init/main.c: Give init_task a canary
[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 #include <linux/magic.h>
61
62 #include <asm/processor.h>
63
64 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
65
66 /*
67 * Extended scheduling parameters data structure.
68 *
69 * This is needed because the original struct sched_param can not be
70 * altered without introducing ABI issues with legacy applications
71 * (e.g., in sched_getparam()).
72 *
73 * However, the possibility of specifying more than just a priority for
74 * the tasks may be useful for a wide variety of application fields, e.g.,
75 * multimedia, streaming, automation and control, and many others.
76 *
77 * This variant (sched_attr) is meant at describing a so-called
78 * sporadic time-constrained task. In such model a task is specified by:
79 * - the activation period or minimum instance inter-arrival time;
80 * - the maximum (or average, depending on the actual scheduling
81 * discipline) computation time of all instances, a.k.a. runtime;
82 * - the deadline (relative to the actual activation time) of each
83 * instance.
84 * Very briefly, a periodic (sporadic) task asks for the execution of
85 * some specific computation --which is typically called an instance--
86 * (at most) every period. Moreover, each instance typically lasts no more
87 * than the runtime and must be completed by time instant t equal to
88 * the instance activation time + the deadline.
89 *
90 * This is reflected by the actual fields of the sched_attr structure:
91 *
92 * @size size of the structure, for fwd/bwd compat.
93 *
94 * @sched_policy task's scheduling policy
95 * @sched_flags for customizing the scheduler behaviour
96 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
97 * @sched_priority task's static priority (SCHED_FIFO/RR)
98 * @sched_deadline representative of the task's deadline
99 * @sched_runtime representative of the task's runtime
100 * @sched_period representative of the task's period
101 *
102 * Given this task model, there are a multiplicity of scheduling algorithms
103 * and policies, that can be used to ensure all the tasks will make their
104 * timing constraints.
105 *
106 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
107 * only user of this new interface. More information about the algorithm
108 * available in the scheduling class file or in Documentation/.
109 */
110 struct sched_attr {
111 u32 size;
112
113 u32 sched_policy;
114 u64 sched_flags;
115
116 /* SCHED_NORMAL, SCHED_BATCH */
117 s32 sched_nice;
118
119 /* SCHED_FIFO, SCHED_RR */
120 u32 sched_priority;
121
122 /* SCHED_DEADLINE */
123 u64 sched_runtime;
124 u64 sched_deadline;
125 u64 sched_period;
126 };
127
128 struct exec_domain;
129 struct futex_pi_state;
130 struct robust_list_head;
131 struct bio_list;
132 struct fs_struct;
133 struct perf_event_context;
134 struct blk_plug;
135 struct filename;
136
137 #define VMACACHE_BITS 2
138 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
139 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
140
141 /*
142 * These are the constant used to fake the fixed-point load-average
143 * counting. Some notes:
144 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
145 * a load-average precision of 10 bits integer + 11 bits fractional
146 * - if you want to count load-averages more often, you need more
147 * precision, or rounding will get you. With 2-second counting freq,
148 * the EXP_n values would be 1981, 2034 and 2043 if still using only
149 * 11 bit fractions.
150 */
151 extern unsigned long avenrun[]; /* Load averages */
152 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
153
154 #define FSHIFT 11 /* nr of bits of precision */
155 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
156 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
157 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
158 #define EXP_5 2014 /* 1/exp(5sec/5min) */
159 #define EXP_15 2037 /* 1/exp(5sec/15min) */
160
161 #define CALC_LOAD(load,exp,n) \
162 load *= exp; \
163 load += n*(FIXED_1-exp); \
164 load >>= FSHIFT;
165
166 extern unsigned long total_forks;
167 extern int nr_threads;
168 DECLARE_PER_CPU(unsigned long, process_counts);
169 extern int nr_processes(void);
170 extern unsigned long nr_running(void);
171 extern unsigned long nr_iowait(void);
172 extern unsigned long nr_iowait_cpu(int cpu);
173 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
174
175 extern void calc_global_load(unsigned long ticks);
176 extern void update_cpu_load_nohz(void);
177
178 extern unsigned long get_parent_ip(unsigned long addr);
179
180 extern void dump_cpu_task(int cpu);
181
182 struct seq_file;
183 struct cfs_rq;
184 struct task_group;
185 #ifdef CONFIG_SCHED_DEBUG
186 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
187 extern void proc_sched_set_task(struct task_struct *p);
188 extern void
189 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
190 #endif
191
192 /*
193 * Task state bitmask. NOTE! These bits are also
194 * encoded in fs/proc/array.c: get_task_state().
195 *
196 * We have two separate sets of flags: task->state
197 * is about runnability, while task->exit_state are
198 * about the task exiting. Confusing, but this way
199 * modifying one set can't modify the other one by
200 * mistake.
201 */
202 #define TASK_RUNNING 0
203 #define TASK_INTERRUPTIBLE 1
204 #define TASK_UNINTERRUPTIBLE 2
205 #define __TASK_STOPPED 4
206 #define __TASK_TRACED 8
207 /* in tsk->exit_state */
208 #define EXIT_DEAD 16
209 #define EXIT_ZOMBIE 32
210 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
211 /* in tsk->state again */
212 #define TASK_DEAD 64
213 #define TASK_WAKEKILL 128
214 #define TASK_WAKING 256
215 #define TASK_PARKED 512
216 #define TASK_STATE_MAX 1024
217
218 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
219
220 extern char ___assert_task_state[1 - 2*!!(
221 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
222
223 /* Convenience macros for the sake of set_task_state */
224 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
225 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
226 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
227
228 /* Convenience macros for the sake of wake_up */
229 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
230 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
231
232 /* get_task_state() */
233 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
234 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
235 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
236
237 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
238 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
239 #define task_is_stopped_or_traced(task) \
240 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
241 #define task_contributes_to_load(task) \
242 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
243 (task->flags & PF_FROZEN) == 0)
244
245 #define __set_task_state(tsk, state_value) \
246 do { (tsk)->state = (state_value); } while (0)
247 #define set_task_state(tsk, state_value) \
248 set_mb((tsk)->state, (state_value))
249
250 /*
251 * set_current_state() includes a barrier so that the write of current->state
252 * is correctly serialised wrt the caller's subsequent test of whether to
253 * actually sleep:
254 *
255 * set_current_state(TASK_UNINTERRUPTIBLE);
256 * if (do_i_need_to_sleep())
257 * schedule();
258 *
259 * If the caller does not need such serialisation then use __set_current_state()
260 */
261 #define __set_current_state(state_value) \
262 do { current->state = (state_value); } while (0)
263 #define set_current_state(state_value) \
264 set_mb(current->state, (state_value))
265
266 /* Task command name length */
267 #define TASK_COMM_LEN 16
268
269 #include <linux/spinlock.h>
270
271 /*
272 * This serializes "schedule()" and also protects
273 * the run-queue from deletions/modifications (but
274 * _adding_ to the beginning of the run-queue has
275 * a separate lock).
276 */
277 extern rwlock_t tasklist_lock;
278 extern spinlock_t mmlist_lock;
279
280 struct task_struct;
281
282 #ifdef CONFIG_PROVE_RCU
283 extern int lockdep_tasklist_lock_is_held(void);
284 #endif /* #ifdef CONFIG_PROVE_RCU */
285
286 extern void sched_init(void);
287 extern void sched_init_smp(void);
288 extern asmlinkage void schedule_tail(struct task_struct *prev);
289 extern void init_idle(struct task_struct *idle, int cpu);
290 extern void init_idle_bootup_task(struct task_struct *idle);
291
292 extern int runqueue_is_locked(int cpu);
293
294 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
295 extern void nohz_balance_enter_idle(int cpu);
296 extern void set_cpu_sd_state_idle(void);
297 extern int get_nohz_timer_target(int pinned);
298 #else
299 static inline void nohz_balance_enter_idle(int cpu) { }
300 static inline void set_cpu_sd_state_idle(void) { }
301 static inline int get_nohz_timer_target(int pinned)
302 {
303 return smp_processor_id();
304 }
305 #endif
306
307 /*
308 * Only dump TASK_* tasks. (0 for all tasks)
309 */
310 extern void show_state_filter(unsigned long state_filter);
311
312 static inline void show_state(void)
313 {
314 show_state_filter(0);
315 }
316
317 extern void show_regs(struct pt_regs *);
318
319 /*
320 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
321 * task), SP is the stack pointer of the first frame that should be shown in the back
322 * trace (or NULL if the entire call-chain of the task should be shown).
323 */
324 extern void show_stack(struct task_struct *task, unsigned long *sp);
325
326 void io_schedule(void);
327 long io_schedule_timeout(long timeout);
328
329 extern void cpu_init (void);
330 extern void trap_init(void);
331 extern void update_process_times(int user);
332 extern void scheduler_tick(void);
333
334 extern void sched_show_task(struct task_struct *p);
335
336 #ifdef CONFIG_LOCKUP_DETECTOR
337 extern void touch_softlockup_watchdog(void);
338 extern void touch_softlockup_watchdog_sync(void);
339 extern void touch_all_softlockup_watchdogs(void);
340 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
341 void __user *buffer,
342 size_t *lenp, loff_t *ppos);
343 extern unsigned int softlockup_panic;
344 void lockup_detector_init(void);
345 #else
346 static inline void touch_softlockup_watchdog(void)
347 {
348 }
349 static inline void touch_softlockup_watchdog_sync(void)
350 {
351 }
352 static inline void touch_all_softlockup_watchdogs(void)
353 {
354 }
355 static inline void lockup_detector_init(void)
356 {
357 }
358 #endif
359
360 #ifdef CONFIG_DETECT_HUNG_TASK
361 void reset_hung_task_detector(void);
362 #else
363 static inline void reset_hung_task_detector(void)
364 {
365 }
366 #endif
367
368 /* Attach to any functions which should be ignored in wchan output. */
369 #define __sched __attribute__((__section__(".sched.text")))
370
371 /* Linker adds these: start and end of __sched functions */
372 extern char __sched_text_start[], __sched_text_end[];
373
374 /* Is this address in the __sched functions? */
375 extern int in_sched_functions(unsigned long addr);
376
377 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
378 extern signed long schedule_timeout(signed long timeout);
379 extern signed long schedule_timeout_interruptible(signed long timeout);
380 extern signed long schedule_timeout_killable(signed long timeout);
381 extern signed long schedule_timeout_uninterruptible(signed long timeout);
382 asmlinkage void schedule(void);
383 extern void schedule_preempt_disabled(void);
384
385 struct nsproxy;
386 struct user_namespace;
387
388 #ifdef CONFIG_MMU
389 extern void arch_pick_mmap_layout(struct mm_struct *mm);
390 extern unsigned long
391 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
392 unsigned long, unsigned long);
393 extern unsigned long
394 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
395 unsigned long len, unsigned long pgoff,
396 unsigned long flags);
397 #else
398 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
399 #endif
400
401 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
402 #define SUID_DUMP_USER 1 /* Dump as user of process */
403 #define SUID_DUMP_ROOT 2 /* Dump as root */
404
405 /* mm flags */
406
407 /* for SUID_DUMP_* above */
408 #define MMF_DUMPABLE_BITS 2
409 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
410
411 extern void set_dumpable(struct mm_struct *mm, int value);
412 /*
413 * This returns the actual value of the suid_dumpable flag. For things
414 * that are using this for checking for privilege transitions, it must
415 * test against SUID_DUMP_USER rather than treating it as a boolean
416 * value.
