdef CONFIG_FANOTIFY
748 atomic_t fanotify_listeners;
749 #endif
750 #ifdef CONFIG_EPOLL
751 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
752 #endif
753 #ifdef CONFIG_POSIX_MQUEUE
754 /* protected by mq_lock */
755 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
756 #endif
757 unsigned long locked_shm; /* How many pages of mlocked shm ? */
758
759 #ifdef CONFIG_KEYS
760 struct key *uid_keyring; /* UID specific keyring */
761 struct key *session_keyring; /* UID's default session keyring */
762 #endif
763
764 /* Hash table maintenance information */
765 struct hlist_node uidhash_node;
766 kuid_t uid;
767
768 #ifdef CONFIG_PERF_EVENTS
769 atomic_long_t locked_vm;
770 #endif
771 };
772
773 extern int uids_sysfs_init(void);
774
775 extern struct user_struct *find_user(kuid_t);
776
777 extern struct user_struct root_user;
778 #define INIT_USER (&root_user)
779
780
781 struct backing_dev_info;
782 struct reclaim_state;
783
784 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
785 struct sched_info {
786 /* cumulative counters */
787 unsigned long pcount; /* # of times run on this cpu */
788 unsigned long long run_delay; /* time spent waiting on a runqueue */
789
790 /* timestamps */
791 unsigned long long last_arrival,/* when we last ran on a cpu */
792 last_queued; /* when we were last queued to run */
793 };
794 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
795
796 #ifdef CONFIG_TASK_DELAY_ACCT
797 struct task_delay_info {
798 spinlock_t lock;
799 unsigned int flags; /* Private per-task flags */
800
801 /* For each stat XXX, add following, aligned appropriately
802 *
803 * struct timespec XXX_start, XXX_end;
804 * u64 XXX_delay;
805 * u32 XXX_count;
806 *
807 * Atomicity of updates to XXX_delay, XXX_count protected by
808 * single lock above (split into XXX_lock if contention is an issue).
809 */
810
811 /*
812 * XXX_count is incremented on every XXX operation, the delay
813 * associated with the operation is added to XXX_delay.
814 * XXX_delay contains the accumulated delay time in nanoseconds.
815 */
816 u64 blkio_start; /* Shared by blkio, swapin */
817 u64 blkio_delay; /* wait for sync block io completion */
818 u64 swapin_delay; /* wait for swapin block io completion */
819 u32 blkio_count; /* total count of the number of sync block */
820 /* io operations performed */
821 u32 swapin_count; /* total count of the number of swapin block */
822 /* io operations performed */
823
824 u64 freepages_start;
825 u64 freepages_delay; /* wait for memory reclaim */
826 u32 freepages_count; /* total count of memory reclaim */
827 };
828 #endif /* CONFIG_TASK_DELAY_ACCT */
829
830 static inline int sched_info_on(void)
831 {
832 #ifdef CONFIG_SCHEDSTATS
833 return 1;
834 #elif defined(CONFIG_TASK_DELAY_ACCT)
835 extern int delayacct_on;
836 return delayacct_on;
837 #else
838 return 0;
839 #endif
840 }
841
842 enum cpu_idle_type {
843 CPU_IDLE,
844 CPU_NOT_IDLE,
845 CPU_NEWLY_IDLE,
846 CPU_MAX_IDLE_TYPES
847 };
848
849 /*
850 * Increase resolution of cpu_capacity calculations
851 */
852 #define SCHED_CAPACITY_SHIFT 10
853 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
854
855 /*
856 * sched-domains (multiprocessor balancing) declarations:
857 */
858 #ifdef CONFIG_SMP
859 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
860 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
861 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
862 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
863 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
864 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
865 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
866 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
867 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
868 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
869 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
870 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
871 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
872 #define SD_NUMA 0x4000 /* cross-node balancing */
873
874 #ifdef CONFIG_SCHED_SMT
875 static inline int cpu_smt_flags(void)
876 {
877 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
878 }
879 #endif
880
881 #ifdef CONFIG_SCHED_MC
882 static inline int cpu_core_flags(void)
883 {
884 return SD_SHARE_PKG_RESOURCES;
885 }
886 #endif
887
888 #ifdef CONFIG_NUMA
889 static inline int cpu_numa_flags(void)
890 {
891 return SD_NUMA;
892 }
893 #endif
894
895 struct sched_domain_attr {
896 int relax_domain_level;
897 };
898
899 #define SD_ATTR_INIT (struct sched_domain_attr) { \
900 .relax_domain_level = -1, \
901 }
902
903 extern int sched_domain_level_max;
904
905 struct sched_group;
906
907 struct sched_domain {
908 /* These fields must be setup */
909 struct sched_domain *parent; /* top domain must be null terminated */
910 struct sched_domain *child; /* bottom domain must be null terminated */
911 struct sched_group *groups; /* the balancing groups of the domain */
912 unsigned long min_interval; /* Minimum balance interval ms */
913 unsigned long max_interval; /* Maximum balance interval ms */
914 unsigned int busy_factor; /* less balancing by factor if busy */
915 unsigned int imbalance_pct; /* No balance until over watermark */
916 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
917 unsigned int busy_idx;
918 unsigned int idle_idx;
919 unsigned int newidle_idx;
920 unsigned int wake_idx;
921 unsigned int forkexec_idx;
922 unsigned int smt_gain;
923
924 int nohz_idle; /* NOHZ IDLE status */
925 int flags; /* See SD_* */
926 int level;
927
928 /* Runtime fields. */
929 unsigned long last_balance; /* init to jiffies. units in jiffies */
930 unsigned int balance_interval; /* initialise to 1. units in ms. */
931 unsigned int nr_balance_failed; /* initialise to 0 */
932
933 /* idle_balance() stats */
934 u64 max_newidle_lb_cost;
935 unsigned long next_decay_max_lb_cost;
936
937 #ifdef CONFIG_SCHEDSTATS
938 /* load_balance() stats */
939 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
940 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
941 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
942 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
943 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
944 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
945 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
946 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
947
948 /* Active load balancing */
949 unsigned int alb_count;
950 unsigned int alb_failed;
951 unsigned int alb_pushed;
952
953 /* SD_BALANCE_EXEC stats */
954 unsigned int sbe_count;
955 unsigned int sbe_balanced;
956 unsigned int sbe_pushed;
957
958 /* SD_BALANCE_FORK stats */
959 unsigned int sbf_count;
960 unsigned int sbf_balanced;
961 unsigned int sbf_pushed;
962
963 /* try_to_wake_up() stats */
964 unsigned int ttwu_wake_remote;
965 unsigned int ttwu_move_affine;
966 unsigned int ttwu_move_balance;
967 #endif
968 #ifdef CONFIG_SCHED_DEBUG
969 char *name;
970 #endif
971 union {
972 void *private; /* used during construction */
973 struct rcu_head rcu; /* used during destruction */
974 };
975
976 unsigned int span_weight;
977 /*
978 * Span of all CPUs in this domain.
979 *
980 * NOTE: this field is variable length. (Allocated dynamically
981 * by attaching extra space to the end of the structure,
982 * depending on how many CPUs the kernel has booted up with)
983 */
984 unsigned long span[0];
985 };
986
987 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
988 {
989 return to_cpumask(sd->span);
990 }
991
992 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
993 struct sched_domain_attr *dattr_new);
994
995 /* Allocate an array of sched domains, for partition_sched_domains(). */
996 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
997 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
998
999 bool cpus_share_cache(int this_cpu, int that_cpu);
1000
1001 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1002 typedef int (*sched_domain_flags_f)(void);
1003
1004 #define SDTL_OVERLAP 0x01
1005
1006 struct sd_data {
1007 struct sched_domain **__percpu sd;
1008 struct sched_group **__percpu sg;
1009 struct sched_group_capacity **__percpu sgc;
1010 };
1011
1012 struct sched_domain_topology_level {
1013 sched_domain_mask_f mask;
1014 sched_domain_flags_f sd_flags;
1015 int flags;
1016 int numa_level;
1017 struct sd_data data;
1018 #ifdef CONFIG_SCHED_DEBUG
1019 char *name;
1020 #endif
1021 };
1022
1023 extern struct sched_domain_topology_level *sched_domain_topology;
1024
1025 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1026
1027 #ifdef CONFIG_SCHED_DEBUG
1028 # define SD_INIT_NAME(type) .name = #type
1029 #else
1030 # define SD_INIT_NAME(type)
1031 #endif
1032
1033 #else /* CONFIG_SMP */
1034
1035 struct sched_domain_attr;
1036
1037 static inline void
1038 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1039 struct sched_domain_attr *dattr_new)
1040 {
1041 }
1042
1043 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1044 {
1045 return true;
1046 }
1047
1048 #endif /* !CONFIG_SMP */
1049
1050
1051 struct io_context; /* See blkdev.h */
1052
1053
1054 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1055 extern void prefetch_stack(struct task_struct *t);
1056 #else
1057 static inline void prefetch_stack(struct task_struct *t) { }
1058 #endif
1059
1060 struct audit_context; /* See audit.c */
1061 struct mempolicy;
1062 struct pipe_inode_info;
1063 struct uts_namespace;
1064
1065 struct load_weight {
1066 unsigned long weight;
1067 u32 inv_weight;
1068 };
1069
1070 struct sched_avg {
1071 /*
1072 * These sums represent an infinite geometric series and so are bound
1073 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1074 * choices of y < 1-2^(-32)*1024.
1075 */
1076 u32 runnable_avg_sum, runnable_avg_period;
1077 u64 last_runnable_update;
1078 s64 decay_count;
1079 unsigned long load_avg_contrib;
1080 };
1081
1082 #ifdef CONFIG_SCHEDSTATS
1083 struct sched_statistics {
1084 u64 wait_start;
1085 u64 wait_max;
1086 u64 wait_count;
1087 u64 wait_sum;
1088 u64 iowait_count;
1089 u64 iowait_sum;
1090
1091 u64 sleep_start;
1092 u64 sleep_max;
1093 s64 sum_sleep_runtime;
1094
1095 u64 block_start;
1096 u64 block_max;
1097 u64 exec_max;
1098 u64 slice_max;
1099
1100 u64 nr_migrations_cold;
1101 u64 nr_failed_migrations_affine;
1102 u64 nr_failed_migrations_running;
1103 u64 nr_failed_migrations_hot;
1104 u64 nr_forced_migrations;
1105
1106 u64 nr_wakeups;
1107 u64 nr_wakeups_sync;
1108 u64 nr_wakeups_migrate;
1109 u64 nr_wakeups_local;
1110 u64 nr_wakeups_remote;
1111 u64 nr_wakeups_affine;
1112 u64 nr_wakeups_affine_attempts;
1113 u64 nr_wakeups_passive;
1114 u64 nr_wakeups_idle;
1115 };
1116 #endif
1117
1118 struct sched_entity {
1119 struct load_weight load; /* for load-balancing */
1120 struct rb_node run_node;
1121 struct list_head group_node;
1122 unsigned int on_rq;
1123
1124 u64 exec_start;
1125 u64 sum_exec_runtime;
1126 u64 vruntime;
1127 u64 prev_sum_exec_runtime;
1128
1129 u64 nr_migrations;
1130
1131 #ifdef CONFIG_SCHEDSTATS
1132 struct sched_statistics statistics;
1133 #endif
1134
1135 #ifdef CONFIG_FAIR_GROUP_SCHED
1136 int depth;
1137 struct sched_entity *parent;
1138 /* rq on which this entity is (to be) queued: */
1139 struct cfs_rq *cfs_rq;
1140 /* rq "owned" by this entity/group: */
1141 struct cfs_rq *my_q;
1142 #endif
1143
1144 #ifdef CONFIG_SMP
1145 /* Per-entity load-tracking */
1146 struct sched_avg avg;
1147 #endif
1148 };
1149
1150 struct sched_rt_entity {
1151 struct list_head run_list;
1152 unsigned long timeout;
1153 unsigned long watchdog_stamp;
1154 unsigned int time_slice;
1155
1156 struct sched_rt_entity *back;
1157 #ifdef CONFIG_RT_GROUP_SCHED
1158 struct sched_rt_entity *parent;
1159 /* rq on which this entity is (to be) queued: */
1160 struct rt_rq *rt_rq;
1161 /* rq "owned" by this entity/group: */
1162 struct rt_rq *my_q;
1163 #endif
1164 };
1165
1166 struct sched_dl_entity {
1167 struct rb_node rb_node;
1168
1169 /*
1170 * Original scheduling parameters. Copied here from sched_attr
1171 * during sched_setattr(), they will remain the same until
1172 * the next sched_setattr().
1173 */
1174 u64 dl_runtime; /* maximum runtime for each instance */
1175 u64 dl_deadline; /* relative deadline of each instance */
1176 u64 dl_period; /* separation of two instances (period) */
1177 u64 dl_bw; /* dl_runtime / dl_deadline */
1178
1179 /*
1180 * Actual scheduling parameters. Initialized with the values above,
1181 * they are continously updated during task execution. Note that
1182 * the remaining runtime could be < 0 in case we are in overrun.
1183 */
1184 s64 runtime; /* remaining runtime for this instance */
1185 u64 deadline; /* absolute deadline for this instance */
1186 unsigned int flags; /* specifying the scheduler behaviour */
1187
1188 /*
1189 * Some bool flags:
1190 *
1191 * @dl_throttled tells if we exhausted the runtime. If so, the
1192 * task has to wait for a replenishment to be performed at the
1193 * next firing of dl_timer.
1194 *
1195 * @dl_new tells if a new instance arrived. If so we must
1196 * start executing it with full runtime and reset its absolute
1197 * deadline;
1198 *
1199 * @dl_boosted tells if we are boosted due to DI. If so we are
1200 * outside bandwidth enforcement mechanism (but only until we
1201 * exit the critical section);
1202 *
1203 * @dl_yielded tells if task gave up the cpu before consuming
1204 * all its available runtime during the last job.
1205 */
1206 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1207
1208 /*
1209 * Bandwidth enforcement timer. Each -deadline task has its
1210 * own bandwidth to be enforced, thus we need one timer per task.
1211 */
1212 struct hrtimer dl_timer;
1213 };
1214
1215 union rcu_special {
1216 struct {
1217 bool blocked;
1218 bool need_qs;
1219 } b;
1220 short s;
1221 };
1222 struct rcu_node;
1223
1224 enum perf_event_task_context {
1225 perf_invalid_context = -1,
1226 perf_hw_context = 0,
1227 perf_sw_context,
1228 perf_nr_task_contexts,
1229 };
1230
1231 struct task_struct {
1232 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1233 void *stack;
1234 atomic_t usage;
1235 unsigned int flags; /* per process flags, defined below */
1236 unsigned int ptrace;
1237
1238 #ifdef CONFIG_SMP
1239 struct llist_node wake_entry;
1240 int on_cpu;
1241 struct task_struct *last_wakee;
1242 unsigned long wakee_flips;
1243 unsigned long wakee_flip_decay_ts;
1244
1245 int wake_cpu;
1246 #endif
1247 int on_rq;
1248
1249 int prio, static_prio, normal_prio;
1250 unsigned int rt_priority;
1251 const struct sched_class *sched_class;
1252 struct sched_entity se;
1253 struct sched_rt_entity rt;
1254 #ifdef CONFIG_CGROUP_SCHED
1255 struct task_group *sched_task_group;
1256 #endif
1257 struct sched_dl_entity dl;
1258
1259 #ifdef CONFIG_PREEMPT_NOTIFIERS
1260 /* list of struct preempt_notifier: */
1261 struct hlist_head preempt_notifiers;
1262 #endif
1263
1264 #ifdef CONFIG_BLK_DEV_IO_TRACE
1265 unsigned int btrace_seq;
1266 #endif
1267
1268 unsigned int policy;
1269 int nr_cpus_allowed;
1270 cpumask_t cpus_allowed;
1271
1272 #ifdef CONFIG_PREEMPT_RCU
1273 int rcu_read_lock_nesting;
1274 union rcu_special rcu_read_unlock_special;
1275 struct list_head rcu_node_entry;
1276 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1277 #ifdef CONFIG_TREE_PREEMPT_RCU
1278 struct rcu_node *rcu_blocked_node;
1279 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1280 #ifdef CONFIG_TASKS_RCU
1281 unsigned long rcu_tasks_nvcsw;
1282 bool rcu_tasks_holdout;
1283 struct list_head rcu_tasks_holdout_list;
1284 int rcu_tasks_idle_cpu;
1285 #endif /* #ifdef CONFIG_TASKS_RCU */
1286
1287 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1288 struct sched_info sched_info;
1289 #endif
1290
1291 struct list_head tasks;
1292 #ifdef CONFIG_SMP
1293 struct plist_node pushable_tasks;
1294 struct rb_node pushable_dl_tasks;
1295 #endif
1296
1297 struct mm_struct *mm, *active_mm;
1298 #ifdef CONFIG_COMPAT_BRK
1299 unsigned brk_randomized:1;
1300 #endif
1301 /* per-thread vma caching */
1302 u32 vmacache_seqnum;
1303 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1304 #if defined(SPLIT_RSS_COUNTING)
1305 struct task_rss_stat rss_stat;
1306 #endif
1307 /* task state */
1308 int exit_state;
1309 int exit_code, exit_signal;
1310 int pdeath_signal; /* The signal sent when the parent dies */
1311 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1312
1313 /* Used for emulating ABI behavior of previous Linux versions */
1314 unsigned int personality;
1315
1316 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1317 * execve */
1318 unsigned in_iowait:1;
1319
1320 /* Revert to default priority/policy when forking */
1321 unsigned sched_reset_on_fork:1;
1322 unsigned sched_contributes_to_load:1;
1323
1324 unsigned long atomic_flags; /* Flags needing atomic access. */
1325
1326 pid_t pid;
1327 pid_t tgid;
1328
1329 #ifdef CONFIG_CC_STACKPROTECTOR
1330 /* Canary value for the -fstack-protector gcc feature */
1331 unsigned long stack_canary;
1332 #endif
1333 /*
1334 * pointers to (original) parent process, youngest child, younger sibling,
1335 * older sibling, respectively. (p->father can be replaced with
1336 * p->real_parent->pid)
1337 */
1338 struct task_struct __rcu *real_parent; /* real parent process */
1339 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1340 /*
1341 * children/sibling forms the list of my natural children
1342 */
1343 struct list_head children; /* list of my children */
1344 struct list_head sibling; /* linkage in my parent's children list */
1345 struct task_struct *group_leader; /* threadgroup leader */
1346
1347 /*
1348 * ptraced is the list of tasks this task is using ptrace on.