417 */
418 static inline int __get_dumpable(unsigned long mm_flags)
419 {
420 return mm_flags & MMF_DUMPABLE_MASK;
421 }
422
423 static inline int get_dumpable(struct mm_struct *mm)
424 {
425 return __get_dumpable(mm->flags);
426 }
427
428 /* coredump filter bits */
429 #define MMF_DUMP_ANON_PRIVATE 2
430 #define MMF_DUMP_ANON_SHARED 3
431 #define MMF_DUMP_MAPPED_PRIVATE 4
432 #define MMF_DUMP_MAPPED_SHARED 5
433 #define MMF_DUMP_ELF_HEADERS 6
434 #define MMF_DUMP_HUGETLB_PRIVATE 7
435 #define MMF_DUMP_HUGETLB_SHARED 8
436
437 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
438 #define MMF_DUMP_FILTER_BITS 7
439 #define MMF_DUMP_FILTER_MASK \
440 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
441 #define MMF_DUMP_FILTER_DEFAULT \
442 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
443 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
444
445 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
446 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
447 #else
448 # define MMF_DUMP_MASK_DEFAULT_ELF 0
449 #endif
450 /* leave room for more dump flags */
451 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
452 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
453 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
454
455 #define MMF_HAS_UPROBES 19 /* has uprobes */
456 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
457
458 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
459
460 struct sighand_struct {
461 atomic_t count;
462 struct k_sigaction action[_NSIG];
463 spinlock_t siglock;
464 wait_queue_head_t signalfd_wqh;
465 };
466
467 struct pacct_struct {
468 int ac_flag;
469 long ac_exitcode;
470 unsigned long ac_mem;
471 cputime_t ac_utime, ac_stime;
472 unsigned long ac_minflt, ac_majflt;
473 };
474
475 struct cpu_itimer {
476 cputime_t expires;
477 cputime_t incr;
478 u32 error;
479 u32 incr_error;
480 };
481
482 /**
483 * struct cputime - snaphsot of system and user cputime
484 * @utime: time spent in user mode
485 * @stime: time spent in system mode
486 *
487 * Gathers a generic snapshot of user and system time.
488 */
489 struct cputime {
490 cputime_t utime;
491 cputime_t stime;
492 };
493
494 /**
495 * struct task_cputime - collected CPU time counts
496 * @utime: time spent in user mode, in &cputime_t units
497 * @stime: time spent in kernel mode, in &cputime_t units
498 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
499 *
500 * This is an extension of struct cputime that includes the total runtime
501 * spent by the task from the scheduler point of view.
502 *
503 * As a result, this structure groups together three kinds of CPU time
504 * that are tracked for threads and thread groups. Most things considering
505 * CPU time want to group these counts together and treat all three
506 * of them in parallel.
507 */
508 struct task_cputime {
509 cputime_t utime;
510 cputime_t stime;
511 unsigned long long sum_exec_runtime;
512 };
513 /* Alternate field names when used to cache expirations. */
514 #define prof_exp stime
515 #define virt_exp utime
516 #define sched_exp sum_exec_runtime
517
518 #define INIT_CPUTIME \
519 (struct task_cputime) { \
520 .utime = 0, \
521 .stime = 0, \
522 .sum_exec_runtime = 0, \
523 }
524
525 #ifdef CONFIG_PREEMPT_COUNT
526 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
527 #else
528 #define PREEMPT_DISABLED PREEMPT_ENABLED
529 #endif
530
531 /*
532 * Disable preemption until the scheduler is running.
533 * Reset by start_kernel()->sched_init()->init_idle().
534 *
535 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
536 * before the scheduler is active -- see should_resched().
537 */
538 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
539
540 /**
541 * struct thread_group_cputimer - thread group interval timer counts
542 * @cputime: thread group interval timers.
543 * @running: non-zero when there are timers running and
544 * @cputime receives updates.
545 * @lock: lock for fields in this struct.
546 *
547 * This structure contains the version of task_cputime, above, that is
548 * used for thread group CPU timer calculations.
549 */
550 struct thread_group_cputimer {
551 struct task_cputime cputime;
552 int running;
553 raw_spinlock_t lock;
554 };
555
556 #include <linux/rwsem.h>
557 struct autogroup;
558
559 /*
560 * NOTE! "signal_struct" does not have its own
561 * locking, because a shared signal_struct always
562 * implies a shared sighand_struct, so locking
563 * sighand_struct is always a proper superset of
564 * the locking of signal_struct.
565 */
566 struct signal_struct {
567 atomic_t sigcnt;
568 atomic_t live;
569 int nr_threads;
570 struct list_head thread_head;
571
572 wait_queue_head_t wait_chldexit; /* for wait4() */
573
574 /* current thread group signal load-balancing target: */
575 struct task_struct *curr_target;
576
577 /* shared signal handling: */
578 struct sigpending shared_pending;
579
580 /* thread group exit support */
581 int group_exit_code;
582 /* overloaded:
583 * - notify group_exit_task when ->count is equal to notify_count
584 * - everyone except group_exit_task is stopped during signal delivery
585 * of fatal signals, group_exit_task processes the signal.
586 */
587 int notify_count;
588 struct task_struct *group_exit_task;
589
590 /* thread group stop support, overloads group_exit_code too */
591 int group_stop_count;
592 unsigned int flags; /* see SIGNAL_* flags below */
593
594 /*
595 * PR_SET_CHILD_SUBREAPER marks a process, like a service
596 * manager, to re-parent orphan (double-forking) child processes
597 * to this process instead of 'init'. The service manager is
598 * able to receive SIGCHLD signals and is able to investigate
599 * the process until it calls wait(). All children of this
600 * process will inherit a flag if they should look for a
601 * child_subreaper process at exit.
602 */
603 unsigned int is_child_subreaper:1;
604 unsigned int has_child_subreaper:1;
605
606 /* POSIX.1b Interval Timers */
607 int posix_timer_id;
608 struct list_head posix_timers;
609
610 /* ITIMER_REAL timer for the process */
611 struct hrtimer real_timer;
612 struct pid *leader_pid;
613 ktime_t it_real_incr;
614
615 /*
616 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
617 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
618 * values are defined to 0 and 1 respectively
619 */
620 struct cpu_itimer it[2];
621
622 /*
623 * Thread group totals for process CPU timers.
624 * See thread_group_cputimer(), et al, for details.
625 */
626 struct thread_group_cputimer cputimer;
627
628 /* Earliest-expiration cache. */
629 struct task_cputime cputime_expires;
630
631 struct list_head cpu_timers[3];
632
633 struct pid *tty_old_pgrp;
634
635 /* boolean value for session group leader */
636 int leader;
637
638 struct tty_struct *tty; /* NULL if no tty */
639
640 #ifdef CONFIG_SCHED_AUTOGROUP
641 struct autogroup *autogroup;
642 #endif
643 /*
644 * Cumulative resource counters for dead threads in the group,
645 * and for reaped dead child processes forked by this group.
646 * Live threads maintain their own counters and add to these
647 * in __exit_signal, except for the group leader.
648 */
649 seqlock_t stats_lock;
650 cputime_t utime, stime, cutime, cstime;
651 cputime_t gtime;
652 cputime_t cgtime;
653 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
654 struct cputime prev_cputime;
655 #endif
656 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
657 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
658 unsigned long inblock, oublock, cinblock, coublock;
659 unsigned long maxrss, cmaxrss;
660 struct task_io_accounting ioac;
661
662 /*
663 * Cumulative ns of schedule CPU time fo dead threads in the
664 * group, not including a zombie group leader, (This only differs
665 * from jiffies_to_ns(utime + stime) if sched_clock uses something
666 * other than jiffies.)
667 */
668 unsigned long long sum_sched_runtime;
669
670 /*
671 * We don't bother to synchronize most readers of this at all,
672 * because there is no reader checking a limit that actually needs
673 * to get both rlim_cur and rlim_max atomically, and either one
674 * alone is a single word that can safely be read normally.
675 * getrlimit/setrlimit use task_lock(current->group_leader) to
676 * protect this instead of the siglock, because they really
677 * have no need to disable irqs.
678 */
679 struct rlimit rlim[RLIM_NLIMITS];
680
681 #ifdef CONFIG_BSD_PROCESS_ACCT
682 struct pacct_struct pacct; /* per-process accounting information */
683 #endif
684 #ifdef CONFIG_TASKSTATS
685 struct taskstats *stats;
686 #endif
687 #ifdef CONFIG_AUDIT
688 unsigned audit_tty;
689 unsigned audit_tty_log_passwd;
690 struct tty_audit_buf *tty_audit_buf;
691 #endif
692 #ifdef CONFIG_CGROUPS
693 /*
694 * group_rwsem prevents new tasks from entering the threadgroup and
695 * member tasks from exiting,a more specifically, setting of
696 * PF_EXITING. fork and exit paths are protected with this rwsem
697 * using threadgroup_change_begin/end(). Users which require
698 * threadgroup to remain stable should use threadgroup_[un]lock()
699 * which also takes care of exec path. Currently, cgroup is the
700 * only user.
701 */
702 struct rw_semaphore group_rwsem;
703 #endif
704
705 oom_flags_t oom_flags;
706 short oom_score_adj; /* OOM kill score adjustment */
707 short oom_score_adj_min; /* OOM kill score adjustment min value.
708 * Only settable by CAP_SYS_RESOURCE. */
709
710 struct mutex cred_guard_mutex; /* guard against foreign influences on
711 * credential calculations
712 * (notably. ptrace) */
713 };
714
715 /*
716 * Bits in flags field of signal_struct.
717 */
718 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
719 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
720 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
721 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
722 /*
723 * Pending notifications to parent.
724 */
725 #define SIGNAL_CLD_STOPPED 0x00000010
726 #define SIGNAL_CLD_CONTINUED 0x00000020
727 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
728
729 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
730
731 /* If true, all threads except ->group_exit_task have pending SIGKILL */
732 static inline int signal_group_exit(const struct signal_struct *sig)
733 {
734 return (sig->flags & SIGNAL_GROUP_EXIT) ||
735 (sig->group_exit_task != NULL);
736 }
737
738 /*
739 * Some day this will be a full-fledged user tracking system..
740 */
741 struct user_struct {
742 atomic_t __count; /* reference count */
743 atomic_t processes; /* How many processes does this user have? */
744 atomic_t sigpending; /* How many pending signals does this user have? */
745 #ifdef CONFIG_INOTIFY_USER
746 atomic_t inotify_watches; /* How many inotify watches does this user have? */
747 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
748 #endif
749 #ifdef CONFIG_FANOTIFY
750 atomic_t fanotify_listeners;
751 #endif
752 #ifdef CONFIG_EPOLL
753 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
754 #endif
755 #ifdef CONFIG_POSIX_MQUEUE
756 /* protected by mq_lock */
757 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
758 #endif
759 unsigned long locked_shm; /* How many pages of mlocked shm ? */
760
761 #ifdef CONFIG_KEYS
762 struct key *uid_keyring; /* UID specific keyring */
763 struct key *session_keyring; /* UID's default session keyring */
764 #endif
765
766 /* Hash table maintenance information */
767 struct hlist_node uidhash_node;
768 kuid_t uid;
769
770 #ifdef CONFIG_PERF_EVENTS
771 atomic_long_t locked_vm;
772 #endif
773 };
774
775 extern int uids_sysfs_init(void);
776
777 extern struct user_struct *find_user(kuid_t);
778
779 extern struct user_struct root_user;
780 #define INIT_USER (&root_user)
781
782
783 struct backing_dev_info;
784 struct reclaim_state;
785
786 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
787 struct sched_info {
788 /* cumulative counters */
789 unsigned long pcount; /* # of times run on this cpu */
790 unsigned long long run_delay; /* time spent waiting on a runqueue */
791
792 /* timestamps */
793 unsigned long long last_arrival,/* when we last ran on a cpu */
794 last_queued; /* when we were last queued to run */
795 };
796 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
797
798 #ifdef CONFIG_TASK_DELAY_ACCT
799 struct task_delay_info {
800 spinlock_t lock;
801 unsigned int flags; /* Private per-task flags */
802
803 /* For each stat XXX, add following, aligned appropriately
804 *
805 * struct timespec XXX_start, XXX_end;
806 * u64 XXX_delay;
807 * u32 XXX_count;
808 *
809 * Atomicity of updates to XXX_delay, XXX_count protected by
810 * single lock above (split into XXX_lock if contention is an issue).
811 */
812
813 /*
814 * XXX_count is incremented on every XXX operation, the delay
815 * associated with the operation is added to XXX_delay.
816 * XXX_delay contains the accumulated delay time in nanoseconds.
817 */
818 u64 blkio_start; /* Shared by blkio, swapin */
819 u64 blkio_delay; /* wait for sync block io completion */
820 u64 swapin_delay; /* wait for swapin block io completion */
821 u32 blkio_count; /* total count of the number of sync block */
822 /* io operations performed */
823 u32 swapin_count; /* total count of the number of swapin block */
824 /* io operations performed */
825
826 u64 freepages_start;
827 u64 freepages_delay; /* wait for memory reclaim */
828 u32 freepages_count; /* total count of memory reclaim */
829 };
830 #endif /* CONFIG_TASK_DELAY_ACCT */
831
832 static inline int sched_info_on(void)
833 {
834 #ifdef CONFIG_SCHEDSTATS
835 return 1;
836 #elif defined(CONFIG_TASK_DELAY_ACCT)
837 extern int delayacct_on;
838 return delayacct_on;
839 #else
840 return 0;
841 #endif
842 }
843
844 enum cpu_idle_type {
845 CPU_IDLE,
846 CPU_NOT_IDLE,
847 CPU_NEWLY_IDLE,
848 CPU_MAX_IDLE_TYPES
849 };
850
851 /*
852 * Increase resolution of cpu_capacity calculations
853 */
854 #define SCHED_CAPACITY_SHIFT 10
855 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
856
857 /*
858 * sched-domains (multiprocessor balancing) declarations:
859 */
860 #ifdef CONFIG_SMP
861 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
862 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
863 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
864 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
865 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
866 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
867 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
868 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
869 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
870 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
871 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
872 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
873 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
874 #define SD_NUMA 0x4000 /* cross-node balancing */
875
876 #ifdef CONFIG_SCHED_SMT
877 static inline int cpu_smt_flags(void)
878 {
879 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
880 }
881 #endif
882
883 #ifdef CONFIG_SCHED_MC
884 static inline int cpu_core_flags(void)
885 {
886 return SD_SHARE_PKG_RESOURCES;
887 }
888 #endif
889
890 #ifdef CONFIG_NUMA
891 static inline int cpu_numa_flags(void)
892 {
893 return SD_NUMA;
894 }
895 #endif
896
897 struct sched_domain_attr {
898 int relax_domain_level;
899 };
900
901 #define SD_ATTR_INIT (struct sched_domain_attr) { \
902 .relax_domain_level = -1, \
903 }
904
905 extern int sched_domain_level_max;
906
907 struct sched_group;
908
909 struct sched_domain {
910 /* These fields must be setup */
911 struct sched_domain *parent; /* top domain must be null terminated */
912 struct sched_domain *child; /* bottom domain must be null terminated */
913 struct sched_group *groups; /* the balancing groups of the domain */
914 unsigned long min_interval; /* Minimum balance interval ms */
915 unsigned long max_interval; /* Maximum balance interval ms */
916 unsigned int busy_factor; /* less balancing by factor if busy */
917 unsigned int imbalance_pct; /* No balance until over watermark */
918 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
919 unsigned int busy_idx;
920 unsigned int idle_idx;
921 unsigned int newidle_idx;
922 unsigned int wake_idx;
923 unsigned int forkexec_idx;
924 unsigned int smt_gain;
925
926 int nohz_idle; /* NOHZ IDLE status */
927 int flags; /* See SD_* */
928 int level;
929
930 /* Runtime fields. */
931 unsigned long last_balance; /* init to jiffies. units in jiffies */
932 unsigned int balance_interval; /* initialise to 1. units in ms. */
933 unsigned int nr_balance_failed; /* initialise to 0 */
934
935 /* idle_balance() stats */
936 u64 max_newidle_lb_cost;
937 unsigned long next_decay_max_lb_cost;
938
939 #ifdef CONFIG_SCHEDSTATS
940 /* load_balance() stats */
941 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
942 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
943 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
944 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
945 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
946 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
947 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
948 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
949
950 /* Active load balancing */
951 unsigned int alb_count;
952 unsigned int alb_failed;
953 unsigned int alb_pushed;
954
955 /* SD_BALANCE_EXEC stats */
956 unsigned int sbe_count;
957 unsigned int sbe_balanced;
958 unsigned int sbe_pushed;
959
960 /* SD_BALANCE_FORK stats */
961 unsigned int sbf_count;
962 unsigned int sbf_balanced;
963 unsigned int sbf_pushed;
964
965 /* try_to_wake_up() stats */
966 unsigned int ttwu_wake_remote;
967 unsigned int ttwu_move_affine;
968 unsigned int ttwu_move_balance;
969 #endif
970 #ifdef CONFIG_SCHED_DEBUG
971 char *name;
972 #endif
973 union {
974 void *private; /* used during construction */
975 struct rcu_head rcu; /* used during destruction */
976 };
977
978 unsigned int span_weight;
979 /*
980 * Span of all CPUs in this domain.
981 *
982 * NOTE: this field is variable length. (Allocated dynamically
983 * by attaching extra space to the end of the structure,
984 * depending on how many CPUs the kernel has booted up with)
985 */
986 unsigned long span[0];
987 };
988
989 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
990 {
991 return to_cpumask(sd->span);
992 }
993
994 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
995 struct sched_domain_attr *dattr_new);
996
997 /* Allocate an array of sched domains, for partition_sched_domains(). */
998 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
999 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1000
1001 bool cpus_share_cache(int this_cpu, int that_cpu);
1002
1003 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1004 typedef int (*sched_domain_flags_f)(void);
1005
1006 #define SDTL_OVERLAP 0x01
1007
1008 struct sd_data {
1009 struct sched_domain **__percpu sd;
1010 struct sched_group **__percpu sg;
1011 struct sched_group_capacity **__percpu sgc;
1012 };
1013
1014 struct sched_domain_topology_level {
1015 sched_domain_mask_f mask;
1016 sched_domain_flags_f sd_flags;
1017 int flags;
1018 int numa_level;
1019 struct sd_data data;
1020 #ifdef CONFIG_SCHED_DEBUG
1021 char *name;
1022 #endif
1023 };
1024
1025 extern struct sched_domain_topology_level *sched_domain_topology;
1026
1027 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1028 extern void wake_up_if_idle(int cpu);
1029
1030 #ifdef CONFIG_SCHED_DEBUG
1031 # define SD_INIT_NAME(type) .name = #type
1032 #else
1033 # define SD_INIT_NAME(type)
1034 #endif
1035
1036 #else /* CONFIG_SMP */
1037
1038 struct sched_domain_attr;
1039
1040 static inline void
1041 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1042 struct sched_domain_attr *dattr_new)
1043 {
1044 }
1045
1046 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1047 {
1048 return true;
1049 }
1050
1051 #endif /* !CONFIG_SMP */
1052
1053
1054 struct io_context; /* See blkdev.h */
1055
1056
1057 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1058 extern void prefetch_stack(struct task_struct *t);
1059 #else
1060 static inline void prefetch_stack(struct task_struct *t) { }
1061 #endif
1062
1063 struct audit_context; /* See audit.c */
1064 struct mempolicy;
1065 struct pipe_inode_info;
1066 struct uts_namespace;
1067
1068 struct load_weight {
1069 unsigned long weight;
1070 u32 inv_weight;
1071 };
1072
1073 struct sched_avg {
1074 /*
1075 * These sums represent an infinite geometric series and so are bound
1076 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1077 * choices of y < 1-2^(-32)*1024.
1078 */
1079 u32 runnable_avg_sum, runnable_avg_period;
1080 u64 last_runnable_update;
1081 s64 decay_count;
1082 unsigned long load_avg_contrib;
1083 };
1084
1085 #ifdef CONFIG_SCHEDSTATS
1086 struct sched_statistics {
1087 u64 wait_start;
1088 u64 wait_max;
1089 u64 wait_count;
1090 u64 wait_sum;
1091 u64 iowait_count;
1092 u64 iowait_sum;
1093
1094 u64 sleep_start;
1095 u64 sleep_max;
1096 s64 sum_sleep_runtime;
1097
1098 u64 block_start;
1099 u64 block_max;
1100 u64 exec_max;
1101 u64 slice_max;
1102
1103 u64 nr_migrations_cold;
1104 u64 nr_failed_migrations_affine;
1105 u64 nr_failed_migrations_running;
1106 u64 nr_failed_migrations_hot;
1107 u64 nr_forced_migrations;
1108
1109 u64 nr_wakeups;
1110 u64 nr_wakeups_sync;
1111 u64 nr_wakeups_migrate;
1112 u64 nr_wakeups_local;
1113 u64 nr_wakeups_remote;
1114 u64 nr_wakeups_affine;
1115 u64 nr_wakeups_affine_attempts;
1116 u64 nr_wakeups_passive;
1117 u64 nr_wakeups_idle;
1118 };
1119 #endif
1120
1121 struct sched_entity {
1122 struct load_weight load; /* for load-balancing */
1123 struct rb_node run_node;
1124 struct list_head group_node;
1125 unsigned int on_rq;
1126
1127 u64 exec_start;
1128 u64 sum_exec_runtime;
1129 u64 vruntime;
1130 u64 prev_sum_exec_runtime;
1131
1132 u64 nr_migrations;
1133
1134 #ifdef CONFIG_SCHEDSTATS
1135 struct sched_statistics statistics;
1136 #endif
1137
1138 #ifdef CONFIG_FAIR_GROUP_SCHED
1139 int depth;
1140 struct sched_entity *parent;
1141 /* rq on which this entity is (to be) queued: */
1142 struct cfs_rq *cfs_rq;
1143 /* rq "owned" by this entity/group: */
1144 struct cfs_rq *my_q;
1145 #endif
1146
1147 #ifdef CONFIG_SMP
1148 /* Per-entity load-tracking */
1149 struct sched_avg avg;
1150 #endif
1151 };
1152
1153 struct sched_rt_entity {
1154 struct list_head run_list;
1155 unsigned long timeout;
1156 unsigned long watchdog_stamp;
1157 unsigned int time_slice;
1158
1159 struct sched_rt_entity *back;
1160 #ifdef CONFIG_RT_GROUP_SCHED
1161 struct sched_rt_entity *parent;
1162 /* rq on which this entity is (to be) queued: */
1163 struct rt_rq *rt_rq;
1164 /* rq "owned" by this entity/group: */
1165 struct rt_rq *my_q;
1166 #endif
1167 };
1168
1169 struct sched_dl_entity {
1170 struct rb_node rb_node;
1171
1172 /*
1173 * Original scheduling parameters. Copied here from sched_attr
1174 * during sched_setattr(), they will remain the same until
1175 * the next sched_setattr().