1349 * This includes both natural children and PTRACE_ATTACH targets.
1350 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1351 */
1352 struct list_head ptraced;
1353 struct list_head ptrace_entry;
1354
1355 /* PID/PID hash table linkage. */
1356 struct pid_link pids[PIDTYPE_MAX];
1357 struct list_head thread_group;
1358 struct list_head thread_node;
1359
1360 struct completion *vfork_done; /* for vfork() */
1361 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1362 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1363
1364 cputime_t utime, stime, utimescaled, stimescaled;
1365 cputime_t gtime;
1366 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1367 struct cputime prev_cputime;
1368 #endif
1369 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1370 seqlock_t vtime_seqlock;
1371 unsigned long long vtime_snap;
1372 enum {
1373 VTIME_SLEEPING = 0,
1374 VTIME_USER,
1375 VTIME_SYS,
1376 } vtime_snap_whence;
1377 #endif
1378 unsigned long nvcsw, nivcsw; /* context switch counts */
1379 u64 start_time; /* monotonic time in nsec */
1380 u64 real_start_time; /* boot based time in nsec */
1381 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1382 unsigned long min_flt, maj_flt;
1383
1384 struct task_cputime cputime_expires;
1385 struct list_head cpu_timers[3];
1386
1387 /* process credentials */
1388 const struct cred __rcu *real_cred; /* objective and real subjective task
1389 * credentials (COW) */
1390 const struct cred __rcu *cred; /* effective (overridable) subjective task
1391 * credentials (COW) */
1392 char comm[TASK_COMM_LEN]; /* executable name excluding path
1393 - access with [gs]et_task_comm (which lock
1394 it with task_lock())
1395 - initialized normally by setup_new_exec */
1396 /* file system info */
1397 int link_count, total_link_count;
1398 #ifdef CONFIG_SYSVIPC
1399 /* ipc stuff */
1400 struct sysv_sem sysvsem;
1401 struct sysv_shm sysvshm;
1402 #endif
1403 #ifdef CONFIG_DETECT_HUNG_TASK
1404 /* hung task detection */
1405 unsigned long last_switch_count;
1406 #endif
1407 /* CPU-specific state of this task */
1408 struct thread_struct thread;
1409 /* filesystem information */
1410 struct fs_struct *fs;
1411 /* open file information */
1412 struct files_struct *files;
1413 /* namespaces */
1414 struct nsproxy *nsproxy;
1415 /* signal handlers */
1416 struct signal_struct *signal;
1417 struct sighand_struct *sighand;
1418
1419 sigset_t blocked, real_blocked;
1420 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1421 struct sigpending pending;
1422
1423 unsigned long sas_ss_sp;
1424 size_t sas_ss_size;
1425 int (*notifier)(void *priv);
1426 void *notifier_data;
1427 sigset_t *notifier_mask;
1428 struct callback_head *task_works;
1429
1430 struct audit_context *audit_context;
1431 #ifdef CONFIG_AUDITSYSCALL
1432 kuid_t loginuid;
1433 unsigned int sessionid;
1434 #endif
1435 struct seccomp seccomp;
1436
1437 /* Thread group tracking */
1438 u32 parent_exec_id;
1439 u32 self_exec_id;
1440 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1441 * mempolicy */
1442 spinlock_t alloc_lock;
1443
1444 /* Protection of the PI data structures: */
1445 raw_spinlock_t pi_lock;
1446
1447 #ifdef CONFIG_RT_MUTEXES
1448 /* PI waiters blocked on a rt_mutex held by this task */
1449 struct rb_root pi_waiters;
1450 struct rb_node *pi_waiters_leftmost;
1451 /* Deadlock detection and priority inheritance handling */
1452 struct rt_mutex_waiter *pi_blocked_on;
1453 #endif
1454
1455 #ifdef CONFIG_DEBUG_MUTEXES
1456 /* mutex deadlock detection */
1457 struct mutex_waiter *blocked_on;
1458 #endif
1459 #ifdef CONFIG_TRACE_IRQFLAGS
1460 unsigned int irq_events;
1461 unsigned long hardirq_enable_ip;
1462 unsigned long hardirq_disable_ip;
1463 unsigned int hardirq_enable_event;
1464 unsigned int hardirq_disable_event;
1465 int hardirqs_enabled;
1466 int hardirq_context;
1467 unsigned long softirq_disable_ip;
1468 unsigned long softirq_enable_ip;
1469 unsigned int softirq_disable_event;
1470 unsigned int softirq_enable_event;
1471 int softirqs_enabled;
1472 int softirq_context;
1473 #endif
1474 #ifdef CONFIG_LOCKDEP
1475 # define MAX_LOCK_DEPTH 48UL
1476 u64 curr_chain_key;
1477 int lockdep_depth;
1478 unsigned int lockdep_recursion;
1479 struct held_lock held_locks[MAX_LOCK_DEPTH];
1480 gfp_t lockdep_reclaim_gfp;
1481 #endif
1482
1483 /* journalling filesystem info */
1484 void *journal_info;
1485
1486 /* stacked block device info */
1487 struct bio_list *bio_list;
1488
1489 #ifdef CONFIG_BLOCK
1490 /* stack plugging */
1491 struct blk_plug *plug;
1492 #endif
1493
1494 /* VM state */
1495 struct reclaim_state *reclaim_state;
1496
1497 struct backing_dev_info *backing_dev_info;
1498
1499 struct io_context *io_context;
1500
1501 unsigned long ptrace_message;
1502 siginfo_t *last_siginfo; /* For ptrace use. */
1503 struct task_io_accounting ioac;
1504 #if defined(CONFIG_TASK_XACCT)
1505 u64 acct_rss_mem1; /* accumulated rss usage */
1506 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1507 cputime_t acct_timexpd; /* stime + utime since last update */
1508 #endif
1509 #ifdef CONFIG_CPUSETS
1510 nodemask_t mems_allowed; /* Protected by alloc_lock */
1511 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1512 int cpuset_mem_spread_rotor;
1513 int cpuset_slab_spread_rotor;
1514 #endif
1515 #ifdef CONFIG_CGROUPS
1516 /* Control Group info protected by css_set_lock */
1517 struct css_set __rcu *cgroups;
1518 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1519 struct list_head cg_list;
1520 #endif
1521 #ifdef CONFIG_FUTEX
1522 struct robust_list_head __user *robust_list;
1523 #ifdef CONFIG_COMPAT
1524 struct compat_robust_list_head __user *compat_robust_list;
1525 #endif
1526 struct list_head pi_state_list;
1527 struct futex_pi_state *pi_state_cache;
1528 #endif
1529 #ifdef CONFIG_PERF_EVENTS
1530 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1531 struct mutex perf_event_mutex;
1532 struct list_head perf_event_list;
1533 #endif
1534 #ifdef CONFIG_DEBUG_PREEMPT
1535 unsigned long preempt_disable_ip;
1536 #endif
1537 #ifdef CONFIG_NUMA
1538 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1539 short il_next;
1540 short pref_node_fork;
1541 #endif
1542 #ifdef CONFIG_NUMA_BALANCING
1543 int numa_scan_seq;
1544 unsigned int numa_scan_period;
1545 unsigned int numa_scan_period_max;
1546 int numa_preferred_nid;
1547 unsigned long numa_migrate_retry;
1548 u64 node_stamp; /* migration stamp */
1549 u64 last_task_numa_placement;
1550 u64 last_sum_exec_runtime;
1551 struct callback_head numa_work;
1552
1553 struct list_head numa_entry;
1554 struct numa_group *numa_group;
1555
1556 /*
1557 * Exponential decaying average of faults on a per-node basis.
1558 * Scheduling placement decisions are made based on the these counts.
1559 * The values remain static for the duration of a PTE scan
1560 */
1561 unsigned long *numa_faults_memory;
1562 unsigned long total_numa_faults;
1563
1564 /*
1565 * numa_faults_buffer records faults per node during the current
1566 * scan window. When the scan completes, the counts in
1567 * numa_faults_memory decay and these values are copied.
1568 */
1569 unsigned long *numa_faults_buffer_memory;
1570
1571 /*
1572 * Track the nodes the process was running on when a NUMA hinting
1573 * fault was incurred.
1574 */
1575 unsigned long *numa_faults_cpu;
1576 unsigned long *numa_faults_buffer_cpu;
1577
1578 /*
1579 * numa_faults_locality tracks if faults recorded during the last
1580 * scan window were remote/local. The task scan period is adapted
1581 * based on the locality of the faults with different weights
1582 * depending on whether they were shared or private faults
1583 */
1584 unsigned long numa_faults_locality[2];
1585
1586 unsigned long numa_pages_migrated;
1587 #endif /* CONFIG_NUMA_BALANCING */
1588
1589 struct rcu_head rcu;
1590
1591 /*
1592 * cache last used pipe for splice
1593 */
1594 struct pipe_inode_info *splice_pipe;
1595
1596 struct page_frag task_frag;
1597
1598 #ifdef CONFIG_TASK_DELAY_ACCT
1599 struct task_delay_info *delays;
1600 #endif
1601 #ifdef CONFIG_FAULT_INJECTION
1602 int make_it_fail;
1603 #endif
1604 /*
1605 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1606 * balance_dirty_pages() for some dirty throttling pause
1607 */
1608 int nr_dirtied;
1609 int nr_dirtied_pause;
1610 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1611
1612 #ifdef CONFIG_LATENCYTOP
1613 int latency_record_count;
1614 struct latency_record latency_record[LT_SAVECOUNT];
1615 #endif
1616 /*
1617 * time slack values; these are used to round up poll() and
1618 * select() etc timeout values. These are in nanoseconds.
1619 */
1620 unsigned long timer_slack_ns;
1621 unsigned long default_timer_slack_ns;
1622
1623 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1624 /* Index of current stored address in ret_stack */
1625 int curr_ret_stack;
1626 /* Stack of return addresses for return function tracing */
1627 struct ftrace_ret_stack *ret_stack;
1628 /* time stamp for last schedule */
1629 unsigned long long ftrace_timestamp;
1630 /*
1631 * Number of functions that haven't been traced
1632 * because of depth overrun.
1633 */
1634 atomic_t trace_overrun;
1635 /* Pause for the tracing */
1636 atomic_t tracing_graph_pause;
1637 #endif
1638 #ifdef CONFIG_TRACING
1639 /* state flags for use by tracers */
1640 unsigned long trace;
1641 /* bitmask and counter of trace recursion */
1642 unsigned long trace_recursion;
1643 #endif /* CONFIG_TRACING */
1644 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1645 unsigned int memcg_kmem_skip_account;
1646 struct memcg_oom_info {
1647 struct mem_cgroup *memcg;
1648 gfp_t gfp_mask;
1649 int order;
1650 unsigned int may_oom:1;
1651 } memcg_oom;
1652 #endif
1653 #ifdef CONFIG_UPROBES
1654 struct uprobe_task *utask;
1655 #endif
1656 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1657 unsigned int sequential_io;
1658 unsigned int sequential_io_avg;
1659 #endif
1660 };
1661
1662 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1663 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1664
1665 #define TNF_MIGRATED 0x01
1666 #define TNF_NO_GROUP 0x02
1667 #define TNF_SHARED 0x04
1668 #define TNF_FAULT_LOCAL 0x08
1669
1670 #ifdef CONFIG_NUMA_BALANCING
1671 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1672 extern pid_t task_numa_group_id(struct task_struct *p);
1673 extern void set_numabalancing_state(bool enabled);
1674 extern void task_numa_free(struct task_struct *p);
1675 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1676 int src_nid, int dst_cpu);
1677 #else
1678 static inline void task_numa_fault(int last_node, int node, int pages,
1679 int flags)
1680 {
1681 }
1682 static inline pid_t task_numa_group_id(struct task_struct *p)
1683 {
1684 return 0;
1685 }
1686 static inline void set_numabalancing_state(bool enabled)
1687 {
1688 }
1689 static inline void task_numa_free(struct task_struct *p)
1690 {
1691 }
1692 static inline bool should_numa_migrate_memory(struct task_struct *p,
1693 struct page *page, int src_nid, int dst_cpu)
1694 {
1695 return true;
1696 }
1697 #endif
1698
1699 static inline struct pid *task_pid(struct task_struct *task)
1700 {
1701 return task->pids[PIDTYPE_PID].pid;
1702 }
1703
1704 static inline struct pid *task_tgid(struct task_struct *task)
1705 {
1706 return task->group_leader->pids[PIDTYPE_PID].pid;
1707 }
1708
1709 /*
1710 * Without tasklist or rcu lock it is not safe to dereference
1711 * the result of task_pgrp/task_session even if task == current,
1712 * we can race with another thread doing sys_setsid/sys_setpgid.
1713 */
1714 static inline struct pid *task_pgrp(struct task_struct *task)
1715 {
1716 return task->group_leader->pids[PIDTYPE_PGID].pid;
1717 }
1718
1719 static inline struct pid *task_session(struct task_struct *task)
1720 {
1721 return task->group_leader->pids[PIDTYPE_SID].pid;
1722 }
1723
1724 struct pid_namespace;
1725
1726 /*
1727 * the helpers to get the task's different pids as they are seen
1728 * from various namespaces
1729 *
1730 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1731 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1732 * current.
1733 * task_xid_nr_ns() : id seen from the ns specified;
1734 *
1735 * set_task_vxid() : assigns a virtual id to a task;
1736 *
1737 * see also pid_nr() etc in include/linux/pid.h
1738 */
1739 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1740 struct pid_namespace *ns);
1741
1742 static inline pid_t task_pid_nr(struct task_struct *tsk)
1743 {
1744 return tsk->pid;
1745 }
1746
1747 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1748 struct pid_namespace *ns)
1749 {
1750 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1751 }
1752
1753 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1754 {
1755 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1756 }
1757
1758
1759 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1760 {
1761 return tsk->tgid;
1762 }
1763
1764 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1765
1766 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1767 {
1768 return pid_vnr(task_tgid(tsk));
1769 }
1770
1771
1772 static inline int pid_alive(const struct task_struct *p);
1773 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1774 {
1775 pid_t pid = 0;
1776
1777 rcu_read_lock();
1778 if (pid_alive(tsk))
1779 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1780 rcu_read_unlock();
1781
1782 return pid;
1783 }
1784
1785 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1786 {
1787 return task_ppid_nr_ns(tsk, &init_pid_ns);
1788 }
1789
1790 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1791 struct pid_namespace *ns)
1792 {
1793 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1794 }
1795
1796 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1797 {
1798 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1799 }
1800
1801
1802 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1803 struct pid_namespace *ns)
1804 {
1805 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1806 }
1807
1808 static inline pid_t task_session_vnr(struct task_struct *tsk)
1809 {
1810 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1811 }
1812
1813 /* obsolete, do not use */
1814 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1815 {
1816 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1817 }
1818
1819 /**
1820 * pid_alive - check that a task structure is not stale
1821 * @p: Task structure to be checked.
1822 *
1823 * Test if a process is not yet dead (at most zombie state)
1824 * If pid_alive fails, then pointers within the task structure
1825 * can be stale and must not be dereferenced.