1176 */
1177 u64 dl_runtime; /* maximum runtime for each instance */
1178 u64 dl_deadline; /* relative deadline of each instance */
1179 u64 dl_period; /* separation of two instances (period) */
1180 u64 dl_bw; /* dl_runtime / dl_deadline */
1181
1182 /*
1183 * Actual scheduling parameters. Initialized with the values above,
1184 * they are continously updated during task execution. Note that
1185 * the remaining runtime could be < 0 in case we are in overrun.
1186 */
1187 s64 runtime; /* remaining runtime for this instance */
1188 u64 deadline; /* absolute deadline for this instance */
1189 unsigned int flags; /* specifying the scheduler behaviour */
1190
1191 /*
1192 * Some bool flags:
1193 *
1194 * @dl_throttled tells if we exhausted the runtime. If so, the
1195 * task has to wait for a replenishment to be performed at the
1196 * next firing of dl_timer.
1197 *
1198 * @dl_new tells if a new instance arrived. If so we must
1199 * start executing it with full runtime and reset its absolute
1200 * deadline;
1201 *
1202 * @dl_boosted tells if we are boosted due to DI. If so we are
1203 * outside bandwidth enforcement mechanism (but only until we
1204 * exit the critical section);
1205 *
1206 * @dl_yielded tells if task gave up the cpu before consuming
1207 * all its available runtime during the last job.
1208 */
1209 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1210
1211 /*
1212 * Bandwidth enforcement timer. Each -deadline task has its
1213 * own bandwidth to be enforced, thus we need one timer per task.
1214 */
1215 struct hrtimer dl_timer;
1216 };
1217
1218 struct rcu_node;
1219
1220 enum perf_event_task_context {
1221 perf_invalid_context = -1,
1222 perf_hw_context = 0,
1223 perf_sw_context,
1224 perf_nr_task_contexts,
1225 };
1226
1227 struct task_struct {
1228 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1229 void *stack;
1230 atomic_t usage;
1231 unsigned int flags; /* per process flags, defined below */
1232 unsigned int ptrace;
1233
1234 #ifdef CONFIG_SMP
1235 struct llist_node wake_entry;
1236 int on_cpu;
1237 struct task_struct *last_wakee;
1238 unsigned long wakee_flips;
1239 unsigned long wakee_flip_decay_ts;
1240
1241 int wake_cpu;
1242 #endif
1243 int on_rq;
1244
1245 int prio, static_prio, normal_prio;
1246 unsigned int rt_priority;
1247 const struct sched_class *sched_class;
1248 struct sched_entity se;
1249 struct sched_rt_entity rt;
1250 #ifdef CONFIG_CGROUP_SCHED
1251 struct task_group *sched_task_group;
1252 #endif
1253 struct sched_dl_entity dl;
1254
1255 #ifdef CONFIG_PREEMPT_NOTIFIERS
1256 /* list of struct preempt_notifier: */
1257 struct hlist_head preempt_notifiers;
1258 #endif
1259
1260 #ifdef CONFIG_BLK_DEV_IO_TRACE
1261 unsigned int btrace_seq;
1262 #endif
1263
1264 unsigned int policy;
1265 int nr_cpus_allowed;
1266 cpumask_t cpus_allowed;
1267
1268 #ifdef CONFIG_PREEMPT_RCU
1269 int rcu_read_lock_nesting;
1270 char rcu_read_unlock_special;
1271 struct list_head rcu_node_entry;
1272 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1273 #ifdef CONFIG_TREE_PREEMPT_RCU
1274 struct rcu_node *rcu_blocked_node;
1275 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1276
1277 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1278 struct sched_info sched_info;
1279 #endif
1280
1281 struct list_head tasks;
1282 #ifdef CONFIG_SMP
1283 struct plist_node pushable_tasks;
1284 struct rb_node pushable_dl_tasks;
1285 #endif
1286
1287 struct mm_struct *mm, *active_mm;
1288 #ifdef CONFIG_COMPAT_BRK
1289 unsigned brk_randomized:1;
1290 #endif
1291 /* per-thread vma caching */
1292 u32 vmacache_seqnum;
1293 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1294 #if defined(SPLIT_RSS_COUNTING)
1295 struct task_rss_stat rss_stat;
1296 #endif
1297 /* task state */
1298 int exit_state;
1299 int exit_code, exit_signal;
1300 int pdeath_signal; /* The signal sent when the parent dies */
1301 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1302
1303 /* Used for emulating ABI behavior of previous Linux versions */
1304 unsigned int personality;
1305
1306 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1307 * execve */
1308 unsigned in_iowait:1;
1309
1310 /* Revert to default priority/policy when forking */
1311 unsigned sched_reset_on_fork:1;
1312 unsigned sched_contributes_to_load:1;
1313
1314 unsigned long atomic_flags; /* Flags needing atomic access. */
1315
1316 pid_t pid;
1317 pid_t tgid;
1318
1319 #ifdef CONFIG_CC_STACKPROTECTOR
1320 /* Canary value for the -fstack-protector gcc feature */
1321 unsigned long stack_canary;
1322 #endif
1323 /*
1324 * pointers to (original) parent process, youngest child, younger sibling,
1325 * older sibling, respectively. (p->father can be replaced with
1326 * p->real_parent->pid)
1327 */
1328 struct task_struct __rcu *real_parent; /* real parent process */
1329 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1330 /*
1331 * children/sibling forms the list of my natural children
1332 */
1333 struct list_head children; /* list of my children */
1334 struct list_head sibling; /* linkage in my parent's children list */
1335 struct task_struct *group_leader; /* threadgroup leader */
1336
1337 /*
1338 * ptraced is the list of tasks this task is using ptrace on.
1339 * This includes both natural children and PTRACE_ATTACH targets.
1340 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1341 */
1342 struct list_head ptraced;
1343 struct list_head ptrace_entry;
1344
1345 /* PID/PID hash table linkage. */
1346 struct pid_link pids[PIDTYPE_MAX];
1347 struct list_head thread_group;
1348 struct list_head thread_node;
1349
1350 struct completion *vfork_done; /* for vfork() */
1351 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1352 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1353
1354 cputime_t utime, stime, utimescaled, stimescaled;
1355 cputime_t gtime;
1356 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1357 struct cputime prev_cputime;
1358 #endif
1359 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1360 seqlock_t vtime_seqlock;
1361 unsigned long long vtime_snap;
1362 enum {
1363 VTIME_SLEEPING = 0,
1364 VTIME_USER,
1365 VTIME_SYS,
1366 } vtime_snap_whence;
1367 #endif
1368 unsigned long nvcsw, nivcsw; /* context switch counts */
1369 u64 start_time; /* monotonic time in nsec */
1370 u64 real_start_time; /* boot based time in nsec */
1371 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1372 unsigned long min_flt, maj_flt;
1373
1374 struct task_cputime cputime_expires;
1375 struct list_head cpu_timers[3];
1376
1377 /* process credentials */
1378 const struct cred __rcu *real_cred; /* objective and real subjective task
1379 * credentials (COW) */
1380 const struct cred __rcu *cred; /* effective (overridable) subjective task
1381 * credentials (COW) */
1382 char comm[TASK_COMM_LEN]; /* executable name excluding path
1383 - access with [gs]et_task_comm (which lock
1384 it with task_lock())
1385 - initialized normally by setup_new_exec */
1386 /* file system info */
1387 int link_count, total_link_count;
1388 #ifdef CONFIG_SYSVIPC
1389 /* ipc stuff */
1390 struct sysv_sem sysvsem;
1391 struct sysv_shm sysvshm;
1392 #endif
1393 #ifdef CONFIG_DETECT_HUNG_TASK
1394 /* hung task detection */
1395 unsigned long last_switch_count;
1396 #endif
1397 /* CPU-specific state of this task */
1398 struct thread_struct thread;
1399 /* filesystem information */
1400 struct fs_struct *fs;
1401 /* open file information */
1402 struct files_struct *files;
1403 /* namespaces */
1404 struct nsproxy *nsproxy;
1405 /* signal handlers */
1406 struct signal_struct *signal;
1407 struct sighand_struct *sighand;
1408
1409 sigset_t blocked, real_blocked;
1410 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1411 struct sigpending pending;
1412
1413 unsigned long sas_ss_sp;
1414 size_t sas_ss_size;
1415 int (*notifier)(void *priv);
1416 void *notifier_data;
1417 sigset_t *notifier_mask;
1418 struct callback_head *task_works;
1419
1420 struct audit_context *audit_context;
1421 #ifdef CONFIG_AUDITSYSCALL
1422 kuid_t loginuid;
1423 unsigned int sessionid;
1424 #endif
1425 struct seccomp seccomp;
1426
1427 /* Thread group tracking */
1428 u32 parent_exec_id;
1429 u32 self_exec_id;
1430 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1431 * mempolicy */
1432 spinlock_t alloc_lock;
1433
1434 /* Protection of the PI data structures: */
1435 raw_spinlock_t pi_lock;
1436
1437 #ifdef CONFIG_RT_MUTEXES
1438 /* PI waiters blocked on a rt_mutex held by this task */
1439 struct rb_root pi_waiters;
1440 struct rb_node *pi_waiters_leftmost;
1441 /* Deadlock detection and priority inheritance handling */
1442 struct rt_mutex_waiter *pi_blocked_on;
1443 #endif
1444
1445 #ifdef CONFIG_DEBUG_MUTEXES
1446 /* mutex deadlock detection */
1447 struct mutex_waiter *blocked_on;
1448 #endif
1449 #ifdef CONFIG_TRACE_IRQFLAGS
1450 unsigned int irq_events;
1451 unsigned long hardirq_enable_ip;
1452 unsigned long hardirq_disable_ip;
1453 unsigned int hardirq_enable_event;
1454 unsigned int hardirq_disable_event;
1455 int hardirqs_enabled;
1456 int hardirq_context;
1457 unsigned long softirq_disable_ip;
1458 unsigned long softirq_enable_ip;
1459 unsigned int softirq_disable_event;
1460 unsigned int softirq_enable_event;
1461 int softirqs_enabled;
1462 int softirq_context;
1463 #endif
1464 #ifdef CONFIG_LOCKDEP
1465 # define MAX_LOCK_DEPTH 48UL
1466 u64 curr_chain_key;
1467 int lockdep_depth;
1468 unsigned int lockdep_recursion;
1469 struct held_lock held_locks[MAX_LOCK_DEPTH];
1470 gfp_t lockdep_reclaim_gfp;
1471 #endif
1472
1473 /* journalling filesystem info */
1474 void *journal_info;
1475
1476 /* stacked block device info */
1477 struct bio_list *bio_list;
1478
1479 #ifdef CONFIG_BLOCK
1480 /* stack plugging */
1481 struct blk_plug *plug;
1482 #endif
1483
1484 /* VM state */
1485 struct reclaim_state *reclaim_state;
1486
1487 struct backing_dev_info *backing_dev_info;
1488
1489 struct io_context *io_context;
1490
1491 unsigned long ptrace_message;
1492 siginfo_t *last_siginfo; /* For ptrace use. */
1493 struct task_io_accounting ioac;
1494 #if defined(CONFIG_TASK_XACCT)
1495 u64 acct_rss_mem1; /* accumulated rss usage */
1496 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1497 cputime_t acct_timexpd; /* stime + utime since last update */
1498 #endif
1499 #ifdef CONFIG_CPUSETS
1500 nodemask_t mems_allowed; /* Protected by alloc_lock */
1501 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1502 int cpuset_mem_spread_rotor;
1503 int cpuset_slab_spread_rotor;
1504 #endif
1505 #ifdef CONFIG_CGROUPS
1506 /* Control Group info protected by css_set_lock */
1507 struct css_set __rcu *cgroups;
1508 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1509 struct list_head cg_list;
1510 #endif
1511 #ifdef CONFIG_FUTEX
1512 struct robust_list_head __user *robust_list;
1513 #ifdef CONFIG_COMPAT
1514 struct compat_robust_list_head __user *compat_robust_list;
1515 #endif
1516 struct list_head pi_state_list;
1517 struct futex_pi_state *pi_state_cache;
1518 #endif
1519 #ifdef CONFIG_PERF_EVENTS
1520 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1521 struct mutex perf_event_mutex;
1522 struct list_head perf_event_list;
1523 #endif
1524 #ifdef CONFIG_DEBUG_PREEMPT
1525 unsigned long preempt_disable_ip;
1526 #endif
1527 #ifdef CONFIG_NUMA
1528 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1529 short il_next;
1530 short pref_node_fork;
1531 #endif
1532 #ifdef CONFIG_NUMA_BALANCING
1533 int numa_scan_seq;
1534 unsigned int numa_scan_period;
1535 unsigned int numa_scan_period_max;
1536 int numa_preferred_nid;
1537 unsigned long numa_migrate_retry;
1538 u64 node_stamp; /* migration stamp */
1539 u64 last_task_numa_placement;
1540 u64 last_sum_exec_runtime;
1541 struct callback_head numa_work;
1542
1543 struct list_head numa_entry;
1544 struct numa_group *numa_group;
1545
1546 /*
1547 * Exponential decaying average of faults on a per-node basis.