1826 *
1827 * Return: 1 if the process is alive. 0 otherwise.
1828 */
1829 static inline int pid_alive(const struct task_struct *p)
1830 {
1831 return p->pids[PIDTYPE_PID].pid != NULL;
1832 }
1833
1834 /**
1835 * is_global_init - check if a task structure is init
1836 * @tsk: Task structure to be checked.
1837 *
1838 * Check if a task structure is the first user space task the kernel created.
1839 *
1840 * Return: 1 if the task structure is init. 0 otherwise.
1841 */
1842 static inline int is_global_init(struct task_struct *tsk)
1843 {
1844 return tsk->pid == 1;
1845 }
1846
1847 extern struct pid *cad_pid;
1848
1849 extern void free_task(struct task_struct *tsk);
1850 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1851
1852 extern void __put_task_struct(struct task_struct *t);
1853
1854 static inline void put_task_struct(struct task_struct *t)
1855 {
1856 if (atomic_dec_and_test(&t->usage))
1857 __put_task_struct(t);
1858 }
1859
1860 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1861 extern void task_cputime(struct task_struct *t,
1862 cputime_t *utime, cputime_t *stime);
1863 extern void task_cputime_scaled(struct task_struct *t,
1864 cputime_t *utimescaled, cputime_t *stimescaled);
1865 extern cputime_t task_gtime(struct task_struct *t);
1866 #else
1867 static inline void task_cputime(struct task_struct *t,
1868 cputime_t *utime, cputime_t *stime)
1869 {
1870 if (utime)
1871 *utime = t->utime;
1872 if (stime)
1873 *stime = t->stime;
1874 }
1875
1876 static inline void task_cputime_scaled(struct task_struct *t,
1877 cputime_t *utimescaled,
1878 cputime_t *stimescaled)
1879 {
1880 if (utimescaled)
1881 *utimescaled = t->utimescaled;
1882 if (stimescaled)
1883 *stimescaled = t->stimescaled;
1884 }
1885
1886 static inline cputime_t task_gtime(struct task_struct *t)
1887 {
1888 return t->gtime;
1889 }
1890 #endif
1891 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1892 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1893
1894 /*
1895 * Per process flags
1896 */
1897 #define PF_EXITING 0x00000004 /* getting shut down */
1898 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1899 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1900 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1901 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1902 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1903 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1904 #define PF_DUMPCORE 0x00000200 /* dumped core */
1905 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1906 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1907 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1908 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1909 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1910 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1911 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1912 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1913 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1914 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1915 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1916 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1917 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1918 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1919 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1920 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1921 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1922 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1923 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1924 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1925 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1926
1927 /*
1928 * Only the _current_ task can read/write to tsk->flags, but other
1929 * tasks can access tsk->flags in readonly mode for example
1930 * with tsk_used_math (like during threaded core dumping).
1931 * There is however an exception to this rule during ptrace
1932 * or during fork: the ptracer task is allowed to write to the
1933 * child->flags of its traced child (same goes for fork, the parent
1934 * can write to the child->flags), because we're guaranteed the
1935 * child is not running and in turn not changing child->flags
1936 * at the same time the parent does it.
1937 */
1938 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1939 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1940 #define clear_used_math() clear_stopped_child_used_math(current)
1941 #define set_used_math() set_stopped_child_used_math(current)
1942 #define conditional_stopped_child_used_math(condition, child) \
1943 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1944 #define conditional_used_math(condition) \
1945 conditional_stopped_child_used_math(condition, current)
1946 #define copy_to_stopped_child_used_math(child) \
1947 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1948 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1949 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1950 #define used_math() tsk_used_math(current)
1951
1952 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1953 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1954 {
1955 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1956 flags &= ~__GFP_IO;
1957 return flags;
1958 }
1959
1960 static inline unsigned int memalloc_noio_save(void)
1961 {
1962 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1963 current->flags |= PF_MEMALLOC_NOIO;
1964 return flags;
1965 }
1966
1967 static inline void memalloc_noio_restore(unsigned int flags)
1968 {
1969 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1970 }
1971
1972 /* Per-process atomic flags. */
1973 #define PFA_NO_NEW_PRIVS 0x00000001 /* May not gain new privileges. */
1974
1975 static inline bool task_no_new_privs(struct task_struct *p)
1976 {
1977 return test_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1978 }
1979
1980 static inline void task_set_no_new_privs(struct task_struct *p)
1981 {
1982 set_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1983 }
1984
1985 /*
1986 * task->jobctl flags
1987 */
1988 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1989
1990 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1991 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1992 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1993 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1994 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1995 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1996 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1997
1998 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1999 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
2000 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
2001 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
2002 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
2003 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
2004 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
2005
2006 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2007 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2008
2009 extern bool task_set_jobctl_pending(struct task_struct *task,
2010 unsigned int mask);
2011 extern void task_clear_jobctl_trapping(struct task_struct *task);
2012 extern void task_clear_jobctl_pending(struct task_struct *task,
2013 unsigned int mask);
2014
2015 static inline void rcu_copy_process(struct task_struct *p)
2016 {
2017 #ifdef CONFIG_PREEMPT_RCU
2018 p->rcu_read_lock_nesting = 0;
2019 p->rcu_read_unlock_special.s = 0;
2020 p->rcu_blocked_node = NULL;
2021 INIT_LIST_HEAD(&p->rcu_node_entry);
2022 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2023 #ifdef CONFIG_TASKS_RCU
2024 p->rcu_tasks_holdout = false;
2025 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2026 p->rcu_tasks_idle_cpu = -1;
2027 #endif /* #ifdef CONFIG_TASKS_RCU */
2028 }
2029
2030 static inline void tsk_restore_flags(struct task_struct *task,
2031 unsigned long orig_flags, unsigned long flags)
2032 {
2033 task->flags &= ~flags;
2034 task->flags |= orig_flags & flags;
2035 }
2036
2037 #ifdef CONFIG_SMP
2038 extern void do_set_cpus_allowed(struct task_struct *p,
2039 const struct cpumask *new_mask);
2040
2041 extern int set_cpus_allowed_ptr(struct task_struct *p,
2042 const struct cpumask *new_mask);
2043 #else
2044 static inline void do_set_cpus_allowed(struct task_struct *p,
2045 const struct cpumask *new_mask)
2046 {
2047 }
2048 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2049 const struct cpumask *new_mask)
2050 {
2051 if (!cpumask_test_cpu(0, new_mask))
2052 return -EINVAL;
2053 return 0;
2054 }
2055 #endif
2056
2057 #ifdef CONFIG_NO_HZ_COMMON
2058 void calc_load_enter_idle(void);
2059 void calc_load_exit_idle(void);
2060 #else
2061 static inline void calc_load_enter_idle(void) { }
2062 static inline void calc_load_exit_idle(void) { }
2063 #endif /* CONFIG_NO_HZ_COMMON */
2064
2065 #ifndef CONFIG_CPUMASK_OFFSTACK
2066 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2067 {
2068 return set_cpus_allowed_ptr(p, &new_mask);
2069 }
2070 #endif
2071
2072 /*
2073 * Do not use outside of architecture code which knows its limitations.
2074 *
2075 * sched_clock() has no promise of monotonicity or bounded drift between
2076 * CPUs, use (which you should not) requires disabling IRQs.
2077 *
2078 * Please use one of the three interfaces below.
2079 */
2080 extern unsigned long long notrace sched_clock(void);
2081 /*
2082 * See the comment in kernel/sched/clock.c
2083 */
2084 extern u64 cpu_clock(int cpu);
2085 extern u64 local_clock(void);
2086 extern u64 sched_clock_cpu(int cpu);
2087
2088
2089 extern void sched_clock_init(void);
2090
2091 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2092 static inline void sched_clock_tick(void)
2093 {
2094 }
2095
2096 static inline void sched_clock_idle_sleep_event(void)
2097 {
2098 }
2099
2100 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2101 {
2102 }
2103 #else
2104 /*
2105 * Architectures can set this to 1 if they have specified
2106 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2107 * but then during bootup it turns out that sched_clock()
2108 * is reliable after all:
2109 */
2110 extern int sched_clock_stable(void);
2111 extern void set_sched_clock_stable(void);
2112 extern void clear_sched_clock_stable(void);
2113
2114 extern void sched_clock_tick(void);
2115 extern void sched_clock_idle_sleep_event(void);
2116 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2117 #endif
2118
2119 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2120 /*
2121 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2122 * The reason for this explicit opt-in is not to have perf penalty with
2123 * slow sched_clocks.