1548 * Scheduling placement decisions are made based on the these counts.
1549 * The values remain static for the duration of a PTE scan
1550 */
1551 unsigned long *numa_faults_memory;
1552 unsigned long total_numa_faults;
1553
1554 /*
1555 * numa_faults_buffer records faults per node during the current
1556 * scan window. When the scan completes, the counts in
1557 * numa_faults_memory decay and these values are copied.
1558 */
1559 unsigned long *numa_faults_buffer_memory;
1560
1561 /*
1562 * Track the nodes the process was running on when a NUMA hinting
1563 * fault was incurred.
1564 */
1565 unsigned long *numa_faults_cpu;
1566 unsigned long *numa_faults_buffer_cpu;
1567
1568 /*
1569 * numa_faults_locality tracks if faults recorded during the last
1570 * scan window were remote/local. The task scan period is adapted
1571 * based on the locality of the faults with different weights
1572 * depending on whether they were shared or private faults
1573 */
1574 unsigned long numa_faults_locality[2];
1575
1576 unsigned long numa_pages_migrated;
1577 #endif /* CONFIG_NUMA_BALANCING */
1578
1579 struct rcu_head rcu;
1580
1581 /*
1582 * cache last used pipe for splice
1583 */
1584 struct pipe_inode_info *splice_pipe;
1585
1586 struct page_frag task_frag;
1587
1588 #ifdef CONFIG_TASK_DELAY_ACCT
1589 struct task_delay_info *delays;
1590 #endif
1591 #ifdef CONFIG_FAULT_INJECTION
1592 int make_it_fail;
1593 #endif
1594 /*
1595 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1596 * balance_dirty_pages() for some dirty throttling pause
1597 */
1598 int nr_dirtied;
1599 int nr_dirtied_pause;
1600 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1601
1602 #ifdef CONFIG_LATENCYTOP
1603 int latency_record_count;
1604 struct latency_record latency_record[LT_SAVECOUNT];
1605 #endif
1606 /*
1607 * time slack values; these are used to round up poll() and
1608 * select() etc timeout values. These are in nanoseconds.
1609 */
1610 unsigned long timer_slack_ns;
1611 unsigned long default_timer_slack_ns;
1612
1613 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1614 /* Index of current stored address in ret_stack */
1615 int curr_ret_stack;
1616 /* Stack of return addresses for return function tracing */
1617 struct ftrace_ret_stack *ret_stack;
1618 /* time stamp for last schedule */
1619 unsigned long long ftrace_timestamp;
1620 /*
1621 * Number of functions that haven't been traced
1622 * because of depth overrun.
1623 */
1624 atomic_t trace_overrun;
1625 /* Pause for the tracing */
1626 atomic_t tracing_graph_pause;
1627 #endif
1628 #ifdef CONFIG_TRACING
1629 /* state flags for use by tracers */
1630 unsigned long trace;
1631 /* bitmask and counter of trace recursion */
1632 unsigned long trace_recursion;
1633 #endif /* CONFIG_TRACING */
1634 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1635 unsigned int memcg_kmem_skip_account;
1636 struct memcg_oom_info {
1637 struct mem_cgroup *memcg;
1638 gfp_t gfp_mask;
1639 int order;
1640 unsigned int may_oom:1;
1641 } memcg_oom;
1642 #endif
1643 #ifdef CONFIG_UPROBES
1644 struct uprobe_task *utask;
1645 #endif
1646 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1647 unsigned int sequential_io;
1648 unsigned int sequential_io_avg;
1649 #endif
1650 };
1651
1652 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1653 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1654
1655 #define TNF_MIGRATED 0x01
1656 #define TNF_NO_GROUP 0x02
1657 #define TNF_SHARED 0x04
1658 #define TNF_FAULT_LOCAL 0x08
1659
1660 #ifdef CONFIG_NUMA_BALANCING
1661 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1662 extern pid_t task_numa_group_id(struct task_struct *p);
1663 extern void set_numabalancing_state(bool enabled);
1664 extern void task_numa_free(struct task_struct *p);
1665 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1666 int src_nid, int dst_cpu);
1667 #else
1668 static inline void task_numa_fault(int last_node, int node, int pages,
1669 int flags)
1670 {
1671 }
1672 static inline pid_t task_numa_group_id(struct task_struct *p)
1673 {
1674 return 0;
1675 }
1676 static inline void set_numabalancing_state(bool enabled)
1677 {
1678 }
1679 static inline void task_numa_free(struct task_struct *p)
1680 {
1681 }
1682 static inline bool should_numa_migrate_memory(struct task_struct *p,
1683 struct page *page, int src_nid, int dst_cpu)
1684 {
1685 return true;
1686 }
1687 #endif
1688
1689 static inline struct pid *task_pid(struct task_struct *task)
1690 {
1691 return task->pids[PIDTYPE_PID].pid;
1692 }
1693
1694 static inline struct pid *task_tgid(struct task_struct *task)
1695 {
1696 return task->group_leader->pids[PIDTYPE_PID].pid;
1697 }
1698
1699 /*
1700 * Without tasklist or rcu lock it is not safe to dereference
1701 * the result of task_pgrp/task_session even if task == current,
1702 * we can race with another thread doing sys_setsid/sys_setpgid.
1703 */
1704 static inline struct pid *task_pgrp(struct task_struct *task)
1705 {
1706 return task->group_leader->pids[PIDTYPE_PGID].pid;
1707 }
1708
1709 static inline struct pid *task_session(struct task_struct *task)
1710 {
1711 return task->group_leader->pids[PIDTYPE_SID].pid;
1712 }
1713
1714 struct pid_namespace;
1715
1716 /*
1717 * the helpers to get the task's different pids as they are seen
1718 * from various namespaces
1719 *
1720 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1721 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1722 * current.
1723 * task_xid_nr_ns() : id seen from the ns specified;
1724 *
1725 * set_task_vxid() : assigns a virtual id to a task;
1726 *
1727 * see also pid_nr() etc in include/linux/pid.h
1728 */
1729 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1730 struct pid_namespace *ns);
1731
1732 static inline pid_t task_pid_nr(struct task_struct *tsk)
1733 {
1734 return tsk->pid;
1735 }
1736
1737 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1738 struct pid_namespace *ns)
1739 {
1740 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1741 }
1742
1743 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1744 {
1745 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1746 }
1747
1748
1749 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1750 {
1751 return tsk->tgid;
1752 }
1753
1754 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1755
1756 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1757 {
1758 return pid_vnr(task_tgid(tsk));
1759 }
1760
1761
1762 static inline int pid_alive(const struct task_struct *p);
1763 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1764 {
1765 pid_t pid = 0;
1766
1767 rcu_read_lock();
1768 if (pid_alive(tsk))
1769 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1770 rcu_read_unlock();
1771
1772 return pid;
1773 }
1774
1775 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1776 {
1777 return task_ppid_nr_ns(tsk, &init_pid_ns);
1778 }
1779
1780 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1781 struct pid_namespace *ns)
1782 {
1783 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1784 }
1785
1786 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1787 {
1788 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1789 }
1790
1791
1792 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1793 struct pid_namespace *ns)
1794 {
1795 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1796 }
1797
1798 static inline pid_t task_session_vnr(struct task_struct *tsk)
1799 {
1800 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1801 }
1802
1803 /* obsolete, do not use */
1804 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1805 {
1806 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1807 }
1808
1809 /**
1810 * pid_alive - check that a task structure is not stale
1811 * @p: Task structure to be checked.
1812 *
1813 * Test if a process is not yet dead (at most zombie state)
1814 * If pid_alive fails, then pointers within the task structure
1815 * can be stale and must not be dereferenced.
1816 *
1817 * Return: 1 if the process is alive. 0 otherwise.
1818 */
1819 static inline int pid_alive(const struct task_struct *p)
1820 {
1821 return p->pids[PIDTYPE_PID].pid != NULL;
1822 }
1823
1824 /**
1825 * is_global_init - check if a task structure is init
1826 * @tsk: Task structure to be checked.
1827 *
1828 * Check if a task structure is the first user space task the kernel created.
1829 *
1830 * Return: 1 if the task structure is init. 0 otherwise.
1831 */
1832 static inline int is_global_init(struct task_struct *tsk)
1833 {
1834 return tsk->pid == 1;
1835 }
1836
1837 extern struct pid *cad_pid;
1838
1839 extern void free_task(struct task_struct *tsk);
1840 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1841
1842 extern void __put_task_struct(struct task_struct *t);
1843
1844 static inline void put_task_struct(struct task_struct *t)
1845 {
1846 if (atomic_dec_and_test(&t->usage))
1847 __put_task_struct(t);
1848 }
1849
1850 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1851 extern void task_cputime(struct task_struct *t,
1852 cputime_t *utime, cputime_t *stime);
1853 extern void task_cputime_scaled(struct task_struct *t,
1854 cputime_t *utimescaled, cputime_t *stimescaled);
1855 extern cputime_t task_gtime(struct task_struct *t);
1856 #else
1857 static inline void task_cputime(struct task_struct *t,
1858 cputime_t *utime, cputime_t *stime)
1859 {
1860 if (utime)
1861 *utime = t->utime;
1862 if (stime)
1863 *stime = t->stime;
1864 }
1865
1866 static inline void task_cputime_scaled(struct task_struct *t,
1867 cputime_t *utimescaled,
1868 cputime_t *stimescaled)
1869 {
1870 if (utimescaled)
1871 *utimescaled = t->utimescaled;
1872 if (stimescaled)
1873 *stimescaled = t->stimescaled;
1874 }
1875
1876 static inline cputime_t task_gtime(struct task_struct *t)
1877 {
1878 return t->gtime;
1879 }
1880 #endif
1881 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1882 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1883
1884 /*
1885 * Per process flags
1886 */
1887 #define PF_EXITING 0x00000004 /* getting shut down */
1888 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1889 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1890 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1891 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1892 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1893 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1894 #define PF_DUMPCORE 0x00000200 /* dumped core */
1895 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1896 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1897 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1898 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1899 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1900 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1901 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1902 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1903 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1904 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1905 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1906 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1907 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1908 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1909 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1910 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1911 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1912 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1913 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1914 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1915 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1916
1917 /*
1918 * Only the _current_ task can read/write to tsk->flags, but other
1919 * tasks can access tsk->flags in readonly mode for example
1920 * with tsk_used_math (like during threaded core dumping).
1921 * There is however an exception to this rule during ptrace
1922 * or during fork: the ptracer task is allowed to write to the
1923 * child->flags of its traced child (same goes for fork, the parent
1924 * can write to the child->flags), because we're guaranteed the
1925 * child is not running and in turn not changing child->flags
1926 * at the same time the parent does it.
1927 */
1928 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1929 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1930 #define clear_used_math() clear_stopped_child_used_math(current)
1931 #define set_used_math() set_stopped_child_used_math(current)
1932 #define conditional_stopped_child_used_math(condition, child) \
1933 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1934 #define conditional_used_math(condition) \
1935 conditional_stopped_child_used_math(condition, current)
1936 #define copy_to_stopped_child_used_math(child) \
1937 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1938 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1939 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1940 #define used_math() tsk_used_math(current)
1941
1942 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1943 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1944 {
1945 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1946 flags &= ~__GFP_IO;
1947 return flags;
1948 }
1949
1950 static inline unsigned int memalloc_noio_save(void)
1951 {
1952 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1953 current->flags |= PF_MEMALLOC_NOIO;
1954 return flags;
1955 }
1956
1957 static inline void memalloc_noio_restore(unsigned int flags)
1958 {
1959 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1960 }
1961
1962 /* Per-process atomic flags. */
1963 #define PFA_NO_NEW_PRIVS 0x00000001 /* May not gain new privileges. */
1964
1965 static inline bool task_no_new_privs(struct task_struct *p)
1966 {
1967 return test_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1968 }
1969
1970 static inline void task_set_no_new_privs(struct task_struct *p)
1971 {
1972 set_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1973 }
1974
1975 /*
1976 * task->jobctl flags
1977 */
1978 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1979
1980 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1981 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1982 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1983 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1984 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1985 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1986 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1987
1988 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1989 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1990 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1991 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1992 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1993 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1994 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1995
1996 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1997 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1998
1999 extern bool task_set_jobctl_pending(struct task_struct *task,
2000 unsigned int mask);
2001 extern void task_clear_jobctl_trapping(struct task_struct *task);
2002 extern void task_clear_jobctl_pending(struct task_struct *task,
2003 unsigned int mask);
2004
2005 #ifdef CONFIG_PREEMPT_RCU
2006
2007 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
2008 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
2009
2010 static inline void rcu_copy_process(struct task_struct *p)
2011 {
2012 p->rcu_read_lock_nesting = 0;
2013 p->rcu_read_unlock_special = 0;
2014 #ifdef CONFIG_TREE_PREEMPT_RCU
2015 p->rcu_blocked_node = NULL;
2016 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
2017 INIT_LIST_HEAD(&p->rcu_node_entry);
2018 }
2019
2020 #else
2021
2022 static inline void rcu_copy_process(struct task_struct *p)
2023 {
2024 }
2025
2026 #endif
2027
2028 static inline void tsk_restore_flags(struct task_struct *task,
2029 unsigned long orig_flags, unsigned long flags)
2030 {
2031 task->flags &= ~flags;
2032 task->flags |= orig_flags & flags;
2033 }
2034
2035 #ifdef CONFIG_SMP
2036 extern void do_set_cpus_allowed(struct task_struct *p,
2037 const struct cpumask *new_mask);
2038
2039 extern int set_cpus_allowed_ptr(struct task_struct *p,
2040 const struct cpumask *new_mask);
2041 #else
2042 static inline void do_set_cpus_allowed(struct task_struct *p,
2043 const struct cpumask *new_mask)
2044 {
2045 }
2046 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2047 const struct cpumask *new_mask)
2048 {
2049 if (!cpumask_test_cpu(0, new_mask))
2050 return -EINVAL;
2051 return 0;
2052 }
2053 #endif
2054
2055 #ifdef CONFIG_NO_HZ_COMMON
2056 void calc_load_enter_idle(void);
2057 void calc_load_exit_idle(void);
2058 #else
2059 static inline void calc_load_enter_idle(void) { }
2060 static inline void calc_load_exit_idle(void) { }
2061 #endif /* CONFIG_NO_HZ_COMMON */
2062
2063 #ifndef CONFIG_CPUMASK_OFFSTACK
2064 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2065 {
2066 return set_cpus_allowed_ptr(p, &new_mask);
2067 }
2068 #endif
2069
2070 /*
2071 * Do not use outside of architecture code which knows its limitations.
2072 *
2073 * sched_clock() has no promise of monotonicity or bounded drift between
2074 * CPUs, use (which you should not) requires disabling IRQs.
2075 *
2076 * Please use one of the three interfaces below.
2077 */
2078 extern unsigned long long notrace sched_clock(void);
2079 /*
2080 * See the comment in kernel/sched/clock.c
2081 */
2082 extern u64 cpu_clock(int cpu);
2083 extern u64 local_clock(void);
2084 extern u64 sched_clock_cpu(int cpu);
2085
2086
2087 extern void sched_clock_init(void);
2088
2089 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2090 static inline void sched_clock_tick(void)
2091 {
2092 }
2093
2094 static inline void sched_clock_idle_sleep_event(void)
2095 {
2096 }
2097
2098 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2099 {
2100 }
2101 #else
2102 /*
2103 * Architectures can set this to 1 if they have specified
2104 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2105 * but then during bootup it turns out that sched_clock()
2106 * is reliable after all:
2107 */
2108 extern int sched_clock_stable(void);
2109 extern void set_sched_clock_stable(void);
2110 extern void clear_sched_clock_stable(void);
2111
2112 extern void sched_clock_tick(void);
2113 extern void sched_clock_idle_sleep_event(void);
2114 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2115 #endif
2116
2117 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2118 /*
2119 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2120 * The reason for this explicit opt-in is not to have perf penalty with
2121 * slow sched_clocks.
2122 */
2123 extern void enable_sched_clock_irqtime(void);
2124 extern void disable_sched_clock_irqtime(void);
2125 #else
2126 static inline void enable_sched_clock_irqtime(void) {}
2127 static inline void disable_sched_clock_irqtime(void) {}
2128 #endif
2129
2130 extern unsigned long long
2131 task_sched_runtime(struct task_struct *task);
2132
2133 /* sched_exec is called by processes performing an exec */
2134 #ifdef CONFIG_SMP
2135 extern void sched_exec(void);
2136 #else
2137 #define sched_exec() {}
2138 #endif
2139
2140 extern void sched_clock_idle_sleep_event(void);
2141 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2142
2143 #ifdef CONFIG_HOTPLUG_CPU
2144 extern void idle_task_exit(void);
2145 #else
2146 static inline void idle_task_exit(void) {}
2147 #endif
2148
2149 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2150 extern void wake_up_nohz_cpu(int cpu);
2151 #else
2152 static inline void wake_up_nohz_cpu(int cpu) { }
2153 #endif
2154
2155 #ifdef CONFIG_NO_HZ_FULL
2156 extern bool sched_can_stop_tick(void);
2157 extern u64 scheduler_tick_max_deferment(void);
2158 #else
2159 static inline bool sched_can_stop_tick(void) { return false; }
2160 #endif
2161
2162 #ifdef CONFIG_SCHED_AUTOGROUP
2163 extern void sched_autogroup_create_attach(struct task_struct *p);
2164 extern void sched_autogroup_detach(struct task_struct *p);
2165 extern void sched_autogroup_fork(struct signal_struct *sig);
2166 extern void sched_autogroup_exit(struct signal_struct *sig);
2167 #ifdef CONFIG_PROC_FS
2168 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2169 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2170 #endif
2171 #else
2172 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2173 static inline void sched_autogroup_detach(struct task_struct *p) { }
2174 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2175 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2176 #endif
2177
2178 extern int yield_to(struct task_struct *p, bool preempt);
2179 extern void set_user_nice(struct task_struct *p, long nice);
2180 extern int task_prio(const struct task_struct *p);
2181 /**
2182 * task_nice - return the nice value of a given task.
2183 * @p: the task in question.
2184 *
2185 * Return: The nice value [ -20 ... 0 ... 19 ].
2186 */
2187 static inline int task_nice(const struct task_struct *p)
2188 {
2189 return PRIO_TO_NICE((p)->static_prio);
2190 }
2191 extern int can_nice(const struct task_struct *p, const int nice);
2192 extern int task_curr(const struct task_struct *p);
2193 extern int idle_cpu(int cpu);
2194 extern int sched_setscheduler(struct task_struct *, int,
2195 const struct sched_param *);
2196 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2197 const struct sched_param *);
2198 extern int sched_setattr(struct task_struct *,
2199 const struct sched_attr *);
2200 extern struct task_struct *idle_task(int cpu);
2201 /**
2202 * is_idle_task - is the specified task an idle task?
2203 * @p: the task in question.