2124 */
2125 extern void enable_sched_clock_irqtime(void);
2126 extern void disable_sched_clock_irqtime(void);
2127 #else
2128 static inline void enable_sched_clock_irqtime(void) {}
2129 static inline void disable_sched_clock_irqtime(void) {}
2130 #endif
2131
2132 extern unsigned long long
2133 task_sched_runtime(struct task_struct *task);
2134
2135 /* sched_exec is called by processes performing an exec */
2136 #ifdef CONFIG_SMP
2137 extern void sched_exec(void);
2138 #else
2139 #define sched_exec() {}
2140 #endif
2141
2142 extern void sched_clock_idle_sleep_event(void);
2143 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2144
2145 #ifdef CONFIG_HOTPLUG_CPU
2146 extern void idle_task_exit(void);
2147 #else
2148 static inline void idle_task_exit(void) {}
2149 #endif
2150
2151 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2152 extern void wake_up_nohz_cpu(int cpu);
2153 #else
2154 static inline void wake_up_nohz_cpu(int cpu) { }
2155 #endif
2156
2157 #ifdef CONFIG_NO_HZ_FULL
2158 extern bool sched_can_stop_tick(void);
2159 extern u64 scheduler_tick_max_deferment(void);
2160 #else
2161 static inline bool sched_can_stop_tick(void) { return false; }
2162 #endif
2163
2164 #ifdef CONFIG_SCHED_AUTOGROUP
2165 extern void sched_autogroup_create_attach(struct task_struct *p);
2166 extern void sched_autogroup_detach(struct task_struct *p);
2167 extern void sched_autogroup_fork(struct signal_struct *sig);
2168 extern void sched_autogroup_exit(struct signal_struct *sig);
2169 #ifdef CONFIG_PROC_FS
2170 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2171 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2172 #endif
2173 #else
2174 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2175 static inline void sched_autogroup_detach(struct task_struct *p) { }
2176 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2177 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2178 #endif
2179
2180 extern int yield_to(struct task_struct *p, bool preempt);
2181 extern void set_user_nice(struct task_struct *p, long nice);
2182 extern int task_prio(const struct task_struct *p);
2183 /**
2184 * task_nice - return the nice value of a given task.
2185 * @p: the task in question.
2186 *
2187 * Return: The nice value [ -20 ... 0 ... 19 ].
2188 */
2189 static inline int task_nice(const struct task_struct *p)
2190 {
2191 return PRIO_TO_NICE((p)->static_prio);
2192 }
2193 extern int can_nice(const struct task_struct *p, const int nice);
2194 extern int task_curr(const struct task_struct *p);
2195 extern int idle_cpu(int cpu);
2196 extern int sched_setscheduler(struct task_struct *, int,
2197 const struct sched_param *);
2198 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2199 const struct sched_param *);
2200 extern int sched_setattr(struct task_struct *,
2201 const struct sched_attr *);
2202 extern struct task_struct *idle_task(int cpu);
2203 /**
2204 * is_idle_task - is the specified task an idle task?
2205 * @p: the task in question.
2206 *
2207 * Return: 1 if @p is an idle task. 0 otherwise.
2208 */
2209 static inline bool is_idle_task(const struct task_struct *p)
2210 {
2211 return p->pid == 0;
2212 }
2213 extern struct task_struct *curr_task(int cpu);
2214 extern void set_curr_task(int cpu, struct task_struct *p);
2215
2216 void yield(void);
2217
2218 /*
2219 * The default (Linux) execution domain.
2220 */
2221 extern struct exec_domain default_exec_domain;
2222
2223 union thread_union {
2224 struct thread_info thread_info;
2225 unsigned long stack[THREAD_SIZE/sizeof(long)];
2226 };
2227
2228 #ifndef __HAVE_ARCH_KSTACK_END
2229 static inline int kstack_end(void *addr)
2230 {
2231 /* Reliable end of stack detection:
2232 * Some APM bios versions misalign the stack
2233 */
2234 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2235 }
2236 #endif
2237
2238 extern union thread_union init_thread_union;
2239 extern struct task_struct init_task;
2240
2241 extern struct mm_struct init_mm;
2242
2243 extern struct pid_namespace init_pid_ns;
2244
2245 /*
2246 * find a task by one of its numerical ids
2247 *
2248 * find_task_by_pid_ns():
2249 * finds a task by its pid in the specified namespace
2250 * find_task_by_vpid():
2251 * finds a task by its virtual pid
2252 *
2253 * see also find_vpid() etc in include/linux/pid.h
2254 */
2255
2256 extern struct task_struct *find_task_by_vpid(pid_t nr);
2257 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2258 struct pid_namespace *ns);
2259
2260 /* per-UID process charging. */
2261 extern struct user_struct * alloc_uid(kuid_t);
2262 static inline struct user_struct *get_uid(struct user_struct *u)
2263 {
2264 atomic_inc(&u->__count);
2265 return u;
2266 }
2267 extern void free_uid(struct user_struct *);
2268
2269 #include <asm/current.h>
2270
2271 extern void xtime_update(unsigned long ticks);
2272
2273 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2274 extern int wake_up_process(struct task_struct *tsk);
2275 extern void wake_up_new_task(struct task_struct *tsk);
2276 #ifdef CONFIG_SMP
2277 extern void kick_process(struct task_struct *tsk);
2278 #else
2279 static inline void kick_process(struct task_struct *tsk) { }
2280 #endif
2281 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2282 extern void sched_dead(struct task_struct *p);
2283
2284 extern void proc_caches_init(void);
2285 extern void flush_signals(struct task_struct *);
2286 extern void __flush_signals(struct task_struct *);
2287 extern void ignore_signals(struct task_struct *);
2288 extern void flush_signal_handlers(struct task_struct *, int force_default);
2289 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2290
2291 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2292 {
2293 unsigned long flags;
2294 int ret;
2295
2296 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2297 ret = dequeue_signal(tsk, mask, info);
2298 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2299
2300 return ret;
2301 }
2302
2303 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2304 sigset_t *mask);
2305 extern void unblock_all_signals(void);
2306 extern void release_task(struct task_struct * p);
2307 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2308 extern int force_sigsegv(int, struct task_struct *);
2309 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2310 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2311 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2312 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2313 const struct cred *, u32);
2314 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2315 extern int kill_pid(struct pid *pid, int sig, int priv);
2316 extern int kill_proc_info(int, struct siginfo *, pid_t);
2317 extern __must_check bool do_notify_parent(struct task_struct *, int);
2318 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2319 extern void force_sig(int, struct task_struct *);
2320 extern int send_sig(int, struct task_struct *, int);
2321 extern int zap_other_threads(struct task_struct *p);
2322 extern struct sigqueue *sigqueue_alloc(void);
2323 extern void sigqueue_free(struct sigqueue *);
2324 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2325 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2326
2327 static inline void restore_saved_sigmask(void)
2328 {
2329 if (test_and_clear_restore_sigmask())
2330 __set_current_blocked(&current->saved_sigmask);
2331 }
2332
2333 static inline sigset_t *sigmask_to_save(void)
2334 {
2335 sigset_t *res = &current->blocked;
2336 if (unlikely(test_restore_sigmask()))
2337 res = &current->saved_sigmask;
2338 return res;
2339 }
2340
2341 static inline int kill_cad_pid(int sig, int priv)
2342 {
2343 return kill_pid(cad_pid, sig, priv);
2344 }
2345
2346 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2347 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2348 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2349 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2350
2351 /*
2352 * True if we are on the alternate signal stack.
2353 */
2354 static inline int on_sig_stack(unsigned long sp)
2355 {
2356 #ifdef CONFIG_STACK_GROWSUP
2357 return sp >= current->sas_ss_sp &&
2358 sp - current->sas_ss_sp < current->sas_ss_size;
2359 #else
2360 return sp > current->sas_ss_sp &&
2361 sp - current->sas_ss_sp <= current->sas_ss_size;
2362 #endif
2363 }
2364
2365 static inline int sas_ss_flags(unsigned long sp)
2366 {
2367 if (!current->sas_ss_size)
2368 return SS_DISABLE;
2369
2370 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2371 }
2372
2373 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2374 {
2375 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2376 #ifdef CONFIG_STACK_GROWSUP
2377 return current->sas_ss_sp;
2378 #else
2379 return current->sas_ss_sp + current->sas_ss_size;
2380 #endif
2381 return sp;
2382 }
2383
2384 /*
2385 * Routines for handling mm_structs
2386 */
2387 extern struct mm_struct * mm_alloc(void);
2388
2389 /* mmdrop drops the mm and the page tables */
2390 extern void __mmdrop(struct mm_struct *);
2391 static inline void mmdrop(struct mm_struct * mm)
2392 {
2393 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2394 __mmdrop(mm);
2395 }
2396
2397 /* mmput gets rid of the mappings and all user-space */
2398 extern void mmput(struct mm_struct *);
2399 /* Grab a reference to a task's mm, if it is not already going away */
2400 extern struct mm_struct *get_task_mm(struct task_struct *task);
2401 /*
2402 * Grab a reference to a task's mm, if it is not already going away
2403 * and ptrace_may_access with the mode parameter passed to it
2404 * succeeds.