2204 *
2205 * Return: 1 if @p is an idle task. 0 otherwise.
2206 */
2207 static inline bool is_idle_task(const struct task_struct *p)
2208 {
2209 return p->pid == 0;
2210 }
2211 extern struct task_struct *curr_task(int cpu);
2212 extern void set_curr_task(int cpu, struct task_struct *p);
2213
2214 void yield(void);
2215
2216 /*
2217 * The default (Linux) execution domain.
2218 */
2219 extern struct exec_domain default_exec_domain;
2220
2221 union thread_union {
2222 struct thread_info thread_info;
2223 unsigned long stack[THREAD_SIZE/sizeof(long)];
2224 };
2225
2226 #ifndef __HAVE_ARCH_KSTACK_END
2227 static inline int kstack_end(void *addr)
2228 {
2229 /* Reliable end of stack detection:
2230 * Some APM bios versions misalign the stack
2231 */
2232 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2233 }
2234 #endif
2235
2236 extern union thread_union init_thread_union;
2237 extern struct task_struct init_task;
2238
2239 extern struct mm_struct init_mm;
2240
2241 extern struct pid_namespace init_pid_ns;
2242
2243 /*
2244 * find a task by one of its numerical ids
2245 *
2246 * find_task_by_pid_ns():
2247 * finds a task by its pid in the specified namespace
2248 * find_task_by_vpid():
2249 * finds a task by its virtual pid
2250 *
2251 * see also find_vpid() etc in include/linux/pid.h
2252 */
2253
2254 extern struct task_struct *find_task_by_vpid(pid_t nr);
2255 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2256 struct pid_namespace *ns);
2257
2258 /* per-UID process charging. */
2259 extern struct user_struct * alloc_uid(kuid_t);
2260 static inline struct user_struct *get_uid(struct user_struct *u)
2261 {
2262 atomic_inc(&u->__count);
2263 return u;
2264 }
2265 extern void free_uid(struct user_struct *);
2266
2267 #include <asm/current.h>
2268
2269 extern void xtime_update(unsigned long ticks);
2270
2271 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2272 extern int wake_up_process(struct task_struct *tsk);
2273 extern void wake_up_new_task(struct task_struct *tsk);
2274 #ifdef CONFIG_SMP
2275 extern void kick_process(struct task_struct *tsk);
2276 #else
2277 static inline void kick_process(struct task_struct *tsk) { }
2278 #endif
2279 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2280 extern void sched_dead(struct task_struct *p);
2281
2282 extern void proc_caches_init(void);
2283 extern void flush_signals(struct task_struct *);
2284 extern void __flush_signals(struct task_struct *);
2285 extern void ignore_signals(struct task_struct *);
2286 extern void flush_signal_handlers(struct task_struct *, int force_default);
2287 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2288
2289 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2290 {
2291 unsigned long flags;
2292 int ret;
2293
2294 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2295 ret = dequeue_signal(tsk, mask, info);
2296 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2297
2298 return ret;
2299 }
2300
2301 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2302 sigset_t *mask);
2303 extern void unblock_all_signals(void);
2304 extern void release_task(struct task_struct * p);
2305 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2306 extern int force_sigsegv(int, struct task_struct *);
2307 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2308 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2309 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2310 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2311 const struct cred *, u32);
2312 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2313 extern int kill_pid(struct pid *pid, int sig, int priv);
2314 extern int kill_proc_info(int, struct siginfo *, pid_t);
2315 extern __must_check bool do_notify_parent(struct task_struct *, int);
2316 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2317 extern void force_sig(int, struct task_struct *);
2318 extern int send_sig(int, struct task_struct *, int);
2319 extern int zap_other_threads(struct task_struct *p);
2320 extern struct sigqueue *sigqueue_alloc(void);
2321 extern void sigqueue_free(struct sigqueue *);
2322 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2323 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2324
2325 static inline void restore_saved_sigmask(void)
2326 {
2327 if (test_and_clear_restore_sigmask())
2328 __set_current_blocked(&current->saved_sigmask);
2329 }
2330
2331 static inline sigset_t *sigmask_to_save(void)
2332 {
2333 sigset_t *res = &current->blocked;
2334 if (unlikely(test_restore_sigmask()))
2335 res = &current->saved_sigmask;
2336 return res;
2337 }
2338
2339 static inline int kill_cad_pid(int sig, int priv)
2340 {
2341 return kill_pid(cad_pid, sig, priv);
2342 }
2343
2344 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2345 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2346 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2347 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2348
2349 /*
2350 * True if we are on the alternate signal stack.
2351 */
2352 static inline int on_sig_stack(unsigned long sp)
2353 {
2354 #ifdef CONFIG_STACK_GROWSUP
2355 return sp >= current->sas_ss_sp &&
2356 sp - current->sas_ss_sp < current->sas_ss_size;
2357 #else
2358 return sp > current->sas_ss_sp &&
2359 sp - current->sas_ss_sp <= current->sas_ss_size;
2360 #endif
2361 }
2362
2363 static inline int sas_ss_flags(unsigned long sp)
2364 {
2365 if (!current->sas_ss_size)
2366 return SS_DISABLE;
2367
2368 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2369 }
2370
2371 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2372 {
2373 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2374 #ifdef CONFIG_STACK_GROWSUP
2375 return current->sas_ss_sp;
2376 #else
2377 return current->sas_ss_sp + current->sas_ss_size;
2378 #endif
2379 return sp;
2380 }
2381
2382 /*
2383 * Routines for handling mm_structs
2384 */
2385 extern struct mm_struct * mm_alloc(void);
2386
2387 /* mmdrop drops the mm and the page tables */
2388 extern void __mmdrop(struct mm_struct *);
2389 static inline void mmdrop(struct mm_struct * mm)
2390 {
2391 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2392 __mmdrop(mm);
2393 }
2394
2395 /* mmput gets rid of the mappings and all user-space */
2396 extern void mmput(struct mm_struct *);
2397 /* Grab a reference to a task's mm, if it is not already going away */
2398 extern struct mm_struct *get_task_mm(struct task_struct *task);
2399 /*
2400 * Grab a reference to a task's mm, if it is not already going away
2401 * and ptrace_may_access with the mode parameter passed to it
2402 * succeeds.
2403 */
2404 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2405 /* Remove the current tasks stale references to the old mm_struct */
2406 extern void mm_release(struct task_struct *, struct mm_struct *);
2407
2408 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2409 struct task_struct *);
2410 extern void flush_thread(void);
2411 extern void exit_thread(void);
2412
2413 extern void exit_files(struct task_struct *);
2414 extern void __cleanup_sighand(struct sighand_struct *);
2415
2416 extern void exit_itimers(struct signal_struct *);
2417 extern void flush_itimer_signals(void);
2418
2419 extern void do_group_exit(int);
2420
2421 extern int do_execve(struct filename *,
2422 const char __user * const __user *,
2423 const char __user * const __user *);
2424 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2425 struct task_struct *fork_idle(int);
2426 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2427
2428 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2429 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2430 {
2431 __set_task_comm(tsk, from, false);
2432 }
2433 extern char *get_task_comm(char *to, struct task_struct *tsk);
2434
2435 #ifdef CONFIG_SMP
2436 void scheduler_ipi(void);
2437 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2438 #else
2439 static inline void scheduler_ipi(void) { }
2440 static inline unsigned long wait_task_inactive(struct task_struct *p,
2441 long match_state)
2442 {
2443 return 1;
2444 }
2445 #endif
2446
2447 #define next_task(p) \
2448 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2449
2450 #define for_each_process(p) \
2451 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2452
2453 extern bool current_is_single_threaded(void);
2454
2455 /*
2456 * Careful: do_each_thread/while_each_thread is a double loop so
2457 * 'break' will not work as expected - use goto instead.
2458 */
2459 #define do_each_thread(g, t) \
2460 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2461
2462 #define while_each_thread(g, t) \
2463 while ((t = next_thread(t)) != g)
2464
2465 #define __for_each_thread(signal, t) \
2466 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2467
2468 #define for_each_thread(p, t) \
2469 __for_each_thread((p)->signal, t)
2470
2471 /* Careful: this is a double loop, 'break' won't work as expected. */
2472 #define for_each_process_thread(p, t) \
2473 for_each_process(p) for_each_thread(p, t)
2474
2475 static inline int get_nr_threads(struct task_struct *tsk)
2476 {
2477 return tsk->signal->nr_threads;
2478 }
2479
2480 static inline bool thread_group_leader(struct task_struct *p)
2481 {
2482 return p->exit_signal >= 0;
2483 }
2484
2485 /* Do to the insanities of de_thread it is possible for a process
2486 * to have the pid of the thread group leader without actually being
2487 * the thread group leader. For iteration through the pids in proc
2488 * all we care about is that we have a task with the appropriate
2489 * pid, we don't actually care if we have the right task.
2490 */
2491 static inline bool has_group_leader_pid(struct task_struct *p)
2492 {
2493 return task_pid(p) == p->signal->leader_pid;
2494 }
2495
2496 static inline
2497 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2498 {
2499 return p1->signal == p2->signal;
2500 }
2501
2502 static inline struct task_struct *next_thread(const struct task_struct *p)
2503 {
2504 return list_entry_rcu(p->thread_group.next,
2505 struct task_struct, thread_group);
2506 }
2507
2508 static inline int thread_group_empty(struct task_struct *p)
2509 {
2510 return list_empty(&p->thread_group);
2511 }
2512
2513 #define delay_group_leader(p) \
2514 (thread_group_leader(p) && !thread_group_empty(p))
2515
2516 /*
2517 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2518 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2519 * pins the final release of task.io_context. Also protects ->cpuset and
2520 * ->cgroup.subsys[]. And ->vfork_done.
2521 *
2522 * Nests both inside and outside of read_lock(&tasklist_lock).
2523 * It must not be nested with write_lock_irq(&tasklist_lock),
2524 * neither inside nor outside.
2525 */
2526 static inline void task_lock(struct task_struct *p)
2527 {
2528 spin_lock(&p->alloc_lock);
2529 }
2530
2531 static inline void task_unlock(struct task_struct *p)
2532 {
2533 spin_unlock(&p->alloc_lock);
2534 }
2535
2536 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2537 unsigned long *flags);
2538
2539 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2540 unsigned long *flags)
2541 {
2542 struct sighand_struct *ret;
2543
2544 ret = __lock_task_sighand(tsk, flags);
2545 (void)__cond_lock(&tsk->sighand->siglock, ret);
2546 return ret;
2547 }
2548
2549 static inline void unlock_task_sighand(struct task_struct *tsk,
2550 unsigned long *flags)
2551 {
2552 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2553 }
2554
2555 #ifdef CONFIG_CGROUPS
2556 static inline void threadgroup_change_begin(struct task_struct *tsk)
2557 {
2558 down_read(&tsk->signal->group_rwsem);
2559 }
2560 static inline void threadgroup_change_end(struct task_struct *tsk)
2561 {
2562 up_read(&tsk->signal->group_rwsem);
2563 }
2564
2565 /**
2566 * threadgroup_lock - lock threadgroup
2567 * @tsk: member task of the threadgroup to lock
2568 *
2569 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2570 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2571 * change ->group_leader/pid. This is useful for cases where the threadgroup
2572 * needs to stay stable across blockable operations.