2405 */
2406 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2407 /* Remove the current tasks stale references to the old mm_struct */
2408 extern void mm_release(struct task_struct *, struct mm_struct *);
2409
2410 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2411 struct task_struct *);
2412 extern void flush_thread(void);
2413 extern void exit_thread(void);
2414
2415 extern void exit_files(struct task_struct *);
2416 extern void __cleanup_sighand(struct sighand_struct *);
2417
2418 extern void exit_itimers(struct signal_struct *);
2419 extern void flush_itimer_signals(void);
2420
2421 extern void do_group_exit(int);
2422
2423 extern int do_execve(struct filename *,
2424 const char __user * const __user *,
2425 const char __user * const __user *);
2426 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2427 struct task_struct *fork_idle(int);
2428 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2429
2430 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2431 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2432 {
2433 __set_task_comm(tsk, from, false);
2434 }
2435 extern char *get_task_comm(char *to, struct task_struct *tsk);
2436
2437 #ifdef CONFIG_SMP
2438 void scheduler_ipi(void);
2439 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2440 #else
2441 static inline void scheduler_ipi(void) { }
2442 static inline unsigned long wait_task_inactive(struct task_struct *p,
2443 long match_state)
2444 {
2445 return 1;
2446 }
2447 #endif
2448
2449 #define next_task(p) \
2450 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2451
2452 #define for_each_process(p) \
2453 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2454
2455 extern bool current_is_single_threaded(void);
2456
2457 /*
2458 * Careful: do_each_thread/while_each_thread is a double loop so
2459 * 'break' will not work as expected - use goto instead.
2460 */
2461 #define do_each_thread(g, t) \
2462 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2463
2464 #define while_each_thread(g, t) \
2465 while ((t = next_thread(t)) != g)
2466
2467 #define __for_each_thread(signal, t) \
2468 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2469
2470 #define for_each_thread(p, t) \
2471 __for_each_thread((p)->signal, t)
2472
2473 /* Careful: this is a double loop, 'break' won't work as expected. */
2474 #define for_each_process_thread(p, t) \
2475 for_each_process(p) for_each_thread(p, t)
2476
2477 static inline int get_nr_threads(struct task_struct *tsk)
2478 {
2479 return tsk->signal->nr_threads;
2480 }
2481
2482 static inline bool thread_group_leader(struct task_struct *p)
2483 {
2484 return p->exit_signal >= 0;
2485 }
2486
2487 /* Do to the insanities of de_thread it is possible for a process
2488 * to have the pid of the thread group leader without actually being
2489 * the thread group leader. For iteration through the pids in proc
2490 * all we care about is that we have a task with the appropriate
2491 * pid, we don't actually care if we have the right task.
2492 */
2493 static inline bool has_group_leader_pid(struct task_struct *p)
2494 {
2495 return task_pid(p) == p->signal->leader_pid;
2496 }
2497
2498 static inline
2499 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2500 {
2501 return p1->signal == p2->signal;
2502 }
2503
2504 static inline struct task_struct *next_thread(const struct task_struct *p)
2505 {
2506 return list_entry_rcu(p->thread_group.next,
2507 struct task_struct, thread_group);
2508 }
2509
2510 static inline int thread_group_empty(struct task_struct *p)
2511 {
2512 return list_empty(&p->thread_group);
2513 }
2514
2515 #define delay_group_leader(p) \
2516 (thread_group_leader(p) && !thread_group_empty(p))
2517
2518 /*
2519 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2520 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2521 * pins the final release of task.io_context. Also protects ->cpuset and
2522 * ->cgroup.subsys[]. And ->vfork_done.
2523 *
2524 * Nests both inside and outside of read_lock(&tasklist_lock).
2525 * It must not be nested with write_lock_irq(&tasklist_lock),
2526 * neither inside nor outside.
2527 */
2528 static inline void task_lock(struct task_struct *p)
2529 {
2530 spin_lock(&p->alloc_lock);
2531 }
2532
2533 static inline void task_unlock(struct task_struct *p)
2534 {
2535 spin_unlock(&p->alloc_lock);
2536 }
2537
2538 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2539 unsigned long *flags);
2540
2541 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2542 unsigned long *flags)
2543 {
2544 struct sighand_struct *ret;
2545
2546 ret = __lock_task_sighand(tsk, flags);
2547 (void)__cond_lock(&tsk->sighand->siglock, ret);
2548 return ret;
2549 }
2550
2551 static inline void unlock_task_sighand(struct task_struct *tsk,
2552 unsigned long *flags)
2553 {
2554 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2555 }
2556
2557 #ifdef CONFIG_CGROUPS
2558 static inline void threadgroup_change_begin(struct task_struct *tsk)
2559 {
2560 down_read(&tsk->signal->group_rwsem);
2561 }
2562 static inline void threadgroup_change_end(struct task_struct *tsk)
2563 {
2564 up_read(&tsk->signal->group_rwsem);
2565 }
2566
2567 /**
2568 * threadgroup_lock - lock threadgroup
2569 * @tsk: member task of the threadgroup to lock
2570 *
2571 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2572 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2573 * change ->group_leader/pid. This is useful for cases where the threadgroup
2574 * needs to stay stable across blockable operations.
2575 *
2576 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2577 * synchronization. While held, no new task will be added to threadgroup
2578 * and no existing live task will have its PF_EXITING set.
2579 *
2580 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2581 * sub-thread becomes a new leader.
2582 */
2583 static inline void threadgroup_lock(struct task_struct *tsk)
2584 {
2585 down_write(&tsk->signal->group_rwsem);
2586 }
2587
2588 /**
2589 * threadgroup_unlock - unlock threadgroup
2590 * @tsk: member task of the threadgroup to unlock
2591 *
2592 * Reverse threadgroup_lock().
2593 */
2594 static inline void threadgroup_unlock(struct task_struct *tsk)
2595 {
2596 up_write(&tsk->signal->group_rwsem);
2597 }
2598 #else
2599 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2600 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2601 static inline void threadgroup_lock(struct task_struct *tsk) {}
2602 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2603 #endif
2604
2605 #ifndef __HAVE_THREAD_FUNCTIONS
2606
2607 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2608 #define task_stack_page(task) ((task)->stack)
2609
2610 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2611 {
2612 *task_thread_info(p) = *task_thread_info(org);
2613 task_thread_info(p)->task = p;
2614 }
2615
2616 static inline unsigned long *end_of_stack(struct task_struct *p)
2617 {
2618 return (unsigned long *)(task_thread_info(p) + 1);
2619 }
2620
2621 #endif
2622
2623 static inline int object_is_on_stack(void *obj)
2624 {
2625 void *stack = task_stack_page(current);
2626
2627 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2628 }
2629
2630 extern void thread_info_cache_init(void);
2631
2632 #ifdef CONFIG_DEBUG_STACK_USAGE
2633 static inline unsigned long stack_not_used(struct task_struct *p)
2634 {
2635 unsigned long *n = end_of_stack(p);
2636
2637 do { /* Skip over canary */
2638 n++;
2639 } while (!*n);
2640
2641 return (unsigned long)n - (unsigned long)end_of_stack(p);
2642 }
2643 #endif
2644
2645 /* set thread flags in other task's structures
2646 * - see asm/thread_info.h for TIF_xxxx flags available
2647 */
2648 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2649 {
2650 set_ti_thread_flag(task_thread_info(tsk), flag);
2651 }
2652
2653 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2654 {
2655 clear_ti_thread_flag(task_thread_info(tsk), flag);
2656 }
2657
2658 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2659 {
2660 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2661 }
2662
2663 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2664 {
2665 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2666 }
2667
2668 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2669 {
2670 return test_ti_thread_flag(task_thread_info(tsk), flag);
2671 }
2672
2673 static inline void set_tsk_need_resched(struct task_struct *tsk)
2674 {
2675 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2676 }
2677
2678 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2679 {
2680 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2681 }
2682
2683 static inline int test_tsk_need_resched(struct task_struct *tsk)
2684 {
2685 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2686 }
2687
2688 static inline int restart_syscall(void)
2689 {
2690 set_tsk_thread_flag(current, TIF_SIGPENDING);
2691 return -ERESTARTNOINTR;
2692 }
2693
2694 static inline int signal_pending(struct task_struct *p)
2695 {
2696 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2697 }
2698
2699 static inline int __fatal_signal_pending(struct task_struct *p)
2700 {
2701 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2702 }
2703
2704 static inline int fatal_signal_pending(struct task_struct *p)
2705 {
2706 return signal_pending(p) && __fatal_signal_pending(p);
2707 }
2708
2709 static inline int signal_pending_state(long state, struct task_struct *p)
2710 {
2711 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2712 return 0;
2713 if (!signal_pending(p))
2714 return 0;
2715
2716 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2717 }
2718
2719 /*
2720 * cond_resched() and cond_resched_lock(): latency reduction via
2721 * explicit rescheduling in places that are safe. The return
2722 * value indicates whether a reschedule was done in fact.