2573 *
2574 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2575 * synchronization. While held, no new task will be added to threadgroup
2576 * and no existing live task will have its PF_EXITING set.
2577 *
2578 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2579 * sub-thread becomes a new leader.
2580 */
2581 static inline void threadgroup_lock(struct task_struct *tsk)
2582 {
2583 down_write(&tsk->signal->group_rwsem);
2584 }
2585
2586 /**
2587 * threadgroup_unlock - unlock threadgroup
2588 * @tsk: member task of the threadgroup to unlock
2589 *
2590 * Reverse threadgroup_lock().
2591 */
2592 static inline void threadgroup_unlock(struct task_struct *tsk)
2593 {
2594 up_write(&tsk->signal->group_rwsem);
2595 }
2596 #else
2597 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2598 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2599 static inline void threadgroup_lock(struct task_struct *tsk) {}
2600 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2601 #endif
2602
2603 #ifndef __HAVE_THREAD_FUNCTIONS
2604
2605 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2606 #define task_stack_page(task) ((task)->stack)
2607
2608 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2609 {
2610 *task_thread_info(p) = *task_thread_info(org);
2611 task_thread_info(p)->task = p;
2612 }
2613
2614 static inline unsigned long *end_of_stack(struct task_struct *p)
2615 {
2616 return (unsigned long *)(task_thread_info(p) + 1);
2617 }
2618
2619 #endif
2620
2621 static inline int object_is_on_stack(void *obj)
2622 {
2623 void *stack = task_stack_page(current);
2624
2625 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2626 }
2627
2628 extern void thread_info_cache_init(void);
2629
2630 #ifdef CONFIG_DEBUG_STACK_USAGE
2631 static inline unsigned long stack_not_used(struct task_struct *p)
2632 {
2633 unsigned long *n = end_of_stack(p);
2634
2635 do { /* Skip over canary */
2636 n++;
2637 } while (!*n);
2638
2639 return (unsigned long)n - (unsigned long)end_of_stack(p);
2640 }
2641 #endif
2642 extern void set_task_stack_end_magic(struct task_struct *tsk);
2643
2644 /* set thread flags in other task's structures
2645 * - see asm/thread_info.h for TIF_xxxx flags available
2646 */
2647 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2648 {
2649 set_ti_thread_flag(task_thread_info(tsk), flag);
2650 }
2651
2652 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2653 {
2654 clear_ti_thread_flag(task_thread_info(tsk), flag);
2655 }
2656
2657 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2658 {
2659 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2660 }
2661
2662 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2663 {
2664 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2665 }
2666
2667 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2668 {
2669 return test_ti_thread_flag(task_thread_info(tsk), flag);
2670 }
2671
2672 static inline void set_tsk_need_resched(struct task_struct *tsk)
2673 {
2674 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2675 }
2676
2677 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2678 {
2679 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2680 }
2681
2682 static inline int test_tsk_need_resched(struct task_struct *tsk)
2683 {
2684 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2685 }
2686
2687 static inline int restart_syscall(void)
2688 {
2689 set_tsk_thread_flag(current, TIF_SIGPENDING);
2690 return -ERESTARTNOINTR;
2691 }
2692
2693 static inline int signal_pending(struct task_struct *p)
2694 {
2695 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2696 }
2697
2698 static inline int __fatal_signal_pending(struct task_struct *p)
2699 {
2700 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2701 }
2702
2703 static inline int fatal_signal_pending(struct task_struct *p)
2704 {
2705 return signal_pending(p) && __fatal_signal_pending(p);
2706 }
2707
2708 static inline int signal_pending_state(long state, struct task_struct *p)
2709 {
2710 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2711 return 0;
2712 if (!signal_pending(p))
2713 return 0;
2714
2715 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2716 }
2717
2718 /*
2719 * cond_resched() and cond_resched_lock(): latency reduction via
2720 * explicit rescheduling in places that are safe. The return
2721 * value indicates whether a reschedule was done in fact.
2722 * cond_resched_lock() will drop the spinlock before scheduling,
2723 * cond_resched_softirq() will enable bhs before scheduling.
2724 */
2725 extern int _cond_resched(void);
2726
2727 #define cond_resched() ({ \
2728 __might_sleep(__FILE__, __LINE__, 0); \
2729 _cond_resched(); \
2730 })
2731
2732 extern int __cond_resched_lock(spinlock_t *lock);
2733
2734 #ifdef CONFIG_PREEMPT_COUNT
2735 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2736 #else
2737 #define PREEMPT_LOCK_OFFSET 0
2738 #endif
2739
2740 #define cond_resched_lock(lock) ({ \
2741 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2742 __cond_resched_lock(lock); \
2743 })
2744
2745 extern int __cond_resched_softirq(void);
2746
2747 #define cond_resched_softirq() ({ \
2748 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2749 __cond_resched_softirq(); \
2750 })
2751
2752 static inline void cond_resched_rcu(void)
2753 {
2754 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2755 rcu_read_unlock();
2756 cond_resched();
2757 rcu_read_lock();
2758 #endif
2759 }
2760
2761 /*
2762 * Does a critical section need to be broken due to another
2763 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2764 * but a general need for low latency)
2765 */
2766 static inline int spin_needbreak(spinlock_t *lock)
2767 {
2768 #ifdef CONFIG_PREEMPT
2769 return spin_is_contended(lock);
2770 #else
2771 return 0;
2772 #endif
2773 }
2774
2775 /*
2776 * Idle thread specific functions to determine the need_resched
2777 * polling state.
2778 */
2779 #ifdef TIF_POLLING_NRFLAG
2780 static inline int tsk_is_polling(struct task_struct *p)
2781 {
2782 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2783 }
2784
2785 static inline void __current_set_polling(void)
2786 {
2787 set_thread_flag(TIF_POLLING_NRFLAG);
2788 }
2789
2790 static inline bool __must_check current_set_polling_and_test(void)
2791 {
2792 __current_set_polling();
2793
2794 /*
2795 * Polling state must be visible before we test NEED_RESCHED,
2796 * paired by resched_curr()
2797 */
2798 smp_mb__after_atomic();
2799
2800 return unlikely(tif_need_resched());
2801 }
2802
2803 static inline void __current_clr_polling(void)
2804 {
2805 clear_thread_flag(TIF_POLLING_NRFLAG);
2806 }
2807
2808 static inline bool __must_check current_clr_polling_and_test(void)
2809 {
2810 __current_clr_polling();
2811
2812 /*
2813 * Polling state must be visible before we test NEED_RESCHED,
2814 * paired by resched_curr()
2815 */
2816 smp_mb__after_atomic();
2817
2818 return unlikely(tif_need_resched());
2819 }
2820
2821 #else
2822 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2823 static inline void __current_set_polling(void) { }
2824 static inline void __current_clr_polling(void) { }
2825
2826 static inline bool __must_check current_set_polling_and_test(void)
2827 {
2828 return unlikely(tif_need_resched());
2829 }
2830 static inline bool __must_check current_clr_polling_and_test(void)
2831 {
2832 return unlikely(tif_need_resched());
2833 }
2834 #endif
2835
2836 static inline void current_clr_polling(void)
2837 {
2838 __current_clr_polling();
2839
2840 /*
2841 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2842 * Once the bit is cleared, we'll get IPIs with every new
2843 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2844 * fold.
2845 */
2846 smp_mb(); /* paired with resched_curr() */
2847
2848 preempt_fold_need_resched();
2849 }
2850
2851 static __always_inline bool need_resched(void)
2852 {
2853 return unlikely(tif_need_resched());
2854 }
2855
2856 /*
2857 * Thread group CPU time accounting.
2858 */
2859 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2860 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2861
2862 static inline void thread_group_cputime_init(struct signal_struct *sig)
2863 {
2864 raw_spin_lock_init(&sig->cputimer.lock);
2865 }
2866
2867 /*
2868 * Reevaluate whether the task has signals pending delivery.
2869 * Wake the task if so.
2870 * This is required every time the blocked sigset_t changes.
2871 * callers must hold sighand->siglock.
2872 */
2873 extern void recalc_sigpending_and_wake(struct task_struct *t);
2874 extern void recalc_sigpending(void);
2875
2876 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2877
2878 static inline void signal_wake_up(struct task_struct *t, bool resume)
2879 {
2880 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2881 }
2882 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2883 {
2884 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2885 }
2886
2887 /*
2888 * Wrappers for p->thread_info->cpu access. No-op on UP.
2889 */
2890 #ifdef CONFIG_SMP
2891
2892 static inline unsigned int task_cpu(const struct task_struct *p)
2893 {
2894 return task_thread_info(p)->cpu;
2895 }
2896
2897 static inline int task_node(const struct task_struct *p)
2898 {
2899 return cpu_to_node(task_cpu(p));
2900 }
2901
2902 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2903
2904 #else
2905
2906 static inline unsigned int task_cpu(const struct task_struct *p)
2907 {
2908 return 0;
2909 }
2910
2911 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2912 {
2913 }
2914
2915 #endif /* CONFIG_SMP */
2916
2917 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2918 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2919
2920 #ifdef CONFIG_CGROUP_SCHED
2921 extern struct task_group root_task_group;
2922 #endif /* CONFIG_CGROUP_SCHED */
2923
2924 extern int task_can_switch_user(struct user_struct *up,
2925 struct task_struct *tsk);
2926
2927 #ifdef CONFIG_TASK_XACCT
2928 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2929 {
2930 tsk->ioac.rchar += amt;
2931 }
2932
2933 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2934 {
2935 tsk->ioac.wchar += amt;
2936 }
2937
2938 static inline void inc_syscr(struct task_struct *tsk)
2939 {
2940 tsk->ioac.syscr++;
2941 }
2942
2943 static inline void inc_syscw(struct task_struct *tsk)
2944 {
2945 tsk->ioac.syscw++;
2946 }
2947 #else
2948 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2949 {
2950 }
2951
2952 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2953 {
2954 }
2955
2956 static inline void inc_syscr(struct task_struct *tsk)
2957 {
2958 }
2959
2960 static inline void inc_syscw(struct task_struct *tsk)
2961 {
2962 }
2963 #endif
2964
2965 #ifndef TASK_SIZE_OF
2966 #define TASK_SIZE_OF(tsk) TASK_SIZE
2967 #endif
2968
2969 #ifdef CONFIG_MEMCG
2970 extern void mm_update_next_owner(struct mm_struct *mm);
2971 #else
2972 static inline void mm_update_next_owner(struct mm_struct *mm)
2973 {
2974 }
2975 #endif /* CONFIG_MEMCG */
2976
2977 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2978 unsigned int limit)
2979 {
2980 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2981 }
2982
2983 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2984 unsigned int limit)
2985 {
2986 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2987 }
2988
2989 static inline unsigned long rlimit(unsigned int limit)
2990 {
2991 return task_rlimit(current, limit);
2992 }
2993
2994 static inline unsigned long rlimit_max(unsigned int limit)
2995 {
2996 return task_rlimit_max(current, limit);
2997 }
2998
2999 #endif
This page took 0.097621 seconds and 5 git commands to generate.