2723 * cond_resched_lock() will drop the spinlock before scheduling,
2724 * cond_resched_softirq() will enable bhs before scheduling.
2725 */
2726 extern int _cond_resched(void);
2727
2728 #define cond_resched() ({ \
2729 __might_sleep(__FILE__, __LINE__, 0); \
2730 _cond_resched(); \
2731 })
2732
2733 extern int __cond_resched_lock(spinlock_t *lock);
2734
2735 #ifdef CONFIG_PREEMPT_COUNT
2736 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2737 #else
2738 #define PREEMPT_LOCK_OFFSET 0
2739 #endif
2740
2741 #define cond_resched_lock(lock) ({ \
2742 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2743 __cond_resched_lock(lock); \
2744 })
2745
2746 extern int __cond_resched_softirq(void);
2747
2748 #define cond_resched_softirq() ({ \
2749 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2750 __cond_resched_softirq(); \
2751 })
2752
2753 static inline void cond_resched_rcu(void)
2754 {
2755 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2756 rcu_read_unlock();
2757 cond_resched();
2758 rcu_read_lock();
2759 #endif
2760 }
2761
2762 /*
2763 * Does a critical section need to be broken due to another
2764 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2765 * but a general need for low latency)
2766 */
2767 static inline int spin_needbreak(spinlock_t *lock)
2768 {
2769 #ifdef CONFIG_PREEMPT
2770 return spin_is_contended(lock);
2771 #else
2772 return 0;
2773 #endif
2774 }
2775
2776 /*
2777 * Idle thread specific functions to determine the need_resched
2778 * polling state.
2779 */
2780 #ifdef TIF_POLLING_NRFLAG
2781 static inline int tsk_is_polling(struct task_struct *p)
2782 {
2783 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2784 }
2785
2786 static inline void __current_set_polling(void)
2787 {
2788 set_thread_flag(TIF_POLLING_NRFLAG);
2789 }
2790
2791 static inline bool __must_check current_set_polling_and_test(void)
2792 {
2793 __current_set_polling();
2794
2795 /*
2796 * Polling state must be visible before we test NEED_RESCHED,
2797 * paired by resched_curr()
2798 */
2799 smp_mb__after_atomic();
2800
2801 return unlikely(tif_need_resched());
2802 }
2803
2804 static inline void __current_clr_polling(void)
2805 {
2806 clear_thread_flag(TIF_POLLING_NRFLAG);
2807 }
2808
2809 static inline bool __must_check current_clr_polling_and_test(void)
2810 {
2811 __current_clr_polling();
2812
2813 /*
2814 * Polling state must be visible before we test NEED_RESCHED,
2815 * paired by resched_curr()
2816 */
2817 smp_mb__after_atomic();
2818
2819 return unlikely(tif_need_resched());
2820 }
2821
2822 #else
2823 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2824 static inline void __current_set_polling(void) { }
2825 static inline void __current_clr_polling(void) { }
2826
2827 static inline bool __must_check current_set_polling_and_test(void)
2828 {
2829 return unlikely(tif_need_resched());
2830 }
2831 static inline bool __must_check current_clr_polling_and_test(void)
2832 {
2833 return unlikely(tif_need_resched());
2834 }
2835 #endif
2836
2837 static inline void current_clr_polling(void)
2838 {
2839 __current_clr_polling();
2840
2841 /*
2842 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2843 * Once the bit is cleared, we'll get IPIs with every new
2844 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2845 * fold.
2846 */
2847 smp_mb(); /* paired with resched_curr() */
2848
2849 preempt_fold_need_resched();
2850 }
2851
2852 static __always_inline bool need_resched(void)
2853 {
2854 return unlikely(tif_need_resched());
2855 }
2856
2857 /*
2858 * Thread group CPU time accounting.
2859 */
2860 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2861 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2862
2863 static inline void thread_group_cputime_init(struct signal_struct *sig)
2864 {
2865 raw_spin_lock_init(&sig->cputimer.lock);
2866 }
2867
2868 /*
2869 * Reevaluate whether the task has signals pending delivery.
2870 * Wake the task if so.
2871 * This is required every time the blocked sigset_t changes.
2872 * callers must hold sighand->siglock.
2873 */
2874 extern void recalc_sigpending_and_wake(struct task_struct *t);
2875 extern void recalc_sigpending(void);
2876
2877 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2878
2879 static inline void signal_wake_up(struct task_struct *t, bool resume)
2880 {
2881 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2882 }
2883 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2884 {
2885 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2886 }
2887
2888 /*
2889 * Wrappers for p->thread_info->cpu access. No-op on UP.
2890 */
2891 #ifdef CONFIG_SMP
2892
2893 static inline unsigned int task_cpu(const struct task_struct *p)
2894 {
2895 return task_thread_info(p)->cpu;
2896 }
2897
2898 static inline int task_node(const struct task_struct *p)
2899 {
2900 return cpu_to_node(task_cpu(p));
2901 }
2902
2903 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2904
2905 #else
2906
2907 static inline unsigned int task_cpu(const struct task_struct *p)
2908 {
2909 return 0;
2910 }
2911
2912 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2913 {
2914 }
2915
2916 #endif /* CONFIG_SMP */
2917
2918 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2919 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2920
2921 #ifdef CONFIG_CGROUP_SCHED
2922 extern struct task_group root_task_group;
2923 #endif /* CONFIG_CGROUP_SCHED */
2924
2925 extern int task_can_switch_user(struct user_struct *up,
2926 struct task_struct *tsk);
2927
2928 #ifdef CONFIG_TASK_XACCT
2929 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2930 {
2931 tsk->ioac.rchar += amt;
2932 }
2933
2934 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2935 {
2936 tsk->ioac.wchar += amt;
2937 }
2938
2939 static inline void inc_syscr(struct task_struct *tsk)
2940 {
2941 tsk->ioac.syscr++;
2942 }
2943
2944 static inline void inc_syscw(struct task_struct *tsk)
2945 {
2946 tsk->ioac.syscw++;
2947 }
2948 #else
2949 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2950 {
2951 }
2952
2953 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2954 {
2955 }
2956
2957 static inline void inc_syscr(struct task_struct *tsk)
2958 {
2959 }
2960
2961 static inline void inc_syscw(struct task_struct *tsk)
2962 {
2963 }
2964 #endif
2965
2966 #ifndef TASK_SIZE_OF
2967 #define TASK_SIZE_OF(tsk) TASK_SIZE
2968 #endif
2969
2970 #ifdef CONFIG_MEMCG
2971 extern void mm_update_next_owner(struct mm_struct *mm);
2972 #else
2973 static inline void mm_update_next_owner(struct mm_struct *mm)
2974 {
2975 }
2976 #endif /* CONFIG_MEMCG */
2977
2978 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2979 unsigned int limit)
2980 {
2981 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2982 }
2983
2984 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2985 unsigned int limit)
2986 {
2987 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2988 }
2989
2990 static inline unsigned long rlimit(unsigned int limit)
2991 {
2992 return task_rlimit(current, limit);
2993 }
2994
2995 static inline unsigned long rlimit_max(unsigned int limit)
2996 {
2997 return task_rlimit_max(current, limit);
2998 }
2999
3000 #endif
This page took 0.289938 seconds and 5 git commands to generate.